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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6 * expr.c (convert_move): Use them.
7 (convert_modes): Likewise.
8 (store_expr_with_bounds): Likewise.
10 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11 Alan Hayward <alan.hayward@arm.com>
12 David Sherwood <david.sherwood@arm.com>
14 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15 parameter for the mode of "x". Remove the "known_x", "known_mode"
16 and "known_ret" arguments. Change the type of the mode argument
18 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
20 (reg_num_sign_bit_copies_for_combine): Likewise.
21 * rtlanal.c (nonzero_bits1): Likewise.
22 (num_sign_bit_copies1): Likewise.
23 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
24 (reg_num_sign_bit_copies_general): Likewise.
25 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
26 (reg_nonzero_bits_general): Likewise.
28 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
29 Alan Hayward <alan.hayward@arm.com>
30 David Sherwood <david.sherwood@arm.com>
32 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
33 than in subroutines. Return 1 for non-integer modes.
34 (cached_num_sign_bit_copies): Change the type of the mode parameter
36 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
37 classes. Handle CONST_INT_P first and then check whether X also
38 has a scalar integer mode. Check the same thing for inner registers
39 of a SUBREG and for values that are being extended or truncated.
41 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
42 Alan Hayward <alan.hayward@arm.com>
43 David Sherwood <david.sherwood@arm.com>
45 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
46 in subroutines. Return the mode mask for non-integer modes.
47 (cached_nonzero_bits): Change the type of the mode parameter
49 (nonzero_bits1): Likewise. Remove early exit for other mode
50 classes. Handle CONST_INT_P first and then check whether X
51 also has a scalar integer mode.
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 * combine.c (extract_left_shift): Add a mode argument and update
59 (make_compound_operation_int): Change the type of the mode parameter
60 to scalar_int_mode and update the call to extract_left_shift.
62 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
63 Alan Hayward <alan.hayward@arm.com>
64 David Sherwood <david.sherwood@arm.com>
66 * combine.c (simplify_and_const_int): Change the type of the mode
67 parameter to scalar_int_mode.
68 (simplify_and_const_int_1): Likewise. Update recursive call.
70 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * combine.c (simplify_compare_const): Check that the mode is a
75 scalar_int_mode (rather than VOIDmode) before testing its
77 (simplify_comparison): Move COMPARISON_P handling out of the
78 loop and restrict the latter part of the loop to scalar_int_modes.
79 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
80 and when considering SUBREG_REGs. Use is_int_mode instead of
81 checking GET_MODE_CLASS against MODE_INT.
83 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
84 Alan Hayward <alan.hayward@arm.com>
85 David Sherwood <david.sherwood@arm.com>
87 * combine.c (try_widen_shift_mode): Move check for equal modes to...
88 (simplify_shift_const_1): ...here. Use scalar_int_mode for
89 shift_unit_mode and for modes involved in scalar shifts.
91 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
92 Alan Hayward <alan.hayward@arm.com>
93 David Sherwood <david.sherwood@arm.com>
95 * combine.c (force_int_to_mode): New function, split out from...
96 (force_to_mode): ...here. Keep xmode up-to-date and use it
97 instead of GET_MODE (x).
99 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
100 Alan Hayward <alan.hayward@arm.com>
101 David Sherwood <david.sherwood@arm.com>
103 * optabs-query.h (extraction_insn::struct_mode): Change type to
104 opt_scalar_int_mode and update comment.
105 (extraction_insn::field_mode): Change type to scalar_int_mode.
106 (extraction_insn::pos_mode): Likewise.
107 * combine.c (make_extraction): Update accordingly.
108 * optabs-query.c (get_traditional_extraction_insn): Likewise.
109 (get_optab_extraction_insn): Likewise.
110 * recog.c (simplify_while_replacing): Likewise.
111 * expmed.c (narrow_bit_field_mem): Change the type of the mode
112 parameter to opt_scalar_int_mode.
114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
115 Alan Hayward <alan.hayward@arm.com>
116 David Sherwood <david.sherwood@arm.com>
118 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
119 to a scalar_int_mode instead of a machine_mode.
120 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
121 (get_best_mode): Return a boolean and use a pointer argument to store
122 the selected mode. Replace the limit mode parameter with a bit limit.
123 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
124 for the values returned by bit_field_mode_iterator::next_mode.
125 (store_bit_field): Update call to get_best_mode.
126 (store_fixed_bit_field): Likewise.
127 (extract_fixed_bit_field): Likewise.
128 * expr.c (optimize_bitfield_assignment_op): Likewise.
129 * fold-const.c (optimize_bit_field_compare): Likewise.
130 (fold_truth_andor_1): Likewise.
131 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
132 Update for new type of m_mode.
133 (get_best_mode): As above.
135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
136 Alan Hayward <alan.hayward@arm.com>
137 David Sherwood <david.sherwood@arm.com>
139 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
140 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
141 (store_bit_field): Check is_a <scalar_int_mode> before calling
142 strict_volatile_bitfield_p.
143 (extract_bit_field): Likewise.
145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
146 Alan Hayward <alan.hayward@arm.com>
147 David Sherwood <david.sherwood@arm.com>
149 * target.def (cstore_mode): Return a scalar_int_mode.
150 * doc/tm.texi: Regenerate.
151 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
152 * targhooks.h (default_cstore_mode): Likewise.
153 * targhooks.c (default_cstore_mode): Likewise, using a forced
155 * expmed.c (emit_cstore): Expect the target of the cstore to be
158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
159 Alan Hayward <alan.hayward@arm.com>
160 David Sherwood <david.sherwood@arm.com>
162 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
163 * builtins.c (expand_builtin_signbit): Use it.
164 * cfgexpand.c (expand_debug_expr): Likewise.
165 * dojump.c (do_jump): Likewise.
166 (do_compare_and_jump): Likewise.
167 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
168 * expmed.c (make_tree): Likewise.
169 * expr.c (expand_expr_real_2): Likewise.
170 (expand_expr_real_1): Likewise.
171 (try_casesi): Likewise.
172 * fold-const-call.c (fold_const_call_ss): Likewise.
173 * fold-const.c (unextend): Likewise.
174 (extract_muldiv_1): Likewise.
175 (fold_single_bit_test): Likewise.
176 (native_encode_int): Likewise.
177 (native_encode_string): Likewise.
178 (native_interpret_int): Likewise.
179 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
180 * internal-fn.c (expand_addsub_overflow): Likewise.
181 (expand_neg_overflow): Likewise.
182 (expand_mul_overflow): Likewise.
183 (expand_arith_overflow): Likewise.
184 * match.pd: Likewise.
185 * stor-layout.c (layout_type): Likewise.
186 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
187 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
188 * tree-ssanames.c (get_range_info): Likewise.
189 * tree-switch-conversion.c (array_value_type) Likewise.
190 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
191 (vect_recog_divmod_pattern): Likewise.
192 (vect_recog_mixed_size_cond_pattern): Likewise.
193 * tree-vrp.c (extract_range_basic): Likewise.
194 (simplify_float_conversion_using_ranges): Likewise.
195 * tree.c (int_fits_type_p): Likewise.
196 * ubsan.c (instrument_bool_enum_load): Likewise.
197 * varasm.c (mergeable_string_section): Likewise.
198 (narrowing_initializer_constant_valid_p): Likewise.
199 (output_constant): Likewise.
201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
202 Alan Hayward <alan.hayward@arm.com>
203 David Sherwood <david.sherwood@arm.com>
205 * machmode.h (NARROWEST_INT_MODE): New macro.
206 * expr.c (alignment_for_piecewise_move): Use it instead of
207 GET_CLASS_NARROWEST_MODE (MODE_INT).
208 (push_block): Likewise.
209 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
211 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
219 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
220 * optabs.c (expand_unop): Likewise.
222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
223 Alan Hayward <alan.hayward@arm.com>
224 David Sherwood <david.sherwood@arm.com>
226 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
227 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
228 (popcount_loc_descriptor): Likewise.
229 (bswap_loc_descriptor): Likewise.
230 (rotate_loc_descriptor): Likewise.
231 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
232 calling the functions above.
234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
235 Alan Hayward <alan.hayward@arm.com>
236 David Sherwood <david.sherwood@arm.com>
238 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
240 (try_combine): Likewise.
241 (simplify_if_then_else): Likewise.
242 * cse.c (cse_insn): Likewise.
243 * dwarf2out.c (mem_loc_descriptor): Likewise.
244 * emit-rtl.c (gen_lowpart_common): Likewise.
245 * simplify-rtx.c (simplify_truncation): Likewise.
246 (simplify_binary_operation_1): Likewise.
247 (simplify_const_relational_operation): Likewise.
248 (simplify_ternary_operation): Likewise.
249 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
252 Alan Hayward <alan.hayward@arm.com>
253 David Sherwood <david.sherwood@arm.com>
255 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
256 * reload.c (push_reload): Likewise.
257 (find_reloads): Likewise.
259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
260 Alan Hayward <alan.hayward@arm.com>
261 David Sherwood <david.sherwood@arm.com>
263 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
264 (make_compound_operation_int): Likewise.
265 (change_zero_ext): Likewise.
266 * expr.c (convert_move): Likewise.
267 (convert_modes): Likewise.
268 * fwprop.c (forward_propagate_subreg): Likewise.
269 * loop-iv.c (get_biv_step_1): Likewise.
270 * optabs.c (widen_operand): Likewise.
271 * postreload.c (move2add_valid_value_p): Likewise.
272 * recog.c (simplify_while_replacing): Likewise.
273 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
274 (simplify_binary_operation_1): Likewise. Remove redundant
277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
278 Alan Hayward <alan.hayward@arm.com>
279 David Sherwood <david.sherwood@arm.com>
281 * combine.c (combine_simplify_rtx): Add checks for
282 is_a <scalar_int_mode>.
283 (simplify_if_then_else): Likewise.
284 (make_field_assignment): Likewise.
285 (simplify_comparison): Likewise.
286 * ifcvt.c (noce_try_bitop): Likewise.
287 * loop-invariant.c (canonicalize_address_mult): Likewise.
288 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
291 Alan Hayward <alan.hayward@arm.com>
292 David Sherwood <david.sherwood@arm.com>
294 * simplify-rtx.c (simplify_binary_operation_1): Use
295 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
298 Alan Hayward <alan.hayward@arm.com>
299 David Sherwood <david.sherwood@arm.com>
301 * wide-int.h (int_traits<unsigned char>) New class.
302 (int_traits<unsigned short>) Likewise.
303 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
304 Use GET_MODE_UNIT_PRECISION and remove redundant test for
306 * combine.c (set_nonzero_bits_and_sign_copies): Use
307 is_a <scalar_int_mode>.
308 (find_split_point): Likewise.
309 (combine_simplify_rtx): Likewise.
310 (simplify_logical): Likewise.
311 (expand_compound_operation): Likewise.
312 (expand_field_assignment): Likewise.
313 (make_compound_operation): Likewise.
314 (extended_count): Likewise.
315 (change_zero_ext): Likewise.
316 (simplify_comparison): Likewise.
317 * dwarf2out.c (scompare_loc_descriptor): Likewise.
318 (ucompare_loc_descriptor): Likewise.
319 (minmax_loc_descriptor): Likewise.
320 (mem_loc_descriptor): Likewise.
321 (loc_descriptor): Likewise.
322 * expmed.c (init_expmed_one_mode): Likewise.
323 * lra-constraints.c (lra_constraint_offset): Likewise.
324 * optabs.c (prepare_libcall_arg): Likewise.
325 * postreload.c (move2add_note_store): Likewise.
326 * reload.c (operands_match_p): Likewise.
327 * rtl.h (load_extend_op): Likewise.
328 * rtlhooks.c (gen_lowpart_general): Likewise.
329 * simplify-rtx.c (simplify_truncation): Likewise.
330 (simplify_unary_operation_1): Likewise.
331 (simplify_binary_operation_1): Likewise.
332 (simplify_const_binary_operation): Likewise.
333 (simplify_const_relational_operation): Likewise.
334 (simplify_subreg): Likewise.
335 * stor-layout.c (bitwise_mode_for_mode): Likewise.
336 * var-tracking.c (adjust_mems): Likewise.
337 (prepare_call_arguments): Likewise.
339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
340 Alan Hayward <alan.hayward@arm.com>
341 David Sherwood <david.sherwood@arm.com>
343 * machmode.h (is_int_mode): New fuction.
344 * combine.c (find_split_point): Use it.
345 (combine_simplify_rtx): Likewise.
346 (simplify_if_then_else): Likewise.
347 (simplify_set): Likewise.
348 (simplify_shift_const_1): Likewise.
349 (simplify_comparison): Likewise.
350 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
351 * cse.c (notreg_cost): Likewise.
352 (cse_insn): Likewise.
353 * cselib.c (cselib_lookup_1): Likewise.
354 * dojump.c (do_jump_1): Likewise.
355 (do_compare_rtx_and_jump): Likewise.
356 * dse.c (get_call_args): Likewise.
357 * dwarf2out.c (rtl_for_decl_init): Likewise.
358 (native_encode_initializer): Likewise.
359 * expmed.c (emit_store_flag_1): Likewise.
360 (emit_store_flag): Likewise.
361 * expr.c (convert_modes): Likewise.
362 (store_field): Likewise.
363 (expand_expr_real_1): Likewise.
364 * fold-const.c (fold_read_from_constant_string): Likewise.
365 * gimple-ssa-sprintf.c (get_format_string): Likewise.
366 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
367 * optabs.c (expand_binop): Likewise.
368 (expand_unop): Likewise.
369 (expand_abs_nojump): Likewise.
370 (expand_one_cmpl_abs_nojump): Likewise.
371 * simplify-rtx.c (mode_signbit_p): Likewise.
372 (val_signbit_p): Likewise.
373 (val_signbit_known_set_p): Likewise.
374 (val_signbit_known_clear_p): Likewise.
375 (simplify_relational_operation_1): Likewise.
376 * tree.c (vector_type_mode): Likewise.
378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
379 Alan Hayward <alan.hayward@arm.com>
380 David Sherwood <david.sherwood@arm.com>
382 * machmode.h (smallest_mode_for_size): Fix formatting.
383 (smallest_int_mode_for_size): New function.
384 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
385 instead of smallest_mode_for_size.
386 * combine.c (make_extraction): Likewise.
387 * config/arc/arc.c (arc_expand_movmem): Likewise.
388 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
389 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
390 * config/s390/s390.c (s390_expand_insv): Likewise.
391 * config/sparc/sparc.c (assign_int_registers): Likewise.
392 * config/spu/spu.c (spu_function_value): Likewise.
393 (spu_function_arg): Likewise.
394 * coverage.c (get_gcov_type): Likewise.
395 (get_gcov_unsigned_t): Likewise.
396 * dse.c (find_shift_sequence): Likewise.
397 * expmed.c (store_bit_field_1): Likewise.
398 * expr.c (convert_move): Likewise.
399 (store_field): Likewise.
400 * internal-fn.c (expand_arith_overflow): Likewise.
401 * optabs-query.c (get_best_extraction_insn): Likewise.
402 * optabs.c (expand_twoval_binop_libfunc): Likewise.
403 * stor-layout.c (layout_type): Likewise.
404 (initialize_sizetypes): Likewise.
405 * targhooks.c (default_get_mask_mode): Likewise.
406 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
409 Alan Hayward <alan.hayward@arm.com>
410 David Sherwood <david.sherwood@arm.com>
412 * machmode.h (opt_mode::else_blk): New function.
413 (int_mode_for_mode): Declare.
414 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
415 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
417 * cfgexpand.c (expand_debug_expr): Likewise.
418 * combine.c (gen_lowpart_or_truncate): Likewise.
419 (gen_lowpart_for_combine): Likewise.
420 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
421 * config/avr/avr.c (avr_to_int_mode): Likewise.
422 (avr_out_plus_1): Likewise.
423 (avr_out_plus): Likewise.
424 (avr_out_round): Likewise.
425 * config/i386/i386.c (ix86_split_to_parts): Likewise.
426 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
427 (s390_expand_vcond): Likewise.
428 * config/spu/spu.c (spu_split_immediate): Likewise.
429 (spu_expand_mov): Likewise.
430 * dse.c (get_stored_val): Likewise.
431 * expmed.c (store_bit_field_1): Likewise.
432 (convert_extracted_bit_field): Use int_mode_for_mode instead of
434 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
435 (extract_low_bits): Likewise.
436 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
437 handling rather than repeating the check.
438 (emit_group_store): Likewise.
439 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
440 * optabs.c (expand_absneg_bit): Likewise.
441 (expand_copysign_absneg): Likewise.
442 (expand_copysign_bit): Likewise.
443 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
444 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
445 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
446 * var-tracking.c (prepare_call_arguments): Likewise.
447 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
448 int_mode_for_mode instead of mode_for_size.
449 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
452 Alan Hayward <alan.hayward@arm.com>
453 David Sherwood <david.sherwood@arm.com>
455 * machmode.h (int_mode_for_size): New function.
456 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
457 instead of mode_for_size.
458 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
460 * combine.c (expand_field_assignment): Use int_mode_for_size
461 instead of mode_for_size.
462 (make_extraction): Likewise.
463 (simplify_shift_const_1): Likewise.
464 (simplify_comparison): Likewise.
465 * dojump.c (do_jump): Likewise.
466 * dwarf2out.c (mem_loc_descriptor): Likewise.
467 * emit-rtl.c (init_derived_machine_modes): Likewise.
468 * expmed.c (flip_storage_order): Likewise.
469 (convert_extracted_bit_field): Likewise.
470 * expr.c (copy_blkmode_from_reg): Likewise.
471 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
472 * internal-fn.c (expand_mul_overflow): Likewise.
473 * lower-subreg.c (simple_move): Likewise.
474 * optabs-libfuncs.c (init_optabs): Likewise.
475 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
476 * tree.c (vector_type_mode): Likewise.
477 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
478 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
479 * tree-vect-generic.c (expand_vector_parallel): Likewise.
480 * tree-vect-stmts.c (vectorizable_load): Likewise.
481 (vectorizable_store): Likewise.
483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
484 Alan Hayward <alan.hayward@arm.com>
485 David Sherwood <david.sherwood@arm.com>
487 * coretypes.h (pod_mode): New type.
488 (scalar_int_mode_pod): New typedef.
489 * machmode.h (pod_mode): New class.
490 (int_n_data_t::m): Change type to scalar_int_mode_pod.
491 * genmodes.c (emit_mode_int_n): Update accordingly.
492 * lower-subreg.h (target_lower_subreg): Change type to
494 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
498 Alan Hayward <alan.hayward@arm.com>
499 David Sherwood <david.sherwood@arm.com>
501 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
502 machine_mode to scalar_int_mode.
503 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
504 (rs6000_option_override_internal): Remove cast to int.
505 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
506 machine_mode to scalar_int_mode.
507 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
508 (rs6000_option_override_internal): Remove cast to int.
509 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
510 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
512 * config/s390/s390.c (s390_expand_builtin): Likewise.
513 * coretypes.h (scalar_int_mode): New type.
514 (opt_scalar_int_mode): New typedef.
515 * machmode.h (scalar_int_mode): New class.
516 (scalar_int_mode::includes_p): New function.
517 (byte_mode): Change type to scalar_int_mode.
518 (word_mode): Likewise.
519 (ptr_mode): Likewise.
520 * emit-rtl.c (byte_mode): Likewise.
521 (word_mode): Likewise.
522 (ptr_mode): Likewise.
523 (init_derived_machine_modes): Update accordingly.
524 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
525 and MODE_PARTIAL_INT.
526 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * target.def (libgcc_floating_mode_supported_p): Take a
535 * doc/tm.texi: Regenerate.
536 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
538 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
539 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
547 * doc/tm.texi: Regenerate.
548 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
549 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
550 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
551 * targhooks.h (default_floatn_mode): Likewise.
552 * targhooks.c (default_floatn_mode): Likewise.
553 * tree.c (build_common_tree_nodes): Update accordingly.
555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
556 Alan Hayward <alan.hayward@arm.com>
557 David Sherwood <david.sherwood@arm.com>
559 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
560 (mode_iterator::iterate_p): Likewise.
561 (mode_iterator::get_wider): Likewise.
562 * expr.c (init_expr_target): Use opt_scalar_float_mode.
564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
565 Alan Hayward <alan.hayward@arm.com>
566 David Sherwood <david.sherwood@arm.com>
568 * coretypes.h (opt_scalar_float_mode): New typedef.
569 * machmode.h (float_mode_for_size): New function.
570 * emit-rtl.c (double_mode): Delete.
571 (init_emit_once): Use float_mode_for_size.
572 * stor-layout.c (layout_type): Likewise.
573 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
576 Alan Hayward <alan.hayward@arm.com>
577 David Sherwood <david.sherwood@arm.com>
579 * output.h (assemble_real): Take a scalar_float_mode.
580 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
581 * config/arm/arm.md (consttable_4): Likewise.
582 (consttable_8): Likewise.
583 (consttable_16): Likewise.
584 * config/mips/mips.md (consttable_float): Likewise.
585 * config/s390/s390.c (s390_output_pool_entry): Likewise.
586 * varasm.c (assemble_real): Take a scalar_float_mode.
587 (output_constant_pool_2): Update accordingly.
588 (output_constant): Likewise.
590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
591 Alan Hayward <alan.hayward@arm.com>
592 David Sherwood <david.sherwood@arm.com>
594 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
595 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
596 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
597 (native_encode_real): Likewise.
598 (native_interpret_real): Likewise.
599 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
600 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
603 Alan Hayward <alan.hayward@arm.com>
604 David Sherwood <david.sherwood@arm.com>
606 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
607 <scalar_float_mode>. Simplify.
608 (gen_extend_conv_libfunc): Likewise.
610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
611 Alan Hayward <alan.hayward@arm.com>
612 David Sherwood <david.sherwood@arm.com>
614 * coretypes.h (scalar_float_mode): New type.
615 * machmode.h (mode_traits::from_int): Use machine_mode if
616 USE_ENUM_MODES is defined.
617 (is_a): New function.
619 (dyn_cast): Likewise.
620 (scalar_float_mode): New class.
621 (scalar_float_mode::includes_p): New function.
622 (is_float_mode): Likewise.
623 * gdbhooks.py (MachineModePrinter): New class.
624 (build_pretty_printer): Use it for scalar_float_mode.
625 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
626 (format_helper::format_helper): Turn into a template.
627 * genmodes.c (get_mode_class): New function.
628 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
629 or machine_mode if none.
630 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
631 as_a <scalar_float_mode>.
632 * dwarf2out.c (mem_loc_descriptor): Likewise.
633 (insert_float): Likewise.
634 (add_const_value_attribute): Likewise.
635 * simplify-rtx.c (simplify_immed_subreg): Likewise.
636 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
637 (expand_unop): Update accordingly.
638 (expand_abs_nojump): Likewise.
639 (expand_copysign_absneg): Take a scalar_float_mode.
640 (expand_copysign_bit): Likewise.
641 (expand_copysign): Update accordingly.
643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
644 Alan Hayward <alan.hayward@arm.com>
645 David Sherwood <david.sherwood@arm.com>
647 * coretypes.h (opt_mode): New class.
648 * machmode.h (opt_mode): Likewise.
649 (opt_mode::else_void): New function.
650 (opt_mode::require): Likewise.
651 (opt_mode::exists): Likewise.
652 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
653 (GET_MODE_2XWIDER_MODE): Likewise.
654 (mode_iterator::get_wider): Update accordingly.
655 (mode_iterator::get_2xwider): Likewise.
656 (mode_iterator::get_known_wider): Likewise, turning into a template.
657 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
658 forcing a wider mode to exist.
659 * config/cr16/cr16.h (LONG_REG_P): Likewise.
660 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
661 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
662 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
663 * lower-subreg.c (init_lower_subreg): Likewise.
664 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
665 on the final iteration.
666 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
667 a wider mode exists before asking for a move pattern.
668 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
669 forcing a wider mode to exist.
670 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
671 returning false if no such mode exists.
672 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
673 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
674 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
675 Avoid checking for a MODE_INT if we already know the mode is not a
677 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
678 forcing a wider mode to exist.
679 (expmed_mult_highpart_optab): Likewise.
680 (expmed_mult_highpart): Likewise.
681 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
683 * lto-streamer-in.c (lto_input_mode_table): Likewise.
684 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
685 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
686 * internal-fn.c (expand_mul_overflow): Update use of
687 GET_MODE_2XWIDER_MODE.
688 * omp-low.c (omp_clause_aligned_alignment): Likewise.
689 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
691 (convert_plusminus_to_widen): Likewise.
692 * tree-switch-conversion.c (array_value_type): Likewise.
693 * var-tracking.c (emit_note_insn_var_location): Likewise.
694 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
695 Return false inside rather than outside the loop if no wider mode
697 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
698 and GET_MODE_2XWIDER_MODE
699 (can_compare_p): Use else_void.
700 * gdbhooks.py (OptMachineModePrinter): New class.
701 (build_pretty_printer): Use it for opt_mode.
703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
704 Alan Hayward <alan.hayward@arm.com>
705 David Sherwood <david.sherwood@arm.com>
707 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
708 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
711 Alan Hayward <alan.hayward@arm.com>
712 David Sherwood <david.sherwood@arm.com>
714 * machmode.h (mode_traits): New structure.
715 (get_narrowest_mode): New function.
716 (mode_iterator::start): Likewise.
717 (mode_iterator::iterate_p): Likewise.
718 (mode_iterator::get_wider): Likewise.
719 (mode_iterator::get_known_wider): Likewise.
720 (mode_iterator::get_2xwider): Likewise.
721 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
722 (FOR_EACH_MODE): Likewise.
723 (FOR_EACH_MODE_FROM): Likewise.
724 (FOR_EACH_MODE_UNTIL): Likewise.
725 (FOR_EACH_WIDER_MODE): Likewise.
726 (FOR_EACH_2XWIDER_MODE): Likewise.
727 * builtins.c (expand_builtin_strlen): Use new mode iterators.
728 * combine.c (simplify_comparison): Likewise
729 * config/i386/i386.c (type_natural_mode): Likewise.
730 * cse.c (cse_insn): Likewise.
731 * dse.c (find_shift_sequence): Likewise.
732 * emit-rtl.c (init_derived_machine_modes): Likewise.
733 (init_emit_once): Likewise.
734 * explow.c (hard_function_value): Likewise.
735 * expmed.c (extract_fixed_bit_field_1): Likewise.
736 (extract_bit_field_1): Likewise.
737 (expand_divmod): Likewise.
738 (emit_store_flag_1): Likewise.
739 * expr.c (init_expr_target): Likewise.
740 (convert_move): Likewise.
741 (alignment_for_piecewise_move): Likewise.
742 (widest_int_mode_for_size): Likewise.
743 (emit_block_move_via_movmem): Likewise.
744 (copy_blkmode_to_reg): Likewise.
745 (set_storage_via_setmem): Likewise.
746 (compress_float_constant): Likewise.
747 * omp-low.c (omp_clause_aligned_alignment): Likewise.
748 * optabs-query.c (get_best_extraction_insn): Likewise.
749 * optabs.c (expand_binop): Likewise.
750 (expand_twoval_unop): Likewise.
751 (expand_twoval_binop): Likewise.
752 (widen_leading): Likewise.
753 (widen_bswap): Likewise.
754 (expand_parity): Likewise.
755 (expand_unop): Likewise.
756 (prepare_cmp_insn): Likewise.
757 (prepare_float_lib_cmp): Likewise.
758 (expand_float): Likewise.
759 (expand_fix): Likewise.
760 (expand_sfix_optab): Likewise.
761 * postreload.c (move2add_use_add2_insn): Likewise.
762 * reg-stack.c (reg_to_stack): Likewise.
763 * reginfo.c (choose_hard_reg_mode): Likewise.
764 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
765 * stor-layout.c (mode_for_size): Likewise.
766 (smallest_mode_for_size): Likewise.
767 (mode_for_vector): Likewise.
768 (finish_bitfield_representative): Likewise.
769 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
770 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
771 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
772 * var-tracking.c (prepare_call_arguments): Likewise.
774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
775 Alan Hayward <alan.hayward@arm.com>
776 David Sherwood <david.sherwood@arm.com>
778 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
779 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
780 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
781 * machmode.h (mode_size): Move earlier in file.
782 (mode_precision): Likewise.
783 (mode_inner): Likewise.
784 (mode_nunits): Likewise.
785 (mode_unit_size): Likewise.
786 (unit_unit_precision): Likewise.
787 (mode_wider): Likewise.
788 (mode_2xwider): Likewise.
789 (machine_mode): New class.
790 (mode_to_bytes): New function.
791 (mode_to_bits): Likewise.
792 (mode_to_precision): Likewise.
793 (mode_to_inner): Likewise.
794 (mode_to_unit_size): Likewise.
795 (mode_to_unit_precision): Likewise.
796 (mode_to_nunits): Likewise.
797 (GET_MODE_SIZE): Use mode_to_bytes.
798 (GET_MODE_BITSIZE): Use mode_to_bits.
799 (GET_MODE_PRECISION): Use mode_to_precision.
800 (GET_MODE_INNER): Use mode_to_inner.
801 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
802 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
803 (GET_MODE_NUNITS): Use mode_to_nunits.
804 * system.h (ALWAYS_INLINE): New macro.
805 * config/powerpcspe/powerpcspe-c.c
806 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
807 int for arg1_mode and arg2_mode.
809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
810 Alan Hayward <alan.hayward@arm.com>
811 David Sherwood <david.sherwood@arm.com>
813 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
814 Prefix mode names with E_ in case statements.
815 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
816 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
817 (aarch64_split_simd_move): Likewise.
818 (aarch64_gen_storewb_pair): Likewise.
819 (aarch64_gen_loadwb_pair): Likewise.
820 (aarch64_gen_store_pair): Likewise.
821 (aarch64_gen_load_pair): Likewise.
822 (aarch64_get_condition_code_1): Likewise.
823 (aarch64_constant_pool_reload_icode): Likewise.
824 (get_rsqrte_type): Likewise.
825 (get_rsqrts_type): Likewise.
826 (get_recpe_type): Likewise.
827 (get_recps_type): Likewise.
828 (aarch64_gimplify_va_arg_expr): Likewise.
829 (aarch64_simd_container_mode): Likewise.
830 (aarch64_emit_load_exclusive): Likewise.
831 (aarch64_emit_store_exclusive): Likewise.
832 (aarch64_expand_compare_and_swap): Likewise.
833 (aarch64_gen_atomic_cas): Likewise.
834 (aarch64_emit_bic): Likewise.
835 (aarch64_emit_atomic_swap): Likewise.
836 (aarch64_emit_atomic_load_op): Likewise.
837 (aarch64_evpc_trn): Likewise.
838 (aarch64_evpc_uzp): Likewise.
839 (aarch64_evpc_zip): Likewise.
840 (aarch64_evpc_ext): Likewise.
841 (aarch64_evpc_rev): Likewise.
842 (aarch64_evpc_dup): Likewise.
843 (aarch64_gen_ccmp_first): Likewise.
844 (aarch64_gen_ccmp_next): Likewise.
845 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
846 (alpha_emit_xfloating_libcall): Likewise.
847 (emit_insxl): Likewise.
848 (alpha_arg_type): Likewise.
849 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
850 (arc_preferred_simd_mode): Likewise.
851 (arc_secondary_reload): Likewise.
852 (get_arc_condition_code): Likewise.
853 (arc_print_operand): Likewise.
854 (arc_legitimate_constant_p): Likewise.
855 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
856 * config/arc/arc.md (casesi_load): Likewise.
857 (casesi_compact_jump): Likewise.
858 * config/arc/predicates.md (proper_comparison_operator): Likewise.
859 (cc_use_register): Likewise.
860 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
861 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
862 (arm_init_iwmmxt_builtins): Likewise.
863 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
864 (neon_expand_vector_init): Likewise.
865 (arm_attr_length_move_neon): Likewise.
866 (maybe_get_arm_condition_code): Likewise.
867 (arm_emit_vector_const): Likewise.
868 (arm_preferred_simd_mode): Likewise.
869 (arm_output_iwmmxt_tinsr): Likewise.
870 (thumb1_output_casesi): Likewise.
871 (thumb2_output_casesi): Likewise.
872 (arm_emit_load_exclusive): Likewise.
873 (arm_emit_store_exclusive): Likewise.
874 (arm_expand_compare_and_swap): Likewise.
875 (arm_evpc_neon_vuzp): Likewise.
876 (arm_evpc_neon_vzip): Likewise.
877 (arm_evpc_neon_vrev): Likewise.
878 (arm_evpc_neon_vtrn): Likewise.
879 (arm_evpc_neon_vext): Likewise.
880 (arm_validize_comparison): Likewise.
881 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
882 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
883 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
884 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
885 (c6x_preferred_simd_mode): Likewise.
886 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
887 (epiphany_rtx_costs): Likewise.
888 * config/epiphany/predicates.md (proper_comparison_operator):
890 * config/frv/frv.c (condexec_memory_operand): Likewise.
891 (frv_emit_move): Likewise.
892 (output_move_single): Likewise.
893 (output_condmove_single): Likewise.
894 (frv_hard_regno_mode_ok): Likewise.
895 (frv_matching_accg_mode): Likewise.
896 * config/h8300/h8300.c (split_adds_subs): Likewise.
897 (h8300_rtx_costs): Likewise.
898 (h8300_print_operand): Likewise.
899 (compute_mov_length): Likewise.
900 (output_logical_op): Likewise.
901 (compute_logical_op_length): Likewise.
902 (compute_logical_op_cc): Likewise.
903 (h8300_shift_needs_scratch_p): Likewise.
904 (output_a_shift): Likewise.
905 (compute_a_shift_length): Likewise.
906 (compute_a_shift_cc): Likewise.
907 (expand_a_rotate): Likewise.
908 (output_a_rotate): Likewise.
909 * config/i386/i386.c (classify_argument): Likewise.
910 (function_arg_advance_32): Likewise.
911 (function_arg_32): Likewise.
912 (function_arg_64): Likewise.
913 (function_value_64): Likewise.
914 (ix86_gimplify_va_arg): Likewise.
915 (ix86_legitimate_constant_p): Likewise.
916 (put_condition_code): Likewise.
917 (split_double_mode): Likewise.
918 (ix86_avx256_split_vector_move_misalign): Likewise.
919 (ix86_expand_vector_logical_operator): Likewise.
920 (ix86_split_idivmod): Likewise.
921 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
922 (ix86_build_const_vector): Likewise.
923 (ix86_build_signbit_mask): Likewise.
924 (ix86_match_ccmode): Likewise.
925 (ix86_cc_modes_compatible): Likewise.
926 (ix86_expand_branch): Likewise.
927 (ix86_expand_sse_cmp): Likewise.
928 (ix86_expand_sse_movcc): Likewise.
929 (ix86_expand_int_sse_cmp): Likewise.
930 (ix86_expand_vec_perm_vpermi2): Likewise.
931 (ix86_expand_vec_perm): Likewise.
932 (ix86_expand_sse_unpack): Likewise.
933 (ix86_expand_int_addcc): Likewise.
934 (ix86_split_to_parts): Likewise.
935 (ix86_vectorize_builtin_gather): Likewise.
936 (ix86_vectorize_builtin_scatter): Likewise.
937 (avx_vpermilp_parallel): Likewise.
938 (inline_memory_move_cost): Likewise.
939 (ix86_tieable_integer_mode_p): Likewise.
940 (x86_maybe_negate_const_int): Likewise.
941 (ix86_expand_vector_init_duplicate): Likewise.
942 (ix86_expand_vector_init_one_nonzero): Likewise.
943 (ix86_expand_vector_init_one_var): Likewise.
944 (ix86_expand_vector_init_concat): Likewise.
945 (ix86_expand_vector_init_interleave): Likewise.
946 (ix86_expand_vector_init_general): Likewise.
947 (ix86_expand_vector_set): Likewise.
948 (ix86_expand_vector_extract): Likewise.
949 (emit_reduc_half): Likewise.
950 (ix86_emit_i387_round): Likewise.
951 (ix86_mangle_type): Likewise.
952 (ix86_expand_round_sse4): Likewise.
953 (expand_vec_perm_blend): Likewise.
954 (canonicalize_vector_int_perm): Likewise.
955 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
956 (expand_vec_perm_1): Likewise.
957 (expand_vec_perm_interleave3): Likewise.
958 (expand_vec_perm_even_odd_pack): Likewise.
959 (expand_vec_perm_even_odd_1): Likewise.
960 (expand_vec_perm_broadcast_1): Likewise.
961 (ix86_vectorize_vec_perm_const_ok): Likewise.
962 (ix86_expand_vecop_qihi): Likewise.
963 (ix86_expand_mul_widen_hilo): Likewise.
964 (ix86_expand_sse2_abs): Likewise.
965 (ix86_expand_pextr): Likewise.
966 (ix86_expand_pinsr): Likewise.
967 (ix86_preferred_simd_mode): Likewise.
968 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
969 * config/i386/sse.md (*andnot<mode>3): Likewise.
970 (<mask_codefor><code><mode>3<mask_name>): Likewise.
971 (*<code><mode>3): Likewise.
972 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
973 (ia64_expand_atomic_op): Likewise.
974 (ia64_arg_type): Likewise.
975 (ia64_mode_to_int): Likewise.
976 (ia64_scalar_mode_supported_p): Likewise.
977 (ia64_vector_mode_supported_p): Likewise.
978 (expand_vec_perm_broadcast): Likewise.
979 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
980 (iq2000_function_arg_advance): Likewise.
981 (iq2000_function_arg): Likewise.
982 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
983 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
984 (m68k_libcall_value): Likewise.
985 (m68k_function_value): Likewise.
986 (sched_attr_op_type): Likewise.
987 * config/mcore/mcore.c (mcore_output_move): Likewise.
988 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
990 (microblaze_function_arg): Likewise.
991 * config/mips/mips.c (mips16_build_call_stub): Likewise.
992 (mips_print_operand): Likewise.
993 (mips_mode_ok_for_mov_fmt_p): Likewise.
994 (mips_vector_mode_supported_p): Likewise.
995 (mips_preferred_simd_mode): Likewise.
996 (mips_expand_vpc_loongson_even_odd): Likewise.
997 (mips_expand_vec_unpack): Likewise.
998 (mips_expand_vi_broadcast): Likewise.
999 (mips_expand_vector_init): Likewise.
1000 (mips_expand_vec_reduc): Likewise.
1001 (mips_expand_msa_cmp): Likewise.
1002 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1003 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1004 (cc_flags_for_mode): Likewise.
1005 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1006 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1007 (nds32_output_casesi_pc_relative): Likewise.
1008 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1009 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1010 (nvptx_gen_unpack): Likewise.
1011 (nvptx_gen_pack): Likewise.
1012 (nvptx_gen_shuffle): Likewise.
1013 (nvptx_gen_wcast): Likewise.
1014 (nvptx_preferred_simd_mode): Likewise.
1015 * config/pa/pa.c (pa_secondary_reload): Likewise.
1016 * config/pa/predicates.md (base14_operand): Likewise.
1017 * config/powerpcspe/powerpcspe-c.c
1018 (altivec_resolve_overloaded_builtin): Likewise.
1019 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1021 (rs6000_preferred_simd_mode): Likewise.
1022 (output_vec_const_move): Likewise.
1023 (rs6000_expand_vector_extract): Likewise.
1024 (rs6000_split_vec_extract_var): Likewise.
1025 (reg_offset_addressing_ok_p): Likewise.
1026 (rs6000_legitimate_offset_address_p): Likewise.
1027 (rs6000_legitimize_address): Likewise.
1028 (rs6000_emit_set_const): Likewise.
1029 (rs6000_const_vec): Likewise.
1030 (rs6000_emit_move): Likewise.
1031 (spe_build_register_parallel): Likewise.
1032 (rs6000_darwin64_record_arg_recurse): Likewise.
1033 (swap_selector_for_mode): Likewise.
1034 (spe_init_builtins): Likewise.
1035 (paired_init_builtins): Likewise.
1036 (altivec_init_builtins): Likewise.
1037 (do_load_for_compare): Likewise.
1038 (rs6000_generate_compare): Likewise.
1039 (rs6000_expand_float128_convert): Likewise.
1040 (emit_load_locked): Likewise.
1041 (emit_store_conditional): Likewise.
1042 (rs6000_output_function_epilogue): Likewise.
1043 (rs6000_handle_altivec_attribute): Likewise.
1044 (rs6000_function_value): Likewise.
1045 (emit_fusion_gpr_load): Likewise.
1046 (emit_fusion_p9_load): Likewise.
1047 (emit_fusion_p9_store): Likewise.
1048 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1049 (fusion_gpr_mem_load): Likewise.
1050 (fusion_addis_mem_combo_load): Likewise.
1051 (fusion_addis_mem_combo_store): Likewise.
1052 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1053 (fusion_gpr_mem_load): Likewise.
1054 (fusion_addis_mem_combo_load): Likewise.
1055 (fusion_addis_mem_combo_store): Likewise.
1056 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1058 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1059 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1060 (rs6000_preferred_simd_mode): Likewise.
1061 (output_vec_const_move): Likewise.
1062 (rs6000_expand_vector_extract): Likewise.
1063 (rs6000_split_vec_extract_var): Likewise.
1064 (reg_offset_addressing_ok_p): Likewise.
1065 (rs6000_legitimate_offset_address_p): Likewise.
1066 (rs6000_legitimize_address): Likewise.
1067 (rs6000_emit_set_const): Likewise.
1068 (rs6000_const_vec): Likewise.
1069 (rs6000_emit_move): Likewise.
1070 (rs6000_darwin64_record_arg_recurse): Likewise.
1071 (swap_selector_for_mode): Likewise.
1072 (paired_init_builtins): Likewise.
1073 (altivec_init_builtins): Likewise.
1074 (rs6000_expand_float128_convert): Likewise.
1075 (emit_load_locked): Likewise.
1076 (emit_store_conditional): Likewise.
1077 (rs6000_output_function_epilogue): Likewise.
1078 (rs6000_handle_altivec_attribute): Likewise.
1079 (rs6000_function_value): Likewise.
1080 (emit_fusion_gpr_load): Likewise.
1081 (emit_fusion_p9_load): Likewise.
1082 (emit_fusion_p9_store): Likewise.
1083 * config/rx/rx.c (rx_gen_move_template): Likewise.
1084 (flags_from_mode): Likewise.
1085 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1086 (s390_slb_comparison): Likewise.
1087 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1088 (s390_vector_mode_supported_p): Likewise.
1089 (s390_cc_modes_compatible): Likewise.
1090 (s390_match_ccmode_set): Likewise.
1091 (s390_canonicalize_comparison): Likewise.
1092 (s390_emit_compare_and_swap): Likewise.
1093 (s390_branch_condition_mask): Likewise.
1094 (s390_rtx_costs): Likewise.
1095 (s390_secondary_reload): Likewise.
1096 (__SECONDARY_RELOAD_CASE): Likewise.
1097 (s390_expand_cs): Likewise.
1098 (s390_preferred_simd_mode): Likewise.
1099 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1100 * config/sh/sh.c (sh_print_operand): Likewise.
1101 (dump_table): Likewise.
1102 (sh_secondary_reload): Likewise.
1103 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1104 * config/sh/sh.md (casesi_worker_1): Likewise.
1105 (casesi_worker_2): Likewise.
1106 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1107 (fcc_comparison_operator): Likewise.
1108 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1109 (emit_soft_tfmode_cvt): Likewise.
1110 (sparc_preferred_simd_mode): Likewise.
1111 (output_cbranch): Likewise.
1112 (sparc_print_operand): Likewise.
1113 (sparc_expand_vec_perm_bmask): Likewise.
1114 (vector_init_bshuffle): Likewise.
1115 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1116 (spu_vector_mode_supported_p): Likewise.
1117 (spu_expand_insv): Likewise.
1118 (spu_emit_branch_or_set): Likewise.
1119 (spu_handle_vector_attribute): Likewise.
1120 (spu_builtin_splats): Likewise.
1121 (spu_builtin_extract): Likewise.
1122 (spu_builtin_promote): Likewise.
1123 (spu_expand_sign_extend): Likewise.
1124 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1125 (tilegx_simd_int): Likewise.
1126 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1127 (tilepro_simd_int): Likewise.
1128 * config/v850/v850.c (const_double_split): Likewise.
1129 (v850_print_operand): Likewise.
1130 (ep_memory_offset): Likewise.
1131 * config/vax/vax.c (vax_rtx_costs): Likewise.
1132 (vax_output_int_move): Likewise.
1133 (vax_output_int_add): Likewise.
1134 (vax_output_int_subtract): Likewise.
1135 * config/visium/predicates.md (visium_branch_operator): Likewise.
1136 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1137 (visium_print_operand_address): Likewise.
1138 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1139 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1140 (xtensa_expand_conditional_branch): Likewise.
1141 (xtensa_copy_incoming_a7): Likewise.
1142 (xtensa_output_literal): Likewise.
1143 * dfp.c (decimal_real_maxval): Likewise.
1144 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1147 Alan Hayward <alan.hayward@arm.com>
1148 David Sherwood <david.sherwood@arm.com>
1150 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1151 (mode_nunits_inline): Likewise.
1152 (mode_inner_inline): Likewise.
1153 (mode_unit_size_inline): Likewise.
1154 (mode_unit_precision_inline): Likewise.
1155 (emit_insn_modes_h): Likewise. Also emit a #define of the
1157 (emit_mode_wider): Add an E_ prefix to mode names.
1158 (emit_mode_complex): Likewise.
1159 (emit_mode_inner): Likewise.
1160 (emit_mode_adjustments): Likewise.
1161 (emit_mode_int_n): Likewise.
1162 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1163 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1164 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1165 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1166 (CRC32_BUILTIN, ENTRY): Likewise.
1167 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1168 (aarch64_pop_regs): Likewise.
1169 (aarch64_process_components): Likewise.
1170 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1171 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1172 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1173 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1174 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1175 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1177 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1178 (pcmpistr): Likewise.
1179 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1180 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1181 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1183 * config/rl78/rl78.c (mduc_regs): Likewise.
1184 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1185 (htm_expand_builtin): Likewise.
1186 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1187 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1188 * config/xtensa/xtensa.c (print_operand): Likewise.
1189 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1190 (NUM_MODE_VECTOR_INT): Likewise.
1191 * genoutput.c (null_operand): Likewise.
1192 (output_operand_data): Likewise.
1193 * genrecog.c (print_parameter_value): Likewise.
1194 * lra.c (debug_operand_data): Likewise.
1196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1198 * dwarf2out.c (scompare_loc_descriptor_wide)
1199 (scompare_loc_descriptor_narrow): New functions, split out from...
1200 (scompare_loc_descriptor): ...here.
1201 * expmed.c (emit_store_flag_int): New function, split out from...
1202 (emit_store_flag): ...here.
1204 2017-08-30 Richard Biener <rguenther@suse.de>
1206 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1207 (dwarf2out_early_finish): Move setting of AT_pubnames from
1208 early debug output to early finish.
1210 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1212 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1213 and -mdata-region to the assembler.
1215 2017-08-30 Richard Biener <rguenther@suse.de>
1217 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1219 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1220 (dwarf2out_early_global_decl): Only generate a DIE for the
1221 abstract origin if it doesn't already exist or is a declaration DIE.
1222 (resolve_addr): Do not add the linkage name twice when
1223 generating a stub DIE for the DW_TAG_GNU_call_site target.
1225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1227 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1228 Use machine_mode rather than int for arg1_mode.
1230 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1233 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1234 that the second argument of the built-in functions to unpack
1235 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1236 switch statement instead a lot of if statements.
1237 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1238 Allow 64-bit values to be in Altivec registers as well as
1239 traditional floating point registers.
1240 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1242 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1244 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1245 MAX_REGS_PER_ADDRESS == 1.
1247 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1249 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1250 * config/i386/i386.c (ix86_option_override_internal): Simplify
1251 setting of opts->x_flag_entry.
1253 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1254 Jakub Jelinek <jakub@redhat.com>
1255 Richard Biener <rguenther@suse.de>
1257 PR tree-optimization/81503
1258 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1259 folded constant fits in the target type; reorder tests for clarity.
1261 2017-08-29 Martin Liska <mliska@suse.cz>
1263 * passes.def: Include pass_lower_switch.
1264 * stmt.c (dump_case_nodes): Remove and move to
1265 tree-switch-conversion.
1266 (case_values_threshold): Likewise.
1267 (expand_switch_as_decision_tree_p): Likewise.
1268 (emit_case_decision_tree): Likewise.
1269 (expand_case): Likewise.
1270 (balance_case_nodes): Likewise.
1271 (node_has_low_bound): Likewise.
1272 (node_has_high_bound): Likewise.
1273 (node_is_bounded): Likewise.
1274 (emit_case_nodes): Likewise.
1275 (struct simple_case_node): New struct.
1276 (add_case_node): Remove.
1277 (emit_case_dispatch_table): Use vector instead of case_list.
1278 (reset_out_edges_aux): Remove.
1279 (compute_cases_per_edge): Likewise.
1280 (expand_case): Build list of simple_case_node.
1281 (expand_sjlj_dispatch_table): Use it.
1282 * tree-switch-conversion.c (struct case_node): Moved from
1283 stmt.c and adjusted.
1284 (emit_case_nodes): Likewise.
1285 (node_has_low_bound): Likewise.
1286 (node_has_high_bound): Likewise.
1287 (node_is_bounded): Likewise.
1288 (case_values_threshold): Likewise.
1289 (reset_out_edges_aux): Likewise.
1290 (compute_cases_per_edge): Likewise.
1291 (add_case_node): Likewise.
1292 (dump_case_nodes): Likewise.
1293 (balance_case_nodes): Likewise.
1294 (expand_switch_as_decision_tree_p): Likewise.
1295 (emit_jump): Likewise.
1296 (emit_case_decision_tree): Likewise.
1297 (try_switch_expansion): Likewise.
1298 (do_jump_if_equal): Likewise.
1299 (emit_cmp_and_jump_insns): Likewise.
1300 (fix_phi_operands_for_edge): New function.
1301 (record_phi_operand_mapping): Likewise.
1302 (class pass_lower_switch): New pass.
1303 (pass_lower_switch::execute): New function.
1304 (make_pass_lower_switch): Likewise.
1305 (conditional_probability):
1306 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1307 * tree-pass.h: Add make_pass_lower_switch.
1309 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1312 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1315 2017-08-29 Richard Biener <rguenther@suse.de>
1317 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1318 we do not add a DW_AT_inline attribute twice.
1319 (gen_subprogram_die): Remove code setting DW_AT_inline on
1320 DECL_ABSTRACT_P nodes.
1322 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1324 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1325 calls to internal functions.
1326 (gimplify_modify_expr): Likewise.
1327 * tree-call-cdce.c (use_internal_fn): Likewise.
1328 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1329 (convert_to_divmod): Set the nothrow flag.
1330 * tree-if-conv.c (predicate_mem_writes): Likewise.
1331 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1332 (vectorizable_call): Likewise.
1333 (vectorizable_store): Likewise.
1334 (vectorizable_load): Likewise.
1335 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1336 (vect_recog_mask_conversion_pattern): Likewise.
1338 2017-08-29 Martin Liska <mliska@suse.cz>
1341 * gcc.c (driver_handle_option): Add new argument.
1342 * opts-common.c (handle_option): Pass
1343 target_option_override_hook.
1344 * opts-global.c (lang_handle_option): Add new option.
1345 (set_default_handlers): Add new argument.
1346 (decode_options): Likewise.
1347 * opts.c (target_handle_option): Likewise.
1348 (common_handle_option): Call target_option_override_hook.
1349 * opts.h (struct cl_option_handler_func): Add hook for
1350 target option override.
1351 (struct cl_option_handlers): Likewise.
1352 (set_default_handlers): Add new argument.
1353 (decode_options): Likewise.
1354 (common_handle_option): Likewise.
1355 (target_handle_option): Likewise.
1356 * toplev.c (toplev::main): Pass targetm.target_option.override
1359 2017-08-29 Richard Biener <rguenther@suse.de>
1360 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1362 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1363 life to the active subtree.
1365 2017-08-28 Jeff Law <law@redhat.com>
1367 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1368 derive_equivalences.
1369 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1371 (edge_info::derive_equivalences): New private member function
1373 * tree-ssa-dom.c (class edge_info): Changed from a struct
1374 to a class. Add ctor/dtor, methods and data members.
1375 (edge_info::edge_info): Renamed from allocate_edge_info.
1376 Initialize additional members.
1377 (edge_info::~edge_info): New.
1378 (free_dom_edge_info): Delete the edge info.
1379 (record_edge_info): Use new class & associated member functions.
1380 Tighten forms for testing for edge equivalences.
1381 (record_temporary_equivalences): Iterate over the simple
1382 equivalences rather than assuming there's only one per edge.
1383 (cprop_into_successor_phis): Iterate over the simple
1384 equivalences rather than assuming there's only one per edge.
1385 (optimize_stmt): Use operand_equal_p rather than pointer
1386 equality for mini-DSE code.
1388 2017-08-28 Nathan Sidwell <nathan@acm.org>
1390 * gcc.c (execute): Fold SIGPIPE handling into switch
1391 statement. Adjust internal error message.
1393 2017-08-28 Richard Biener <rguenther@suse.de>
1396 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1397 Do nothing for removed DIEs.
1399 2017-08-28 Richard Biener <rguenther@suse.de>
1401 PR tree-optimization/81977
1402 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1405 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1408 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1409 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1410 using targetm.gen_mem_thread_fence.
1412 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1415 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1416 predicate to register_operand. Reorder operands.
1417 (*btr<mode>): Ditto.
1418 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1419 (*btr<mode>_mask): Ditto.
1421 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1423 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1424 * config/rs6000/xmmintrin.h: New file.
1425 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1427 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1430 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1431 parameter and_insn and return it.
1432 (recombine_lvx_pattern): Insert a copy to ensure availability of
1433 the base register of the copied masking operation at the point of
1434 the instruction replacement.
1435 (recombine_stvx_pattern): Likewise.
1437 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1439 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1440 undocumented switches.
1441 (-mpower9-dform-vector): Likewise.
1442 (-mpower9-dform): Likewise.
1443 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1444 comments to delete references to -mpower9-dform* switches.
1445 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1446 Delete reference to -mpower9-dform* switches, test for
1447 -mpower9-vector instead.
1448 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1449 (OTHER_P9_VECTOR_MASKS): Likewise.
1450 (POWERPC_MASKS): Likewise.
1451 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1452 tests against -mpower9-dform* to -mpower9-vector. Delete code
1453 that checked for -mpower9-dform* consistancy with other options.
1454 Add test for -mpower9-misc to enable other power9 switches.
1455 (rs6000_init_hard_regno_mode_ok): Likewise.
1456 (rs6000_option_override_internal): Likewise.
1457 (rs6000_emit_prologue): Likewise.
1458 (rs6000_emit_epilogue): Likewise.
1459 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1460 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1461 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1463 (emit_fusion_p9_store): Likewise.
1464 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1465 resetting these macros if the assembler does not support ISA 3.0
1467 (TARGET_P9_DFORM_VECTOR): Likewise.
1468 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1469 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1471 2017-08-25 Alan Modra <amodra@gmail.com>
1473 PR rtl-optimization/81747
1474 * cse.c (cse_extended_basic_block): Don't attempt to record
1475 equivalences for degenerate conditional branches that branch
1476 to their fall-through.
1478 2017-08-24 Martin Sebor <msebor@redhat.com>
1481 * gimple-fold.c (size_must_be_zero_p): New function.
1482 (gimple_fold_builtin_memory_op): Call it.
1484 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1486 * config/rs6000/mm_malloc.h: New file.
1488 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1490 PR tree-optimization/81913
1491 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1492 analysis when either IVs in condition can wrap.
1494 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1496 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1497 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1499 2017-08-24 Richard Biener <rguenther@suse.de>
1502 * targhooks.c (default_target_can_inline_p): Properly
1503 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1504 is present and always compare.
1505 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1506 infer -mfpmath=sse from TARGET_SSE_P.
1507 (ix86_can_inline_p): Properly use target_option_default_node when
1508 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1510 2017-08-24 Richard Biener <rguenther@suse.de>
1513 * dwarf2out.c (output_die): Handle flag_generate_offload like
1515 (output_comp_unit): Likewise.
1516 (gen_array_type_die): Likewise.
1517 (dwarf2out_early_finish): Likewise.
1518 (note_variable_value_in_expr): Likewise.
1519 (dwarf2out_finish): Likewise. Adjust assert.
1520 * cgraphunit.c (symbol_table::compile): Move setting of
1521 flag_generate_offload earlier ...
1522 (symbol_table::finalize_compilation_unit): ... here, before
1523 early debug finalization.
1525 2017-08-24 Richard Biener <rguenther@suse.de>
1527 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1528 and ipa-fnsummary.h.
1529 (ix86_can_inline_p): When ix86_fpmath flags do not match
1530 check whether the callee uses FP math at all.
1532 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1535 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1536 instead of TYPE_PRECISION.
1538 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1539 Alan Hayward <alan.hayward@arm.com>
1540 David Sherwood <david.sherwood@arm.com>
1542 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1544 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1546 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1548 (ix86_frame::stack_realign_allocate): New field.
1549 (struct machine_frame_state): Modify comments.
1550 (machine_frame_state::sp_realigned_fp_end): New field.
1551 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1553 (sp_valid_at): Add assertion to assure no attempt to access invalid
1554 offset of a realigned stack.
1555 (fp_valid_at): Likewise.
1556 (choose_baseaddr): Modify comments.
1557 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1558 ix86_expand_prologue.
1559 (ix86_expand_prologue): Modify stack realignment and allocation.
1560 (ix86_expand_epilogue): Modify comments.
1561 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1562 avx2_runtime, avx512f, and avx512f_runtime.
1564 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1566 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1567 (mstackrealign): Do not init to -1.
1568 * config/i386/i386.c (ix86_option_override_internal):
1569 Check opts_set, not opts when setting default value of
1570 opts->x_ix86_force_align_arg_pointer.
1572 2017-08-23 Richard Biener <rguenther@suse.de>
1574 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1575 lang_hooks.decl_printable_name.
1576 * print-rtl-function.c (print_rtx_function): Likewise.
1577 * tree-pretty-print.c (dump_function_header): Likewise.
1579 2017-08-23 Richard Biener <rguenther@suse.de>
1582 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1583 -g0 at compile-time.
1585 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1588 * doc/sourcebuild.texi (Other hardware attributes):
1591 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1594 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1595 Fix single-use check.
1597 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1599 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1601 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1603 * doc/install.texi: Modify to add more details on running selected
1606 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1608 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1609 is combined with -mabi=ms.
1610 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1613 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1615 PR tree-optimization/81488
1616 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1617 and cached_basis fields.
1618 (MAX_SPREAD): New constant.
1619 (alloc_cand_and_find_basis): Initialize new fields.
1620 (clear_visited): New function.
1621 (create_phi_basis_1): Rename from create_phi_basis, set visited
1622 and cached_basis fields.
1623 (create_phi_basis): New wrapper function.
1624 (phi_add_costs_1): Rename from phi_add_costs, add spread
1625 parameter, set visited field, short-circuit when limits reached.
1626 (phi_add_costs): New wrapper function.
1627 (record_phi_increments_1): Rename from record_phi_increments, set
1629 (record_phi_increments): New wrapper function.
1630 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1631 (phi_incr_cost): New wrapper function.
1632 (all_phi_incrs_profitable_1): Rename from
1633 all_phi_incrs_profitable, set visited field.
1634 (all_phi_incrs_profitable): New wrapper function.
1636 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1637 Alan Hayward <alan.hayward@arm.com>
1638 David Sherwood <david.sherwood@arm.com>
1640 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1641 that takes the outer and inner modes.
1642 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1643 comparison as the canonical test for a paradoxical subreg.
1644 * combine.c (simplify_set): Use paradoxical_subreg_p.
1645 (make_extraction): Likewise.
1646 (force_to_mode): Likewise.
1647 (rtx_equal_for_field_assignment_p): Likewise.
1648 (gen_lowpart_for_combine): Likewise.
1649 (simplify_comparison): Likewise.
1650 * cse.c (equiv_constant): Likewise.
1651 * expmed.c (store_bit_field_1): Likewise.
1652 * final.c (alter_subreg): Likewise.
1653 * fwprop.c (propagate_rtx): Likewise.
1654 (forward_propagate_subreg): Likewise.
1655 * ira-conflicts.c (ira_build_conflicts): Likewise.
1656 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1657 * lra-constraints.c (curr_insn_transform): Likewise.
1658 (split_reg): Likewise.
1659 * lra-eliminations.c (move_plus_up): Likewise.
1660 (lra_eliminate_regs_1): Likewise.
1661 * recog.c (general_operand): Likewise.
1662 * ree.c (combine_reaching_defs): Likewise.
1663 * reload.c (push_reload): Likewise.
1664 (find_reloads): Likewise.
1665 * reload1.c (elimination_effects): Likewise.
1666 (compute_reload_subreg_offset): Likewise.
1667 (choose_reload_regs): Likewise.
1668 * rtlanal.c (subreg_lsb_1): Likewise.
1669 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1670 (simplify_subreg): Likewise.
1671 * var-tracking.c (track_loc_p): Likewise.
1672 * emit-rtl.c (byte_lowpart_offset): Likewise.
1673 (paradoxical_subreg_p): Delete out-of-line definition.
1675 2017-08-22 Jeff Law <law@redhat.com>
1677 PR tree-optimization/81741
1678 PR tree-optimization/71947
1679 * tree-ssa-dom.c: Include tree-inline.h.
1680 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1681 equivalences if one is more expensive to compute than the other.
1682 * tree-ssa-scopedtables.h (class const_or_copies): Make
1683 record_const_or_copy_raw method private.
1684 (class avail_exprs_stack): New method simplify_binary_operation.
1685 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1686 avail_exprs_stack::simplify_binary_operation as needed.
1687 (avail_exprs_stack::simplify_binary_operation): New function.
1689 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
1691 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
1692 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
1693 (DOT_SYMBOLS): Likewise.
1694 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
1695 (RELOCATABLE_NEEDS_FIXUP): Likewise.
1696 (RS6000_ABI_NAME): Likewise.
1697 (TARGET_CMODEL): Likewise.
1698 (TOC_SECTION_ASM_OP): Likewise.
1699 (SET_CMODEL): New macro.
1700 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
1702 2017-08-22 Richard Biener <rguenther@suse.de>
1704 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
1705 to allow for free-lang-data replacements similar to verify_type_variant.
1707 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1708 Alan Hayward <alan.hayward@arm.com>
1709 David Sherwood <david.sherwood@arm.com>
1711 * config/aarch64/aarch64.md (casesi): Use DImode rather than
1712 VOIDmode for the LABEL_REF.
1714 2017-08-22 Richard Biener <rguenther@suse.de>
1716 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
1718 2017-08-22 Richard Biener <rguenther@suse.de>
1720 * common.opt (feliminate-dwarf2-dups): Ignore.
1721 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
1722 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
1723 same_die_p_wrap, compute_section_prefix,
1724 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
1725 (comdat_symbol_id, comdat_symbol_number): Likewise.
1726 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
1728 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
1729 (output_die): Mark unreachable path unreachable.
1730 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
1731 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
1732 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
1733 (dwarf2out_early_finish): Likewise.
1735 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
1737 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
1739 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
1742 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
1743 not VAR_P. Filter attribute warnings with OPT_Wattributes.
1744 (avr_attribute_table) <io, io_low, address>: Initialize
1745 .decl_required with true.
1747 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
1749 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
1750 undocumented debugging options.
1751 (-mvsx-scalar-double): Likewise.
1752 (-mallow-df-permute): Likewise.
1753 (-mvectorize-builtins): Likewise.
1754 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1755 (rs6000_builtin_vectorized_function): Likewise.
1756 (rs6000_builtin_md_vectorized_function): Likewise.
1757 (rs6000_opt_vars): Likewise.
1759 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
1762 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
1763 (*btrq_imm): Rename from *btrq.
1764 (*btcq_imm): Rename from *btcq.
1765 (btsc): New code attribute.
1766 (*<btsc><mode>): New insn pattern.
1767 (*btr<mode>): Ditto.
1768 (*<btsc><mode>_mask): New insn_and_split pattern.
1769 (*btr<mode>_mask): Ditto.
1771 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1772 Alan Hayward <alan.hayward@arm.com>
1773 David Sherwood <david.sherwood@arm.com>
1775 * function.c (pad_below): Simplify padding calculation.
1777 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1778 Alan Hayward <alan.hayward@arm.com>
1779 David Sherwood <david.sherwood@arm.com>
1781 * target.def (function_prologue): Remove frame size argument.
1782 (function_epilogue): Likewise.
1783 * doc/tm.texi: Regenerate.
1784 * final.c (final_start_function): Update call to function_prologue.
1785 (final_end_function): Update call to function_epilogue.
1786 (default_function_pro_epilogue): Remove frame size argument.
1787 * output.h (default_function_pro_epilogue): Likewise.
1788 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
1789 (arm_output_function_prologue): Likewise.
1790 * config/frv/frv.c (frv_function_prologue): Likewise.
1791 (frv_function_epilogue): Likewise.
1792 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
1793 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
1794 (ia64_output_function_epilogue): Likewise.
1795 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
1796 (m32r_output_function_epilogue): Likewise.
1797 * config/microblaze/microblaze.c (microblaze_function_prologue)
1798 (microblaze_function_epilogue): Likewise.
1799 * config/mips/mips.c (mips_output_function_prologue): Likewise.
1800 (mips_output_function_epilogue): Likewise.
1801 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
1802 (mmix_target_asm_function_epilogue): Likewise.
1803 * config/msp430/msp430.c (msp430_start_function): Likewise.
1804 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
1805 (nds32_asm_function_epilogue): Likewise.
1806 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
1807 * config/pa/pa.c (pa_output_function_prologue): Likewise.
1808 (pa_output_function_epilogue): Likewise.
1809 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
1810 (rs6000_output_function_epilogue): Likewise.
1811 * config/rl78/rl78.c (rl78_start_function): Likewise.
1812 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
1813 (rs6000_output_function_epilogue): Likewise.
1814 * config/rx/rx.c (rx_output_function_prologue): Likewise.
1815 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
1816 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
1817 (sparc_asm_function_epilogue): Likewise.
1819 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1821 * tree.h (type_has_mode_precision_p): New function.
1822 * convert.c (convert_to_integer_1): Use it.
1823 * expr.c (expand_expr_real_2): Likewise.
1824 (expand_expr_real_1): Likewise.
1825 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
1826 * match.pd: Likewise.
1827 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
1828 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
1829 * tree-tailcall.c (process_assignment): Likewise.
1830 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1831 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
1832 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
1833 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1834 (vectorizable_assignment): Likewise.
1835 (vectorizable_shift): Likewise.
1836 (vectorizable_operation): Likewise.
1837 * tree-vrp.c (register_edge_assert_for_2): Likewise.
1839 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
1841 * match.pd: Add pow (C, x) simplification.
1843 2017-08-21 Richard Biener <rguenther@suse.de>
1845 PR tree-optimization/81900
1846 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
1847 for blocks with abnormal predecessors.
1848 (compute_antic): Do not set visited flag prematurely.
1850 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
1853 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
1855 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1857 * stor-layout.h (vector_type_mode): Move to...
1858 * tree.h (vector_type_mode): ...here.
1859 * stor-layout.c (vector_type_mode): Move to...
1860 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
1862 2017-08-21 Richard Biener <rguenther@suse.de>
1864 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
1865 register_external_die hooks.
1866 (debug_false_tree_charstarstar_uhwistar): Declare.
1867 (debug_nothing_tree_charstar_uhwi): Likewise.
1868 * debug.c (do_nothing_debug_hooks): Adjust.
1869 (debug_false_tree_charstarstar_uhwistar): New do nothing.
1870 (debug_nothing_tree_charstar_uhwi): Likewise.
1871 * dbxout.c (dbx_debug_hooks): Adjust.
1872 (xcoff_debug_hooks): Likewise.
1873 * sdbout.c (sdb_debug_hooks): Likewise.
1874 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1875 * dwarf2out.c (macinfo_label_base): New global.
1876 (dwarf2out_register_external_die): New function for the
1877 register_external_die hook.
1878 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
1879 (dwarf2_debug_hooks): Use them.
1880 (dwarf2_lineno_debug_hooks): Adjust.
1881 (struct die_struct): Add with_offset flag.
1882 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
1883 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
1884 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
1885 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
1886 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
1887 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
1888 defining section names for the early LTO debug variants.
1889 (reset_indirect_string): New helper.
1890 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
1891 (print_dw_val): Add support for offsetted symbol references.
1892 (get_ultimate_context): Split out from is_cxx.
1893 (is_cxx): Use get_ultimate_context.
1894 (is_fortran): Add decl overload.
1895 (compute_comp_unit_symbol): Split out worker from
1896 compute_section_prefix.
1897 (compute_section_prefix): Call compute_comp_unit_symbol and
1898 set comdat_type_p here.
1899 (output_die): Skip DIE symbol output for the LTO added one.
1900 Handle DIE symbol references with offset.
1901 (output_comp_unit): Guard section name mangling properly.
1902 For LTO debug sections emit a symbol at the section beginning
1903 which we use to refer to its DIEs.
1904 (add_abstract_origin_attribute): For DIEs registered via
1905 dwarf2out_register_external_die directly refer to the early
1906 DIE rather than indirectly through the shadow one we created.
1907 Remove obsolete call to dwarf2out_abstract_function for
1908 non-function/block origins.
1909 (gen_array_type_die): When generating early LTO debug do
1910 not emit DW_AT_string_length.
1911 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
1912 late when in LTO. As suggested place a gcc_unreachable for
1913 the DECL_ABSTRACT_P case.
1914 (gen_subprogram_die): Avoid another specification DIE
1915 for early built declarations/definitions for the late LTO case.
1916 (gen_variable_die): Add type references for late duplicated VLA dies
1918 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
1919 we have the abstract instance already.
1920 (process_scope_var): Adjust decl DIE contexts in LTO which
1921 first puts them in limbo.
1922 (gen_decl_die): Do not generate type DIEs late apart from
1923 types for VLAs or for decls we do not yet have a DIE. Do not
1924 call dwarf2out_abstract_function late.
1925 (dwarf2out_early_global_decl): Make sure to create DIEs
1926 for abstract instances of a decl first.
1927 (dwarf2out_late_global_decl): Adjust comment.
1928 (output_macinfo_op): With multiple macro sections use
1929 macinfo_label_base to distinguish labels.
1930 (output_macinfo): Likewise. Update macinfo_label_base.
1931 Pass in the line info label.
1932 (note_variable_value_in_expr): When generating LTO resolve
1933 all variable values here by generating DIEs as needed.
1934 (init_sections_and_labels): Add early LTO debug flag parameter
1935 and generate different sections and names if set. Add generation
1936 counter for the labels so we can have multiple of them.
1937 (reset_dies): Helper to allow DIEs to be output multiple times.
1938 (dwarf2out_finish): When outputting DIEs to the fat part of an
1939 LTO object first reset DIEs.
1940 (dwarf2out_early_finish): Output early DIEs when generating LTO.
1941 (modified_type_die): Check for decl_ultimate_origin being self
1943 (gen_type_die_with_usage): Likewise.
1944 (gen_typedef_die): Allow decl_ultimate_origin being self.
1945 (set_decl_abstract_flags): Remove.
1946 (set_block_abstract_flags): Likewise.
1947 (dwarf2out_abstract_function): Treat the early generated DIEs
1948 as the abstract copy and only add DW_AT_inline and
1949 DW_AT_artificial here and call set_decl_origin_self.
1950 If the DIE has an abstract origin don't do anything.
1951 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
1952 if we have none yet (Go fails to build one, PR78628).
1953 (variably_modified_type_p): Prevent endless recursion for Ada
1954 cyclic pointer types.
1955 * lto-streamer-in.c: Include debug.h.
1956 (dref_queue): New global.
1957 (lto_read_tree_1): Stream in DIE references.
1958 (lto_input_tree): Register DIE references.
1959 (input_function): Stream DECL_DEBUG_ARGS.
1960 * lto-streamer-out.c: Include debug.h.
1961 (lto_write_tree_1): Output DIE references.
1962 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
1963 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
1964 (output_function): Stream DECL_DEBUG_ARGS.
1965 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
1966 Stream DECL_ABSTRACT_ORIGIN.
1967 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
1968 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
1969 DECL_CONTEXT for file-scope decls.
1970 * lto-streamer.h (struct dref_entry): Declare.
1971 (dref_queue): Likewise.
1972 * cfgexpand.c (pass_expand::execute): Do not call the
1973 outlining_inline_function hook here.
1974 * lto-wrapper.c (debug_obj): New global.
1975 (tool_cleanup): Unlink it if required.
1976 (debug_objcopy): New function.
1977 (run_gcc): Handle early debug sections in the IL files by
1978 extracting them to separate files, partially linkin them and
1979 feeding the result back as result to the linker.
1980 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
1981 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
1982 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
1983 sections into a separate segment.
1984 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
1986 (darwin_asm_dwarf_section): Likewise.
1987 (darwin_asm_output_dwarf_offset): Likewise.
1988 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
1990 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1991 Alan Hayward <alan.hayward@arm.com>
1992 David Sherwood <david.sherwood@arm.com>
1994 * read-md.h (md_reader::record_potential_iterator_use): Replace
1995 pointer argument with an rtx and an index.
1996 * read-rtl.c (iterator_group::apply_iterator): Likewise.
1997 (apply_mode_iterator): Likewise.
1998 (apply_code_iterator): Likewise.
1999 (apply_int_iterator): Likewise.
2000 (apply_subst_iterator): Likewise.
2001 (record_iterator_use): Likewise.
2002 (record_attribute_use): Likewise.
2003 (md_reader::record_potential_iterator_use): Likewise. Update calls
2004 to record_iterator_use and apply_iterator.
2005 (iterator_use): Replace ptr with x and index.
2006 (attribute_use): Likewise.
2007 (apply_attribute_uses): Update calls to apply_iterator.
2008 (apply_iterators): Likewise. Update initialization of iterator_use.
2009 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2010 and record_potential_iterator_use.
2011 (rtx_reader::read_rtx_operand): Likewise.
2013 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2014 Alan Hayward <alan.hayward@arm.com>
2015 David Sherwood <david.sherwood@arm.com>
2017 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2020 2017-08-21 Richard Biener <rguenther@suse.de>
2023 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2024 at struct end conservatively when comparing common bases.
2026 2017-08-21 Richard Biener <rguenther@suse.de>
2028 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2029 (mem_ref_in_stmt): Remove.
2030 (determine_max_movement): Use ref index to get at the reference.
2031 (invariantness_dom_walker::before_dom_children): Deal with
2032 lim data already initialized.
2033 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2035 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2037 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2038 (TARGET_ISA_ROUND): Ditto.
2039 (TARGET_ROUND): Ditto.
2040 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2041 * config/i386/i386.md: Ditto.
2042 * config/i386/sse.md: Ditto.
2043 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2044 with OPTION_MASK_ISA_SSE4_1.
2046 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2049 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2050 __builtin_ia32_lzcnt_u16.
2052 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2055 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2056 (rs6000_set_current_function): Rewrite function to use it.
2058 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2061 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2062 and TYPE_WARN_IF_NOT_ALIGN.
2063 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2064 (handle_warn_if_not_align): New.
2065 (place_union_field): Call handle_warn_if_not_align.
2066 (place_field): Call handle_warn_if_not_align.
2067 Copy TYPE_WARN_IF_NOT_ALIGN.
2068 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2069 (layout_type): Likewise.
2070 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2072 (tree_decl_common): Add warn_if_not_align.
2073 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2074 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2075 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2076 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2077 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2078 * doc/extend.texi: Document warn_if_not_aligned attribute.
2079 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2081 2017-08-17 Martin Liska <mliska@suse.cz>
2084 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2085 (get_data_dependence): Use it as pointer type.
2086 (distribute_loop): Likewise.
2088 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2090 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2091 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2092 (p8_vmrgow_<mode>_direct): New define_insn.
2093 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2094 handle endianness for vmrgew and vmrgow permute patterns.
2096 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2098 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2099 * config/rs6000/rs6000-cpus.def: Remove comment.
2100 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2101 (POWERPC_MASKS): Likewise.
2102 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2103 use of TARGET_VSX_TIMODE.
2104 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2105 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2106 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2107 (rs6000_option_override_internal): Remove dead code.
2108 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2109 (rs6000_legitimize_reload_address): Likewise.
2110 (rs6000_legitimate_address_p): Likewise.
2111 (rs6000_opt_masks): Delete "vsx-timode".
2112 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2113 from function comment.
2114 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2115 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2116 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2118 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2119 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2120 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2123 (define_peephole2): Likewise.
2125 2017-08-17 Martin Sebor <msebor@redhat.com>
2128 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2129 past the end of an array.
2130 (test_pp_format): Add test cases.
2132 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2134 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2135 missing ECF_NOTHROW flags.
2137 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2140 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2141 operands residing in integer registers.
2142 (*vsx_le_perm_load_<mode>): Likewise.
2143 (*vsx_le_perm_store_<mode>): Likewise.
2144 (define_peephole2): Add peepholes to optimize the above.
2146 2017-08-17 Marek Polacek <polacek@redhat.com>
2149 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2150 to mimic what shorten_compare did. Change the return type to bool.
2151 (fold_cond_expr_with_comparison): Update call to
2152 operand_equal_for_comparison_p.
2153 (fold_ternary_loc): Likewise.
2155 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2157 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2159 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2161 2017-08-17 Richard Biener <rguenther@suse.de>
2163 * tree-ssa-structalias.c (solve_graph): When propagating
2164 to successors update the graphs succ edges and avoid duplicate work.
2166 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2169 * config/i386/i386.c (ix86_option_override_internal): Save target
2170 specific options after ix86_stack_protector_guard_reg was changed.
2172 2017-08-17 Richard Biener <rguenther@suse.de>
2174 PR tree-optimization/81827
2175 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2176 (new_var_info): Initialize it conservatively.
2177 (get_call_vi): Mark register vars.
2178 (new_scalar_tmp_constraint_exp): Likewise.
2179 (handle_rhs_call): Likewise.
2180 (handle_const_call): Likewise.
2181 (create_function_info_for): Likewise.
2182 (solve_constraints): Sort varinfos to separate register from
2183 non-register vars to pack points-to solution bitmaps during
2186 2017-08-17 Marek Polacek <polacek@redhat.com>
2188 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2190 2017-08-17 Richard Biener <rguenther@suse.de>
2192 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2193 to true when overflow is undefined and we saturated the result.
2195 2017-08-17 Alan Modra <amodra@gmail.com>
2198 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2199 Don't use store multiple if only one reg needs saving.
2200 (interesting_frame_related_regno): New function.
2201 (rs6000_frame_related): Don't emit frame info for regs that
2203 (rs6000_emit_epilogue): Likewise.
2205 2017-08-16 Nathan Sidwell <nathan@acm.org>
2207 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2208 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2209 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2210 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2211 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2212 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2213 (verify_type): Adjust for TYPE_BINFO move.
2214 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2215 process TYPE_BINFO directly.
2216 (hash_tree): Likewise.
2217 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2219 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2222 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2224 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2226 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2229 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2230 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2231 Add ix86_binary_operator_ok to insn constraint.
2232 (*iordi_1_bts): Ditto.
2233 (*xordi_1_btc): Ditto.
2234 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2235 Update corresponding peephole2 pattern.
2239 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2241 PR tree-optimization/81832
2242 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2243 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2245 2017-08-16 Marek Polacek <polacek@redhat.com>
2248 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2249 regarding size_zero_node.
2251 2017-08-16 Martin Liska <mliska@suse.cz>
2254 * config.gcc: Respect previously set extra_objs in case
2257 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2259 PR tree-optimization/81835
2260 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2261 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2262 not depend on the phi.
2264 2017-08-16 Alan Modra <amodra@gmail.com>
2266 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2269 2017-08-16 Alan Modra <amodra@gmail.com>
2271 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2272 (save_reg_p): ..into this. Update all callers.
2273 (first_reg_to_save): Simplify.
2275 2017-08-16 Alan Modra <amodra@gmail.com>
2277 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2280 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2284 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2285 constant count if that count is less than 32.
2287 2017-08-15 Nathan Sidwell <nathan@acm.org>
2289 * gcc.c (execute): Emit friendlier message if inferior is killed
2290 by an external cause.
2292 2017-08-15 Richard Biener <rguenther@suse.de>
2294 PR tree-optimization/81790
2295 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2296 CONSTRUCTORs from simplifying and VN.
2298 2017-08-14 Martin Sebor <msebor@redhat.com>
2300 * builtin-attrs.def: Add comments.
2302 2017-08-14 Martin Sebor <msebor@redhat.com>
2305 * doc/extend.texi (attribute nonstring): Document new attribute.
2307 2017-08-14 Martin Sebor <msebor@redhat.com>
2310 * tree-diagnostic.c (default_tree_printer): Handle %G.
2311 * gimple-pretty-print.h (percent_G_format): Declare new function.
2312 * gimple-pretty-print.c (percent_G_format): Define.
2313 * tree-pretty-print.c (percent_K_format): Add argument.
2315 2017-08-14 Martin Sebor <msebor@redhat.com>
2317 PR translation/79998
2318 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2319 Remove a stray space.
2321 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2324 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2325 (*iordi_1_bts): Ditto.
2326 (*xordi_1_btc): Ditto.
2328 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2331 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2334 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2335 quoted strings, and make more translator-friendly.
2336 (darwin_rs6000_override_options): Likewise.
2337 (rs6000_option_override_internal): Likewise.
2338 (rs6000_return_in_memory): Fix overlong line.
2339 (init_cmulative_args): Use quoted strings, and make more
2340 translator-friendly.
2341 (rs6000_pass_by_reference): Fix overlong line.
2342 (def_builtin): Use quoted strings.
2343 (altivec_expand_predicate_builtin): Use quoted strings, and make
2344 more translator-friendly.
2345 (htm_expand_builtin): Use quoted strings.
2346 (cpu_expand_builtin): Use quoted strings, and make more
2347 translator-friendly.
2348 (altivec_expand_builtin): Likewise.
2349 (paired_expand_predicate_builtin): Likewise.
2350 (rs6000_invalid_builtin): Likewise.
2351 (builtin_function_type): Use quoted strings.
2352 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2353 more translator-friendly.
2354 (rs6000_trampoline_init): Likewise.
2355 (rs6000_handle_altivec_attribute): Likewise.
2356 (rs6000_inner_target_options): Use quoted strings.
2357 (rs6000_disable_incompatible_switches): Likewise.
2358 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2359 strings, and make more translator-friendly.
2360 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2362 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2364 PR tree-optimization/81799
2365 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2366 cond_expr to simple gimple operand.
2368 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2371 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2373 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2377 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2378 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2379 TARGET_GASISR_PROLOGUES.
2380 * config/avr/avr.c (avr_option_override): Same.
2381 (avr_pass_pre_proep::execute): Same.
2383 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2386 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2387 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2389 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2391 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2392 define_insn_and_split. Split to a memory load from 0 in
2393 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2394 using PTR mode iterator.
2395 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2396 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2397 (*add_tp_<mode>): Redefine as define_insn_and_split.
2398 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2399 address space. Merge with *add_tp_x32 using PTR mode iterator.
2400 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2401 Split to an add with a memory load from 0 in
2402 DEFAULT_TLS_SEG_REG address space.
2404 2017-08-12 Andrew Pinski <apinski@cavium.com>
2406 * config/aarch64/aarch64-option-extensions.def (rdma):
2407 Fix feature string to what Linux prints out in /proc/cpuinfo.
2409 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2412 * dwarf2out.c (modified_type_die): For C typedef types that have
2413 an ultimate origin, process the ultimate origin instead of the
2415 (gen_typedef_die): Assert that input DECLs have no ultimate
2417 (gen_type_die_with_usage): For typedef variants that have an
2418 ultimate origin, just call gen_decl_die on the original DECL.
2419 (process_scope_var): Avoid creating DIEs for local typedefs and
2420 concrete static variables.
2422 2017-08-12 Alan Modra <amodra@gmail.com>
2426 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2427 match gnu-user.h startfile.
2428 (ENDFILE_LINUX_SPEC): Similarly.
2430 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2433 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2435 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2437 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2438 * config/aarch64/aarch64.md (mov<mode>): Change.
2439 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2440 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2441 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2443 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2445 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2446 for storage order barriers.
2448 2017-08-11 Martin Liska <mliska@suse.cz>
2451 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2452 variables of void type.
2454 2017-08-11 Martin Liska <mliska@suse.cz>
2456 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2457 TARGET_SUPPORTS_ALIASES.
2458 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2459 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2460 (optimize_weakref): Likewise.
2461 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2462 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2463 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2465 2017-08-11 Martin Liska <mliska@suse.cz>
2468 * config/i386/i386.c (make_resolver_func): Do complete
2469 refactoring of the function.
2471 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2474 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2475 * config/i386/i386.c (ix86_stack_protect_guard): Use
2476 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2477 * doc/invoke.texi (x86 Options): Document
2478 -mstack-protector-guard-symbol= option.
2480 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2482 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2483 * config/i386/i386.c (ix86_split_stack_guard): New function.
2484 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2485 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2486 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2487 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2488 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2489 (split_stack_space_check): Call ix86_split_stack_guard.
2491 2017-08-10 Martin Sebor <msebor@redhat.com>
2493 * print-tree.c (print_node): Print location using the established
2495 Replace spaces with colons.
2496 (debug_raw, debug): Ditto.
2498 2017-08-10 Martin Sebor <msebor@redhat.com>
2501 * pretty-print.c (pp_format): Correct the handling of %s precision.
2503 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2506 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2508 (ix86_finalize_stack_frame_flags): This. Also clear
2509 frame_pointer_needed if -fno-omit-frame-pointer is used without
2511 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2512 with ix86_finalize_stack_frame_flags.
2513 (ix86_expand_epilogue): Likewise.
2514 (ix86_expand_split_stack_prologue): Likewise.
2515 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2517 2017-08-10 Martin Liska <mliska@suse.cz>
2520 * c-attribs.c (handle_target_attribute):
2521 Report warning for an empty string argument of target attribute.
2523 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2526 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2528 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2529 or DECL_NONLOCAL labels.
2530 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2531 or DECL_NONLOCAL labels here.
2533 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2536 to indicate when early gimple folding has been disabled.
2537 (rs6000_gimple_fold_builtin): Add debug content.
2538 (rs6000_invalid_builtin): Fix whitespace.
2539 (rs6000_expand_builtin): Fix whitespace.
2540 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2542 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2545 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2546 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2548 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2550 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2551 (qdf24xx): Likewise.
2552 * config/aarch64/aarch64.md: Include falkor.md.
2553 * config/aarch64/falkor.md: New.
2555 2017-08-09 Marek Polacek <polacek@redhat.com>
2558 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2559 * diagnostic.c (emit_diagnostic): Add a comment.
2560 (emit_diagnostic_valist): New function.
2562 2017-08-09 Marek Polacek <polacek@redhat.com>
2565 * input.c (make_location): New overload.
2566 * input.h (make_location): Declare.
2568 2017-08-08 Alan Modra <amodra@gmail.com>
2569 H.J. Lu <hongjiu.lu@intel.com>
2572 * gcc.c (NO_PIE_SPEC): Delete.
2573 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2575 (LINK_PIE_SPEC): ..to here.
2576 (LINK_COMMAND_SPEC): Support -no-pie.
2577 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2578 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2580 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2581 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2582 (ENDFILE_CRTEND_SPEC): Similarly.
2584 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2587 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2588 (mstack-protector-guard-offset=): Ditto.
2589 * config/i386/i386.c (ix86_option_override): Handle
2590 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2592 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2593 ix86_stack_protect_guard_offset variables.
2594 (TARGET_STACK_PROTECT_GUARD): Always define.
2595 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2596 and -mstack-protector-guard-offset= options.
2598 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2600 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2601 boundary case for the last candidate.
2603 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2605 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2606 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2608 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2611 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2612 * config/aarch64/aarch64-builtins.c
2613 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2614 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2615 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2617 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2618 Andrew Pinski <pinskia@gmail.com>
2621 * internal-fn.def (XORSIGN): New.
2622 * optabs.def (xorsign_optab): New.
2623 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2624 (convert_expand_mult_copysign): New.
2625 (pass_optimize_widening_mul::execute): Call
2626 convert_expand_mult_copysign.
2628 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2630 PR tree-optimization/81354
2631 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2632 Insert on edges rather than explicitly creating landing pads.
2633 (analyze_candidates_and_replace): Commit edge inserts.
2635 2017-08-08 Richard Biener <rguenther@suse.de>
2638 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2639 (expand_simple_operations): Also look through ADDR_EXPRs with
2640 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2642 2017-08-08 Richard Biener <rguenther@suse.de>
2644 PR tree-optimization/81723
2645 * tree-vect-slp.c (struct bst_traits): New hash traits.
2646 (bst_fail): New global.
2647 (vect_build_slp_tree_2): New worker, split out from ...
2648 (vect_build_slp_tree): ... this now wrapping it with using
2649 bst_fail set to cache SLP tree build fails. Properly handle
2651 (vect_analyze_slp_instance): Allocate and free bst_fail.
2653 2017-08-08 Martin Liska <mliska@suse.cz>
2656 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2657 LABEL_DECLs that can be from a different function.
2659 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2661 PR tree-optimization/81744
2662 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2663 loop's number of iterations.
2665 2017-08-08 Martin Liska <mliska@suse.cz>
2667 * asan.c: Include header files.
2668 * attribs.c (build_decl_attribute_variant): New function moved
2670 (build_type_attribute_qual_variant): Likewise.
2671 (cmp_attrib_identifiers): Likewise.
2672 (simple_cst_list_equal): Likewise.
2673 (omp_declare_simd_clauses_equal): Likewise.
2674 (attribute_value_equal): Likewise.
2675 (comp_type_attributes): Likewise.
2676 (build_type_attribute_variant): Likewise.
2677 (lookup_ident_attribute): Likewise.
2678 (remove_attribute): Likewise.
2679 (merge_attributes): Likewise.
2680 (merge_type_attributes): Likewise.
2681 (merge_decl_attributes): Likewise.
2682 (merge_dllimport_decl_attributes): Likewise.
2683 (handle_dll_attribute): Likewise.
2684 (attribute_list_equal): Likewise.
2685 (attribute_list_contained): Likewise.
2686 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2687 (lookup_attribute_by_prefix): Likewise.
2688 * bb-reorder.c: Include header files.
2689 * builtins.c: Likewise.
2690 * calls.c: Likewise.
2691 * cfgexpand.c: Likewise.
2692 * cgraph.c: Likewise.
2693 * cgraphunit.c: Likewise.
2694 * convert.c: Likewise.
2695 * dwarf2out.c: Likewise.
2696 * final.c: Likewise.
2697 * fold-const.c: Likewise.
2698 * function.c: Likewise.
2699 * gimple-expr.c: Likewise.
2700 * gimple-fold.c: Likewise.
2701 * gimple-pretty-print.c: Likewise.
2702 * gimple.c: Likewise.
2703 * gimplify.c: Likewise.
2704 * hsa-common.c: Likewise.
2705 * hsa-gen.c: Likewise.
2706 * internal-fn.c: Likewise.
2707 * ipa-chkp.c: Likewise.
2708 * ipa-cp.c: Likewise.
2709 * ipa-devirt.c: Likewise.
2710 * ipa-fnsummary.c: Likewise.
2711 * ipa-inline.c: Likewise.
2712 * ipa-visibility.c: Likewise.
2714 * lto-cgraph.c: Likewise.
2715 * omp-expand.c: Likewise.
2716 * omp-general.c: Likewise.
2717 * omp-low.c: Likewise.
2718 * omp-offload.c: Likewise.
2719 * omp-simd-clone.c: Likewise.
2720 * opts-global.c: Likewise.
2721 * passes.c: Likewise.
2722 * predict.c: Likewise.
2723 * sancov.c: Likewise.
2724 * sanopt.c: Likewise.
2725 * symtab.c: Likewise.
2726 * toplev.c: Likewise.
2727 * trans-mem.c: Likewise.
2728 * tree-chkp.c: Likewise.
2729 * tree-eh.c: Likewise.
2730 * tree-into-ssa.c: Likewise.
2731 * tree-object-size.c: Likewise.
2732 * tree-parloops.c: Likewise.
2733 * tree-profile.c: Likewise.
2734 * tree-ssa-ccp.c: Likewise.
2735 * tree-ssa-live.c: Likewise.
2736 * tree-ssa-loop.c: Likewise.
2737 * tree-ssa-sccvn.c: Likewise.
2738 * tree-ssa-structalias.c: Likewise.
2739 * tree-ssa.c: Likewise.
2740 * tree-streamer-in.c: Likewise.
2741 * tree-vectorizer.c: Likewise.
2742 * tree-vrp.c: Likewise.
2744 * ubsan.c: Likewise.
2745 * varasm.c: Likewise.
2746 * varpool.c: Likewise.
2747 * tree.c: Remove functions moved to attribs.[ch].
2749 * config/aarch64/aarch64.c: Add attrs.h header file.
2750 * config/alpha/alpha.c: Likewise.
2751 * config/arc/arc.c: Likewise.
2752 * config/arm/arm.c: Likewise.
2753 * config/avr/avr.c: Likewise.
2754 * config/bfin/bfin.c: Likewise.
2755 * config/c6x/c6x.c: Likewise.
2756 * config/cr16/cr16.c: Likewise.
2757 * config/cris/cris.c: Likewise.
2758 * config/darwin.c: Likewise.
2759 * config/epiphany/epiphany.c: Likewise.
2760 * config/fr30/fr30.c: Likewise.
2761 * config/frv/frv.c: Likewise.
2762 * config/ft32/ft32.c: Likewise.
2763 * config/h8300/h8300.c: Likewise.
2764 * config/i386/winnt.c: Likewise.
2765 * config/ia64/ia64.c: Likewise.
2766 * config/iq2000/iq2000.c: Likewise.
2767 * config/lm32/lm32.c: Likewise.
2768 * config/m32c/m32c.c: Likewise.
2769 * config/m32r/m32r.c: Likewise.
2770 * config/m68k/m68k.c: Likewise.
2771 * config/mcore/mcore.c: Likewise.
2772 * config/microblaze/microblaze.c: Likewise.
2773 * config/mips/mips.c: Likewise.
2774 * config/mmix/mmix.c: Likewise.
2775 * config/mn10300/mn10300.c: Likewise.
2776 * config/moxie/moxie.c: Likewise.
2777 * config/msp430/msp430.c: Likewise.
2778 * config/nds32/nds32-isr.c: Likewise.
2779 * config/nds32/nds32.c: Likewise.
2780 * config/nios2/nios2.c: Likewise.
2781 * config/nvptx/nvptx.c: Likewise.
2782 * config/pa/pa.c: Likewise.
2783 * config/pdp11/pdp11.c: Likewise.
2784 * config/powerpcspe/powerpcspe.c: Likewise.
2785 * config/riscv/riscv.c: Likewise.
2786 * config/rl78/rl78.c: Likewise.
2787 * config/rx/rx.c: Likewise.
2788 * config/s390/s390.c: Likewise.
2789 * config/sh/sh.c: Likewise.
2790 * config/sol2.c: Likewise.
2791 * config/sparc/sparc.c: Likewise.
2792 * config/spu/spu.c: Likewise.
2793 * config/stormy16/stormy16.c: Likewise.
2794 * config/tilegx/tilegx.c: Likewise.
2795 * config/tilepro/tilepro.c: Likewise.
2796 * config/v850/v850.c: Likewise.
2797 * config/vax/vax.c: Likewise.
2798 * config/visium/visium.c: Likewise.
2799 * config/xtensa/xtensa.c: Likewise.
2801 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2804 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
2805 constraints since the -mupper-regs-* switches have been
2807 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
2808 into a vector from a double word element that was extracted from
2809 another vector, and eliminate extra XXPERMDI instructions.
2810 (vsx_concat_<mode>_2): Likewise.
2811 (vsx_concat_<mode>_3): Likewise.
2812 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
2813 concat to allow optimizing inserts from previous extracts.
2815 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
2817 * config/i386/i386.c (ix86_stack_protect_guard): Generate
2818 memory reference to a SSP offset in TLS address space.
2819 (ix86_print_operand) <case '@'>: Remove.
2820 (ix86_print_operand_punct_valid_p): Remove '@' code.
2821 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
2823 (stack_tls_protect_set_<mode>): Remove.
2824 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
2825 (stack_tls_protect_test_<mode>): Remove.
2826 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
2828 2017-08-07 Olivier Hainque <hainque@adacore.com>
2831 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
2833 2017-08-07 Douglas Rupp <rupp@adacore.com>
2835 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
2836 variable was referenced as multidir in command.
2838 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2841 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
2844 2017-08-07 Martin Liska <mliska@suse.cz>
2846 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
2847 * config/rl78/rl78.c: Add include of attribs.h.
2848 * config/sh/sh.c: Likewise.
2849 * config/v850/v850.c: Likewise.
2851 2017-08-07 Tom de Vries <tom@codesourcery.com>
2854 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
2856 2017-08-07 Martin Liska <mliska@suse.cz>
2858 * config/mips/mips.c: Include attribs.h.
2860 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
2863 * doc/invoke.texi: Document change in behvaior for -Ofast for
2866 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
2868 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
2870 (aarch64_pop_regs): Likewise.
2871 (aarch64_gen_load_pair): Likewise.
2872 (aarch64_save_callee_saves): Likewise.
2873 (aarch64_restore_callee_saves): Likewise.
2875 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2877 * config/i386/i386.c: Revert the last change.
2879 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2882 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2884 (ix86_finalize_stack_frame_flags): This. Also clear
2885 frame_pointer_needed if -fno-omit-frame-pointer is used without
2887 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2888 with ix86_finalize_stack_frame_flags.
2889 (ix86_expand_epilogue): Likewise.
2890 (ix86_expand_split_stack_prologue): Likewise.
2892 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2895 * config/i386/i386.c (get_builtin_code_for_version): Set priority
2896 to P_AES for Westmere.
2898 2017-08-07 Jonathan Yong <10walls@gmail.com>
2900 * config/i386/mingw.opt (fset-stack-executable): Removed.
2901 * config/i386/cygming.opt (fset-stack-executable): Moved
2903 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
2905 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
2907 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
2909 2017-08-07 Marek Polacek <polacek@redhat.com>
2912 * fold-const.c (fold_indirect_ref_1): Check type_domain.
2914 2017-08-07 Martin Liska <mliska@suse.cz>
2916 * attribs.h (canonicalize_attr_name): New function.
2917 (cmp_attribs): Move from c-format.c and adjusted.
2918 (is_attribute_p): Moved from tree.h.
2919 * tree-inline.c: Add new includes.
2920 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
2921 (private_is_attribute_p): Remove.
2922 (private_lookup_attribute): Likewise.
2923 (private_lookup_attribute_by_prefix): Simplify.
2924 (remove_attribute): Use is_attribute_p.
2925 * tree.h: Remove removed declarations.
2927 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2930 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
2931 instead of computing it in the function. Formatting fix.
2932 (expand_case): Don't rely on default_edge being the first edge,
2933 clear it if removing it, pass default_edge to
2934 emit_case_dispatch_table.
2935 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
2938 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
2940 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
2941 insn in the function, emit NOP after the insn.
2943 2017-08-06 Tom de Vries <tom@codesourcery.com>
2945 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
2948 2017-08-06 Tom de Vries <tom@codesourcery.com>
2950 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
2953 2017-08-04 Yury Gribov <tetra2005@gmail.com>
2955 PR tree-optimization/57371
2956 * match.pd: New pattern.
2958 2017-08-04 Marek Polacek <polacek@redhat.com>
2961 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
2962 perform the computation in offset_int.
2964 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2966 PR tree-optimization/81136
2967 * tree-vectorizer.h: Include tree-hash-traits.h.
2968 (vec_base_alignments): New typedef.
2969 (vec_info): Add a base_alignments field.
2970 (vect_record_base_alignments): Declare.
2971 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
2973 (DR_IS_CONDITIONAL_IN_STMT): New macro.
2974 (create_data_ref): Add an is_conditional_in_stmt argument.
2975 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
2976 the is_conditional_in_stmt field.
2977 (data_ref_loc): Add an is_conditional_in_stmt field.
2978 (get_references_in_stmt): Set the is_conditional_in_stmt field.
2979 (find_data_references_in_stmt): Update call to create_data_ref.
2980 (graphite_find_data_references_in_stmt): Likewise.
2981 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
2982 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
2983 (vect_record_base_alignment): New function.
2984 (vect_record_base_alignments): Likewise.
2985 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
2986 for nested statements even if we fail to compute a misalignment.
2987 Use pooled base alignments for unconditional references.
2988 (vect_find_same_alignment_drs): Compare base addresses instead
2990 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
2991 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
2993 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2995 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
2996 Add an explicit name for the enum. Use auto_vec for slp_instances
2998 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3000 (_bb_vec_info): Add a constructor and destructor.
3001 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3002 (destroy_loop_vec_info): Delete.
3003 (vect_destroy_datarefs): Likewise.
3004 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3005 (vec_info::vec_info): New function.
3006 (vec_info::~vec_info): Likewise.
3007 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3008 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3009 destroy_loop_vec_info.
3010 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3011 (_loop_vec_info::_loop_vec_info): ...this.
3012 (destroy_loop_vec_info): Replace with...
3013 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3014 the stmt_vec_infos. Leave handling of vec_info information to its
3015 destructor. Remove explicit vector releases.
3016 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3017 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3018 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3019 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3020 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3021 (destroy_bb_vec_info): Replace with...
3022 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3023 information to its destructor.
3024 (vect_slp_analyze_bb_1): Use new and delete instead of
3025 new_bb_vec_info and destroy_bb_vec_info.
3026 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3029 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3031 * tree-data-ref.h (subscript): Add access_fn field.
3032 (data_dependence_relation): Add could_be_independent_p.
3033 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3034 (same_access_functions): Move to tree-data-ref.c.
3035 * tree-data-ref.c (ref_contains_union_access_p): New function.
3036 (access_fn_component_p): Likewise.
3037 (access_fn_components_comparable_p): Likewise.
3038 (dr_analyze_indices): Add a reference to access_fn_component_p.
3039 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3041 (constant_access_functions): Likewise.
3042 (add_other_self_distances): Likewise.
3043 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3044 (initialize_data_dependence_relation): Use XCNEW and remove
3045 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3046 of access functions that have the same type. Allow the
3047 subsequence to end with different bases in some circumstances.
3048 Record the chosen access functions in SUB_ACCESS_FN.
3049 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3050 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3051 (subscript_dependence_tester_1): Likewise dra and drb.
3052 (build_classic_dist_vector): Update calls accordingly.
3053 (subscript_dependence_tester): Likewise.
3054 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3055 DDR_COULD_BE_INDEPENDENT_P.
3056 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3057 comp_alias_ddrs instead of may_alias_ddrs.
3058 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3060 (vect_analyze_data_ref_dependence): Use it if
3061 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3062 distance vectors if that fails.
3063 (dependence_distance_ge_vf): New function.
3064 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3065 LOOP_VINFO_MAY_ALIAS_DDRS.
3067 2017-08-04 Richard Biener <rguenther@suse.de>
3070 * fold-const.c (fold_binary_loc): Properly restrict
3071 minus_var0 && minus_var1 case when associating undefined overflow
3074 2017-08-04 Tom de Vries <tom@codesourcery.com>
3076 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3078 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3080 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3081 Don't start diagnostic messages with a capital letter.
3082 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3084 (rs6000_invalid_builtin): Likewise.
3085 (rs6000_trampoline_init): Likewise.
3087 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3090 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3091 after setting changeable df flags.
3093 2017-08-03 Richard Biener <rguenther@suse.de>
3095 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3096 up if the use is in USE - X.
3098 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3100 * toplev.c (dumpfile.h): New include.
3101 (internal_error_reentered): New static function. Use it...
3102 (internal_error_function): ...here to handle reentered internal_error.
3104 2017-08-03 Richard Biener <rguenther@suse.de>
3107 * fold-const.c (split_tree): Add minus_var and minus_con
3108 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3109 here but always use minus_*.
3110 (associate_trees): Assert we never associate with MINUS_EXPR
3111 and NULL first operand. Do not recurse for PLUS_EXPR operands
3112 when associating as MINUS_EXPR either.
3113 (fold_binary_loc): Track minus_var and minus_con.
3115 2017-08-03 Tom de Vries <tom@codesourcery.com>
3118 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3119 ACCEL_COMPILER, apply finish_options on
3120 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3122 2017-08-03 Tom de Vries <tom@codesourcery.com>
3125 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3126 function_entry_patch_area_size > 0.
3128 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3131 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3132 instead of 10??LU, perform unit multiplication in wide_int,
3133 don't change alloc_object_size_limit if the limit is larger
3136 PR tree-optimization/81655
3137 PR tree-optimization/81588
3138 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3139 the case when ranges[i].low and high are 1 for unsigned type with
3143 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3144 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3146 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3148 * tree-vrp.h: Add include guard.
3150 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3153 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3154 (ud2): New insn pattern.
3155 * config/i386/i386.c (ix86_expand_epilogue):
3156 For naked functions, generate ud2 instead of trap insn.
3158 2017-08-02 Marek Polacek <polacek@redhat.com>
3161 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3163 2017-08-02 Tom de Vries <tom@codesourcery.com>
3164 Cesar Philippidis <cesar@codesourcery.com>
3166 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3167 Add missing edge probabilities.
3169 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3171 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3174 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3177 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3178 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3179 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3180 prologue_seq sequences - if any.
3182 2017-08-02 Richard Biener <rguenther@suse.de>
3184 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3185 via vectors if supported, integer extracts via punning if supported
3186 or otherwise vector extracts.
3188 2017-08-02 Richard Biener <rguenther@suse.de>
3190 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3192 (bitmap_insert_into_set): ... this.
3194 2017-08-02 Richard Biener <rguenther@suse.de>
3196 PR tree-optimization/81633
3198 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3200 PR tree-optimization/71752
3201 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3203 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3205 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3206 (machine_function::call_ms2sysv_pad_out): Remove field.
3207 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3208 (ix86_compute_frame_layout): Likewise.
3210 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3213 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3214 attribute with interrupt attribute.
3216 2017-08-01 Andrew Pinski <apinski@cavium.com>
3218 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3219 BIT_INSERT_EXPR's operand 1
3220 to see if the types precision matches.
3222 2017-08-01 Martin Liska <mliska@suse.cz>
3225 * builtins.c (expand_builtin_memcpy_args): Remove.
3226 (expand_builtin_memcpy): Call newly added function
3227 expand_builtin_memory_copy_args.
3228 (expand_builtin_memcpy_with_bounds): Likewise.
3229 (expand_builtin_mempcpy): Remove last argument.
3230 (expand_builtin_mempcpy_with_bounds): Likewise.
3231 (expand_builtin_memory_copy_args): New function created from
3232 expand_builtin_mempcpy_args with small modifications.
3233 (expand_builtin_mempcpy_args): Remove.
3234 (expand_builtin_stpcpy): Remove unused argument.
3235 (expand_builtin): Likewise.
3236 (expand_builtin_with_bounds): Likewise.
3238 2017-08-01 Martin Liska <mliska@suse.cz>
3241 Make mempcpy more optimal (PR middle-end/70140).
3243 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3246 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3247 __builtin_vec_cmpne verify both arguments are compatible vectors
3248 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3249 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3250 move computation of aligned to after checking the argument types.
3254 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3255 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3258 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3259 Doug Rupp <rupp@adacore.com>
3260 Olivier Hainque <hainque@adacore.com>
3262 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3263 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3265 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3266 for TARGET_OS_CPP_BUILTIN.
3267 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3268 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3270 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3271 passing required abi options to the assembler for EABI configurations.
3272 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3273 of .text.hot and .text.unlikely sections for kernel modules when
3274 using ARM style exceptions.
3275 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3276 options. Add EXTRA_CC1_SPEC.
3277 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3279 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3281 (ARM_TARGET2_DWARF_FORMAT): Define.
3282 * config/arm/t-vxworks: Adjust multilib control to removal of the
3283 Diab command line options.
3285 2017-08-01 Martin Liska <mliska@suse.cz>
3287 PR gcov-profile/81561
3288 * gcov.c (unblock): Make unblocking safe as we need to preserve
3289 index correspondence of blocks and block_lists.
3291 2017-08-01 Richard Biener <rguenther@suse.de>
3293 PR tree-optimization/81181
3294 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3295 (compute_antic): ... end of iteration here.
3297 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3299 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3300 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3301 (ftree-slp-vectorize): Likewise.
3302 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3303 can no longer be set independent of flag_tree_loop_vectorize.
3304 * omp-general.c (emp_max_vf): Likewise.
3305 * opts.c (enable_fdo_optimizations): Remove references to
3306 flag_tree_vectorize, these are now implicit.
3307 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3308 and leave it for the options machinery.
3310 2017-08-01 Martin Liska <mliska@suse.cz>
3313 * builtins.c (expand_builtin_memcpy_args): Remove.
3314 (expand_builtin_memcpy): Call newly added function
3315 expand_builtin_memory_copy_args.
3316 (expand_builtin_memcpy_with_bounds): Likewise.
3317 (expand_builtin_mempcpy): Remove last argument.
3318 (expand_builtin_mempcpy_with_bounds): Likewise.
3319 (expand_builtin_memory_copy_args): New function created from
3320 expand_builtin_mempcpy_args with small modifications.
3321 (expand_builtin_mempcpy_args): Remove.
3322 (expand_builtin_stpcpy): Remove unused argument.
3323 (expand_builtin): Likewise.
3324 (expand_builtin_with_bounds): Likewise.
3326 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3329 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3330 print "ds:" only for immediates in generic address space.
3332 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3335 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3336 (ix86_function_ok_for_sibcall): Return false for naked functions.
3338 2017-08-01 Richard Biener <rguenther@suse.de>
3340 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3341 (compute_antic): Seed worklist with exit block predecessors.
3342 * cfganal.c (dfs_find_deadend): For a cycle return the source
3343 of the edge closing it.
3345 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3347 * config/aarch64/aarch64.c
3348 (aarch64_can_const_movi_rtx_p): Move 0 check.
3350 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3352 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3353 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3355 * match.pd: Ditto in address comparison pattern.
3357 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3359 PR tree-optimization/81627
3360 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3361 closed ssa form for store-store chain.
3363 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3365 PR tree-optimization/81620
3366 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3367 for store-store chain.
3369 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3371 PR tree-optimization/81588
3372 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3373 ranges[i].in_p, invert comparison code ccode. For >/>=,
3374 swap rhs1 and rhs2 and comparison code unconditionally,
3375 for </<= don't do that. Don't swap rhs1/rhs2 again if
3376 ranges[i].in_p, instead invert comparison code ccode if
3377 opcode or oe->rank is BIT_IOR_EXPR.
3380 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3381 a direct optab to conversion optab.
3382 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3383 with GET_MODE_INNER as last argument instead of optab_handler.
3384 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3385 vector extraction if possible and optab is available.
3386 * expr.c (store_constructor): Use convert_optab_handler instead
3387 of optab_handler. Use vector initialization from smaller
3388 vectors if possible and optab is available.
3389 * tree-vect-stmts.c (vectorizable_load): Likewise.
3390 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3392 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3393 of vec_init from half-sized vectors with the same element mode.
3394 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3395 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3396 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3397 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3398 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3399 after mode in gen_vec_extract* calls.
3400 (vec_extract<mode>): Renamed to ...
3401 (vec_extract<mode><ssescalarmodelower>): ... this.
3402 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3403 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3404 element mode after mode in gen_vec_init* calls.
3405 (VEC_INIT_HALF_MODE): New mode iterator.
3406 (vec_init<mode>): Renamed to ...
3407 (vec_init<mode><ssescalarmodelower>): ... this.
3408 (vec_init<mode><ssehalfvecmodelower>): New expander.
3409 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3410 (vec_extractv2sfsf): ... this.
3411 (vec_initv2sf): Renamed to ...
3412 (vec_initv2sfsf): ... this.
3413 (vec_extractv2si): Renamed to ...
3414 (vec_extractv2sisi): ... this.
3415 (vec_initv2si): Renamed to ...
3416 (vec_initv2sisi): ... this.
3417 (vec_extractv4hi): Renamed to ...
3418 (vec_extractv4hihi): ... this.
3419 (vec_initv4hi): Renamed to ...
3420 (vec_initv4hihi): ... this.
3421 (vec_extractv8qi): Renamed to ...
3422 (vec_extractv8qiqi): ... this.
3423 (vec_initv8qi): Renamed to ...
3424 (vec_initv8qiqi): ... this.
3425 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3426 (vec_init<mode>): Renamed to ...
3427 (vec_init<mode><VEC_base_l>): ... this.
3428 (vec_extract<mode>): Renamed to ...
3429 (vec_extract<mode><VEC_base_l>): ... this.
3430 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3431 (vec_initv2sfsf): ... this.
3432 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3433 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3434 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3435 element mode after mode in gen_vec_init* calls.
3436 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3437 (vec_init<mode><Vel>): ... this.
3438 (vec_extract<mode>): Renamed to ...
3439 (vec_extract<mode><Vel>): ... this.
3440 * config/aarch64/iterators.md (Vel): New mode attribute.
3441 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3442 Add element mode after mode in gen_vec_extract* calls.
3443 * config/s390/vector.md (non_vec_l): New mode attribute.
3444 (vec_extract<mode>): Renamed to ...
3445 (vec_extract<mode><non_vec_l>): ... this.
3446 (vec_init<mode>): Renamed to ...
3447 (vec_init<mode><non_vec_l>): ... this.
3448 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3449 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3451 * config/arm/iterators.md (V_elem_l): New mode attribute.
3452 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3453 (vec_extract<mode><V_elem_l>): ... this.
3454 (vec_extractv2di): Renamed to ...
3455 (vec_extractv2didi): ... this.
3456 (vec_init<mode>): Renamed to ...
3457 (vec_init<mode><V_elem_l>): ... this.
3458 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3459 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3460 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3461 Add element mode after gen_vec_extract* calls.
3462 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3463 (vec_init<mode><unitmode>): ... this.
3464 (vec_extract<mode>): Renamed to ...
3465 (vec_extract<mode><unitmode>): ... this.
3466 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3467 (vec_init<mode><unitmode>): ... this.
3468 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3469 (vec_initv2sfsf): ... this.
3470 (vec_extractv2sf): Renamed to ...
3471 (vec_extractv2sfsf): ... this.
3472 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3473 Add element mode after gen_vec_extract* calls.
3474 * config/mips/mips.md (unitmode): New mode iterator.
3475 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3476 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3477 * config/spu/spu.md (inner_l): New mode attribute.
3478 (vec_init<mode>): Renamed to ...
3479 (vec_init<mode><inner_l>): ... this.
3480 (vec_extract<mode>): Renamed to ...
3481 (vec_extract<mode><inner_l>): ... this.
3482 * config/sparc/sparc.md (veltmode): New mode iterator.
3483 (vec_init<VMALL:mode>): Renamed to ...
3484 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3485 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3486 (vec_initv2sisi): ... this.
3487 (vec_initv2sf): Renamed to ...
3488 (vec_initv2sfsf): ... this.
3489 (vec_extractv2sf): Renamed to ...
3490 (vec_extractv2sfsf): ... this.
3491 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3492 (vec_init<mode>): Renamed to ...
3493 (vec_init<mode><VEC_base_l>): ... this.
3494 (vec_extract<mode>): Renamed to ...
3495 (vec_extract<mode><VEC_base_l>): ... this.
3496 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3497 (vec_initv2sfsf): ... this.
3498 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3499 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3500 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3501 gen_vec_init* calls.
3503 2017-08-01 Richard Biener <rguenther@suse.de>
3505 PR tree-optimization/81297
3506 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3507 TREE_OVERFLOW from INTEGER_CSTs.
3509 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3511 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3513 2017-07-31 Carl Love <cel@us.ibm.com>
3515 * config/rs6000/rs6000-c: Add support for built-in functions
3516 vector signed char vec_xl_be (signed long long, signed char *);
3517 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3518 vector signed int vec_xl_be (signed long long, signed int *);
3519 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3520 vector signed long long vec_xl_be (signed long long, signed long long *);
3521 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3522 vector signed short vec_xl_be (signed long long, signed short *);
3523 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3524 vector double vec_xl_be (signed long long, double *);
3525 vector float vec_xl_be (signed long long, float *);
3526 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3527 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3528 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3530 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3531 (altivec_expand_builtin): Add switch statement to call
3532 altivec_expand_xl_be for each builtin.
3533 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3534 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3535 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3536 __builtin_vsx_le_be_v16qi.
3537 * doc/extend.texi: Update the built-in documentation file for the
3538 new built-in functions.
3540 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3543 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3545 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3547 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3549 * config.gcc: Add z14.
3550 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3551 CPU model numbers for z13s and z14.
3552 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3554 * config/s390/s390-opts.h (enum processor_type): Rename
3555 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3556 * config/s390/s390.c (processor_table): Add field for CPU name to
3557 be passed to Binutils.
3558 (s390_asm_output_machine_for_arch): Use the new field in
3559 processor_table for Binutils.
3560 (s390_expand_builtin): Replace arch12 with z14.
3561 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3562 (s390_get_sched_attrmask): Likewise.
3563 (s390_get_unit_mask): Likewise.
3564 * config/s390/s390.opt: Add z14 to processor_type enum.
3566 2017-07-31 Martin Jambor <mjambor@suse.cz>
3569 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3570 regardless of optimization level.
3572 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3573 Martin Liska <mliska@suse.cz>
3575 * predict.def: Remove old comment and adjust probability.
3576 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3579 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3582 * config/i386/i386.c (ix86_function_naked): New function.
3583 (ix86_can_use_return_insn_p): Return false for naked functions.
3584 (ix86_expand_prologue): Skip prologue for naked functions.
3585 (ix86_expand_epilogue): Skip epilogue for naked functions
3586 and emit trap instruction.
3587 (ix86_warn_func_return): New function.
3588 (ix86_attribute_table): Add "naked" attribute specification.
3589 (TARGET_WARN_FUNC_RETURN): Define.
3590 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3592 2017-07-31 Martin Liska <mliska@suse.cz>
3594 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3595 (dump_gimple_bb_header): Always dump BB info.
3596 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3598 2017-07-31 Martin Liska <mliska@suse.cz>
3601 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3602 also with current_function_decl non-null equality.
3604 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3607 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3608 change type to the element type, instead add eltype variable and
3609 use it where we are interested in the element type.
3611 PR tree-optimization/81603
3612 * ipa-polymorphic-call.c
3613 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3614 offset arithmetic in offset_int, bail out if the resulting bit offset
3615 doesn't fit into shwi.
3617 2017-07-31 Martin Liska <mliska@suse.cz>
3619 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3620 (gimplify_save_expr): Fix comment.
3622 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3625 * config/i386/i386.c (ix86_function_arg): Update arguments for
3627 (ix86_compute_frame_layout): Set the initial stack offset to
3628 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3629 INCOMING_FRAME_SP_OFFSET.
3630 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3631 stack before exception handler returns.
3632 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3633 the 'ERROR_CODE' for exception handler.
3635 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3637 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3638 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3639 (ASM_OUTPUT_REG_POP): Ditto.
3640 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3641 instead of asm_fprintf to output pure string.
3643 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3645 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3646 to imported_module_or_decl hook.
3647 (debug_nothing_tree_tree_tree_bool): Remove.
3648 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3649 * debug.c (do_nothing_debug_hooks): Use
3650 debug_nothing_tree_tree_tree_bool_bool instead of
3651 debug_nothing_tree_tree_tree_bool.
3652 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3653 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3654 * sdbout.c (sdb_debug_hooks): Likewise.
3655 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3656 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3658 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3659 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3660 attribute, don't add anything.
3662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3664 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3665 (fold_build2_stat_loc): Likewise.
3666 (fold_build3_stat_loc): Likewise.
3667 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3668 (fold_build1_loc): Remove macro.
3669 (fold_build2_loc): Likewise.
3670 (fold_build3_loc): Likewise.
3672 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3674 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3675 (gimple_build_debug_bind_source_stat): Likewise.
3676 * gimple.h (gimple_build_debug_bind): Remove macro.
3677 (gimple_build_debug_bind_source): Likewise.
3679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3681 * bitmap.c (bitmap_alloc): Adjust.
3682 (bitmap_gc_alloc): Likewise.
3683 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3685 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3687 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3688 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3689 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
3690 (bitmap_gc_alloc_stat): Likewise.
3691 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
3693 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3695 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
3696 * rtl.h (shallow_copy_rtx): Remove macro.
3698 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3700 * emit-rtl.c (gen_raw_REG): Adjust.
3701 * gengenrtl.c (gendef): Likewise.
3702 * rtl.c (rtx_alloc_stat): Remove _stat from name.
3703 * rtl.h (rtx_alloc): Remove macro.
3705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3707 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
3708 (build_tree_list_stat): Likewise.
3709 * tree.h (build_tree_list): Remove macro.
3710 (build_tree_list_vec): Likewise.
3712 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3714 * tree.c (make_vector_stat): Remove _stat from name.
3715 (build_vector_stat): Likewise.
3716 * tree.h (make_vector_stat): Remove macro.
3717 (build_vector_stat): Likewise.
3719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3721 * tree.h (build_var_debug_value): Remove prototype.
3723 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3725 * tree.c (tree_cons_stat): Remove _stat from name.
3726 * tree.h (tree_cons): Remove macro.
3728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3730 * tree.c (build_vl_exp_stat): Remove _stat from name.
3731 * tree.h (build_vl_exp): Remove macro.
3733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3735 * tree.c (build_decl_stat): Remove _stat from name.
3736 * tree.h (build_decl): Remove macro.
3738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3740 * gimple.c (gimple_build_with_ops_stat): Adjust.
3741 (gimple_alloc_stat): Remove _stat from name.
3742 * gimple.h (gimple_alloc): Remove macro.
3744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3746 * tree.c (make_tree_vec_stat): Remove _stat from name.
3747 (grow_tree_vec_stat): Likewise.
3748 * tree.h (make_tree_vec_stat): Adjust prototype.
3749 (grow_tree_vec_stat): Likewise.
3750 (make_tree_vec): Remove macro.
3751 (grow_tree_vec): Likewise.
3753 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3755 * fold-const.c (fold_build1_stat_loc): Adjust.
3756 (fold_build2_stat_loc): Likewise.
3757 (fold_build3_stat_loc): Likewise.
3758 * tree.c (build0_stat): Remove _stat from name.
3759 (build1_stat): Likewise.
3760 (build2_stat): Likewise.
3761 (build3_stat): Likewise.
3762 (build4_stat): Likewise.
3763 (build5_stat): Likewise.
3764 * tree.h (build1_loc): Remove macro, and rename _stat function
3766 (build2_loc): Likewise.
3767 (build3_loc): Likewise.
3768 (build4_loc): Likewise.
3769 (build5_loc): Likewise.
3771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3773 * tree.c (make_int_cst_stat): Remove _stat from name.
3774 * tree.h (make_int_cst_stat): Adjust prototype.
3775 (make_int_cst): Remove macro.
3777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3779 * tree.c (make_tre_binfo_stat): Remove _stat from name.
3780 * tree.h (make_tree_binfo_stat): Adjust prototype.
3781 (make_tree_binfo): Remove.
3783 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3785 * tree.c (copy_node_stat): Rename to copy_node.
3786 (build_distinct_type_copy): Adjust.
3787 * tree.h (copy_node_stat): Adjust prototype.
3788 (copy_node): Remove macro.
3790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3792 * tree.c (make_node_stat): rename to make_node.
3793 (build_tree_list_stat): Adjust.
3794 (build0_stat): Likewise.
3795 (build2_stat): Likewise.
3796 (build3_stat): Likewise.
3797 (build4_stat): Likewise.
3798 (build5_stat): Likewise.
3799 (build_decl_stat): Likewise.
3800 * tree.h (make_node_stat): Adjust prototype.
3801 (make_node): remove macro.
3803 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
3805 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
3806 (PPC_FEATURE2_SCV): Likewise.
3807 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
3809 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3811 * config/aarch64/aarch64.c
3812 (aarch64_internal_mov_immediate): Add new special pattern.
3813 * config/aarch64/aarch64.md (*movdi_aarch64):
3814 Add reg/32bit const mov case.
3816 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3817 Richard Sandiford <richard.sandiford@linaro.org>
3819 * config/aarch64/aarch64.md (mov<mode>): Generalize.
3820 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3821 Add integer and movi cases.
3822 (movi-split-hf-df-sf split, fp16): New.
3823 (enabled): Added TARGET_FP_F16INST.
3824 * config/aarch64/iterators.md (GPF_HF): New.
3825 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
3827 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3829 * config/aarch64/aarch64.c
3830 (aarch64_simd_container_mode): Add prototype.
3831 (aarch64_expand_mov_immediate): Add HI support.
3832 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
3833 (aarch64_can_const_movi_rtx_p): New.
3834 (aarch64_preferred_reload_class):
3835 Remove restrictions of using FP registers for certain SIMD operations.
3836 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
3837 (aarch64_valid_floating_const): Add integer move validation.
3838 (aarch64_simd_imm_scalar_p): Remove.
3839 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
3840 (aarch64_legitimate_constant_p): Expand list of supported cases.
3841 * config/aarch64/aarch64-protos.h
3842 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
3843 (aarch64_reinterpret_float_as_int): New.
3844 (aarch64_simd_imm_scalar_p): Remove.
3845 * config/aarch64/constraints.md (Uvi): New.
3846 (Dd): Split into Ds and new Dd.
3847 * config/aarch64/aarch64.md (*movsi_aarch64):
3849 (*movdi_aarch64): Add SIMD mov case.
3851 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3853 * tree-predcom.c: (struct chain): Handle store-store chain in which
3854 stores for elimination only store loop invariant values.
3855 (execute_pred_commoning_chain): Ditto.
3856 (prepare_initializers_chain_store_elim): Ditto.
3857 (prepare_finalizers): Ditto.
3858 (is_inv_store_elimination_chain): New function.
3859 (initialize_root_vars_store_elim_1): New function.
3861 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3863 * tree-predcom.c: Revise general description of the pass.
3864 (enum chain_type): New enum type for store elimination.
3865 (struct chain): New field supporting store elimination.
3866 (struct component): Ditto.
3867 (dump_chain): Dump store-stores chain.
3868 (release_chain): Release resources.
3869 (split_data_refs_to_components): Compute and create component
3870 contains only stores for elimination.
3871 (get_chain_last_ref_at): New function.
3872 (make_invariant_chain): Initialization.
3873 (make_rooted_chain): Specify chain type in parameter and record it.
3874 (add_looparound_copies): Skip for store-stores chain.
3875 (determine_roots_comp): Compute type of chain and pass it to
3877 (initialize_root_vars_store_elim_2): New function.
3878 (finalize_eliminated_stores): New function.
3879 (remove_stmt): Handle store for elimination.
3880 (execute_pred_commoning_chain): Execute predictive commoning on
3882 (determine_unroll_factor): Skip unroll for store-stores chain.
3883 (prepare_initializers_chain_store_elim): New function.
3884 (prepare_initializers_chain): Hanlde store-store chain.
3885 (prepare_finalizers_chain, prepare_finalizers): New function.
3886 (tree_predictive_commoning_loop): Return integer value indicating
3887 if loop is unrolled or lcssa form is corrupted.
3888 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
3890 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3892 * tree-predcom.c (initialize_root): Delete.
3893 (execute_pred_commoning_chain): Initialize root vars and replace
3894 reference of non-combined chain directly, rather than call above
3897 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3899 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
3900 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
3902 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3904 * tree-predcom.c (struct chain): New field init_seq.
3905 (release_chain): Release init_seq.
3906 (prepare_initializers_chain): Record intialization stmts in above
3908 (insert_init_seqs): New function.
3909 (tree_predictive_commoning_loop): Call insert_init_seqs.
3911 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3913 * tree-predcom.c (determine_roots_comp): Skip trivial components.
3915 2017-07-28 Richard Biener <rguenther@suse.de>
3917 * match.pd: Remove superfluous :c.
3918 * genmatch.c (simplify::id): Add member.
3919 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
3921 (current_id): New global.
3922 (dt_node::parent): Move from ...
3923 (dt_operand::parent): ... here. Add for_id member.
3924 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
3925 (decision_tree::find_node): Relax order requirement when
3926 merging DT_TRUE nodes to ones inbetween the current simplify
3927 and the one we try to merge with. Add diagnostic whenever
3928 we need to enforce pattern order by not merging.
3929 (decision_tree::insert): Set current_id.
3930 (decision_tree::print_node): Dump parent node and for_id.
3931 (parser::last_id): Add member.
3932 (parser::push_simplify): Assign unique id.
3933 (parser::parser): Initialize last_id.
3935 2017-07-28 Martin Liska <mliska@suse.cz>
3938 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
3939 gimple_build_debug_bind.
3941 2017-07-28 Richard Biener <rguenther@suse.de>
3943 PR tree-optimization/81502
3944 * match.pd: Add pattern combining BIT_INSERT_EXPR with
3946 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
3948 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
3949 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
3950 for BIT_FIELD_REF args.
3951 * fold-const.c (make_bit_field_ref): Likewise.
3952 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3954 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3957 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
3958 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
3959 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
3960 Or it into SANITIZER_UNDEFINED.
3961 * ubsan.c: Include gimple-fold.h and varasm.h.
3962 (ubsan_expand_ptr_ifn): New function.
3963 (instrument_pointer_overflow): New function.
3964 (maybe_instrument_pointer_overflow): New function.
3965 (instrument_object_size): Formatting fix.
3966 (pass_ubsan::execute): Call instrument_pointer_overflow
3967 and maybe_instrument_pointer_overflow.
3968 * internal-fn.c (expand_UBSAN_PTR): New function.
3969 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
3970 * sanitizer.def (__ubsan_handle_pointer_overflow,
3971 __ubsan_handle_pointer_overflow_abort): New builtins.
3972 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
3973 * internal-fn.def (UBSAN_PTR): New internal function.
3974 * opts.c (sanitizer_opts): Add pointer-overflow.
3975 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
3976 * fold-const.c (build_range_check): Compute pointer range check in
3977 integral type if pointer arithmetics would be needed. Formatting
3980 2017-07-28 Martin Liska <mliska@suse.cz>
3983 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
3984 parameters that are of a variable-length.
3986 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
3988 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
3990 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
3991 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
3992 (CRT_CALL_STATIC_FUNCTION): Likewise.
3993 (ASM_DEFAULT_SPEC): New define.
3994 (ASM_SPEC32): Likewise.
3995 (ASM_SPEC64): Likewise.
3996 (ASM_SPEC_COMMON): Likewise.
3997 (ASM_SPEC): Likewise.
3998 (INVALID_64BIT): Likewise.
3999 (LINK_OS_DEFAULT_SPEC): Likewise.
4000 (LINK_OS_SPEC32): Likewise.
4001 (LINK_OS_SPEC64): Likewise.
4002 (POWERPC_LINUX): Likewise.
4003 (PTRDIFF_TYPE): Likewise.
4004 (RESTORE_FP_PREFIX): Likewise.
4005 (RESTORE_FP_SUFFIX): Likewise.
4006 (SAVE_FP_PREFIX): Likewise.
4007 (SAVE_FP_SUFFIX): Likewise.
4008 (SIZE_TYPE): Likewise.
4009 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4010 (TARGET_64BIT): Likewise.
4011 (TARGET_64BIT): Likewise.
4012 (TARGET_AIX): Likewise.
4013 (WCHAR_TYPE_SIZE): Likewise.
4014 (WCHAR_TYPE): Undefine.
4015 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4016 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4017 (CPP_OS_RTEMS_SPEC): Delete.
4018 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4019 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4021 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4023 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4025 PR tree-optimization/81578
4026 * tree-parloops.c (build_new_reduction): Bail out if
4027 reduction_code isn't one of the standard OpenMP reductions.
4028 Move the details printing after that decision.
4030 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4032 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4033 related to reload_in_progress.
4034 (splat_input_operand): Likewise.
4035 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4037 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4039 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4040 (TARGET_INSTANTIATE_DECLS): Likewise.
4041 (legitimate_indexed_address_p): Delete reload_in_progress code.
4042 (rs6000_debug_legitimate_address_p): Likewise.
4043 (rs6000_eliminate_indexed_memrefs): Likewise.
4044 (rs6000_emit_le_vsx_store): Likewise.
4045 (rs6000_emit_move_si_sf_subreg): Likewise.
4046 (rs6000_emit_move): Likewise.
4047 (register_to_reg_type): Likewise.
4048 (rs6000_pre_atomic_barrier): Likewise.
4049 (rs6000_machopic_legitimize_pic_address): Likewise.
4050 (rs6000_allocate_stack_temp): Likewise.
4051 (rs6000_address_for_fpconvert): Likewise.
4052 (rs6000_address_for_altivec): Likewise.
4053 (rs6000_secondary_memory_needed_rtx): Delete function.
4054 (rs6000_check_sdmode): Likewise.
4055 (rs6000_alloc_sdmode_stack_slot): Likewise.
4056 (rs6000_instantiate_decls): Likewise.
4057 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4058 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4059 Delete reload_in_progress.
4060 (*vec_reload_and_plus_<mptrsize>): Likewise.
4061 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4062 (vsx_div_v2di): Likewise.
4063 (vsx_udiv_v2di): Likewise.
4065 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4067 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4068 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4069 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4070 (rs6000_debug_reg_global): Delete print of LRA status.
4071 (rs6000_option_override_internal): Delete dead LRA related code.
4072 (rs6000_lra_p): Delete function.
4073 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4075 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4077 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4078 * config/riscv/rtems.h: New file.
4080 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4081 Sudakshina Das <sudi.das@arm.com>
4083 * config/aarch64/aarch64.md
4084 (define_split for and<mode>3nr_compare): Move
4085 non aarch64_logical_operand to a register.
4086 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4087 register immediate operand to a register.
4088 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4090 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4093 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4095 2017-07-27 Richard Biener <rguenther@suse.de>
4097 PR tree-optimization/81573
4098 PR tree-optimization/81494
4099 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4100 multi defuse cycle case.
4102 2017-07-27 Richard Biener <rguenther@suse.de>
4104 PR tree-optimization/81571
4105 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4108 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4110 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4111 earlier and only if MASK_FPU is set. Adjust formatting.
4113 2017-07-27 Martin Liska <mliska@suse.cz>
4115 * opt-functions.awk: Add validation of value of Init.
4116 * optc-gen.awk: Pass new argument.
4118 2017-07-27 Martin Liska <mliska@suse.cz>
4120 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4121 Fix wrong condition.
4123 2017-07-27 Martin Liska <mliska@suse.cz>
4125 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4126 BBs and edges seen by autoFDO.
4128 2017-07-27 Richard Biener <rguenther@suse.de>
4130 PR tree-optimization/81502
4131 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4132 with incompatible but same sized type.
4133 (execute_update_addresses_taken): Likewise.
4135 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4137 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4138 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4140 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4143 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4144 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4145 Change s_operand to memory_operand.
4147 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4149 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4150 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4151 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4152 Emit instructions rather than returning an expression. Handle TFmode
4153 and KFmode by casting to TImode.
4154 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4155 (rs6000_emit_le_vsx_store): Likewise.
4156 * config/rs6000/vsx.md (VSX_TI): New iterator.
4157 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4158 (*vsx_le_undo_permute_<mode>): Likewise.
4159 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4160 emit the split sequence.
4161 (*vsx_le_perm_store_<mode>): Likewise.
4163 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4165 PR tree-optimization/81555
4166 PR tree-optimization/81556
4167 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4168 if true, force CHANGED for the recursive invocation.
4169 (reassociate_bb): Remember original length of ops array, pass
4170 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4172 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4173 attributes for noipa attribute. For naked attribute use
4174 lookup_attribute first before lookup_attribute_spec.
4175 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4177 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4178 for functions with noipa attribute.
4179 (cgraph_externally_visible_p): Return true for functions with noipa
4181 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4182 for functions with noipa attribute.
4183 * doc/extend.texi: Document noipa function attribute.
4184 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4185 also for functions with noipa attribute.
4186 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4188 2017-07-26 Andrew Pinski <apinski@cavium.com>
4190 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4191 vec_unalign_load_cost and vec_unalign_store_cost.
4193 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4195 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4196 -mvsx-small-integer option.
4197 (ISA_3_0_MASKS_IEEE): Likewise.
4198 (OTHER_VSX_VECTOR_MASKS): Likewise.
4199 (POWERPC_MASKS): Likewise.
4200 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4201 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4202 code, only testing for DImode being allowed in non-VSX floating
4204 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4205 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4207 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4208 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4209 TARGET_P8_VECTOR test.
4210 (rs6000_secondary_reload_simple_move): Likewise.
4211 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4212 since TARGET_P9_VECTOR was already tested.
4213 (rs6000_opt_masks): Remove -mvsx-small-integer.
4214 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4215 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4217 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4218 test for TARGET_VEXTRACTUB was used, and that uses
4220 (p9 extract splitter): Likewise.
4221 (vsx_extract_<mode>_di_p9): Likewise.
4222 (vsx_extract_<mode>_store_p9): Likewise.
4223 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4224 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4225 the elimination of TARGET_VSX_SMALL_INTEGER.
4226 (vsx_extract_<mode>_p8): Likewise.
4227 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4228 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4229 (vsx_set_<mode>_p9): Likewise.
4230 (vsx_set_v4sf_p9): Likewise.
4231 (vsx_set_v4sf_p9_zero): Likewise.
4232 (vsx_insert_extract_v4sf_p9): Likewise.
4233 (vsx_insert_extract_v4sf_p9_2): Likewise.
4234 * config/rs6000/rs6000.md (sign extend splitter): Change
4235 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4236 (floatsi<mode>2_lfiwax_mem): Likewise.
4237 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4238 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4239 since a test for TARGET_P9_VECTOR was used.
4240 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4241 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4242 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4243 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4244 TARGET_P8_VECTOR test.
4245 (fix_trunc<mode>si2_stfiwx): Likewise.
4246 (fix_trunc<mode>si2_internal): Likewise.
4247 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4248 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4250 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4251 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4252 TARGET_P8_VECTOR test.
4253 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4254 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4255 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4257 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4258 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4259 since a test for TARGET_P9_VECTOR was used.
4260 (splitter for loading small constants): Likewise.
4262 2017-07-26 Andrew Pinski <apinski@cavium.com>
4264 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4267 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4270 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4271 (fp_valid_at): Likewise.
4273 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4275 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4276 (qdf24xx_addrcost_table): Likewise.
4277 (cortexa57_tunings): Update to use generic_branch_cost.
4278 (cortexa72_tunings): Likewise.
4279 (cortexa73_tunings): Likewise.
4280 (qdf24xx_tunings): Likewise.
4282 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4284 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4285 (thunderx2t99_branch_cost): Likewise.
4286 (cortexa35_tunings): Update to use generic_branch_cost.
4287 (cortexa53_tunings): Likewise.
4288 (cortexa57_tunings): Likewise.
4289 (cortexa72_tunings): Likewise.
4290 (cortexa73_tunings): Likewise.
4291 (thunderx2t99_tunings): Likewise.
4293 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4295 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4296 (sparc_option_override): Honour MASK_FSMULD.
4297 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4298 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4299 * config/sparc/sparc.opt (mfsmuld): New option.
4300 * doc/invoke.texi (mfsmuld): Document option.
4302 2017-07-26 Marek Polacek <polacek@redhat.com>
4305 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4307 2017-07-26 Richard Biener <rguenther@suse.de>
4309 * gimple-match-head.c (do_valueize): Return OP if valueize
4311 (get_def): New helper to get at the def stmt of a SSA name
4313 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4314 do_valueize to get at the def stmt.
4315 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4317 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4320 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4322 2017-07-26 Martin Liska <mliska@suse.cz>
4325 * function.c (expand_function_start): Make expansion of
4326 nonlocal_goto_save_area after parm_birth_insn.
4328 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4330 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4331 from all CPU target flags enable members.
4333 2017-07-26 Richard Biener <rguenther@suse.de>
4335 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4336 (decision_tree::gen): Make 'type' const.
4337 (write_predicate): Likewise.
4339 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4341 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4342 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4343 (rs6000_option_override_internal): Likewise.
4344 (rs6000_expand_vector_set): Likewise.
4345 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4346 (TARGET_UPPER_REGS_SF): Likewise.
4347 (TARGET_UPPER_REGS_DI): Likewise.
4348 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4349 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4350 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4351 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4352 (Splitters for DI constants in Altivec registers): Likewise.
4353 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4354 (vsx_set_v4sf_p9): Likewise.
4355 (vsx_set_v4sf_p9_zero): Likewise.
4356 (vsx_insert_extract_v4sf_p9): Likewise.
4357 (vsx_insert_extract_v4sf_p9_2): Likewise.
4359 2017-07-25 Carl Love <cel@us.ibm.com>
4361 * doc/extend.texi: Update the built-in documentation file for the
4362 existing built-in functions
4363 vector signed char vec_cnttz (vector signed char);
4364 vector unsigned char vec_cnttz (vector unsigned char);
4365 vector signed short vec_cnttz (vector signed short);
4366 vector unsigned short vec_cnttz (vector unsigned short);
4367 vector signed int vec_cnttz (vector signed int);
4368 vector unsigned int vec_cnttz (vector unsigned int);
4369 vector signed long long vec_cnttz (vector signed long long);
4370 vector unsigned long long vec_cnttz (vector unsigned long long);
4372 2017-07-25 Andrew Pinski <apinski@cavium.com>
4374 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4375 accesses where the use is for the first operand of a BIT_INSERT.
4377 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4380 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4381 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4383 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4385 * config/i386/gstabs.h: Delete.
4386 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4388 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4390 * config/i386/i386.c (ix86_decompose_address): Do not check for
4391 register RTX when looking at index_reg or base_reg.
4392 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4394 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4396 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4397 to update EH info here.
4399 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4401 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4403 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4405 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4407 2017-07-25 Torsten Duwe <duwe@suse.de>
4409 * common.opt: Introduce -fpatchable-function-entry
4410 command line option, and its variables function_entry_patch_area_size
4411 and function_entry_patch_area_start.
4412 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4413 including a two-value parser.
4414 * target.def (print_patchable_function_entry): New target hook.
4415 * targhooks.h (default_print_patchable_function_entry): New function.
4416 * targhooks.c (default_print_patchable_function_entry): Likewise.
4417 * toplev.c (process_options): Switch off IPA-RA if
4418 patchable function entries are being generated.
4419 * varasm.c (assemble_start_function): Look at the
4420 patchable-function-entry command line switch and current
4421 function attributes and maybe generate NOP instructions by
4422 calling the print_patchable_function_entry hook.
4423 * doc/extend.texi: Document patchable_function_entry attribute.
4424 * doc/invoke.texi: Document -fpatchable_function_entry
4425 command line option.
4426 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4428 * doc/tm.texi: Re-generate.
4430 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4433 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4434 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4436 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4438 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4440 2017-07-25 Richard Biener <rguenther@suse.de>
4442 PR tree-optimization/81455
4443 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4444 not walk in cycles when looking for guards.
4446 2017-07-25 Richard Biener <rguenther@suse.de>
4448 PR tree-optimization/81529
4449 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4450 when optimizing backedge uses.
4452 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4454 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4456 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4457 to dl_section_ref. On AIX, append an expression to subtract
4458 the size of the section length to dl_section_ref.
4460 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4462 * configure.ac: If any of the config.* scripts fail, exit 1.
4463 * configure: Regenerate.
4465 2017-07-25 Richard Biener <rguenther@suse.de>
4468 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4469 of immediate uses, be more verbose on errors.
4471 2017-07-25 Richard Biener <rguenther@suse.de>
4473 PR tree-optimization/81510
4474 * tree-vect-loop.c (vect_is_simple_reduction): When the
4475 reduction stmt is not inside the loop bail out.
4477 2017-07-25 Richard Biener <rguenther@suse.de>
4479 PR tree-optimization/81303
4480 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4481 profitability check against LOOP_VINFO_NITERSM1.
4483 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4485 * domwalk.c (cmp_bb_postorder): Simplify.
4486 (sort_bbs_postorder): New function. Use it...
4487 (dom_walker::walk): ...here to optimize common cases.
4489 2017-07-25 Martin Liska <mliska@suse.cz>
4492 * ipa-visibility.c (function_and_variable_visibility): Make the
4493 redirection just on target that supports aliasing.
4494 Fix GNU coding style.
4496 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4499 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4501 * config/arm/rtems.h: Likewise.
4502 * config/bfin/rtems.h: Likewise.
4503 * config/i386/rtemself.h: Likewise.
4504 * config/lm32/rtems.h: Likewise.
4505 * config/m32c/rtems.h: Likewise.
4506 * config/m68k/rtemself.h: Likewise.
4507 * config/microblaze/rtems.h: Likewise.
4508 * config/mips/rtems.h: Likewise.
4509 * config/moxie/rtems.h: Likewise.
4510 * config/nios2/rtems.h: Likewise.
4511 * config/powerpcspe/rtems.h: Likewise.
4512 * config/rs6000/rtems.h: Likewise.
4513 * config/rtems.h: Likewise.
4514 * config/sh/rtems.h: Likewise.
4515 * config/sh/rtemself.h: Likewise.
4516 * config/sparc/rtemself.h: Likewise.
4518 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4521 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4522 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4523 * tree-ssa-structalias.c (alias_get_name): Same.
4525 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4528 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4529 instructions if no du chain is found.
4531 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4533 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4535 2017-07-25 Richard Biener <rguenther@suse.de>
4538 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4541 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4543 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4545 (ISA_2_7_MASKS_SERVER): Likewise.
4546 (ISA_3_0_MASKS_IEEE): Likewise.
4547 (OTHER_P8_VECTOR_MASKS): Likewise.
4548 (OTHER_VSX_VECTOR_MASKS): Likewise.
4549 (POWERPC_MASKS): Likewise.
4550 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4551 duplicate list of options.
4552 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4553 explicit -mupper-regs options.
4554 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4555 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4556 alias for -mupper-regs-df.
4557 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4558 (rs6000_init_hard_regno_mode_ok): Likewise.
4559 (rs6000_option_override_internal): Likewise.
4560 (rs6000_opt_masks): Likewise.
4561 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4562 options in terms of whether -mvsx or -mpower8-vector was used.
4563 (TARGET_UPPER_REGS_DI): Likewise.
4564 (TARGET_UPPER_REGS_SF): Likewise.
4565 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4566 -mupper-regs-* options.
4568 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4570 * passes.c (emergency_dump_function): Print some empty lines and a
4571 header before the RTL dump.
4573 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4575 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4577 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4580 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4581 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4583 2017-07-24 Carl Love <cel@us.ibm.com>
4585 * config/rs6000/rs6000-c.c: Add support for built-in functions
4586 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4587 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4588 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4589 vec_extract_fp_from_shortl): Add defines for the two builtins.
4590 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4591 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4593 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4594 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4595 * doc/extend.texi: Update the built-in documentation file for the
4596 new built-in function.
4598 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4601 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4603 * doc/generic.texi: Likewise.
4604 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4605 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4607 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4609 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4610 (aarch64_mls_elt_merge<mode>): Likewise.
4612 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4614 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4615 having __cxa_atexit.
4617 2017-07-23 Michael Collison <michael.collison@arm.com>
4619 * config/arm/arm.c (arm_option_override): Deprecate
4620 use of -mstructure-size-boundary.
4621 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4622 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4624 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4627 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4628 Reduce cost estimate for direct moves.
4630 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4633 * config/i386/i386.c (ix86_option_override_internal): Disable
4634 BMI, BMI2 and TBM instructions for -m16.
4636 2017-07-21 Carl Love <cel@us.ibm.com>
4638 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4639 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4640 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4641 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4642 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4643 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4644 VMULOSW): New enum "unspec" values.
4645 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4646 altivec_vmulosw): New patterns.
4647 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4648 VMULOSW): Add definitions.
4650 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4652 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4653 (qdf24xx): Likewise.
4654 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4655 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4656 (AARCH64_FL_V8_1): Renumber.
4657 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4658 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4659 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4660 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4661 rdma to feature modifiers list.
4663 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4666 * ipa-visibility (function_and_variable_visibility): Convert
4667 recursive PLT call to direct call if appropriate.
4669 2017-07-21 Andrew Pinski <apinski@cavium.com>
4671 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4672 operand 1 to see if the types precision matches.
4673 * fold-const.c (operand_equal_p): Likewise.
4675 2017-07-21 Richard Biener <rguenther@suse.de>
4677 PR tree-optimization/81303
4678 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4679 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4680 (vect_peeling_hash_get_lowest_cost): Adjust.
4681 (vect_enhance_data_refs_alignment): Likewise. Use
4682 vect_get_peeling_costs_all_drs to compute the penalty for no
4683 peeling to match up costs.
4685 2017-07-21 Richard Biener <rguenther@suse.de>
4687 PR tree-optimization/81500
4688 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4689 we didn't identify a reduction path.
4691 2017-07-21 Tom de Vries <tom@codesourcery.com>
4692 Cesar Philippidis <cesar@codesourcery.com>
4694 PR gcov-profile/81442
4695 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
4698 2017-07-21 Tom de Vries <tom@codesourcery.com>
4701 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
4703 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
4704 nvptx_override_options_after_change.
4706 2017-07-21 Ulrich Drepper <drepper@redhat.com>
4708 * dwarf2out.c (output_file_names): Avoid double testing for
4711 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
4713 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
4715 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4717 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
4719 (try_crossjump_to_edge): Do not punt on partitioned functions.
4721 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4723 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4724 Put all BBs reachable only via paths crossing cold region to cold
4726 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
4728 2016-07-21 Richard Biener <rguenther@suse.de>
4730 PR tree-optimization/81303
4731 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
4732 into account prologue and epilogue iterations when raising
4733 min_profitable_iters to sth at least covering one vector iteration.
4735 2017-07-21 Tamar Christina <tamar.christina@arm.com>
4737 * config/arm/arm.c (arm_test_cpu_arch_dat):
4740 2017-07-20 Nathan Sidwell <nathan@acm.org>
4742 Remove TYPE_METHODS.
4743 * tree.h (TYPE_METHODS): Delete.
4744 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
4745 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
4746 (dbxout_type_methods): Scan TYPE_FIELDS.
4747 (dbxout_type): Don't check TYPE_METHODS here.
4748 * function.c (use_register_for_decl): Always ignore register for
4749 class types when not optimizing.
4750 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
4751 * tree.c (free_lang_data_in_type): Stitch out member functions and
4752 templates from TYPE_FIELDS.
4753 (build_distinct_type_copy, verify_type_variant,
4754 verify_type): Member fns are on TYPE_FIELDS.
4755 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
4756 * tree-pretty-print.c (dump_generic_node): Likewise.
4758 2017-07-20 Jakub Jelinek <jakub@redhat.com>
4761 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
4762 V2TImode and V4TImode.
4763 (ix86_expand_vector_extract): Likewise.
4764 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
4765 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
4766 (ssescalarmode): Handle V4TImode and V2TImode.
4767 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
4768 (*vec_extractv2ti, *vec_extractv4ti): New insns.
4769 (VEXTRACTI128_MODE): New mode iterator.
4770 (splitter for *vec_extractv?ti first element): New.
4771 (VEC_INIT_MODE): New mode iterator.
4772 (vec_init<mode>): Consolidate 3 expanders into one using
4773 VEC_INIT_MODE mode iterator.
4775 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4777 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
4778 non_spilled_static_chain_regno_p.
4780 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4782 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
4784 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
4786 * bb-reorder.c (connect_traces): Allow copying of blocks within
4789 2017-07-20 Richard Biener <rguenther@suse.de>
4791 * gimple.h (gimple_phi_result): Add gphi * overload.
4792 (gimple_phi_result_ptr): Likewise.
4793 (gimple_phi_arg): Likewise. Adjust index assert to only
4794 allow actual argument accesses rather than all slots available
4796 (gimple_phi_arg_def): Add gphi * overload.
4797 * tree-phinodes.c (make_phi_node): Initialize only actual
4799 (resize_phi_node): Clear memory not covered by old node,
4800 do not initialize excess argument slots.
4801 (reserve_phi_args_for_new_edge): Initialize new argument slot
4804 2017-07-20 Bin Cheng <bin.cheng@arm.com>
4806 PR tree-optimization/81388
4808 2016-07-21 Bin Cheng <bin.cheng@arm.com>
4810 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
4811 by removing computation of may_be_zero.
4813 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4814 Tom de Vries <tom@codesourcery.com>
4817 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
4818 when gimple level profile disagrees with what RTL expander did.
4820 2017-07-20 Richard Biener <rguenther@suse.de>
4822 PR tree-optimization/61171
4823 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
4824 (vect_analyze_stmt): Add slp instance parameter.
4825 (vectorizable_reduction): Likewise.
4826 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
4827 (vect_is_simple_reduction): Deal with chains not detected
4828 as SLP reduction chain, specifically not properly associated
4829 chains containing a mix of plus/minus.
4830 (get_reduction_op): Remove.
4831 (get_initial_defs_for_reduction): Simplify, pass in whether
4832 this is a reduction chain, pass in the SLP node for the PHIs.
4833 (vect_create_epilog_for_reduction): Get the SLP instance as
4835 (vectorizable_reduction): Get the SLP instance as arg.
4836 During analysis remember the SLP node with the PHIs in the
4837 instance. Simplify getting at the vectorized reduction PHIs.
4838 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
4839 through SLP instance.
4840 (vect_slp_analyze_operations): Likewise.
4841 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
4842 (vect_transform_stmt): Likewise.
4844 2017-07-20 Tom de Vries <tom@codesourcery.com>
4846 PR tree-optimization/81489
4847 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
4848 read of phi arg location to before loop that modifies phi.
4850 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4852 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
4855 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4858 * except.c (execute): Fix ordering issue.
4860 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4862 PR rtl-optimization/81423
4863 * combine.c (make_compound_operation_int): Don't try to optimize
4864 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
4866 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4868 PR rtl-optimization/81423
4869 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
4870 with a constant that is -1 in the truncated to mode.
4872 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4874 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
4875 (determine_unlikely_bbs): ... here.
4876 * predict.h (propagate_unlikely_bbs_forward): Declare.
4877 * cfgexpand.c (pass_expand::execute): Use it.
4878 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
4880 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
4881 propagate_unlikely_bbs_forward.
4883 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4886 * except.c (maybe_add_nop_after_section_switch): New function.
4889 2017-07-19 Tom de Vries <tom@codesourcery.com>
4891 * gimple.h (gimple_phi_set_arg): Make assert more strict.
4893 2017-07-19 Tom de Vries <tom@codesourcery.com>
4895 * gimple.h (gimple_phi_arg): Make assert more strict.
4897 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
4899 * config.gcc (powerpc*-*-*): Add mmintrin.h.
4900 * config/rs6000/mmintrin.h: New file.
4901 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
4903 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4905 PR tree-optimization/81346
4906 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
4908 2017-07-19 Tom de Vries <tom@codesourcery.com>
4910 * config/nvptx/nvptx.md (VECIM): Add V2DI.
4912 2017-07-19 Tom de Vries <tom@codesourcery.com>
4914 * config/nvptx/nvptx-modes.def: Add V2DImode.
4915 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
4916 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
4917 (nvptx_output_mov_insn): Handle lack of mov.b128.
4918 (nvptx_print_operand): Handle 'H' and 'L' codes.
4919 (nvptx_vector_mode_supported): Allow V2DImode.
4920 (nvptx_preferred_simd_mode): New function.
4921 (nvptx_data_alignment): New function.
4922 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
4923 nvptx_preferred_simd_mode.
4924 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
4926 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
4928 2017-07-19 Tom de Vries <tom@codesourcery.com>
4930 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
4931 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
4932 (nvptx_vector_mode_supported): New function. Allow V2SImode.
4933 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
4934 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
4935 (mov<VECIM>_insn): New define_insn.
4936 (define_expand "mov<VECIM>): New define_expand.
4938 2017-07-19 Tom de Vries <tom@codesourcery.com>
4940 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
4942 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4944 PR tree-optimization/81346
4945 * fold-const.h (fold_div_compare, range_check_type): Declare.
4946 * fold-const.c (range_check_type): New function.
4947 (build_range_check): Use range_check_type.
4948 (fold_div_compare): No longer static, rewritten into
4949 a match.pd helper function.
4950 (fold_comparison): Don't call fold_div_compare here.
4951 * match.pd (X / C1 op C2): New optimization using fold_div_compare
4954 2017-07-19 Nathan Sidwell <nathan@acm.org>
4956 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
4957 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
4958 * tree.c (find_decls_types_r, verify_type): Use
4959 TYPE_{MIN,MAX}_VALUE_RAW.
4960 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
4961 (hash_tree): Likewise.
4962 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4964 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4967 2017-07-18 Tom de Vries <tom@codesourcery.com>
4970 * omp-expand.c (expand_omp_for_static_chunk): Handle
4971 equal-argument loop exit phi.
4973 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
4976 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
4977 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
4978 operand 2 predicate.
4979 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
4980 operand 2 predicate.
4981 (ror,rol -> rorx splitters): Use const_int_operand as
4982 operand 2 predicate.
4984 2017-06-18 Richard Biener <rguenther@suse.de>
4986 PR tree-optimization/81410
4987 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
4988 the gap in the ! slp_perm SLP case after each group.
4990 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4993 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
4996 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4999 * predict.c (set_even_probabilities): Cleanup; do not affect
5000 probabilities that are already known.
5001 (combine_predictions_for_bb): Call even when count is set.
5003 2017-07-18 Nathan Sidwell <nathan@acm.org>
5005 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5008 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5011 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5012 optimization for loop niter analysis.
5014 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5017 * config/avr/avr.c (avr_optimize_casesi): Don't use
5018 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5020 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5022 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5023 body_cost_vec from _vect_peel_extended_info.
5024 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5025 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5028 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5030 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5032 2017-07-18 Richard Biener <rguenther@suse.de>
5034 PR tree-optimization/80620
5035 PR tree-optimization/81403
5036 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5037 info when re-using a VN table entry.
5039 2017-07-18 Richard Biener <rguenther@suse.de>
5041 PR tree-optimization/81418
5042 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5043 vectype_in. Verify that with lane-reducing reduction operations
5044 we have a single def-use cycle.
5046 2017-07-17 Carl Love <cel@us.ibm.com>
5048 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5051 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5052 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5053 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5054 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5055 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5056 VMULOSW): New enum "unspec" values.
5057 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5058 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5059 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5060 altivec_vmulosw): New patterns.
5061 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5062 VMULOSW): Add definitions.
5063 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5065 * config/alpha/alpha.c: Include predict.h.
5067 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5069 * tree-vrp.c (compare_assert_loc): Fix comparison function
5070 to return predictable results.
5072 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5074 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5077 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5078 * doc/invoke.texi (mexpand-adddi): Update text.
5080 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5082 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5083 that also clobbers the CC register. The old expand code is moved
5085 (*arc_clzsi2): ... here.
5086 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5087 the CC register. The old expand code is moved to ...
5088 (arc_ctzsi2): ... here.
5090 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5092 * config/arc/arc.opt (mindexed-loads): Use initial value
5093 TARGET_INDEXED_LOADS_DEFAULT.
5094 (mauto-modify-reg): Use initial value
5095 TARGET_AUTO_MODIFY_REG_DEFAULT.
5096 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5097 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5098 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5099 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5101 2017-07-17 Martin Liska <mliska@suse.cz>
5104 * opts.c (finish_options): Do not allow -fgnu-tm
5105 w/ -fsanitize={kernel-,}address. Say sorry.
5107 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5110 * tree-loop-distribution.c (classify_partition): Only assert on
5111 numer of iterations.
5112 (merge_dep_scc_partitions): Delete prameter. Update function call.
5113 (distribute_loop): Remove code handling loop with unknown niters.
5114 (pass_loop_distribution::execute): Skip loop with unknown niters.
5116 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5119 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5120 function sort_partitions_by_post_order.
5122 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5124 PR tree-optimization/81374
5125 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5126 the max index of basic blocks, rather than number of basic blocks.
5128 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5130 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5132 (arc_legitimate_pic_operand_p): Likewise.
5133 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5135 (arc_needs_pcl_p): Likewise.
5136 (arc_legitimate_pc_offset_p): Likewise.
5137 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5138 function is also used in constrains.md.
5139 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5140 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5141 PLUS. Only return true/false in known cases, otherwise assert.
5142 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5143 is already called in arc_legitimate_constant_p.
5144 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5146 (LEGITIMATE_PIC_OPERAND_P): Use
5147 arc_raw_symbolic_reference_mentioned_p function.
5148 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5153 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5154 Andrew Burgess <andrew.burgess@embecosm.com>
5156 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5157 (arc_return_address_register): New function.
5158 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5159 (arc_handle_fndecl_attribute): Add naked attribute.
5160 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5161 (TARGET_WARN_FUNC_RETURN): Likewise.
5162 (arc_allocate_stack_slots_for_args): New function.
5163 (arc_warn_func_return): Likewise.
5164 (machine_function): Change type fn_type.
5165 (arc_compute_function_type): Consider new naked function type,
5166 change function return type.
5167 (arc_must_save_register): Adapt to handle new
5168 arc_compute_function_type's return type.
5169 (arc_expand_prologue): Likewise.
5170 (arc_expand_epilogue): Likewise.
5171 (arc_return_address_regs): Delete.
5172 (arc_return_address_register): New function.
5173 (arc_epilogue_uses): Use above function.
5174 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5175 (arc_function_type): Change encoding, add naked type.
5176 (ARC_INTERRUPT_P): Change to handle the new encoding.
5177 (ARC_FAST_INTERRUPT_P): Likewise.
5178 (ARC_NORMAL_P): Define.
5179 (ARC_NAKED_P): Likewise.
5180 (arc_compute_function_type): Delete prototype.
5181 * config/arc/arc.md (in_ret_delay_slot): Use
5182 arc_return_address_register function.
5183 (simple_return): Likewise.
5184 (p_return_i): Likewise.
5186 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5188 PR tree-optimization/81428
5189 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5190 can't be built for those types.
5192 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5194 Remove stuff dead since r239246.
5196 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5197 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5198 (avr_inform_devices): Remove dead stuff.
5200 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5202 * config/arm/arm_neon.h: Fix softp typo.
5204 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5206 PR tree-optimization/81365
5207 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5208 aggregate moves onto bb predecessor edges, make sure there are no
5209 loads that could alias the lhs in between the start of bb and the
5212 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5215 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5216 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5217 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5219 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5221 PR tree-optimization/81396
5222 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5223 (init_symbolic_number): Initialize it to 1.
5224 (perform_symbolic_merge): Add n_ops from both operands into the new
5226 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5227 without base_addr as useless if they need more than one operation.
5228 (bswap_replace): Handle !bswap case for NULL base_addr.
5230 2017-07-17 Tom de Vries <tom@codesourcery.com>
5233 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5236 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5238 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5239 conditional builtin define __FIX_LEON3FT_B2BST.
5241 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5243 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5244 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5247 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5249 PR rtl-optimization/81424
5250 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5251 to remove potential trapping from operands if -fnon-call-exceptions.
5253 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5255 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5256 profile_proability for scalling.
5257 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5259 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5261 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5263 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5265 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5266 fixpoint arithmetics.
5268 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5270 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5271 fixpoint arithmetics.
5273 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5275 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5276 fixpoint arithmetics.
5278 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5280 * profile-count.h (profile_probability::from_reg_br_prob_note,
5281 profile_probability::to_reg_br_prob_note): New functions.
5282 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5283 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5284 * predict.c (probability_reliable_p): Update.
5285 (edge_probability_reliable_p): Update.
5286 (br_prob_note_reliable_p): Update.
5287 (invert_br_probabilities): Update.
5288 (add_reg_br_prob_note): New function.
5289 (combine_predictions_for_insn): Update.
5290 * asan.c (asan_clear_shadow): Update.
5291 * cfgbuild.c (compute_outgoing_frequencies): Update.
5292 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5293 (update_br_prob_note): Update.
5294 (rtl_verify_edges): Update.
5295 (purge_dead_edges): Update.
5296 (fixup_reorder_chain): Update.
5297 * emit-rtl.c (try_split): Update.
5298 * ifcvt.c (cond_exec_process_insns): Update.
5299 (cond_exec_process_if_block): Update.
5300 (dead_or_predicable): Update.
5301 * internal-fn.c (expand_addsub_overflow): Update.
5302 (expand_neg_overflow): Update.
5303 (expand_mul_overflow): Update.
5304 * loop-doloop.c (doloop_modify): Update.
5305 * loop-unroll.c (compare_and_jump_seq): Update.
5306 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5307 * predict.h: Update.
5308 * reorg.c (mostly_true_jump): Update.
5310 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5311 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5312 * config/arc/arc.c: (emit_unlikely_jump): Update.
5313 * config/arm/arm.c: (emit_unlikely_jump): Update.
5314 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5315 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5316 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5317 (ix86_print_operand): Update.
5318 (ix86_split_fp_branch): Update.
5319 (predict_jump): Update.
5320 * config/ia64/ia64.c (ia64_print_operand): Update.
5321 * config/mmix/mmix.c (mmix_print_operand): Update.
5322 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5323 (rs6000_expand_split_stack_prologue): Update.
5324 * config/rs6000/rs6000.c: Update.
5325 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5326 (s390_expand_vec_movstr): Update.
5327 (s390_expand_cs_tdsi): Update.
5328 (s390_expand_split_stack_prologue): Update.
5329 * config/sh/sh.c (sh_print_operand): Update.
5330 (expand_cbranchsi4): Update.
5331 (expand_cbranchdi4): Update.
5332 * config/sparc/sparc.c (output_v9branch): Update.
5333 * config/spu/spu.c (get_branch_target): Update.
5334 (ea_load_store_inline): Update.
5335 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5336 * config/tilepro/tilepro.c: Update.
5338 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5340 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5341 (gimplify_save_expr): Likewise.
5343 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5345 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5347 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5349 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5351 * timevar.def (TV_IPA_FNSUMMARY): Define.
5353 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5355 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5356 to back store errata sensitive sequence from being generated.
5357 (sqrtdf2_fix): Likewise.
5359 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5361 * tree-ssa-threadupdate.c (compute_path_counts,
5362 update_joiner_offpath_counts): Use profile_probability.
5364 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5367 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5369 * config/arm/arm-c.c (arm_cpu_builtins): Define
5370 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5372 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5374 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5375 array entries to represent __ieee128 versions of the
5376 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5377 scalar_extract_sig, and scalar_insert_exp built-in functions.
5378 (altivec_resolve_overloaded_builtin): Add special case handling
5379 for the __builtin_scalar_insert_exp function, as represented by
5380 the P9V_BUILTIN_VEC_VSIEDP constant.
5381 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5382 exponent support for __ieee128 argument.
5383 (VSESQP): Add scalar extract signature support for __ieee128
5385 (VSTDCNQP): Add scalar test negative support for __ieee128
5387 (VSIEQP): Add scalar insert exponent support for __int128 argument
5388 with __ieee128 result.
5389 (VSIEQPF): Add scalar insert exponent support for __ieee128
5390 argument with __ieee128 result.
5391 (VSTDCQP): Add scalar test data class support for __ieee128
5393 (VSTDCNQP): Add overload support for scalar test negative with
5395 (VSTDCQP): Add overload support for scalar test data class
5397 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5399 (UNSPEC_VSX_SIEXPQP): New constant.
5400 (xsxexpqp): New insn for VSX scalar extract exponent quad
5402 (xsxsigqp): New insn for VSX scalar extract significand quad
5404 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5405 precision with floating point argument.
5406 (xststdcqp): New expand for VSX scalar test data class quad
5408 (xststdcnegqp): New expand for VSX scalar test negative quad
5410 (xststdcqp): New insn to match expansions for VSX scalar test data
5411 class quad precision and VSX scalar test negative quad precision.
5412 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5413 special case operand checking to enforce that second operand of
5414 VSX scalar test data class with quad precision argument is a 7-bit
5416 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5417 prototypes and descriptions of __ieee128 versions of
5418 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5419 scalar_test_data_class, and scalar_test_neg built-in functions.
5421 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5423 PR tree-optimization/81162
5424 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5425 replace a negate with an add.
5427 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5429 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5431 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5433 * config/arm/arm-c.c (arm_cpu_builtins): Define
5434 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5436 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5438 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5439 (armv8-r): Set ARM Cortex-R52 as default CPU.
5440 * config/arm/arm-tables.opt: Regenerate.
5441 * config/arm/arm-tune.md: Regenerate.
5442 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5444 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5445 extension for -mcpu=cortex-r52.
5447 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5449 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5450 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5451 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5452 (fp-armv8): Define it as FP_ARMv8 only.
5453 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5454 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5456 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5457 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5458 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5459 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5460 than TARGET_FPU_ARMV8.
5461 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5462 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5463 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5465 * config/arm/neon.md (neon_vrint): Likewise.
5466 (neon_vcvt): Likewise.
5467 (neon_<fmaxmin_op><mode>): Likewise.
5468 (<fmaxmin><mode>3): Likewise.
5469 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5470 * config/arm/predicates.md (arm_cond_move_operator): Check against
5471 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5473 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5475 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5478 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5480 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5481 loop in comment with memset.
5483 2017-07-14 Martin Liska <mliska@suse.cz>
5485 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5486 * dwarf2out.c (is_java): Remove the function.
5487 (output_pubname): Remove usage of the function.
5488 (lower_bound_default): Remove usage of DW_LANG_Java.
5489 (gen_compile_unit_die): Likewise.
5490 * gcc.c: Remove compiler defaults for .java and .zip files.
5491 * gimple-expr.c (remove_suffix): Change as there's no longer
5492 extension than 4-letter one.
5493 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5494 (gimplify_save_expr): Likewise.
5495 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5496 as it's possible even for other languages than Java.
5497 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5498 * lto-opts.c (lto_write_options): Remove reference to Java.
5499 * opts.c (strip_off_ending): Update file extension handling.
5500 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5501 * tree-eh.c (lower_resx): Likewise.
5502 * tree.c (free_lang_data_in_type): Remove dead code.
5503 (find_decls_types_r): Likewise.
5504 (build_common_builtin_nodes): Remove Java from a comment.
5505 (verify_type): Remove dead code.
5506 * varasm.c (assemble_external): Remove Java from a comment.
5508 2017-07-14 Martin Liska <mliska@suse.cz>
5510 * opts.c (finish_options): Add quotes.
5511 (common_handle_option): Likewise.
5513 2017-07-14 Martin Liska <mliska@suse.cz>
5515 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5516 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5518 * dwarf2out.c (lower_bound_default): Do not handle
5520 (gen_compile_unit_die): Likewise.
5521 * gcc.c: Remove default extension binding for GNU Pascal.
5522 * stmt.c: Remove Pascal language from a comment.
5523 * xcoffout.c: Likewise.
5525 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5528 * diagnostic-show-locus.c (fixit_cmp): New function.
5529 (layout::layout): Sort m_fixit_hints.
5530 (column_range::column_range): Assert that the values are valid.
5531 (struct char_span): New struct.
5532 (correction::overwrite): New method.
5533 (struct source_line): New struct.
5534 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5535 calls in terms of classes source_line and char_span, and
5536 correction::overwrite.
5537 (selftest::test_overlapped_fixit_printing_2): New function.
5538 (selftest::diagnostic_show_locus_c_tests): Call it.
5540 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5542 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5543 early if there is no lhs.
5545 2017-07-13 Martin Liska <mliska@suse.cz>
5547 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5548 (gen_reference_type_die): Likewise.
5549 * stor-layout.c: Remove Pascal-related comment.
5551 2017-07-13 Martin Liska <mliska@suse.cz>
5553 * opts.c (finish_options): Add quotes to error messages.
5554 (parse_sanitizer_options): Likewise.
5556 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5558 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5560 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5562 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5564 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5566 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5568 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5570 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5573 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5574 provides the hardware capability bits, define the macro
5575 __BUILTIN_CPU_SUPPORTS__.
5576 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5577 if GLIBC does not provide the hardware capability bits. Add a
5578 gcc_unreachable call if the built-in cpu function is neither
5579 __builtin_cpu_is nor __builtin_cpu_supports.
5580 (rs6000_get_function_versions_dispatcher): Change the warning
5581 that an old GLIBC is used which does not export the capability
5582 bits to be an error.
5583 * doc/extend.texi (target_clones attribute): Document the
5584 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5585 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5586 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5587 the macros defined by GCC if the newer GLIBC is available.
5589 2017-07-12 Jeff Law <law@redhat.com>
5591 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5592 remaining includes slightly.
5593 * config/riscv/riscv-builtins.c: Include profile-count.h.
5595 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5598 * config/avr/avr.c (avr_set_current_function): In diagnostic
5599 messages: Quote keywords and (parts of) identifiers.
5600 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5603 2017-07-12 Carl Love <cel@us.ibm.com>
5605 * config/rs6000/rs6000-c.c: Add support for built-in functions
5606 vector bool char vec_revb (vector bool char);
5607 vector bool short vec_revb (vector short char);
5608 vector bool int vec_revb (vector bool int);
5609 vector bool long long vec_revb (vector bool long long);
5610 * doc/extend.texi: Update the built-in documentation file for the
5611 new built-in functions.
5613 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5615 * config/s390/s390.md: Remove movcc splitter.
5617 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5619 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5620 load/store on condition.
5622 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5625 * config/avr/avr.c (avr_encode_section_info)
5626 [progmem && !TREE_READONLY]: Error if progmem object needs
5629 2017-07-11 Michael Collison <michael.collison@arm.com>
5631 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5634 2017-07-11 Carl Love <cel@us.ibm.com>
5636 * config/rs6000/rs6000-c.c: Add support for builtins
5637 vector unsigned int vec_parity_lsbb (vector signed int);
5638 vector unsigned int vec_parity_lsbb (vector unsigned int);
5639 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5640 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5641 vector unsigned long long vec_parity_lsbb (vector signed long long);
5642 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5643 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5644 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5645 * doc/extend.texi: Update the built-in documentation file for the
5646 new built-in functions.
5648 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5650 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5651 (layout::m_primary_loc): New field.
5652 (layout::layout): Initialize new field. Move location filtering
5653 logic from here to...
5654 (layout::maybe_add_location_range): ...this new method. Add
5655 support for filtering to just the lines already specified by other
5657 (layout::will_show_line_p): New method.
5658 (gcc_rich_location::add_location_if_nearby): New method.
5659 (selftest::test_add_location_if_nearby): New test function.
5660 (selftest::diagnostic_show_locus_c_tests): Call it.
5661 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5664 2017-07-11 Tom de Vries <tom@codesourcery.com>
5666 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5667 (bb_first_real_insn): New function.
5668 (nvptx_single): Add extra initialization of broadcasted condition
5671 2017-07-11 Nathan Sidwell <nathan@acm.org>
5673 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5675 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5677 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5678 Binutils doc as TEXI will mess them up.
5679 * doc/invoke.texi (AVR Options): Same here.
5681 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5683 * config/sparc/sparc.opt (mfix-ut700): New option.
5684 (mfix-gr712rc): Likewise.
5685 (sparc_fix_b2bst): New variable.
5686 * doc/invoke.texi (SPARC options): Document them.
5687 (ARM options): Fix warnings.
5688 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5689 instructions to prevent sequences that can trigger the store-store
5690 errata for certain LEON3FT processors.
5691 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
5692 (sparc_option_override): Set sparc_fix_b2bst appropriately.
5693 * config/sparc/sparc.md (fix_b2bst): New attribute.
5694 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
5696 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
5699 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
5701 (*rsqrtsf2_sse): Ditto.
5703 (div<mode>3): Macroize insn from divdf3 and divsf3
5704 using MODEF mode iterator.
5706 2017-07-10 Martin Sebor <msebor@redhat.com>
5708 PR tree-optimization/80397
5709 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
5710 instead of testing for equality to INTEGER_TYPE.
5712 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
5714 * config.gcc: Remove uclibc from arc target spec.
5716 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
5718 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
5720 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5723 * lto-wrapper.c (remove_option): New function.
5724 (merge_and_complain): Merge PIC/PIE options more realistically.
5726 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5728 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
5732 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
5733 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
5734 * config.in: Regenerate.
5735 * configure: Regenerate.
5736 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
5737 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
5738 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
5739 (TARGET_GASISR_PROLOGUES): ...target mask.
5740 * common/config/avr/avr-common.c
5741 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
5742 Set -mgas-isr-prologues.
5743 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
5744 INSERT_PASS_BEFORE for it.
5745 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
5746 * config/avr/avr.c (avr_option_override)
5747 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
5748 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
5749 (avr_attribute_table) <no_gccisr>: Add new function attribute.
5750 (avr_set_current_function) <is_no_gccisr>: Init machine field.
5751 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
5753 (make_avr_pass_pre_proep): New function.
5754 (emit_push_sfr) <treg>: Add argument to function and use it
5756 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
5757 and set machine->gasisr.yes.
5758 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
5759 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
5760 __gcc_isr.n_pushed to .L__stack_usage.
5761 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
5762 (avr_asm_final_postscan_insn): ...this new static function.
5763 * config/avr/avr.h (machine_function)
5764 <is_no_gccisr, use_L__stack_usage>: New fields.
5765 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
5766 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
5767 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
5768 (gasisr, *gasisr): New expander and insn.
5769 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5770 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
5771 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
5773 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
5775 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
5778 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5780 Move jump-tables out of .text again.
5783 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
5784 (ASM_OUTPUT_ADDR_VEC): New function.
5785 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
5786 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
5787 INSN_ADDRESSes as asm comment.
5788 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
5789 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
5790 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
5791 * config/avr/avr.md (*tablejump): Adjust comment.
5792 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
5793 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
5795 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
5796 (avr_output_addr_vec): New proto.
5797 (avr_log_t) <insn_addresses>: New field.
5799 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
5802 * config/i386/i386.c (ix86_function_arg_advance): Set
5803 outgoing_args_on_stack to true if there are outgoing arguments
5805 (ix86_function_arg): Likewise.
5806 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
5807 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
5808 * config/i386/i386.h (machine_function): Add
5809 outgoing_args_on_stack.
5811 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
5813 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5815 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
5817 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
5819 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
5820 (REAL_H): Remove $(MACHMODE_H).
5821 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
5823 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
5824 $(MACHMODE_H) and double-int.h.
5825 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
5827 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
5830 2017-07-07 Andrew Pinski <apinski@cavium.com>
5832 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
5833 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
5835 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5837 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
5838 Add warning if GCC was not configured to link against a GLIBC that
5839 exports the hardware capability bits.
5840 (make_resolver_func): Make resolver function private and not a
5841 COMDAT function. Create the name with clone_function_name instead
5842 of make_unique_name.
5845 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
5846 correct operand in doing the split.
5848 2017-07-07 Carl Love <cel@us.ibm.com>
5850 * config/rs6000/rs6000-c: Add support for built-in function
5851 vector unsigned short vec_pack_to_short_fp32 (vector float,
5853 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
5854 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
5855 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
5856 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
5857 (convert_4f32_8i16): Add define_expand.
5858 * doc/extend.texi: Update the built-in documentation file for the
5859 new built-in function.
5861 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5863 * config/sparc/m8.md: New file.
5864 * config/sparc/sparc.md: Include m8.md.
5866 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5868 * config/sparc/sparc.opt: New option -mvis4b.
5869 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
5870 (sparc_option_override): Handle VIS4B.
5871 (enum sparc_builtins): Define
5872 SPARC_BUILTIN_DICTUNPACK{8,16,32},
5873 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
5874 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
5875 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
5876 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
5877 (check_constant_argument): New function.
5878 (sparc_vis_init_builtins): Define builtins
5879 __builtin_vis_dictunpack{8,16,32},
5880 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
5881 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
5882 __builtin_vis_fpcmpde{8,16,32}shl and
5883 __builtin_vis_fpcmpur{8,16,32}shl.
5884 (sparc_expand_builtin): Check that the constant operands to
5885 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
5886 constant and in range.
5887 * config/sparc/sparc-c.c (sparc_target_macros): Handle
5889 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
5890 (SPARC_IMM5_P): Likewise.
5891 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
5892 (enabled): Handle vis4b.
5893 (UNSPEC_DICTUNPACK): New unspec.
5894 (UNSPEC_FPCMPSHL): Likewise.
5895 (UNSPEC_FPUCMPSHL): Likewise.
5896 (UNSPEC_FPCMPDESHL): Likewise.
5897 (UNSPEC_FPCMPURSHL): Likewise.
5898 (cpu_feature): New CPU feature `vis4b'.
5899 (dictunpack{8,16,32}): New insns.
5900 (FPCSMODE): New mode iterator.
5901 (fpcscond): New code iterator.
5902 (fpcsucond): Likewise.
5903 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
5904 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
5905 (fpcmpde{8,16,32}{si,di}shl): Likewise.
5906 (fpcmpur{8,16,32}{si,di}shl): Likewise.
5907 * config/sparc/constraints.md: Define constraints `q' for unsigned
5908 2-bit integer constants and `t' for unsigned 5-bit integer
5910 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
5912 (imm5_operand_dictunpack16): Likewise.
5913 (imm5_operand_dictunpack32): Likewise.
5914 (imm2_operand): Likewise.
5915 * doc/invoke.texi (SPARC Options): Document -mvis4b.
5916 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
5917 ditunpack* and fpcmp*shl builtins.
5919 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5921 * config.gcc: Handle m8 in --with-{cpu,tune} options.
5922 * config.in: Add HAVE_AS_SPARC6 define.
5923 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
5925 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
5927 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
5928 (CPP_CPU_SPEC): Handle m8.
5929 (ASM_CPU_SPEC): Likewise.
5930 * config/sparc/sparc-opts.h (enum processor_type): Add
5932 * config/sparc/sparc.c (m8_costs): New struct.
5933 (sparc_option_override): Handle TARGET_CPU_m8.
5934 (sparc32_initialize_trampoline): Likewise.
5935 (sparc64_initialize_trampoline): Likewise.
5936 (sparc_issue_rate): Likewise.
5937 (sparc_register_move_cost): Likewise.
5938 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
5939 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
5940 (ASM_CPU64_DEFAULT_SPEC): Likewise.
5941 (CPP_CPU_SPEC): Handle M8.
5942 (ASM_CPU_SPEC): Likewise.
5943 (AS_M8_FLAG): Define.
5944 * config/sparc/sparc.md: Add m8 to the cpu attribute.
5945 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
5946 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
5948 * configure: Regenerate.
5949 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
5952 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5954 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
5956 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
5957 ("*movdi_insn_sp32"): Do not set v3pipe.
5958 ("*movsi_insn"): Likewise.
5959 ("*movdi_insn_sp64"): Likewise.
5960 ("*movsf_insn"): Likewise.
5961 ("*movdf_insn_sp32"): Likewise.
5962 ("*movdf_insn_sp64"): Likewise.
5963 ("*zero_extendsidi2_insn_sp64"): Likewise.
5964 ("*sign_extendsidi2_insn"): Likewise.
5965 ("*mov<VM32:mode>_insn"): Likewise.
5966 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5967 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5968 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5969 ("<vlop:code><VL:mode>3"): Likewise.
5970 ("*not_<vlop:code><VL:mode>3"): Likewise.
5971 ("*nand<VL:mode>_vis"): Likewise.
5972 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
5973 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
5974 ("one_cmpl<VL:mode>2"): Likewise.
5975 ("faligndata<VM64:mode>_vis"): Likewise.
5976 ("alignaddrsi_vis"): Likewise.
5977 ("alignaddrdi_vis"): Likweise.
5978 ("alignaddrlsi_vis"): Likewise.
5979 ("alignaddrldi_vis"): Likewise.
5980 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5981 ("bmaskdi_vis"): Likewise.
5982 ("bmasksi_vis"): Likewise.
5983 ("bshuffle<VM64:mode>_vis"): Likewise.
5984 ("cmask8<P:mode>_vis"): Likewise.
5985 ("cmask16<P:mode>_vis"): Likewise.
5986 ("cmask32<P:mode>_vis"): Likewise.
5987 ("pdistn<P:mode>_vis"): Likewise.
5988 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5990 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5992 * config/sparc/sparc.md ("subtype"): New insn attribute.
5993 ("*wrgsr_sp64"): Set insn subtype.
5994 ("*rdgsr_sp64"): Likewise.
5995 ("alignaddrsi_vis"): Likewise.
5996 ("alignaddrdi_vis"): Likewise.
5997 ("alignaddrlsi_vis"): Likewise.
5998 ("alignaddrldi_vis"): Likewise.
5999 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6000 ("fexpand_vis"): Likewise.
6001 ("fpmerge_vis"): Likewise.
6002 ("faligndata<VM64:mode>_vis"): Likewise.
6003 ("bshuffle<VM64:mode>_vis"): Likewise.
6004 ("cmask8<P:mode>_vis"): Likewise.
6005 ("cmask16<P:mode>_vis"): Likewise.
6006 ("cmask32<P:mode>_vis"): Likewise.
6007 ("fchksm16_vis"): Likewise.
6008 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6009 ("fmean16_vis"): Likewise.
6010 ("fp<plusminus_insn>64_vis"): Likewise.
6011 ("<plusminus_insn>v8qi3"): Likewise.
6012 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6013 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6014 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6015 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6016 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6017 ("*movqi_insn"): Likewise.
6018 ("*movhi_insn"): Likewise.
6019 ("*movsi_insn"): Likewise.
6020 ("movsi_pic_gotdata_op"): Likewise.
6021 ("*movdi_insn_sp32"): Likewise.
6022 ("*movdi_insn_sp64"): Likewise.
6023 ("movdi_pic_gotdata_op"): Likewise.
6024 ("*movsf_insn"): Likewise.
6025 ("*movdf_insn_sp32"): Likewise.
6026 ("*movdf_insn_sp64"): Likewise.
6027 ("*zero_extendhisi2_insn"): Likewise.
6028 ("*zero_extendqihi2_insn"): Likewise.
6029 ("*zero_extendqisi2_insn"): Likewise.
6030 ("*zero_extendqidi2_insn"): Likewise.
6031 ("*zero_extendhidi2_insn"): Likewise.
6032 ("*zero_extendsidi2_insn_sp64"): Likewise.
6033 ("ldfsr"): Likewise.
6034 ("prefetch_64"): Likewise.
6035 ("prefetch_32"): Likewise.
6036 ("tie_ld32"): Likewise.
6037 ("tie_ld64"): Likewise.
6038 ("*tldo_ldub_sp32"): Likewise.
6039 ("*tldo_ldub1_sp32"): Likewise.
6040 ("*tldo_ldub2_sp32"): Likewise.
6041 ("*tldo_ldub_sp64"): Likewise.
6042 ("*tldo_ldub1_sp64"): Likewise.
6043 ("*tldo_ldub2_sp64"): Likewise.
6044 ("*tldo_ldub3_sp64"): Likewise.
6045 ("*tldo_lduh_sp32"): Likewise.
6046 ("*tldo_lduh1_sp32"): Likewise.
6047 ("*tldo_lduh_sp64"): Likewise.
6048 ("*tldo_lduh1_sp64"): Likewise.
6049 ("*tldo_lduh2_sp64"): Likewise.
6050 ("*tldo_lduw_sp32"): Likewise.
6051 ("*tldo_lduw_sp64"): Likewise.
6052 ("*tldo_lduw1_sp64"): Likewise.
6053 ("*tldo_ldx_sp64"): Likewise.
6054 ("*mov<VM32:mode>_insn"): Likewise.
6055 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6056 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6058 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6060 * config/sparc/sparc.md ("type"): New insn type viscmp.
6061 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6063 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6064 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6065 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6066 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6068 ("n7_vis_logical_11cycle"): Likewise.
6069 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6070 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6071 * config/sparc/niagara.md ("niag_vis"): Likewise.
6072 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6073 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6075 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6077 * config/sparc/sparc.md: New instruction type `bmask'.
6078 (bmaskdi_vis): Use the `bmask' type.
6079 (bmasksi_vis): Likewise.
6080 * config/sparc/ultra3.md (us3_array): Likewise.
6081 * config/sparc/niagara7.md (n7_array): Likewise.
6082 * config/sparc/niagara4.md (n4_array): Likewise.
6083 * config/sparc/niagara2.md (niag2_vis): Likewise.
6084 (niag3_vis): Likewise.
6085 * config/sparc/niagara.md (niag_vis): Likewise.
6087 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6089 * ipa-comdats.c: Remove optimize check from gate.
6090 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6091 for functions not optimized.
6092 (ipa_fn_summary_read): Skip optimize check.
6093 (ipa_fn_summary_write): Likewise.
6094 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6096 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6098 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6099 (check_callers): Give up on caller which is not optimized.
6100 (inline_small_functions): Likewise.
6101 (ipa_inline): Do not give up when not optimizing.
6102 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6103 away unoptimizes cdtors.
6104 (whole_program_function_and_variable_visibility): Do
6105 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6106 * ipa.c (process_references): Do not check optimize.
6107 (symbol_table::remove_unreachable_nodes): Update optimize check.
6108 (set_writeonly_bit): Update optimize check.
6109 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6110 (pass_ipa_single_use::gate): Remove.
6112 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6114 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6115 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6116 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6117 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6118 permute_load, permute_store, adjust_extract, adjust_splat,
6119 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6120 replace_swap_with_copy, dump_swap_insn_table,
6121 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6122 recombine_lvx_pattern, recombine_stvx_pattern,
6123 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6124 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6125 to file rs6000-p8swap.c.
6126 * config/rs6000/rs6000-p8swap.c: New file.
6127 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6128 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6129 and rs6000*-*-* targets.
6131 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6133 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6135 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6137 * lto-wrapper.c (merge_and_complain): Do not merge
6138 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6139 fsigned_zeros, ftrapping_math, fwrapv.
6140 (append_compiler_options): Do not track these options.
6141 (append_linker_options): Likewie
6143 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6145 * cgraphunit.c (cgraph_node::finalize_function): When
6146 !flag_toplevel_reorde set no_reorder flag.
6147 (varpool_node::finalize_decl): Likewise.
6148 (symbol_table::compile): Drop no toplevel reorder path.
6150 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6152 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6153 edges; zero probability is not better than uninitialized.
6155 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6157 * asan.h (asan_sanitize_allocas_p): Declare.
6158 * asan.c (asan_sanitize_allocas_p): New function.
6159 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6160 (handle_builtin_alloca): Likewise.
6161 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6162 if !asan_sanitize_allocas_p.
6163 * params.def (asan-instrument-allocas): Add new option.
6164 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6165 * opts.c (common_handle_option): Disable allocas sanitization for
6168 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6170 * asan.c: Include gimple-fold.h.
6171 (get_last_alloca_addr): New function.
6172 (handle_builtin_stackrestore): Likewise.
6173 (handle_builtin_alloca): Likewise.
6174 (asan_emit_allocas_unpoison): Likewise.
6175 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6176 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6177 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6178 (instrument_builtin_call): Pass gimple iterator to
6179 get_mem_refs_of_builtin_call.
6180 (last_alloca_addr): New global.
6181 * asan.h (asan_emit_allocas_unpoison): Declare.
6182 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6183 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6184 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6185 if function calls alloca.
6186 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6187 * gimple-fold.h (replace_call_with_value): Declare.
6188 * internal-fn.c: Include asan.h.
6189 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6190 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6192 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6194 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6195 (C_SELFTEST_FLAGS): New.
6196 (CPP_SELFTEST_FLAGS): New.
6197 (SELFTEST_DEPS): New, from deps of s-selftest.
6198 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6199 (CPP_SELFTEST_DEPS): New.
6200 (selftest): Add dependency on s-selftest-c++.
6201 (s-selftest): Rename to...
6202 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6203 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6204 than SELFTEST_FLAGS.
6205 (selftest-gdb): Rename to...
6206 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6208 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6209 (selftest-valgrind): Rename to...
6210 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6212 (selftest-valgrind): Reintroduce as an alias for
6213 selftest-c-valgrind.
6214 (s-selftest-c++): New.
6215 (selftest-c++-gdb): New.
6216 (selftest-c++-valgrind): New.
6218 2017-07-06 Olivier Hainque <hainque@adacore.com>
6220 * gcc.c (process_command): When deciding if undefined variables
6221 should be ignored when processing specs, accept "gcc -v" as well.
6223 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6225 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6226 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6228 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6230 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6231 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6232 * config/arm/arm-tables.opt: Regenerate.
6233 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6235 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6237 2017-07-06 Carl Love <cel@us.ibm.com>
6239 * ChangeLog: Clean up from mid air collision
6241 2017-07-06 Carl Love <cel@us.ibm.com>
6243 * config/rs6000/rs6000-c.c: Add support for built-in functions
6244 vector signed int vec_subc (vector signed int, vector signed int);
6245 vector signed __int128 vec_subc (vector signed __int128,
6246 vector signed __int128);
6247 vector unsigned __int128 vec_subc (vector unsigned __int128,
6248 vector unsigned __int128);
6249 vector signed int vec_sube (vector signed int, vector signed int,
6251 vector unsigned int vec_sube (vector unsigned int,
6252 vector unsigned int,
6253 vector unsigned int);
6254 vector signed __int128 vec_sube (vector signed __int128,
6255 vector signed __int128,
6256 vector signed__int128);
6257 vector unsigned __int128 vec_sube (vector unsigned __int128,
6258 vector unsigned __int128,
6259 vector unsigned __int128);
6260 vector signed int vec_subec (vector signed int, vector signed int,
6262 vector unsigned int vec_subec (vector unsigned int,
6263 vector unsigned int,
6264 vector unsigned int);
6265 vector signed __int128 vec_subec (vector signed __int128,
6266 vector signed __int128,
6267 vector signed__int128);
6268 vector unsigned __int128 vec_subec (vector unsigned __int128,
6269 vector unsigned __int128,
6270 vector unsigned __int128);
6271 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6272 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6273 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6274 BU_ALTIVEC_OVERLOAD_X definitions.
6275 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6276 * doc/extend.texi: Update the built-in documentation file for the new
6279 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6282 * diagnostic-show-locus.c (layout::layout): Use start and finish
6283 spelling location for the start and finish of each range.
6284 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6285 Add unused aspect param.
6286 * input.c (expand_location_1): Add "aspect" param, and use it
6287 to access the correct part of the location.
6288 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6290 (expand_location_to_spelling_point): Likewise.
6291 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6292 param, and pass it to expand_location_1.
6294 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6296 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6297 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6298 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6299 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6300 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6301 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6302 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6303 _mm_maskz_getmant_ss): New intrinsics.
6304 (__builtin_ia32_getexpss128_mask): Changed to ...
6305 __builtin_ia32_getexpss128_round ... this.
6306 (__builtin_ia32_getexpsd128_mask): Changed to ...
6307 __builtin_ia32_getexpsd128_round ... this.
6308 * config/i386/i386-builtin-types.def
6309 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6310 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6311 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6312 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6313 __builtin_ia32_getmantss_mask_round): New builtins.
6314 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6315 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6316 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6317 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6318 * config/i386/sse.md
6319 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6320 avx512f_sgetexp<mode><mask_scalar_name>
6321 <round_saeonly_scalar_name> ... this.
6322 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6323 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6324 vgetexp<ssescalarmodesuffix>
6325 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6326 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6327 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6328 avx512f_vgetmant<mode><mask_scalar_name>
6329 <round_saeonly_scalar_name> ... this.
6330 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6331 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6332 vgetmant<ssescalarmodesuffix>
6333 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6334 %0<mask_scalar_operand4>, %1, %2
6335 <round_saeonly_scalar_mask_op4>, %3} ... this.
6336 * config/i386/subst.md (mask_scalar_operand4,
6337 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6338 round_saeonly_scalar_nimm_predicate): New subst attributes.
6340 2017-07-06 Julia Koval <julia.koval@intel.com>
6342 * config/i386/i386.c (ix86_erase_embedded_rounding):
6343 Remove code for old rounding pattern.
6345 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6347 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6349 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6351 * doc/sourcebuild.texi (Test Directives, Variants of
6352 dg-require-support): Add documentation for dg-require-stack-check.
6354 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6356 * config/i386/subst.md (mask_scalar, round_scalar,
6357 round_saeonly_scalar): New meta-templates.
6358 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6359 round_scalar_mask_operand3, round_scalar_mask_op3,
6360 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6361 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6362 round_saeonly_scalar_constraint,
6363 round_saeonly_scalar_prefix): New subst attribute.
6364 * config/i386/sse.md
6365 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6366 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6367 <round_scalar_name> ... this.
6368 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6369 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6370 <round_scalar_name> ... this.
6371 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6372 <sse>_vm<code><mode>3<mask_scalar_name>
6373 <round_saeonly_scalar_name> ... this.
6374 (v<plusminus_mnemonic><ssescalarmodesuffix>
6375 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6376 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6377 v<plusminus_mnemonic><ssescalarmodesuffix>
6378 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6379 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6380 (v<multdiv_mnemonic><ssescalarmodesuffix>
6381 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6382 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6383 v<multdiv_mnemonic><ssescalarmodesuffix>
6384 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6385 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6386 (v<maxmin_float><ssescalarmodesuffix>
6387 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6388 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6389 v<maxmin_float><ssescalarmodesuffix>
6390 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6391 %0<mask_scalar_operand3>, %1, %<iptr>2
6392 <round_saeonly_scalar_mask_op3>} ... this.
6394 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6396 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6397 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6399 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6400 Alan Hayward <alan.hayward@arm.com>
6401 David Sherwood <david.sherwood@arm.com>
6403 * combine.c (simplify_if_then_else): Remove "enum" before
6405 * compare-elim.c (can_eliminate_compare): Likewise.
6406 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6408 (aarch64_lookup_simd_builtin_type): Likewise.
6409 (aarch64_simd_builtin_type): Likewise.
6410 (aarch64_init_simd_builtin_types): Likewise.
6411 (aarch64_simd_expand_args): Likewise.
6412 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6414 (aarch64_reverse_mask): Likewise.
6415 (aarch64_simd_emit_reg_reg_move): Likewise.
6416 (aarch64_gen_adjusted_ldpstp): Likewise.
6417 (aarch64_ccmp_mode_to_code): Likewise.
6418 (aarch64_operands_ok_for_ldpstp): Likewise.
6419 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6420 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6422 (aarch64_min_divisions_for_recip_mul): Likewise.
6423 (aarch64_reassociation_width): Likewise.
6424 (aarch64_get_condition_code_1): Likewise.
6425 (aarch64_simd_emit_reg_reg_move): Likewise.
6426 (aarch64_simd_attr_length_rglist): Likewise.
6427 (aarch64_reverse_mask): Likewise.
6428 (aarch64_operands_ok_for_ldpstp): Likewise.
6429 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6430 (aarch64_gen_adjusted_ldpstp): Likewise.
6431 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6433 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6434 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6435 (arm_lookup_simd_builtin_type): Likewise.
6436 (arm_simd_builtin_type): Likewise.
6437 (arm_init_simd_builtin_types): Likewise.
6438 (arm_expand_builtin_args): Likewise.
6439 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6440 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6441 (ft32_setup_incoming_varargs): Likewise.
6442 (ft32_function_arg): Likewise.
6443 (ft32_function_arg_advance): Likewise.
6444 (ft32_pass_by_reference): Likewise.
6445 (ft32_arg_partial_bytes): Likewise.
6446 (ft32_valid_pointer_mode): Likewise.
6447 (ft32_addr_space_pointer_mode): Likewise.
6448 (ft32_addr_space_legitimate_address_p): Likewise.
6449 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6451 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6452 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6453 (iamcu_alignment): Likewise.
6454 (canonicalize_vector_int_perm): Likewise.
6455 (ix86_noce_conversion_profitable_p): Likewise.
6456 (ix86_mpx_bound_mode): Likewise.
6457 (ix86_operands_ok_for_move_multiple): Likewise.
6458 * config/microblaze/microblaze-protos.h
6459 (microblaze_expand_conditional_branch_reg): Likewise.
6460 * config/microblaze/microblaze.c
6461 (microblaze_expand_conditional_branch_reg): Likewise.
6462 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6464 (rs6000_reassociation_width): Likewise.
6465 (rs6000_invalid_binary_op): Likewise.
6466 (fusion_p9_p): Likewise.
6467 (emit_fusion_p9_load): Likewise.
6468 (emit_fusion_p9_store): Likewise.
6469 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6471 (riscv_hard_regno_mode_ok_p): Likewise.
6472 (riscv_address_insns): Likewise.
6473 (riscv_split_symbol): Likewise.
6474 (riscv_legitimize_move): Likewise.
6475 (riscv_function_value): Likewise.
6476 (riscv_hard_regno_nregs): Likewise.
6477 (riscv_expand_builtin): Likewise.
6478 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6479 (riscv_build_integer): Likewise.
6480 (riscv_split_integer): Likewise.
6481 (riscv_legitimate_constant_p): Likewise.
6482 (riscv_cannot_force_const_mem): Likewise.
6483 (riscv_regno_mode_ok_for_base_p): Likewise.
6484 (riscv_valid_base_register_p): Likewise.
6485 (riscv_valid_offset_p): Likewise.
6486 (riscv_valid_lo_sum_p): Likewise.
6487 (riscv_classify_address): Likewise.
6488 (riscv_legitimate_address_p): Likewise.
6489 (riscv_address_insns): Likewise.
6490 (riscv_load_store_insns): Likewise.
6491 (riscv_force_binary): Likewise.
6492 (riscv_split_symbol): Likewise.
6493 (riscv_force_address): Likewise.
6494 (riscv_legitimize_address): Likewise.
6495 (riscv_move_integer): Likewise.
6496 (riscv_legitimize_const_move): Likewise.
6497 (riscv_legitimize_move): Likewise.
6498 (riscv_address_cost): Likewise.
6499 (riscv_subword): Likewise.
6500 (riscv_output_move): Likewise.
6501 (riscv_canonicalize_int_order_test): Likewise.
6502 (riscv_emit_int_order_test): Likewise.
6503 (riscv_function_arg_boundary): Likewise.
6504 (riscv_pass_mode_in_fpr_p): Likewise.
6505 (riscv_pass_fpr_single): Likewise.
6506 (riscv_pass_fpr_pair): Likewise.
6507 (riscv_get_arg_info): Likewise.
6508 (riscv_function_arg): Likewise.
6509 (riscv_function_arg_advance): Likewise.
6510 (riscv_arg_partial_bytes): Likewise.
6511 (riscv_function_value): Likewise.
6512 (riscv_pass_by_reference): Likewise.
6513 (riscv_setup_incoming_varargs): Likewise.
6514 (riscv_print_operand): Likewise.
6515 (riscv_elf_select_rtx_section): Likewise.
6516 (riscv_save_restore_reg): Likewise.
6517 (riscv_for_each_saved_reg): Likewise.
6518 (riscv_register_move_cost): Likewise.
6519 (riscv_hard_regno_mode_ok_p): Likewise.
6520 (riscv_hard_regno_nregs): Likewise.
6521 (riscv_class_max_nregs): Likewise.
6522 (riscv_memory_move_cost): Likewise.
6523 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6524 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6525 (rl78_addr_space_address_mode): Likewise.
6526 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6528 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6529 (rs6000_reassociation_width): Likewise.
6530 (rs6000_invalid_binary_op): Likewise.
6531 (fusion_p9_p): Likewise.
6532 (emit_fusion_p9_load): Likewise.
6533 (emit_fusion_p9_store): Likewise.
6534 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6535 (ok_for_simple_move_operands): Likewise.
6536 (ok_for_simple_move_strict_operands): Likewise.
6537 (ok_for_simple_arith_logic_operands): Likewise.
6538 (visium_legitimize_reload_address): Likewise.
6539 (visium_select_cc_mode): Likewise.
6540 (output_cbranch): Likewise.
6541 (visium_split_double_move): Likewise.
6542 (visium_expand_copysign): Likewise.
6543 (visium_expand_int_cstore): Likewise.
6544 (visium_expand_fp_cstore): Likewise.
6545 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6546 (visium_function_arg): Likewise.
6547 (visium_function_arg_advance): Likewise.
6548 (visium_libcall_value): Likewise.
6549 (visium_setup_incoming_varargs): Likewise.
6550 (visium_legitimate_constant_p): Likewise.
6551 (visium_legitimate_address_p): Likewise.
6552 (visium_legitimize_address): Likewise.
6553 (visium_secondary_reload): Likewise.
6554 (visium_register_move_cost): Likewise.
6555 (visium_memory_move_cost): Likewise.
6556 (prepare_move_operands): Likewise.
6557 (ok_for_simple_move_operands): Likewise.
6558 (ok_for_simple_move_strict_operands): Likewise.
6559 (ok_for_simple_arith_logic_operands): Likewise.
6560 (visium_function_value_1): Likewise.
6561 (rtx_ok_for_offset_p): Likewise.
6562 (visium_legitimize_reload_address): Likewise.
6563 (visium_split_double_move): Likewise.
6564 (visium_expand_copysign): Likewise.
6565 (visium_expand_int_cstore): Likewise.
6566 (visium_expand_fp_cstore): Likewise.
6567 (visium_split_cstore): Likewise.
6568 (visium_select_cc_mode): Likewise.
6569 (visium_split_cbranch): Likewise.
6570 (output_cbranch): Likewise.
6571 (visium_print_operand_address): Likewise.
6572 * expmed.c (flip_storage_order): Likewise.
6573 * expmed.h (emit_cstore): Likewise.
6574 (flip_storage_order): Likewise.
6575 * genrecog.c (validate_pattern): Likewise.
6576 * hsa-gen.c (gen_hsa_addr): Likewise.
6577 * internal-fn.c (expand_arith_overflow): Likewise.
6578 * ira-color.c (allocno_copy_cost_saving): Likewise.
6579 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6580 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6581 (process_invariant_for_inheritance): Likewise.
6582 * lra-eliminations.c (move_plus_up): Likewise.
6583 * omp-low.c (lower_oacc_reductions): Likewise.
6584 * simplify-rtx.c (simplify_subreg): Likewise.
6585 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6586 (TARGET_CHKP_BOUND_MODE): Likewise..
6587 * targhooks.c (default_chkp_bound_mode): Likewise.
6588 (default_setup_incoming_vararg_bounds): Likewise.
6589 * targhooks.h (default_chkp_bound_mode): Likewise.
6590 (default_setup_incoming_vararg_bounds): Likewise.
6591 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6592 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6593 (have_whole_vector_shift): Likewise.
6594 * tree-vect-stmts.c (vectorizable_load): Likewise.
6595 * doc/tm.texi: Regenerate.
6597 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6599 Graceful degrade if Binutils PR21472 is not available.
6602 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6603 .rodata in flash test fails.
6604 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6605 * confgure: Regenerate.
6606 * config.in: Regenerate.
6607 * config/avr/avr.c (avr_asm_named_section)
6608 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6609 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6610 (avr_asm_init_sections): Same.
6612 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6614 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6615 (fma<VH:mode>4_intrinsic): Likewise.
6616 (*fmsub<VCVTF:mode>4): Likewise.
6617 (*fmsub<VH:mode>4_intrinsic): Likewise.
6619 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6622 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6623 Don't depend on "optimize > 0".
6624 (out_movhi_r_mr, out_movqi_mr_r): Same.
6625 (out_movhi_mr_r, out_movqi_r_mr): Same.
6626 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6627 io_address_operand on "optimize > 0".
6629 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6631 * tree-loop-distribution.c: Add general explanantion on the pass.
6632 (generate_loops_for_partition): Mark distributed loop.
6633 (pg_add_dependence_edges): New parameter. Handle alias data
6634 dependence specially and record it in the parameter if asked.
6635 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6636 (init_partition_graph_vertices, add_partition_graph_edge): New.
6637 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6638 (free_partition_graph_vdata, build_partition_graph): New.
6639 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6640 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6641 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6642 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6643 (version_for_distribution_p, finalize_partitions): New.
6644 (distribute_loop): Handle alias data dependence specially. Factor
6645 out loop fusion code as functions and call these functions.
6647 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6649 * tree-loop-distribution.c (classify_partition): New parameter and
6650 better handle reduction statement.
6651 (rdg_build_partitions): Revise comment.
6652 (distribute_loop): Compute statements in all partitions and pass it
6653 to classify_partition.
6655 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6657 * tree-loop-distribution.c (enum partition_type): New.
6658 (struct partition): New field type.
6659 (partition_merge_into): Add parameter. Update partition type.
6660 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6661 (build_rdg_partition_for_vertex): Compute partition type.
6662 (rdg_build_partitions): Dump partition type.
6663 (distribute_loop): Update calls to partition_merge_into.
6665 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6667 * tree-loop-distribution.c (struct ddr_hasher): New.
6668 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6670 (classify_partition): Call get_data_dependence.
6671 (pg_add_dependence_edges): Ditto.
6672 (distribute_loop): Release data dependence hash table.
6674 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6676 * tree-loop-distribution.c (ref_base_address): Delete.
6677 (similar_memory_accesses): Rename ...
6678 (share_memory_accesses): ... to this. Check if partitions access
6679 the same memory reference.
6680 (distribute_loop): Call share_memory_accesses.
6682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6684 * tree-loop-distribution.c (struct partition): New field recording
6686 (partition_alloc, partition_free): Init and release data refs.
6687 (partition_merge_into): Merge data refs.
6688 (build_rdg_partition_for_vertex): Collect data refs for partition.
6689 (pg_add_dependence_edges): Change parameters from vector to bitmap.
6691 (distribute_loop): Remve data refs from vertice data of partition
6694 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6696 * tree-loop-distribution.c (params.h): Include header file.
6697 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
6698 (datarefs_vec): New global var.
6699 (create_rdg_vertices): Use datarefs_vec directly.
6700 (free_rdg): Don't free data references.
6701 (build_rdg): Update use. Don't free data references.
6702 (distribute_loop): Compute global variable for data references.
6703 Bail out if there are too many data references.
6705 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6707 * tree-loop-distribution.c (loop_nest): New global var.
6708 (build_rdg): Use loop directly, rather than loop nest.
6709 (pg_add_dependence_edges): Remove loop nest parameter. Use global
6711 (distribute_loop): Compute global variable loop nest. Update use.
6713 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6715 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
6716 (partition_merge_into): New parameter. Dump reason for fusion.
6717 (distribute_loop): Update use of partition_merge_into.
6719 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6721 * tree-loop-distribution.c (bb_top_order_index): New.
6722 (bb_top_order_index_size, bb_top_order_cmp): New.
6723 (stmts_from_loop): Use topological order.
6724 (pass_loop_distribution::execute): Compute and release topological
6725 order for basic blocks.
6727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6729 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
6732 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6734 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
6735 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
6736 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
6737 * internal-fn.def (LOOP_DIST_ALIAS): New.
6738 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
6739 (fold_loop_internal_call): ... this.
6740 (vect_loop_dist_alias_call): New function.
6741 (set_uid_loop_bbs): Call fold_loop_internal_call.
6742 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
6745 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6748 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
6749 Require dead FLAGS_REG at the beginning of a peephole.
6751 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6754 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
6755 arguments in the call to __builtin_ia32_sbb_u32.
6756 (_subborrow_u64): Swap _X and _Y arguments in the call to
6757 __builtin_ia32_sbb_u64.
6759 2017-07-04 Jakub Jelinek <jakub@redhat.com>
6762 * tree-vrp.c (compare_assert_loc): Turn into a function template
6763 with stable template parameter. Only test if a->e is NULL,
6764 !a->e == !b->e has been verified already. Use e == NULL or
6765 e != NULL instead of e or ! e tests. If stable is true, don't use
6766 iterative_hash_expr, on the other side allow a or b or both NULL
6767 and sort the NULLs last.
6768 (process_assert_insertions): Sort using compare_assert_loc<false>
6769 instead of compare_assert_loc, later sort using
6770 compare_assert_loc<true> before calling process_assert_insertions_for
6771 in a loop. Use break instead of continue once seen NULL pointer.
6773 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6775 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6776 Cortex-R7 and Cortex-R8 processors.
6778 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6780 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
6781 uninitialized while src is not.
6783 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
6785 * common/config/arm/arm-common.c: Adjust include path for
6787 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
6788 (arm-cpu.h): Create in build directory. Adjust dependency rules.
6789 (arm-cpu-data.h): Likewise.
6790 (arm-cpu-cdata.h): Likewise.
6791 * config/arm/arm-cpu.h: Delete.
6792 * config/arm/arm-cpu-cdata.h: Delete.
6793 * config/arm/arm-cpu-data.h: Delete.
6795 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
6797 * config/arm/arm-cpus.in (cortex-a55): New.
6798 (cortex-a75): Likewise.
6799 (cortex-a75.cortex-a55): Likewise.
6800 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
6802 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
6803 * config/arm/arm-cpu-cdata.h: Regenerate.
6804 * config/arm/arm-cpu-data.h: Regenerate.
6805 * config/arm/arm-cpu.h: Regenerate.
6806 * config/arm/arm-tables.opt: Regenerate.
6807 * config/arm/arm-tune.md: Regenerate.
6809 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6811 * haifa-sched.c (sched_create_recovery_edges): Update profile.
6813 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6815 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
6818 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
6820 PR tree-optimization/81292
6821 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
6822 full_string_p, also call adjust_related_strinfos if the adjustment
6823 is simple, otherwise invalidate related strinfos.
6825 2017-07-04 Martin Liska <mliska@suse.cz>
6828 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
6829 newly created variable as DECL_IGNORED_P.
6831 2017-07-04 Martin Liska <mliska@suse.cz>
6834 * ipa-inline.c (inline_small_functions):
6835 Use xstrdup_for_dump.
6837 2017-07-04 Tom de Vries <tom@codesourcery.com>
6839 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
6841 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6844 * config/darwin.c (darwin_function_switched_text_sections):
6847 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
6849 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
6851 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6853 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
6855 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6857 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
6858 min_profitable_iters, and th as inclusive lower bounds.
6859 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
6860 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
6861 for min_profitable_iters and min_profitable_estimate.
6862 (vect_transform_loop): Treat th as an inclusive lower bound.
6863 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6865 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6868 * config/darwin.c (darwin_function_switched_text_sections):
6869 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
6870 in two pieces, and suppress the use of buf.
6872 2017-07-03 Nathan Sidwell <nathan@acm.org>
6874 * hash-table.h (hash_table_mod1): Fix indentation.
6876 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6879 * predict.c (force_edge_cold): Be more careful about propagation
6881 * profile-count.h (profile_probability::guessed,
6882 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
6884 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
6886 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
6888 * doc/invoke.texi (rcpc architecture extension): Document it.
6890 2017-07-03 Richard Biener <rguenther@suse.de>
6892 PR tree-optimization/60510
6893 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
6894 the scalar reduction PHI and use it.
6895 (vectorizable_reduction): Properly guard the single_defuse_cycle
6896 path for non-SLP reduction chains where we cannot use it.
6897 Rework reduc_def/index and vector type deduction. Rework
6898 vector operand gathering during reduction op code-gen.
6899 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
6900 chains dissolve the chain and leave it to non-SLP reduction
6903 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6905 * tree-data-ref.h (dr_alignment): Declare.
6906 * tree-data-ref.c (dr_alignment): New function.
6907 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
6908 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
6910 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
6912 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6914 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
6915 and base_misalignment fields.
6916 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
6917 * tree-data-ref.c: Include builtins.h.
6918 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
6919 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
6920 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
6921 * tree-vect-data-refs.c: Include tree-cfg.h.
6922 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
6923 fields instead of calculating an alignment here.
6924 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
6925 innermost_loop_behavior fields.
6927 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6929 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
6931 (DR_STEP_ALIGNMENT): New macro.
6932 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
6933 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
6934 (create_data_ref): Print it.
6935 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
6936 to tell whether the step preserves vector (mis)alignment.
6937 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6938 Move the check for an integer step and generalise to all INTEGER_CST.
6939 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
6940 Print the outer step alignment.
6942 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6944 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
6945 with offset_alignment.
6946 (DR_ALIGNED_TO): Delete.
6947 (DR_OFFSET_ALIGNMENT): New macro.
6948 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
6949 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
6950 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
6951 (create_data_ref): Likewise.
6952 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6953 (vect_analyze_data_refs): Likewise.
6954 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
6955 creating dummy innermost behavior.
6957 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6959 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
6960 with a "innermost_loop_behavior *" and refeence tree.
6961 * tree-data-ref.c (dr_analyze_innermost): Likewise.
6962 (create_data_ref): Update call accordingly.
6963 * tree-predcom.c (find_looparound_phi): Likewise.
6965 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6967 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
6968 fields with dr_wrt_vec_loop.
6969 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
6970 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
6971 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
6972 (vect_dr_behavior): New function.
6973 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6974 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
6975 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
6976 track whether the step preserves the misalignment.
6977 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6978 Use vect_dr_behavior.
6979 (vect_setup_realignment): Update call accordingly.
6980 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
6981 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
6982 call to vect_create_addr_base_for_vector_ref.
6983 (vect_create_cond_for_align_checks): Likewise.
6984 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
6985 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
6986 (vect_recog_mask_conversion_pattern): Likewise.
6987 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
6988 (new_stmt_vec_info): Remove redundant zeroing.
6990 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6992 * common/config/arm/arm-common.c (arm_be8_option): New function.
6993 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
6994 (ISA_ARMv6): Add isa_bit_be8.
6995 * config/arm/arm.h (arm_be8_option): Add prototype.
6996 (BE8_SPEC_FUNCTION): New define.
6997 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
6998 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
6999 (mlittle-endian): Similarly.
7000 (mbe8, mbe32): New options.
7001 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7002 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7004 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7006 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7008 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7010 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7011 (cleanup_tree_cfg_bb): Use it.
7012 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7014 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7016 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7019 * loop-doloop.c (add_test): Update profile.
7021 2017-07-03 Martin Liska <mliska@suse.cz>
7024 * sanopt.c (rewrite_usage_of_param): New function.
7025 (sanitize_rewrite_addressable_params): Likewise.
7026 (pass_sanopt::execute): Call rewrite_usage_of_param.
7028 2017-07-03 Richard Biener <rguenther@suse.de>
7030 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7031 back to using VIEW_CONVERT_EXPR.
7033 2017-07-03 Martin Liska <mliska@suse.cz>
7036 * doc/extend.texi: Document when a resolver function is
7037 generated for target_clones.
7039 2017-07-03 Martin Liska <mliska@suse.cz>
7041 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7042 and shadow memory of auto variables which are subject of
7043 use-after-scope sanitization.
7044 (asan_expand_mark_ifn): Add do set only when is_poison.
7046 2016-07-03 Richard Biener <rguenther@suse.de>
7048 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7050 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7051 (vectorizable_reduction): Perform reduction PHI creation when
7052 visiting a reduction PHI and adjust and simplify code generation
7053 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7054 (vect_transform_loop): Visit reduction PHIs.
7055 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7056 defs into the SLP tree.
7057 (vect_build_slp_tree): Reduction defs terminate the recursion.
7058 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7060 (vect_get_vec_defs_for_stmt_copy): Export.
7061 (vect_get_vec_defs): Likewise.
7062 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7064 (vect_get_vec_defs_for_stmt_copy): Declare.
7065 (vect_get_vec_defs): Likewise.
7067 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7069 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7070 parameter with a "loop" parameter and use it instead of the
7071 loop containing DR_STMT. Don't check simple_iv when doing
7072 BB analysis. Describe the two analysis modes in the comment.
7074 2017-07-03 Tom de Vries <tom@codesourcery.com>
7076 PR tree-optimization/69468
7077 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7078 (find_same_succ_bb): Handle ignore_edge_flags.
7080 2017-07-03 Tom de Vries <tom@codesourcery.com>
7082 PR tree-optimization/81192
7083 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7085 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7087 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7089 2017-07-03 Tom de Vries <tom@codesourcery.com>
7091 PR tree-optimization/81192
7092 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7093 BB_SAME_SUCC (bb) == NULL.
7095 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7097 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7100 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7102 * dumpfile.c: Include profile-count.h
7103 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7105 (insert_cond_bb): Update profile.
7106 * tree-cfg.h (insert_cond_bb): Update prototype.
7107 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7108 * tree-dump.c: Do not include tree-cfg.
7110 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7112 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7114 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7116 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7119 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7121 * tree-complex.c (expand_complex_div_wide): update profile.
7123 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7124 Alan Hayward <alan.hayward@arm.com>
7125 David Sherwood <david.sherwood@arm.com>
7127 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7128 (CORETYPES_H): New define.
7129 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7130 (insn-modes-inline.h, s-modes-inline-h): New rules.
7131 (generated_files): Add insn-modes-inline.h.
7132 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7133 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7134 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7135 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7136 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7137 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7138 (build/gencodes.o, build/genconditions.o): Likewise.
7139 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7140 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7141 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7142 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7143 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7144 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7145 * coretypes.h: Include everything up to real.h for generators.
7146 Include insn-modes.h first. Include wide-int-print.h after
7147 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7148 * machmode.h: Don't include insn-modes.h here.
7149 * function-tests.c: Remove includes of signop.h, machmode.h,
7150 double-int.h and wide-int.h.
7152 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7154 * optc-save-gen.awk: Likewise.
7155 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7156 * godump.c: Remove include of wide-int-print.h.
7157 * pretty-print.h: Likewise.
7158 * wide-int-print.cc: Likewise.
7159 * wide-int.cc: Likewise.
7160 * hash-map-tests.c: Remove include of signop.h.
7161 * hash-set-tests.c: Likewise.
7162 * rtl-tests.c: Likewise.
7163 * mkconfig.sh: Remove include of machmode.h.
7164 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7166 (emit_insn_modes_inline_h): ...this new function. Emit the code
7167 into an insn-modes-inline.h header file, adding appropriate
7168 include guards and end comments.
7169 (emit_insn_modes_c_header): Remove include of machmode.h.
7170 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7172 (main): Handle -i flag and call emit_insn_modes_inline_h when
7175 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7177 * tree-ssa-strlen.c (strinfo): Rename the length field to
7178 nonzero_chars. Add a full_string_p field.
7179 (compare_nonzero_chars, zero_length_string_p): New functions.
7180 (get_addr_stridx): Add an offset_out parameter.
7181 Use compare_nonzero_chars.
7182 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7183 (new_strinfo): Update after above changes to strinfo.
7184 (set_endptr_and_length): Set full_string_p.
7185 (get_string_length): Update after above changes to strinfo.
7186 (unshare_strinfo): Update call to new_strinfo.
7187 (maybe_invalidate): Likewise.
7188 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7189 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7190 as a uhwi instead of an shwi. Update after above changes to
7191 strinfo and new_strinfo.
7192 (zero_length_string): Assert that chainsi contains full strings.
7193 Use zero_length_string_p. Update call to new_strinfo.
7194 (adjust_related_strinfos): Update after above changes to strinfo.
7195 Copy full_string_p from origsi.
7196 (adjust_last_stmt): Use zero_length_string_p.
7197 (handle_builtin_strlen): Update after above changes to strinfo and
7198 new_strinfo. Install the lhs as the string length if the previous
7199 entry didn't describe a full string.
7200 (handle_builtin_strchr): Update after above changes to strinfo
7202 (handle_builtin_strcpy): Likewise.
7203 (handle_builtin_strcat): Likewise.
7204 (handle_builtin_malloc): Likewise.
7205 (handle_pointer_plus): Likewise.
7206 (handle_builtin_memcpy): Likewise. Track nonzero characters
7207 that aren't necessarily followed by a nul terminator.
7208 (handle_char_store): Likewise.
7210 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7212 PR tree-optimization/80769
7213 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7214 for malloc and calloc. Document the new invariant that all related
7215 strinfos have delayed lengths or none do.
7216 (verify_related_strinfos): Move earlier in file.
7217 (set_endptr_and_length): New function, split out from...
7218 (get_string_length): ...here. Also set the lengths of related
7220 (zero_length_string): Assert that chainsi has known (rather than
7222 (adjust_related_strinfos): Likewise.
7224 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7226 PR tree-optimization/81136
7227 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7228 assert that two references with the same misalignment have the same
7229 compile-time misalignment if those compile-time misalignments
7232 2017-07-01 Andi Kleen <ak@linux.intel.com>
7234 * print-tree.c (print_node): Print all attributes.
7236 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7238 * cfg.c (scale_bbs_frequencies): New function.
7239 * cfg.h (scale_bbs_frequencies): Declare it.
7240 * cfgloopanal.c (single_likely_exit): Cleanup.
7241 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7243 (scale_loop_profile): Likewise.
7244 (loop_version): Likewise.
7245 (create_empty_loop_on_edge): Update.
7246 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7247 scale_loop_frequencies, scale_loop_profile, loopify,
7248 loop_version): Update prototypes.
7249 * modulo-sched.c (sms_schedule): Update.
7250 * predict.c (unlikely_executed_edge_p): Also check probability.
7251 (probably_never_executed_edge_p): Fix typo.
7252 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7253 * tree-parloops.c (gen_parallel_loop): Update.
7254 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7255 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7256 * tree-ssa-loop-split.c (split_loop): Update.
7257 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7258 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7259 (vect_loop_versioning): Update.
7260 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7262 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7264 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7266 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7268 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7269 to keep profile consistent.
7271 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7273 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7274 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7275 * profile-count.h (max_safe_multiplier): Make unsigned.
7276 (profile_count::guessed_zero): New.
7278 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7280 * bb-reorder.c (fix_up_crossing_landing_pad,
7281 fix_crossing_conditional_branches): Use make_single_succ_edge
7282 to keep profile consistent.
7284 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7286 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7289 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7292 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7293 the right scopes, make sure cond_jump isn't preserved between multiple
7294 iterations. Search for fallthru edge whenever there are 3+ edges and
7295 use find_fallthru_edge for it.
7297 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7299 Patch by Alexander Monakov <amonakov@ispras.ru>
7300 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7301 probabilities consistently.
7303 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7305 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7306 emit_cmp_and_jump_insns.
7308 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7311 * tree-inline.c (expand_call_inline): Combine profile statuses.
7313 2017-06-30 Andrew Pinski <apinski@cavium.com>
7315 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7316 fold_stmt returned true.
7318 2017-06-30 Nathan Sidwell <nathan@acm.org>
7320 * ggc.h (empty_string): Delete.
7321 * cfgexpand.c (expand_asm_stmt): Use plain "".
7322 * optabs.c (expand_asm_memory_barrier): Likewise.
7323 * stringpool.c (empty_string): Delete.
7324 (digit_vector, digit_string): Delete.
7325 (ggc_alloc_string): Use plain "", don't optimize single digit
7326 strings. Use ggc_alloc_atomic.
7328 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7330 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7331 comparison set and one other set, use the cost of the non-comparison
7334 2017-06-30 Nathan Sidwell <nathan@acm.org>
7336 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7339 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7341 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7342 loops. Remove now unneeded calls to gimple_switch_set_label() that
7343 just set removed labels to NULL_TREE.
7345 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7347 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7348 (set_range_info): ...here. Only call set_range_info_raw if domain
7350 (set_nonzero_bits): Call set_range_info_raw.
7351 * tree-ssanames.h (set_range_info_raw): New.
7353 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7356 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7357 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7358 of nonimmediate_operand and <store_mask_constraint> instead of m
7359 for the input operand. For V8FI iterator, always split if input
7360 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7361 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7362 <store_mask_predicate> instead of register_operand and
7363 <store_mask_constraint> instead of v for the input operand. Make
7364 sure both operands aren't MEMs for if not <mask_applied>.
7366 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7368 * lto-wrapper.c (copy_file) Close both file descriptors before
7371 2017-06-30 Martin Liska <mliska@suse.cz>
7374 * multiple_target.c (create_dispatcher_calls): Make ifunc
7375 also for function that don't have calls or are not referenced.
7377 2017-06-30 Richard Biener <rguenther@suse.de>
7379 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7380 analyze the first scalar stmt. Move vector type computation
7381 for the BB case here from ...
7382 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7383 live operation processing in the SLP case properly.
7385 2017-06-30 Richard Biener <rguenther@suse.de>
7387 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7389 2017-06-30 Martin Liska <mliska@suse.cz>
7392 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7393 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7395 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7397 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7400 2017-06-30 Richard Biener <rguenther@suse.de>
7402 PR tree-optimization/81249
7403 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7404 condition reduction result to original scalar type.
7406 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7408 * profile-count.h (enum profile_quality): Fix typos and whitespace
7411 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7413 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7414 type for branch probabilities.
7416 2017-06-29 Julian Brown <julian@codesourcery.com>
7417 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7419 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7420 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7421 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7422 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7424 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7426 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7427 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7428 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7429 CC usage from generic code to here.
7430 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7431 CC usage into the target macros.
7433 2017-06-29 Maya Rashish <coypu@sdf.org>
7435 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7438 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7440 * arm/arm-builtins.c: Include profile-count.h
7441 * except.c (sjlj_emit_function_enter): Use
7442 profile_probability::unlikely.
7444 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7446 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7447 and tocrel_offset be pointer args rather than implicitly using
7449 (legitimate_constant_pool_address_p, rs6000_emit_move,
7450 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7451 tocrel_offset and use in toc_relative_expr_p call.
7452 (print_operand, print_operand_address): Use static tocrel_base_oac
7453 and tocrel_offset_oac.
7454 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7457 2017-06-29 Maya Rashish <coypu@sdf.org>
7459 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7461 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7463 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7464 objects, take into account only the alignment of 'op0' and 'mode1' if
7467 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7469 * ccmp.c (ccmp_tree_comparison_p): New function.
7470 (ccmp_candidate_p): Update to use above function.
7471 (get_compare_parts): New function.
7472 (expand_ccmp_next): Update to use new functions.
7473 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7475 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7476 take mode as argument.
7477 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7478 * expr.c (expand_expr_real_1): Pass mode as argument.
7480 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7482 * combine.c (combine_instructions): Print insns to dump_file, together
7485 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7487 * asan.c (asan_emit_stack_protection): Update.
7488 (create_cond_insert_point): Update.
7489 * auto-profile.c (afdo_propagate_circuit): Update.
7490 * basic-block.h (struct edge_def): Turn probability to
7491 profile_probability.
7492 (EDGE_FREQUENCY): Update.
7493 * bb-reorder.c (find_traces_1_round): Update.
7494 (better_edge_p): Update.
7495 (sanitize_hot_paths): Update.
7496 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7497 (make_single_succ_edge): Update.
7498 (check_bb_profile): Update.
7499 (dump_edge_info): Update.
7500 (update_bb_profile_for_threading): Update.
7501 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7502 probabilitycount to 0.
7503 * cfgbuild.c (compute_outgoing_frequencies): Update.
7504 * cfgcleanup.c (try_forward_edges): Update.
7505 (outgoing_edges_match): Update.
7506 (try_crossjump_to_edge): Update.
7507 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7508 (expand_gimple_tailcall): Update.
7509 (construct_init_block): Use make_single_succ_edge.
7510 (construct_exit_block): Use make_single_succ_edge.
7511 * cfghooks.c (verify_flow_info): Update.
7512 (redirect_edge_succ_nodup): Update.
7513 (split_edge): Update.
7514 (account_profile_record): Update.
7515 * cfgloopanal.c (single_likely_exit): Update.
7516 * cfgloopmanip.c (scale_loop_profile): Update.
7517 (set_zero_probability): Remove.
7518 (duplicate_loop_to_header_edge): Update.
7519 * cfgloopmanip.h (loop_version): Update prototype.
7520 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7521 (force_nonfallthru_and_redirect): Update.
7522 (update_br_prob_note): Update.
7523 (rtl_verify_edges): Update.
7524 (purge_dead_edges): Update.
7525 (rtl_lv_add_condition_to_bb): Update.
7526 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7527 * cgraphunit.c (init_lowered_empty_function): Update.
7528 (cgraph_node::expand_thunk): Update.
7529 * cilk-common.c: Include profile-count.h
7530 * dojump.c (inv): Remove.
7531 (jumpifnot): Update.
7532 (jumpifnot_1): Update.
7533 (do_jump_1): Update.
7535 (do_jump_by_parts_greater_rtx): Update.
7536 (do_compare_rtx_and_jump): Update.
7537 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7538 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7539 * dwarf2cfi.c: Include profile-count.h
7540 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7541 (sjlj_emit_dispatch_table): Likewise.
7542 * explow.c: Include profile-count.h
7543 * expmed.c (emit_store_flag_force): Update.
7544 (do_cmp_and_jump): Update.
7545 * expr.c (compare_by_pieces_d::generate): Update.
7546 (compare_by_pieces_d::finish_mode): Update.
7547 (emit_block_move_via_loop): Update.
7548 (store_expr_with_bounds): Update.
7549 (store_constructor): Update.
7550 (expand_expr_real_2): Update.
7551 (expand_expr_real_1): Update.
7552 * expr.h (try_casesi, try_tablejump): Update prototypes.
7553 * gimple-pretty-print.c (dump_probability): Update.
7554 (dump_profile): New.
7555 (dump_gimple_label): Update.
7556 (dump_gimple_bb_header): Update.
7557 * graph.c (draw_cfg_node_succ_edges): Update.
7558 * hsa-gen.c (convert_switch_statements): Update.
7559 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7560 (find_if_case_1): Update.
7561 (find_if_case_2): Update.
7562 * internal-fn.c (expand_arith_overflow_result_store): Update.
7563 (expand_addsub_overflow): Update.
7564 (expand_neg_overflow): Update.
7565 (expand_mul_overflow): Update.
7566 (expand_vector_ubsan_overflow): Update.
7567 * ipa-cp.c (good_cloning_opportunity_p): Update.
7568 * ipa-split.c (split_function): Use make_single_succ_edge.
7569 * ipa-utils.c (ipa_merge_profiles): Update.
7570 * loop-doloop.c (add_test): Update.
7571 (doloop_modify): Update.
7572 * loop-unroll.c (compare_and_jump_seq): Update.
7573 (unroll_loop_runtime_iterations): Update.
7574 * lra-constraints.c (lra_inheritance): Update.
7575 * lto-streamer-in.c (input_cfg): Update.
7576 * lto-streamer-out.c (output_cfg): Update.
7577 * mcf.c (adjust_cfg_counts): Update.
7578 * modulo-sched.c (sms_schedule): Update.
7579 * omp-expand.c (expand_omp_for_init_counts): Update.
7580 (extract_omp_for_update_vars): Update.
7581 (expand_omp_ordered_sink): Update.
7582 (expand_omp_for_ordered_loops): Update.
7583 (expand_omp_for_generic): Update.
7584 (expand_omp_for_static_nochunk): Update.
7585 (expand_omp_for_static_chunk): Update.
7586 (expand_cilk_for): Update.
7587 (expand_omp_simd): Update.
7588 (expand_omp_taskloop_for_outer): Update.
7589 (expand_omp_taskloop_for_inner): Update.
7590 * omp-simd-clone.c (simd_clone_adjust): Update.
7591 * optabs.c (expand_doubleword_shift): Update.
7592 (expand_abs): Update.
7593 (emit_cmp_and_jump_insn_1): Update.
7594 (expand_compare_and_swap_loop): Update.
7595 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7596 * predict.c (predictable_edge_p): Update.
7597 (edge_probability_reliable_p): Update.
7598 (set_even_probabilities): Update.
7599 (combine_predictions_for_insn): Update.
7600 (combine_predictions_for_bb): Update.
7601 (propagate_freq): Update.
7602 (estimate_bb_frequencies): Update.
7603 (force_edge_cold): Update.
7604 * profile-count.c (profile_count::dump): Add missing space into dump.
7605 (profile_count::debug): Add newline.
7606 (profile_count::differs_from_p): Explicitly convert to unsigned.
7607 (profile_count::stream_in): Update.
7608 (profile_probability::dump): New member function.
7609 (profile_probability::debug): New member function.
7610 (profile_probability::differs_from_p): New member function.
7611 (profile_probability::differs_lot_from_p): New member function.
7612 (profile_probability::stream_in): New member function.
7613 (profile_probability::stream_out): New member function.
7614 * profile-count.h (profile_count_quality): Rename to ...
7615 (profile_quality): ... this one.
7616 (profile_probability): New.
7617 (profile_count): Update.
7618 * profile.c (compute_branch_probabilities): Update.
7619 * recog.c (peep2_attempt): Update.
7620 * sched-ebb.c (schedule_ebbs): Update.
7621 * sched-rgn.c (find_single_block_region): Update.
7622 (compute_dom_prob_ps): Update.
7623 (schedule_region): Update.
7624 * sel-sched-ir.c (compute_succs_info): Update.
7625 * stmt.c (struct case_node): Update.
7626 (do_jump_if_equal): Update.
7627 (get_outgoing_edge_probs): Update.
7628 (conditional_probability): Update.
7629 (emit_case_dispatch_table): Update.
7630 (expand_case): Update.
7631 (expand_sjlj_dispatch_table): Update.
7632 (emit_case_nodes): Update.
7633 * targhooks.c: Update.
7634 * tracer.c (better_p): Update.
7635 (find_best_successor): Update.
7636 * trans-mem.c (expand_transaction): Update.
7637 * tree-call-cdce.c: Update.
7638 * tree-cfg.c (gimple_split_edge): Upate.
7639 (move_sese_region_to_fn): Upate.
7640 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7641 * tree-eh.c (lower_resx): Upate.
7642 (cleanup_empty_eh_move_lp): Upate.
7643 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7644 * tree-inline.c (copy_edges_for_bb): Update.
7645 (copy_cfg_body): Update.
7646 * tree-parloops.c (gen_parallel_loop): Update.
7647 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7648 (gimple_gen_time_profiler): Update.
7649 * tree-ssa-dce.c (remove_dead_stmt): Update.
7650 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7651 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7652 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7653 (unloop_loops): Update.
7654 (try_peel_loop): Update.
7655 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7656 * tree-ssa-loop-split.c (connect_loops): Update.
7657 (split_loop): Update.
7658 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7659 (hoist_guard): Update.
7660 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7661 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7662 (value_replacement): Update.
7663 * tree-ssa-reassoc.c (branch_fixup): Update.
7664 * tree-ssa-tail-merge.c (replace_block_by): Update.
7665 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7666 (create_edge_and_update_destination_phis): Update.
7667 (compute_path_counts): Update.
7668 (recompute_probabilities): Update.
7669 (update_joiner_offpath_counts): Update.
7670 (freqs_to_counts_path): Update.
7671 (duplicate_thread_path): Update.
7672 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7673 (struct switch_conv_info): Update.
7674 (gen_inbound_check): Update.
7675 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7676 (vect_do_peeling): Update.
7677 (vect_loop_versioning): Update.
7678 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7679 (optimize_mask_stores): Update.
7680 * ubsan.c (ubsan_expand_null_ifn): Update.
7681 * value-prof.c (gimple_divmod_fixed_value): Update.
7682 (gimple_divmod_fixed_value_transform): Update.
7683 (gimple_mod_pow2): Update.
7684 (gimple_mod_pow2_value_transform): Update.
7685 (gimple_mod_subtract): Update.
7686 (gimple_mod_subtract_transform): Update.
7687 (gimple_ic): Update.
7688 (gimple_stringop_fixed_value): Update.
7689 (gimple_stringops_transform): Update.
7690 * value-prof.h: Update.
7692 2017-06-29 Carl Love <cel@us.ibm.com>
7694 * config/rs6000/rs6000-c.c: Add support for built-in functions
7695 vector signed int vec_signed (vector float);
7696 vector signed long long vec_signed (vector double);
7697 vector signed int vec_signed2 (vector double, vector double);
7698 vector signed int vec_signede (vector double);
7699 vector signed int vec_signedo (vector double);
7700 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
7701 instruction generator.
7702 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7703 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
7704 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
7706 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
7707 vunsignede_v2df): Add define_expands.
7708 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
7709 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
7710 VEC_UNSIGNEDO): Add definitions.
7711 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7712 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
7713 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
7714 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
7715 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
7716 * config/rs6000/altivec.h (vec_signed, vec_signed2,
7717 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
7718 vec_unsignede, vec_unsignedo): Add builtin defines.
7719 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
7721 * doc/extend.texi: Update the built-in documentation file for the
7722 new built-in functions.
7724 2017-06-29 Richard Biener <rguenther@suse.de>
7726 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
7727 reduction chains to LOOP_VINFO_REDUCTIONS.
7728 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
7729 SLP reductions after processing reduction chains.
7731 2017-06-29 Nathan Sidwell <nathan@acm.org>
7733 * builtins.c (fold_builtin_FUNCTION): Use
7734 lang_hooks.decl_printable_name.
7736 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
7739 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
7740 with only one label.
7741 * stmt.c (expand_case): Assert NCASES is greater than one.
7743 2017-06-29 Richard Biener <rguenther@suse.de>
7745 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
7747 (group_case_labels): Likewise.
7748 (find_taken_edge): Push sanity checking on val to workers...
7749 (find_taken_edge_cond_expr): ... here
7750 (find_taken_edge_switch_expr): ... and here, handle cases
7751 with just a default label.
7752 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
7753 (group_case_labels): Likewise.
7754 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
7755 group_case_labels does anything cleanup the CFG again.
7757 2017-06-29 Bin Cheng <bin.cheng@arm.com>
7759 PR tree-optimization/81196
7760 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
7761 exit condition comparing two IVs.
7763 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
7765 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
7766 profile to the dummy entry at the end of the list of architectures.
7767 * config/arm/arm-cpu-cdata.h: Regenerated.
7769 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7770 Michael Collison <michael.collison@arm.com>
7773 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
7775 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
7776 (*aarch64_reg_<mode>3_minus_mask): New pattern.
7777 (*aarch64_<optab>_reg_di3_mask2): New pattern.
7778 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
7779 of shift when the shift amount is masked with constant equal to
7780 the size of the mode.
7781 * config/aarch64/predicates.md (subreg_lowpart_operator): New
7784 2017-06-29 Martin Liska <mliska@suse.cz>
7786 * config/i386/i386.opt: Change range from [1,5] to [0,5].
7788 2017-06-29 Yury Gribov <tetra2005@gmail.com>
7791 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
7794 (edge_growth_cache_entry::edge_growth_cache_entry): New
7796 (reset_edge_growth_cache): Update to use constructor.
7798 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7800 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
7801 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7802 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7804 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
7806 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
7807 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
7809 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
7811 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
7812 (*-linux-uclibc*): Add t-uclibc tmake_file.
7813 * config/t-musl: New.
7814 * config/t-uclibc: New.
7816 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
7818 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
7820 (gen_comm_data): Emit architectural setting of arch_prof.
7821 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
7823 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
7824 (armv8-m.base, armv8-m.main): Likewise.
7825 * arm-protos.h (arm_build_target): Add profile field.
7826 (arch_option): Likewise.
7827 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
7829 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
7830 arm_active_target.profile.
7832 2017-06-28 Richard Biener <rguenther@suse.de>
7835 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
7836 TYPE_OVERFLOW_WRAPS.
7837 * match.pd (negate_expr_p): Likewise.
7838 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
7839 fold_build2, not fold_binary.
7841 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7843 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
7844 Convert memory address to Pmode.
7845 (aarch64_print_operand): Assert MEM operands are always Pmode.
7847 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7850 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
7851 Remove redundant if.
7852 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
7853 * config/arm/aarch-common-protos.h
7854 (aarch_forward_to_shift_is_not_shifted_re): Remove.
7855 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
7857 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
7860 * multiple_target.c (create_dispatcher_calls): Set the default
7861 clone to be static, not public.
7863 2017-06-28 Richard Biener <rguenther@suse.de>
7865 * tree-vect-loop.c (vectorizable_reduction): Move special
7866 cond reduction IV var creation ...
7867 (vect_create_epilog_for_reduction): ... here. Remove induction_index
7868 parameter. Use STMT_VINFO_VECTYPE.
7869 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
7872 2017-06-28 Martin Liska <mliska@suse.cz>
7875 * ipa-visibility.c (non_local_p): Handle visibility.
7877 2017-06-28 Martin Liska <mliska@suse.cz>
7880 * common.opt: Add IntegerRange to various options.
7881 * opt-functions.awk (integer_range_info): New function.
7882 * optc-gen.awk: Add integer_range_info to cl_options struct.
7883 * opts-common.c (decode_cmdline_option): Handle
7884 CL_ERR_INT_RANGE_ARG.
7885 (cmdline_handle_error): Likewise.
7886 * opts.c (print_filtered_help): Show valid interval in
7887 when --help is provided.
7888 * opts.h (struct cl_option): Add range_min and range_max fields.
7889 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
7891 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
7893 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
7894 (x * C EQ/NE y * C): New transformation.
7896 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
7898 * genmultilib (combination_space): Accept '+' in option names.
7900 2017-06-28 Martin Liska <mliska@suse.cz>
7903 * asan.c (instrument_derefs): Bail out inner references
7904 that are hard register variables.
7906 2017-06-28 Jakub Jelinek <jakub@redhat.com>
7909 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
7910 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
7912 2017-06-28 Richard Biener <rguenther@suse.de>
7914 * tree-vectorizer.h (vect_get_vec_defs): Remove.
7915 (vect_get_slp_defs): Adjust.
7916 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
7918 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
7920 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7921 get_initial_defs_for_reduction instead of vect_get_vec_defs.
7922 (vectorizable_reduction): Adjust.
7923 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
7925 (vect_get_slp_defs): Likewise.
7926 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
7927 (vectorizable_bswap): Adjust.
7928 (vectorizable_call): Likewise.
7929 (vectorizable_conversion): Likewise.
7930 (vectorizable_assignment): Likewise.
7931 (vectorizable_shift): Likewise.
7932 (vectorizable_operation): Likewise.
7933 (vectorizable_store): Likewise.
7934 (vectorizable_condition): Likewise.
7935 (vectorizable_comparison): Likewise.
7937 2017-06-28 Michael Collison <michael.collison@arm.com>
7940 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
7942 (arm_gen_movmemqi): Removed unused variable 'i'.
7943 Convert 'for' loop into 'while' loop.
7944 (arm_expand_prologue): Remove last unnecessary set of insn.
7945 (thumb_pop): Remove unused variable 'pushed_words'.
7946 (thumb_exit): Remove last unnecessary set of regs_to_pop.
7948 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7950 * config/s390/predicates.md: Use s390_rel_address_ok_p.
7951 * config/s390/s390-protos.h: Add prototype of
7952 s390_rel_address_ok_p.
7953 * config/s390/s390.c (s390_got_symbol): New function.
7954 (s390_rel_address_ok_p): New function.
7955 (legitimize_pic_address): Use s390_rel_address_ok_p.
7956 (s390_load_got): Use s390_got_symbol.
7957 (s390_option_override): Issue error if
7958 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
7959 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
7961 * config/s390/s390.opt: New option mpic-data-is-text-relative.
7963 2017-06-27 Andrew Pinski <apinski@cavium.com>
7965 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
7966 (X * copysign (1.0, X)): New pattern.
7967 (X * copysign (1.0, -X)): New pattern.
7968 (copysign (-1.0, CST)): New pattern.
7970 2017-06-27 Joseph Myers <joseph@codesourcery.com>
7972 * genmultilib (combination_space): Remove variable.
7973 Validate reuse rules against regular expression for any sequence
7974 of multilib options in any order.
7976 2017-06-27 Michael Collison <michael.collison@arm.com>
7978 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
7979 call aarch64_split_simd_combine.
7980 * (aarch64_combine_internal<mode>): Delete pattern.
7981 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
7982 Allow register and subreg operands.
7984 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7986 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
7987 specific need, just fallback on defaults.
7988 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
7990 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7991 Olivier Hainque <hainque@adacore.com>
7993 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
7995 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
7996 targets. Pick a default if no particular attempt applied.
7997 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8000 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8002 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8003 (x86_64-wrs-vxworks7): Likewise.
8005 2017-06-27 Marek Polacek <polacek@redhat.com>
8008 * ubsan.c (instrument_null): Check get_base_address's result for null.
8010 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8012 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8014 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8016 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8017 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8018 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8019 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8021 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8022 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8023 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8024 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8025 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8026 BUILT_IN_FEUPDATEENV): New builtins.
8027 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8028 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8029 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8030 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8032 (builtin_structptr_types): Adjust size.
8033 * tree.c (builtin_structptr_types): Add four entries.
8035 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8036 Olivier Hainque <hainque@adacore.com>
8038 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8039 (TLS_SYM): New local macro, forcing reference to __tls__ on
8040 link command lines for VxWorks 7 RTPs, triggering initialization
8042 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8043 OS features TLS support, true for RTPs on VxWorks 7.
8044 * config/vxworks.c (vxworks_override_options): Setup emutls
8047 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8049 * predict.c (test_prediction_value_range): Use -1U instead of -1
8050 to avoid narrowing conversion warning.
8051 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8052 to avoid narrowing conversion warning.
8053 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8055 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8057 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8059 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8060 64bit configurations.
8061 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8062 (SIZE_TYPE): Likewise.
8063 * config/vxworks.c (vxworks_emutls_var_fields): Use
8064 long_unsigned_type_node instead of unsigned_type_node as the offset
8065 field type, which is "pointer" mode in emutls.c.
8067 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8070 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8073 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8074 gimple_vuse copying separately from gimple_vdef copying.
8076 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8078 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8081 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8082 Olivier Hainque <hainque@adacore.com>
8084 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8085 for all vxworks7 targets.
8086 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8087 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8088 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8089 variations for VX6/VX7 and 32/64bits later on in ...
8090 (VXWORKS_LIB_SPEC): Leverage new macros.
8091 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8092 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8094 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8096 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8097 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8099 2017-06-26 Carl Love <cel@us.ibm.com>
8101 * config/rs6000/rs6000-c.c: Add support for built-in functions
8102 vector bool char vec_reve (vector bool char);
8103 vector signed char vec_reve (vector signed char);
8104 vector unsigned char vec_reve (vector unsigned char);
8105 vector bool int vec_reve (vector bool int);
8106 vector signed int vec_reve (vector signed int);
8107 vector unsigned int vec_reve (vector unsigned int);
8108 vector bool long long vec_reve (vector bool long long);
8109 vector signed long long vec_reve (vector signed long long);
8110 vector unsigned long long vec_reve (vector unsigned long long);
8111 vector bool short vec_reve (vector bool short);
8112 vector signed short vec_reve (vector signed short);
8113 vector double vec_reve (vector double);
8114 vector float vec_reve (vector float);
8115 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8116 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8117 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8118 (altivec_vreve): New pattern.
8119 * config/rs6000/altivec.h (vec_reve): New define.
8120 * doc/extend.texi (vec_rev): Update the built-in documentation file
8121 for the new built-in functions.
8123 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8125 PR tree-optimization/71815
8126 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8128 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8130 (slsr_process_phi): Likewise.
8131 (replace_uncond_cands_and_profitable_phis): Don't replace a
8132 multiply candidate with a stride of 1 (copy or cast).
8133 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8135 (lowest_cost_path): Likewise.
8136 (total_savings): Likewise.
8138 2017-06-26 Richard Biener <rguenther@suse.de>
8141 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8142 Use def_builtin_pure for all gather builtins.
8144 2017-06-26 Richard Biener <rguenther@suse.de>
8146 PR tree-optimization/81203
8147 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8150 2017-06-26 Marek Polacek <polacek@redhat.com>
8153 * doc/invoke.texi: Document -Wmultistatement-macros.
8155 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8157 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8158 arm_neon_ok_no_float_abi effective target.
8160 2017-06-26 Richard Biener <rguenther@suse.de>
8162 PR tree-optimization/80928
8163 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8164 (copy_bbs): Set BB_DUPLICATED flag early.
8165 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8167 (execute_on_shrinking_pred): Likewise.
8168 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8169 BB_DUPLICATED blocks.
8170 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8171 iterate over all PHIs considering removal of *gsi.
8173 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8175 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8178 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8180 * config/rs6000/rs6000-string.c: (expand_block_clear,
8181 do_load_for_compare, select_block_compare_mode,
8182 compute_current_alignment, expand_block_compare,
8183 expand_strncmp_align_check, expand_strn_compare,
8184 expand_block_move, rs6000_output_load_multiple)
8185 Move functions related to string/block move/compare
8187 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8188 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8189 for this function which is now used in two files.
8190 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8191 * config.gcc: Add rs6000-string.o to extra_objs for
8192 targets powerpc*-*-* and rs6000*-*-*.
8194 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8197 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8198 32-bit, since indexed is not valid for DImode.
8199 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8200 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8201 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8202 (define_peephole2 for Altivec d-form store): Likewise.
8205 * multiple_target.c (create_dispatcher_calls): Only create the
8206 dispatcher call if the function is the default clone of a
8209 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8212 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8213 a call, force the call to not be a tail call.
8215 2017-06-23 Jeff Law <law@redhat.com>
8217 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8220 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8222 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8223 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8224 handling for early expansion of vector shifts (sl,sr,sra,rl).
8225 (builtin_function_type): Add vector shift right instructions
8226 to the unsigned argument list.
8228 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8230 rtl-optimizatoin/79286
8231 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8232 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8233 trap. PIC register plus a const unspec without offset can never trap.
8235 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8237 * tree.h (builtin_structptr_type): New type.
8238 (builtin_structptr_types): Declare new array.
8239 * tree.c (builtin_structptr_types): New array.
8240 (free_lang_data, build_common_tree_nodes): Use it.
8242 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8245 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8248 2017-06-22 Matt Turner <mattst88@gmail.com>
8250 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8251 Lake models to skylake case. Assume skylake for unknown
8252 models with clflushopt.
8254 2017-06-22 Jeff Law <law@redhat.com>
8256 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8257 frame sizes that do not satisfy aarch64_uimm12_shift.
8259 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8261 * profile-count.h (apply_probability,
8262 apply_scale, probability_in): Fix checks for zero.
8264 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8266 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8267 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8269 2016-06-22 Richard Biener <rguenther@suse.de>
8271 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8272 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8273 REDUC_MAX_EXPR support.
8274 (vectorizable_reduction): Likewise.
8275 (vect_create_epilog_for_reduction): Likewise.
8277 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8279 * match.pd (A / (1 << B) -> A >> B): New.
8280 * generic-match-head.c: Include optabs-tree.h.
8281 * gimple-match-head.c: Likewise.
8282 * optabs-tree.h (target_supports_op_p): New.
8283 * optabs-tree.c (target_supports_op_p): New.
8285 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8287 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8288 $gcc_cv_ld --help output.
8289 (gcc_cv_ld_demangle): Likewise.
8290 (gcc_cv_ld_eh_frame_hdr): Likewise.
8291 (gcc_cv_ld_pie): Likewise.
8292 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8293 (gcc_cv_ld_buildid): Likewise.
8294 (gcc_cv_ld_sysroot): Likewise.
8295 (ld_bndplt_support): Likewise.
8296 (ld_pushpopstate_support): Likewise.
8297 * configure: Regenerate.
8298 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8300 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8303 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8304 operands indexes to avoid gap between operands and match_dups.
8306 2017-06-21 Andrew Pinski <apinski@cavium.com>
8308 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8309 Increment Arith_shift and Arith_shift_reg by 1.
8310 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8312 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8313 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8314 (aarch64_strip_extend): Add new argument and test for it.
8315 (aarch64_cheap_mult_shift_p): New function.
8316 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8317 add a cost if it is true.
8318 Update calls to aarch64_strip_extend.
8319 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8321 2017-06-21 Andrew Pinski <apinski@cavium.com>
8323 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8325 (thunderxt88): Likewise.
8326 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8327 (thunderx_prefetch_tune): New variable.
8328 (thunderx2t99_prefetch_tune): Update for the correct values.
8329 (thunderxt88_tunings): New variable.
8330 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8331 generic_prefetch_tune.
8332 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8334 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8336 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8337 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8338 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8339 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8340 (aarch64_atomic_cas<mode>, GPI): Likewise.
8342 2017-06-21 Martin Liska <mliska@suse.cz>
8344 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8345 statements on cold and hot labels.
8346 * predict.c (tree_estimate_probability_bb): Remove the
8347 prediction from this place.
8349 2017-06-21 Martin Liska <mliska@suse.cz>
8351 PR tree-optimization/79489
8352 * gimplify.c (maybe_add_early_return_predict_stmt): New
8354 (gimplify_return_expr): Call the function.
8355 * predict.c (tree_estimate_probability_bb): Remove handling
8357 * predict.def: Update comment about early return predictor.
8358 * gimple-predict.h (is_gimple_predict): New function.
8359 * predict.def: Change default value of early return to 66.
8360 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8362 * passes.def: Put pass_strip_predict_hints to the beginning of
8365 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8367 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8368 FUNCTION_DECL declarations.
8369 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8371 (dwaf2out_decl): Likewise.
8372 * godump.c (go_early_global_decl): Skip call to the real debug hook
8373 for FUNCTION_DECL declarations.
8374 * passes.c (rest_of_decl_compilation): Skip call to the
8375 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8376 -fdump-go-spec is passed.
8378 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8380 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8381 Reorder for compactness.
8382 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8383 (def_builtin_pure, def_builtin_pure2): New functions.
8384 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8386 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8388 * match.pd (nop_convert): New predicate.
8389 ((A +- CST1) +- CST2): Allow some NOP conversions.
8391 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8394 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8395 with ctors/dtors if GOVD_SHARED is set.
8397 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8399 * config/aarch64/aarch64.md (movti_aarch64):
8400 Emit mov rather than orr.
8401 (movtf_aarch64): Likewise.
8402 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8403 Emit mov rather than orr.
8405 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8407 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8408 Swap alternatives, make integer dup more expensive.
8410 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8412 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8413 Return true for non-tls symbols.
8415 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8417 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8418 (cortex-a75): Likewise.
8419 (cortex-a75.cortex-a55): Likewise.
8420 * config/aarch64/aarch64-tune.md: Regenerate.
8421 * doc/invoke.texi (-mtune): Document new values for -mtune.
8423 2017-06-21 Tom de Vries <tom@codesourcery.com>
8425 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8427 (Effective-Target Keywords, Other attributes): Suggest using
8428 dg-add-options stack_size feature to get stack limit in stack_size
8429 effective target documentation.
8431 2017-06-21 Julian Brown <julian@codesourcery.com>
8432 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8434 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8435 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8436 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8438 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8439 attribute type list for neon_multiply.
8440 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8441 attribute type list for neon_multiply.
8442 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8443 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8444 attribute type list for neon_multiply.
8445 * config/arm/types.md (crypto_pmull): Add.
8446 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8447 attribute type list.
8449 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8451 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8454 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8456 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8457 to make sure not to dereference a NULL cost_classes_ptr pointer.
8459 2017-06-20 Carl Love <cel@us.ibm.com>
8461 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8462 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8463 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8464 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8465 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8466 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8467 VMULOSW): New enum "unspec" values.
8468 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8469 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8470 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8471 altivec_vmulosw): New patterns.
8472 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8473 VMULOSW): Add definitions.
8475 2017-06-20 Julia Koval <julia.koval@intel.com>
8477 * config/i386/i386.c: Fix rounding expand for new pattern.
8478 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8480 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8482 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8483 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8485 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8487 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8490 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8492 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8493 architecture, then by implementer ID.
8494 * config/aarch64/aarch64-tune.md: Regenerate.
8496 2017-06-20 Richard Biener <rguenther@suse.de>
8499 * fold-const.c (split_tree): Fold to type before negating.
8501 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8503 * diagnostic-show-locus.c
8504 (selftest::test_fixit_deletion_affecting_newline): New function.
8505 (selftest::diagnostic_show_locus_c_tests): Call it.
8507 2017-06-20 Andreas Schwab <schwab@suse.de>
8510 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8513 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8515 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8516 __ARM_FEATURE_COPROC according to support.
8518 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8520 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8521 Rewritten to avoid overflow for > 32-bit pointers.
8524 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8525 by removing enum keyword.
8526 (ubsan_type_descriptor): Likewise. Formatting fix.
8529 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8530 splitter): Require TARGET_SSE2 in the condition.
8532 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8535 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8536 for doing vector set of SFmode on ISA 3.0.
8537 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8538 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8540 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8541 SFmode value into a V4SF variable that was extracted from another
8542 V4SF variable without converting the element to double precision
8543 and back to single precision vector format.
8544 (vsx_insert_extract_v4sf_p9_2): Likewise.
8546 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8548 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8549 in UWHI to avoid undefined overflow.
8552 * ubsan.h (enum ubsan_encode_value_phase): New.
8553 (ubsan_encode_value): Change second argument to
8554 enum ubsan_encode_value_phase with default value of
8555 UBSAN_ENCODE_VALUE_GENERIC.
8556 * ubsan.c (ubsan_encode_value): Change second argument to
8557 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8558 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8559 create_tmp_var_raw instead of create_tmp_var and use a
8561 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8562 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8563 ubsan_encode_value callers.
8566 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8567 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8568 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8570 2017-06-19 Richard Biener <rguenther@suse.de>
8573 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8574 estimates if we changed anything.
8576 2017-06-19 Richard Biener <rguenther@suse.de>
8578 PR tree-optimization/80887
8579 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8580 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8581 simplified lookups, then reset mprts_hook.
8582 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8584 (try_to_simplify): Likewise.
8586 2017-06-19 Martin Liska <mliska@suse.cz>
8589 * gimplify.c (gimplify_switch_expr):
8590 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8592 2017-06-19 Martin Liska <mliska@suse.cz>
8594 * doc/install.texi: Document that PGO runs in 4 stages.
8596 2017-06-19 Martin Liska <mliska@suse.cz>
8599 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8600 to dispatcher function name.
8601 * multiple_target.c (replace_function_decl): New function.
8602 (create_dispatcher_calls): Redirect both edges and references.
8604 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8606 * profile-count.c (profile_count::dump): Dump quality.
8607 (profile_count::differs_from_p): Update for unsigned val.
8608 * profile-count.h (profile_count_quality): New enum.
8609 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8611 2017-06-19 Richard Biener <rguenther@suse.de>
8613 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8614 struct function as arg.
8615 (estimate_numbers_of_iterations): Export overload with loop arg.
8616 (free_numbers_of_iterations_estimates_loop): Use an overload of
8617 free_numbers_of_iterations_estimates instead.
8618 * tree-cfg.c (remove_bb): Adjust.
8619 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8620 * tree-parloops.c (gen_parallel_loop): Likewise.
8621 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8623 (tree_unroll_loops_completely): Likewise.
8624 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8625 Use an overload instead and export.
8626 (estimated_loop_iterations): Adjust.
8627 (max_loop_iterations): Likewise.
8628 (likely_max_loop_iterations): Likewise.
8629 (estimate_numbers_of_iterations): Take struct function as arg
8631 (loop_exits_before_overflow): Adjust.
8632 (free_numbers_of_iterations_estimates_loop): Use an overload.
8633 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8634 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8636 2017-06-19 Richard Biener <rguenther@suse.de>
8639 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8640 RANGE_EXPR conservatively.
8642 2017-06-16 Carl Love <cel@us.ibm.com>
8644 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8645 definitions for vec_float, vec_float2, vec_floato,
8646 vec_floate built-ins.
8647 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8648 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8650 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8651 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8652 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8653 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8654 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8655 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8656 vec_floato): Add builtin defines.
8657 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8658 Update the built-in documentation file for the new built-in
8661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8663 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8664 (mthumb): Mark as the negative of -marm.
8666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8668 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8670 (ARM Options, -mtune): Document that this accepts the same
8671 extension options as -mcpu.
8672 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8676 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8677 permitted extensions.
8679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8681 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8682 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8683 (armv8-m.main): Add option +nodsp.
8684 * config/arm/arm-cpu-cdata.h: Regenerated.
8686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8688 * config/arm/t-fuchsia: New file.
8689 * config.gcc (arm*-*-fuchsia*): Use it.
8691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8693 * config/arm/t-symbian: Rewrite for new option infrastructure.
8695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8697 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
8698 (MULTILIB_REQUIRED): Likewise.
8700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8702 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
8703 (MULTILIB_RESUE): Likewise.
8704 (MULTILIB_MATCHES): Likewise.
8705 (MULTLIB_REQUIRED): Likewise.
8707 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8709 * config/arm/t-rtems: Rewrite for new option framework.
8711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8713 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
8714 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
8715 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
8716 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
8717 * config/arm/t-multilib: ... here.
8718 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
8719 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
8720 armv7-a and armv8*-a when A-profile libraries have not been built.
8721 * config/arm/t-rmprofile: Rewrite.
8723 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8725 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
8726 with a backslash. Remove the backslash after substituting unescaped
8728 * doc/fragments.texi (MULTILIB_REUSE): Document it.
8730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8732 * config.gcc: (arm*-*-*): When building a-profile libraries, force
8733 the driver to pass through the default setting of -mfloat-abi.
8734 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
8736 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
8737 (all_feat_combs): New rule.
8738 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
8740 * config/arm/t-aprofile: Rewrite.
8742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8744 * config/arm/arm.h (FPUTYPE_AUTO): Define.
8745 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
8746 fpu is not specified by the user/command-line.
8747 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
8748 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
8749 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
8750 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
8751 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
8752 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
8754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8756 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
8757 * config/arm/t-arm-elf: Rewritten.
8759 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8761 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
8762 have some floating-point instructions.
8763 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
8764 (TARGET_MAYBE_HARD_FLOAT): New macro.
8765 * config/arm/arm-builtins.c (arm_init_builtins): Use
8766 TARGET_MAYBE_HARD_FLOAT.
8767 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
8769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8771 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
8772 (configargs.h): Include it.
8773 (arm_print_hint_for_fpu_option): New function.
8774 (arm_parse_fpu_option): New function.
8775 (candidate_extension): New class.
8776 (arm_canon_for_multilib): New function.
8777 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
8778 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
8779 (ARCH_CANONICAL_SPECS): New macro.
8780 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
8782 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8784 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
8785 are set after handling multilib fragments. Set target_cpu_default2
8788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8790 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
8792 (arm*-*-*): Set target_cpu_default2 to a quoted string.
8793 * config/arm/parsecpu.awk (check_cpu): Validate any extension
8795 (check_arch): Likewise.
8796 * config/arm/arm.c (arm_configure_build_target): Handle
8797 TARGET_CPU_DEFAULT being a string constant. Scan any feature
8798 options in the default.
8800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8802 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
8803 when an option is an alias of another.
8804 * config/arm/parsecpu.awk (optalias): New parser token.
8805 (gen_comm_data): Mark non-alias options as such. Emit entries
8806 for extension aliases.
8807 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
8808 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
8809 (armv6kz, armv6zk, armv6t2): Likewise.
8810 (armv7): Make vfpv3-d16 an alias.
8811 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
8813 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
8814 Sort in canonical order.
8815 (armv8-a): Sort in canonical order.
8816 (armv8.1-a, armv8.2-a): Likewise.
8817 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
8819 (cortex-a9): Sort in canonical order.
8820 * config/arm/arm.c (selftests.h): Include it.
8821 (arm_test_cpu_arch_data): New function.
8822 (arm_run_self_tests): New function.
8823 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
8824 (targetm): Move declaration to the end of the file.
8825 * arm-cpu-cdata.h: Regenerated.
8827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8829 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
8830 call to target_mode_check describing the type of option passed.
8831 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
8832 (arm_target_thumb_only): Use arm_parse_arch_option_name or
8833 arm_parse_cpu_option_name to match parameters against list of
8835 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
8836 arm_arch_core_flags data structure.
8837 * config/arm/arm-cpu_cdata.h: Regenerated.
8839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8841 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
8843 (arm_print_hint_for_cpu_option): Likewise.
8844 (arm_print_hint_for_arch_option): Likewise.
8845 (arm_parse_cpu_option_name): Likewise.
8846 (arm_parse_arch_option_name): Likewise.
8847 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
8848 of entries in the all_fpus list.
8849 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
8850 (arm_parse_cpu_option_name): Declare.
8851 (arm_parse_arch_option_name): Declare.
8852 (arm_parse_option_features): Declare.
8853 (arm_intialize_isa): Declare.
8854 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
8856 (gen_comm_data): ... here. Make definitions non-static.
8857 * config/arm/arm-cpu-data.h: Regenerated.
8858 * config/arm/arm-cpu-cdata.h: Regenerated.
8860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8862 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
8863 (cpu_arch_extension): New structure.
8864 (cpu_arch_option, arch_option, cpu_option): New structures.
8865 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
8867 (gen_data): Generate new format data tables.
8868 * config/arm/arm.c (cpu_tune): New structure.
8869 (cpu_option, processors): Delete.
8870 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
8871 (arm_print_hint_for_cpu_option): ... this and ...
8872 (arm_print_hint_for_arch_option): ... this.
8873 (arm_parse_arch_cpu_name): Delete. Replace with ...
8874 (arm_parse_cpu_option_name): ... this and ...
8875 (arm_parse_arch_option_name): ... this.
8876 (arm_unrecognized_feature): Change type of target parameter to
8878 (arm_parse_arch_cpu_features): Delete. Replace with ...
8879 (arm_parse_option_features): ... this.
8880 (arm_configure_build_target): Rework to use new configuration data
8882 (arm_print_tune_info): Rework for new configuration data tables.
8883 * config/arm/arm-cpu-data.h: Regenerated.
8884 * config/arm/arm-cpu.h: Regenerated.
8886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8888 * Makefile.in (OBJS): Move sbitmap.o from here ...
8889 (OBJS-libcommon): ... to here.
8891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8893 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
8894 (ISA_ALL_CRYPTO): New macro.
8895 (ISA_ALL_SIMD): New macro
8896 (ISA_ALL_FP): New macro.
8897 * config/arm/arm.c (fpu_bitlist): Update initializer.
8898 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
8900 (arm9e): Add fpu. Add option for nofp
8901 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
8902 (arm926ej-s, arm1026ej-s): Likewise.
8903 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
8904 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
8905 neon-fp16, neon-vfpv4, nofp and nosimd.
8906 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
8907 (cortex-a8): Add fpu. Add option for nofp.
8908 (cortex-a9): Add fpu. Add options for nosimd and nofp.
8909 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
8910 (cortex-r4f): Add fpu.
8911 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
8912 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
8914 (cortex-r8): Likewise.
8915 (cortex-m4): Add fpu. Add option for nofp.
8916 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
8917 (cortex-a17.cortex-a7): Likewise.
8918 (cortex-a32): Add fpu. Add options for crypto and nofp.
8919 (cortex-a35, cortex-a53): Likewise.
8920 (cortex-a57): Add fpu. Add option for crypto.
8921 (cortex-a72, cortex-a73): Likewise.
8922 (exynos-m1): Likewise.
8923 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
8924 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
8925 (cortex-m33): Add fpu. Add option for nofp.
8926 * config/arm/arm-cpu-cdata.h: Regenerated
8927 * config/arm/arm-cpu-data.h: Regenerated.
8929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8931 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
8932 (armv5te, armv5tej): Likewise.
8933 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
8934 (armv7): Add options fp and vfpv3-d16.
8935 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
8936 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
8938 (armv7ve): Likewise.
8939 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
8940 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
8941 (armv8-a): Add nocrypto option.
8942 (armv8.1-a, armv8.2-a): Likewise.
8943 (armv8-m.main): add options fp, fp.dp and nofp.
8945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8947 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
8949 (armv8-a+crc): Delete.
8950 (armv8.1-a): Add options simd, crypto and nofp.
8951 (armv8.2-a): Add options fp16, simd, crypto and nofp.
8952 (armv8.2-a+fp16): Delete.
8953 (armv8-m.main): Add option dsp.
8954 (armv8-m.main+dsp): Delete.
8955 (cortex-a8): Add fpu. Add nofp option.
8956 (cortex-a9): Add fpu. Add nofp and nosimd options.
8957 * config/arm/parsecpu.awk (gen_data): Generate option tables and
8958 link to main cpu and architecture data structures.
8959 (gen_comm_data): Only put isa attributes from the main architecture
8961 (option): New statement for architecture and CPU entries.
8962 * arm.c (struct cpu_option): New structure.
8963 (struct processors): Add entry for options.
8964 (arm_unrecognized_feature): New function.
8965 (arm_parse_arch_cpu_name): Ignore any characters after the first
8967 (arm_parse_arch_cpu_feature): New function.
8968 (arm_configure_build_target): Separate out any CPU and architecture
8969 features and parse separately. Don't error out if -mfpu=auto is
8970 used with only an architecture string.
8971 (arm_print_asm_arch_directives): New function.
8972 (arm_file_start): Call it.
8973 * config/arm/arm-cpu-cdata.h: Regenerated.
8974 * config/arm/arm-cpu-data.h: Likewise.
8975 * config/arm/arm-tables.opt: Likewise.
8977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8979 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
8980 assembler when it is not -mfpu=auto.
8982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8984 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
8985 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
8986 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
8987 (ASM_CPU_SPEC): Rewrite.
8988 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
8989 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
8990 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
8991 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
8992 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
8993 copied string is NUL-terminated. Also strip any characters prefixed
8995 (arm_rewrite_selected_arch): New function.
8996 (arm_rewrite_march): New function.
8998 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9000 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9001 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9002 (march, mcpu, mtune): Convert to string-based options.
9003 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9004 (arm_parse_arch_cpu_name): New function.
9005 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9006 identify selected architecture or CPU.
9007 (arm_option_save): New function.
9008 (TARGET_OPTION_SAVE): Redefine.
9009 (arm_option_restore): Restore string options.
9010 (arm_option_print): Print string options.
9012 2017-06-16 Martin Sebor <msebor@redhat.com>
9014 PR tree-optimization/80933
9015 PR tree-optimization/80934
9016 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9017 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9018 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9019 (gimple_fold_builtin): Call them.
9021 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9023 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9024 as unlikely; update profile.
9026 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9028 * predict.c (force_edge_cold): Handle declaring edges impossible
9031 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9033 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9035 (try_unroll_loop_completely): Fix reporting.
9037 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9039 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9041 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9044 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9045 if given a non-constant argument for an intrinsic which requires a
9048 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9050 * profile.c (compare_freqs): New function.
9051 (branch_prob): Sort edge list.
9052 (find_spanning_tree): Assume that the list is priority sorted.
9054 2017-06-16 Richard Biener <rguenther@suse.de>
9056 PR tree-optimization/81090
9057 * passes.def (pass_record_bounds): Remove.
9058 * tree-pass.h (make_pass_record_bounds): Likewise.
9059 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9060 make_pass_record_bounds): Likewise.
9061 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9062 not free niter estimates at the beginning but at the end.
9063 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9065 2017-06-16 Richard Biener <rguenther@suse.de>
9067 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9068 initializer to workaround ICE in host GCC 4.8.
9070 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9072 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9074 (clone_inlined_nodes): Update.
9076 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9078 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9079 prefetch settings, and enable prefetching by default at -O3.
9081 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9083 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9084 Set flag_prefetch_loop_arrays according to tuning data.
9086 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9088 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9090 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9091 [Unrelated to main purpose of the patch] Place the pointer field last
9092 to enable type checking errors when tune structure are wrongly merged.
9093 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9094 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9095 (thunderx2t99_prefetch_tune): New tune constants.
9096 (tune_params *_tunings): Update all tunings (no functional change).
9097 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9098 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9099 from tunings structures.
9101 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9104 * ubsan.c (instrument_null): Add T argument, use it instead
9105 of computing it based on IS_LHS.
9106 (instrument_object_size): Likewise.
9107 (pass_ubsan::execute): Adjust instrument_null and
9108 instrument_object_size callers to pass gimple_get_lhs or
9109 gimple_assign_rhs1 result to it. Use instrument_null instead of
9110 calling get_base_address and instrument_mem_ref. Handle
9111 aggregate call arguments for object-size sanitization.
9113 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9115 PR tree-optimization/81089
9116 * tree-vrp.c (is_masked_range_test): Validate operands of
9119 2017-06-15 Martin Sebor <msebor@redhat.com>
9122 * dumpfile.c (dump_register): Avoid calling memset to initialize
9123 a class with a default ctor.
9124 * gcc.c (struct compiler): Remove const qualification.
9125 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9126 * hash-table.h: Ditto.
9127 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9129 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9130 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9132 * params.h (struct param_info): Make struct members non-const.
9133 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9134 with default initialization.
9135 * vec.h (vec_copy_construct, vec_default_construct): New helper
9137 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9138 with vec_copy_construct.
9139 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9140 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9141 * doc/invoke.texi (-Wclass-memaccess): Document.
9143 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9145 * emit-rtl.h (is_leaf): Update comment about local
9148 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9151 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9152 for a variable to have a section before checking if the section has a
9154 Set section to.persistent if persistent attribute is set.
9155 Warn if .persistent attribute is used on an automatic variable.
9157 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9159 PR rtl-optimization/80474
9160 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9162 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9164 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9167 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9169 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9170 (casesi): Emit bounds checking as RTL.
9171 (casesi_internal_mips16_<mode>): Remove bounds checking.
9173 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9175 * config/xtensa/xtensa.c (xtensa_option_override): Append
9176 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9177 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9178 xtensa_doloop_hooks): Define unconditionally.
9179 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9181 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9182 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9183 for it in xtensa_option_override.
9184 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9185 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9187 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9189 * doc/cppopts.texi: Document '-' special value to -MF.
9191 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9193 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9194 (cortex_a53_fconst): Likewise.
9195 (cortex_a53_fpmul): Likewise.
9196 (cortex_a53_f_load_64): Likewise.
9197 (cortex_a53_f_load_many): Likewise.
9198 (cortex_a53_advsimd_alu): Likewise.
9199 (cortex_a53_advsimd_alu_q): Likewise.
9200 (cortex_a53_advsimd_mul): Likewise.
9201 (cortex_a53_advsimd_mul_q): Likewise.
9202 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9203 Add missing fmul, r2f_cvt and fconst cases.
9205 2017-06-14 Richard Biener <rguenther@suse.de>
9208 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9210 (fold_binary_loc): When associating do not treat pre-existing
9211 TREE_OVERFLOW on literal constants as a reason to allow
9212 TREE_OVERFLOW on associated literal constants.
9214 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9216 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9217 (MASK_FEATURES): New macro.
9218 * config/sparc/sparc.c (sparc_option_override): Remove the special
9219 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9221 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9223 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9224 a division of 0 if non-call exceptions are enabled.
9226 2017-06-14 Andrew Pinski <apinski@cavium.com>
9227 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9230 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9231 Improve vector initialization code gen for only variable case.
9233 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9235 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9237 2017-06-14 Richard Biener <rguenther@suse.de>
9239 PR tree-optimization/81083
9240 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9243 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9245 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9246 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9247 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9248 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9249 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9250 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9252 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9254 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9255 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9257 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9259 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9261 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9263 * config/rs6000/t-rtems: Don't handle SPE.
9265 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9267 * config/rs6000/t-linux: Don't handle SPE.
9269 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9271 * config/rs6000/eabispe.h: Delete file.
9273 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9275 * config/rs6000/t-spe: Delete file.
9277 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9279 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9280 (rs6000_legitimate_offset_address_p): Return false for anything in
9281 V2SImode or V2SFmode.
9283 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9285 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9286 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9287 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9289 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9290 (rs6000_legitimate_offset_address_p): Ditto.
9291 (rs6000_emit_move): Ditto.
9292 (rs6000_init_builtins): Remove V4HI_type_node.
9294 2017-06-13 Martin Liska <mliska@suse.cz>
9297 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9298 (gate_asan): Likewise.
9299 * asan.h (asan_no_sanitize_address_p): Remove the function.
9300 (sanitize_flags_p): New function.
9301 * builtins.def: Fix coding style.
9302 * common.opt: Use renamed enum value.
9303 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9304 * doc/extend.texi: Document no_sanitize attribute.
9305 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9306 to SANITIZE_UNDEFINED_NONDEFAULT.
9307 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9308 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9309 Use sanitize_flags_p.
9310 * gimplify.c (gimplify_function_tree): Likewise.
9311 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9312 * opts.c (parse_no_sanitize_attribute): New function.
9313 (common_handle_option): Use renamed enum value.
9314 * opts.h (parse_no_sanitize_attribute): Declare.
9315 * tree.c (sanitize_flags_p): New function.
9316 * tree.h: Declared here.
9317 * tsan.c: Use sanitize_flags_p.
9318 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9319 (instrument_mem_ref): Likewise.
9320 (instrument_bool_enum_load): Likewise.
9321 (do_ubsan_in_current_function): Remove the function.
9322 (pass_ubsan::execute): Use sanitize_flags_p.
9323 * ubsan.h: Remove do_ubsan_in_current_function
9324 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9325 (dump_function_to_file): Use it here.
9327 2017-06-13 Martin Jambor <mjambor@suse.cz>
9329 PR tree-optimization/80803
9330 PR tree-optimization/81063
9331 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9332 (propagate_subaccesses_across_link): Enqueue subtree whenever
9333 necessary instead of relying on the caller.
9335 2017-06-13 Martin Jambor <mjambor@suse.cz>
9337 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9338 that have a first_link.
9339 (sort_and_splice_var_accesses): Do not check first_link before
9341 (subtree_mark_written_and_enqueue): Likewise.
9342 (propagate_all_subaccesses): Likewise and do not stop at first
9343 parent with a first_link.
9345 2017-06-13 Martin Jambor <mjambor@suse.cz>
9347 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9350 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9352 * match.pd: New pattern.
9354 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9356 * tree-vrp.c (is_masked_range_test): New function.
9357 (register_edge_assert_for): Determine ranges for
9360 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9362 PR tree-optimization/67328
9363 * fold-const.c (maskable_range_p): New function.
9364 (build_range_check): Generate bittests if possible.
9366 2017-06-13 Martin Liska <mliska@suse.cz>
9368 * gimple-pretty-print.c (dump_probability): Add new argument.
9369 (dump_edge_probability): Dump both probability and count.
9370 (dump_gimple_label): Likewise.
9371 (dump_gimple_bb_header): Likewise.
9373 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9376 * config/avr/avr-devices.c: Fix indentation.
9377 * config/avr/gen-avr-mmcu-specs.c: Dito.
9379 2017-06-13 Richard Biener <rguenther@suse.de>
9381 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9382 instead get vector type from stmt_info.
9383 (vectorizable_reduction): Adjust. Remove dead code.
9385 2017-06-13 Richard Biener <rguenther@suse.de>
9388 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9389 case of C * (x * C2 + C3).
9390 (fold_addr_of_array_ref_difference): Properly fold index difference.
9392 2017-06-12 David S. Miller <davem@davemloft.net>
9395 * config/sparc/sparc.md (return expander): Emit frame blockage if
9396 function uses alloca.
9398 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9400 * combine.c (make_field_assignment): Check len rather than the mode
9401 precision when calling force_to_mode.
9403 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9405 Support multilibs and devices that see flash in RAM address range.
9408 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9409 (avr_mcu_t) <flash_pm_offset>: New field.
9410 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9411 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9412 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9413 (AVR_TINY_PM_OFFSET): Remove macro.
9414 * config/avr/avr.opt (-mshort-calls): New option.
9415 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9416 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9417 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9418 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9419 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9420 instead of avr_arch->have_jmp_call.
9421 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9422 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9423 avr_arch->flash_pm_offset to define.
9424 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9425 new field flash_pm_offset. Add entry for avrxmega3.
9426 (avr_texinfo): Add entry for avrxmega3.
9427 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9428 attiny212, attiny214,
9429 attiny412, attiny414, attiny416, attiny417,
9430 attiny814, attiny816, attiny817,
9431 attiny1614, attiny1616, attiny1617,
9432 attiny3214, attiny3216, attiny3217.
9433 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9434 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9435 (avr_print_operand_address) [AVR_TINY]: Same.
9436 (avr_asm_init_sections) <readonly_data_section>: Only patch
9437 callback if avr_arch->flash_pm_offset = 0.
9438 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9439 for rodata if avr_arch->flash_pm_offset != 0.
9440 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9441 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9442 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9443 (m_options): Append opt_rcall.
9444 (m_dirnames): Append dir_rcall.
9445 * config/avr/t-multilib: Regenerate.
9447 * configure.ac [target=avr]: Check whether avrxmega3 default
9448 linker description file works as needed.
9449 * configure: Regenerate.
9450 * doc/avr-mmcu.texi: Regenerate.
9451 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9452 <__AVR_ARCH__>: Document avrxmega3 and 103.
9453 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9454 <__AVR_SHORT_CALLS__>: Document it.
9455 <__AVR_PM_BASE_ADDRESS__>: Document it.
9456 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9457 (AVR Variable Attributes) <progmem>: Document this is
9458 not needed for avrxmega3.
9459 (AVR Named Address Spaces) <__flash>: Dito.
9461 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9463 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9465 2017-06-12 Doug Rupp <rupp@adacore.com>
9467 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9468 Append vxworks-stdint.h to the tm_file list.
9469 * config/vxworks-stdint.h: New file.
9471 2017-06-12 Martin Liska <mliska@suse.cz>
9473 PR tree-optimization/81041
9474 * tree-profile.c (gimple_gen_ic_func_profiler):
9475 Create an extra BB in profile-generate
9476 (gimple_gen_time_profiler): Likewise.
9478 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9480 PR tree-optimization/81003
9481 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9482 (update_range_test): Use it instead of force_gimple_operand_gsi.
9484 2017-06-12 Richard Biener <rguenther@suse.de>
9486 PR tree-optimization/81053
9487 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9488 with backedge value not defined in loop. Simplify def stmt
9491 2017-06-11 Tom de Vries <tom@codesourcery.com>
9494 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9496 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9497 nvptx_cannot_force_const_mem.
9499 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9501 * opts.c (finish_options): Move test for flag_split_stack after
9502 it has been initialized.
9504 2017-06-11 Jason Merrill <jason@redhat.com>
9506 * tree.h (id_equal): New.
9507 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9508 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9509 instead of strcmp of IDENTIFIER_POINTER.
9511 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9513 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9514 (mark_all_inlined_calls_cdtor): Fix formating.
9515 (inline_transform): Rescale profile before inlining.
9517 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9519 * cgraph.h (cgraph_edge::clone): Update prototype.
9520 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9521 (cgraph_node::create_clone): Update.
9522 (cgraph_node::create_version_clone): Update.
9523 * tree-inline.c (copy_bb): Update.
9524 (expand_call_inline): Update.
9526 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9528 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9529 factored out from ...
9530 (rs6000_emit_prologue): ... here.
9532 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9534 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9535 factored out from ...
9536 (rs6000_emit_prologue): ... here.
9538 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9540 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9542 (handle_missing_profiles): Fix computation of tp_first_run.
9543 (counts_to_freqs): Do not touch freqs when count is 0.
9545 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9547 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9550 2017-06-10 Tom de Vries <tom@codesourcery.com>
9552 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9553 attributes): Document signal effective target.
9555 2017-06-10 Tom de Vries <tom@codesourcery.com>
9557 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9558 Document effective target stack_size.
9560 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9562 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9563 to the edit_context if they can be auto-applied.
9565 2017-06-9 Ian Lance Taylor <iant@golang.org>
9567 * opts.c (finish_options): If -fsplit-stack, disable implicit
9568 -forder-blocks-and-partition.
9569 * doc/invoke.texi (Optimize Options): Document that when using
9570 -fsplit-stack -forder-blocks-and-partition is not implicitly
9573 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9575 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9576 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9577 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9578 * builtins.def (abort, trap, unreachable): Declare cold.
9579 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9580 * tree-core.h (ECF_COLD): New.
9581 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9582 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9584 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9586 * predict.c (unlikely_executed_stmt_p): Cleanup.
9588 2017-06-09 Richard Biener <rguenther@suse.de>
9590 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9591 model if the ref is always written to.
9593 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9595 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9597 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9599 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9602 2017-06-09 Tom de Vries <tom@codesourcery.com>
9605 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9606 "target cannot support label values" when encountering LABEL_REF.
9608 2017-06-09 Martin Liska <mliska@suse.cz>
9610 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9611 (gimple_gen_ic_func_profiler): Emit direct comparison
9612 of __gcov_indirect_call_callee with NULL.
9613 (gimple_gen_time_profiler): Change probability from
9614 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9616 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9618 * profile.c (edge_gcov_counts): Turn to pointer.
9619 (compute_branch_probabilities, compute_branch_probabilities): Update.
9620 (branch_prob): Do not clear edge_gcov_count.
9621 * profile.h (edge_gcov_counts): Turn to pointer.
9622 (edge_gcov_count): Update.
9624 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9626 * gimple.h (gimple_check_failed): Mark cold.
9628 2017-06-09 Richard Biener <rguenther@suse.de>
9630 PR tree-optimization/66623
9631 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9632 refactor check_reduction into two parts, properly computing
9633 whether we have to check reduction validity for outer loop
9636 2017-06-09 Richard Biener <rguenther@suse.de>
9638 PR tree-optimization/79483
9639 * graphite-scop-detection.c (order): New global.
9640 (get_order): Compute bb to order mapping that satisfies code
9641 generation constraints.
9642 (cmp_pbbs): New helper.
9643 (build_scops): Start domwalk at entry block, sort generated
9646 2017-06-09 Richard Biener <rguenther@suse.de>
9649 * ipa-polymorphic-call.c
9650 (ipa_polymorphic_call_context::restrict_to_inner_class):
9651 Skip FIELD_DECLs with error_mark_node type.
9652 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9655 2017-06-09 Martin Liska <mliska@suse.cz>
9657 * predict.c (struct branch_predictor): New struct.
9658 (test_prediction_value_range): New test.
9659 (predict_c_tests): New function.
9660 * selftest-run-tests.c (selftest::run_tests): Run the function.
9661 * selftest.h: Declare new tests.
9663 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9666 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9667 gen_add3_insn did not fail.
9668 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9669 r0, construct that number in a temporary reg and add that reg to r0.
9670 If asked to put the result in r0 as well, fail.
9672 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9674 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9675 for early expansion of vec_eqv.
9677 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9680 * ubsan.c (instrument_null): Avoid pointless code temporary.
9681 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9683 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9687 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
9690 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
9693 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9695 * predict.c (unlikely_executed_edge_p): Move ahead.
9696 (probably_never_executed_edge_p): Use it.
9698 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9701 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
9702 gimple_call_builtin_p call.
9704 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9706 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
9707 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
9708 rtl_check_failed_type2, rtl_check_failed_code1,
9709 rtl_check_failed_code2, rtl_check_failed_code_mode,
9710 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
9711 rtvec_check_failed_bounds, rtl_check_failed_flag,
9712 _fatal_insn_not_found, _fatal_insn): Likewise.
9713 * tree.h (tree_contains_struct_check_failed,
9714 tree_check_failed, tree_not_check_failed,
9715 tree_class_check_failed, tree_range_check_failed,
9716 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
9717 tree_vec_elt_check_failed, phi_node_elt_check_failed,
9718 tree_operand_check_failed, omp_clause_check_failed,
9719 omp_clause_operand_check_failed, omp_clause_range_check_failed):
9722 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9724 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
9725 flag_branch_probabilities.
9726 * ipa-inline.c (edge_badness): Likewise.
9727 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9728 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
9729 * predict.c (maybe_hot_frequency_p): Likewise.
9730 (probably_never_executed): Likewise.
9731 * sched-ebb.c (schedule_ebbs): Likewise.
9732 * sched-rgn.c (find_single_block_region): Likewise.
9733 * tracer.c (tail_duplicate): Likewise.
9735 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9737 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
9738 longer requires x_flag_profile_use.
9740 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9742 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
9743 instead of flag_reorder_blocks_and_partition.
9744 * dbxout.c (dbxout_function_end): Likewise.
9745 * dwarf2out.c (gen_subprogram_die): Likewise.
9746 * haifa-sched.c (sched_create_recovery_edges): Likewise.
9747 * hw-doloop.c (reorg_loops): Likewise.
9748 * varasm.c (assemble_start_function,
9749 assemble_end_function): Likewise.
9750 (decide_function_section): Do not check for
9751 flag_reorder_blocks_and_partition.
9753 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9755 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9757 (chkp_get_hard_register_fake_addr_expr): Ditto.
9758 (chkp_build_addr_expr): Add check for hard reg case.
9759 (chkp_parse_array_and_component_ref): Ditto.
9760 (chkp_find_bounds_1): Ditto.
9761 (chkp_process_stmt): Don't generate bounds store for
9764 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9766 * predict.c (maybe_hot_bb_p): Do not check profile status.
9767 (maybe_hot_edge_p): Likewise.
9768 (probably_never_executed): Check for zero counts even if profile
9770 (unlikely_executed_edge_p): New function.
9771 (unlikely_executed_stmt_p): New function.
9772 (unlikely_executed_bb_p): New function.
9773 (set_even_probabilities): Use unlikely predicates.
9774 (combine_predictions_for_bb): Likewise.
9775 (predict_paths_for_bb): Likewise.
9776 (predict_paths_leading_to_edge): Likewise.
9777 (determine_unlikely_bbs): New function.
9778 (estimate_bb_frequencies): Use it.
9779 (compute_function_frequency): Use zero counts even if profile is
9781 * profile-count.h: Fix typo.
9783 2017-08-08 Julia Koval <julia.koval@intel.com>
9785 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
9786 _mm512_mask_cvtsepi16_storeu_epi8,
9787 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
9788 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
9789 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
9790 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
9791 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
9792 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
9793 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
9794 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
9795 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
9796 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
9797 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
9798 __builtin_ia32_pmovuswb256mem_mask,
9799 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
9800 __builtin_ia32_pmovwb512mem_mask): New builtins.
9802 2017-08-08 Julia Koval <julia.koval@intel.com>
9804 PR target/73350,80862
9805 * config/i386/subst.md (round): Fix round pattern.
9806 * config/i386/i386.c (ix86_erase_embedded_rounding):
9807 Fix erasing rounding for the fixed pattern.
9809 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9811 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
9813 2017-06-08 Martin Liska <mliska@suse.cz>
9815 PR gcov-profile/80911
9816 * gcov.c (block_info::block_info): New constructor.
9818 2017-06-07 Carl Love <cel@us.ibm.com>
9820 * config/rs6000/rs6000-c: The return type of the following
9821 built-in functions was implemented as int not long long. Fix sign
9822 of return value for the unsigned version of vec_mulo and vec_mule.
9823 vector unsigned long long vec_bperm (vector unsigned long long,
9824 vector unsigned char)
9825 vector signed long long vec_mule (vector signed int,
9827 vector unsigned long long vec_mule (vector unsigned int,
9828 vector unsigned int)
9829 vector signed long long vec_mulo (vector signed int,
9831 vector unsigned long long vec_mulo (vector unsigned int,
9832 vector unsigned int)
9833 * doc/extend.texi: Fix the documentation for the built-in
9836 2017-06-07 Carl Love <cel@us.ibm.com>
9839 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
9842 2017-06-07 Carl Love <cel@us.ibm.com>
9844 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
9845 support, Generate doublehv for signed int/float for BE case only.
9847 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
9849 * doc/invoke.texi (mcx16): Rewrite.
9851 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9853 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
9854 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
9855 *mov<mode>_softfloat, and an anonymous splitter): Use
9856 nonimmediate_operand instead of rs6000_nonimmediate_operand.
9858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9860 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
9862 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9863 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
9864 (rs6000_debug_reg_global): Adjust.
9865 (rs6000_init_hard_regno_mode_ok): Adjust.
9866 (rs6000_dbx_register_number): Adjust.
9867 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
9868 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9869 Remove SPE_ACC and SPEFSCR.
9870 (REG_ALLOC_ORDER): Ditto.
9871 (FRAME_POINTER_REGNUM): Change to 111.
9872 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
9873 (REG_CLASS_NAMES): Ditto.
9874 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
9875 (REGISTER_NAMES): Ditto.
9876 (ADDITIONAL_REG_NAMES): Ditto.
9877 (rs6000_reg_names): Ditto.
9878 * config/rs6000/rs6000.md: Renumber some register number
9881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9883 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
9885 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9886 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
9888 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
9889 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9890 Delete the SPE high registers.
9891 (REG_ALLOC_ORDER): Ditto.
9892 (enum reg_class): Remove SPE_HIGH_REGS.
9893 (REG_CLASS_NAMES): Ditto.
9894 (REG_CLASS_CONTENTS): Delete the SPE high registers.
9895 (REGISTER_NAMES): Ditto.
9896 (rs6000_reg_names): Ditto.
9897 * doc/tm.texi.in: Remove SPE as example.
9898 * doc/tm.texi: Regenerate.
9900 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9902 * config/rs6000/8540.md (ppc8540_brinc): Delete.
9903 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
9904 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
9905 * config/rs6000/rs6000.md (type): Remove "brinc".
9907 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9909 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
9910 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
9911 * config/rs6000/linuxspe.h: Delete file.
9912 * config/rs6000/rs6000.md: Don't include spe.md.
9913 * config/rs6000/spe.h: Delete file.
9914 * config/rs6000/spe.md: Delete file.
9915 * config/rs6000/t-rs6000: Remove spe.md.
9917 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9919 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
9920 (reg_or_none500mem_operand): Delete.
9921 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
9922 instead of reg_or_none500mem_operand.
9924 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9926 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
9927 handling of SPE flags.
9928 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
9930 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9932 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
9934 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
9935 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
9936 paired_divv2sf3): Similar.
9937 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
9938 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
9939 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
9941 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
9942 Rename the paired_* instruction patterns.
9943 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
9945 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
9946 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
9947 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
9948 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
9949 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
9951 (struct machine_function): Delete field spe_insn_chain_scanned_p.
9952 (spe_func_has_64bit_regs_p): Delete.
9953 (spe_expand_predicate_builtin): Delete.
9954 (spe_expand_evsel_builtin): Delete.
9955 (TARGET_DWARF_REGISTER_SPAN): Do not define.
9956 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
9957 (invalid_e500_subreg): Delete.
9958 (rs6000_legitimize_address): Always force_reg op2 as well, for
9959 paired single memory accesses.
9960 (rs6000_member_type_forces_blk): Delete.
9961 (rs6000_spe_function_arg): Delete.
9962 (rs6000_expand_unop_builtin): Delete SPE handling.
9963 (rs6000_expand_binop_builtin): Ditto.
9964 (spe_expand_stv_builtin): Delete.
9965 (bdesc_2arg_spe): Delete.
9966 (spe_expand_builtin): Delete.
9967 (spe_expand_predicate_builtin): Delete.
9968 (spe_expand_evsel_builtin): Delete.
9969 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
9970 (spe_init_builtins): Delete.
9971 (spe_func_has_64bit_regs_p): Delete.
9972 (savres_routine_name): Delete "info" parameter. Adjust callers.
9973 (rs6000_emit_stack_reset): Ditto.
9974 (rs6000_dwarf_register_span): Delete.
9975 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
9976 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
9977 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
9979 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
9981 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
9982 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
9983 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
9984 mulv2sf3, divv2sf3): Delete expanders.
9986 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9988 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
9990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9992 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
9993 * config/rs6000/rs6000.c: Ditto.
9995 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9997 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
9998 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
9999 comparison_operator.
10001 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10003 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10004 * config/rs6000/rs6000.opt: Ditto.
10005 * config/rs6000/t-rtems: Ditto.
10007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10009 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10010 TARGET_E500_SINGLE by 0, simplify.
10011 * config/rs6000/rs6000.c: Ditto.
10012 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10013 (spe_build_register_parallel): Delete.
10014 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10015 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10016 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10017 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10018 (E500_CONVERT): Delete.
10019 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10021 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10023 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10024 * config/rs6000/dfp.md: Ditto.
10025 (negdd2, *negdd2_fpr): Merge.
10026 (absdd2, *absdd2_fpr): Merge.
10027 (negtd2, *negtd2_fpr): Merge.
10028 (abstd2, *abstd2_fpr): Merge.
10029 * config/rs6000/e500.h: Delete file.
10030 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10031 TARGET_FPRS by 1 and simplify.
10032 * config/rs6000/rs6000-c.c: Ditto.
10033 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10034 TARGET_DF_SPE by 0.
10035 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10037 * config/rs6000/rs6000.md: Ditto.
10038 (floatdidf2, *floatdidf2_fpr): Merge.
10039 (move_from_CR_gt_bit): Delete.
10040 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10041 (E500_CR_IOR_COMPARE): Delete.
10042 (All patterns that require !TARGET_FPRS): Delete.
10043 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10045 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10047 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10049 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10051 * graphds.c (add_edge): Intitialize edge's attached data.
10052 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10053 pointer parameter. Call pointed function on each edge during
10054 graph traversing. Skip traversing the edge when the function
10056 (graphds_dfs, graphds_scc): Ditto.
10057 (for_each_edge): New parameter. Pass the new parameter to callback
10059 * graphds.h (skip_edge_callback): New function pointer type.
10060 (graphds_dfs, graphds_scc): New function pointer parameter.
10061 (graphds_edge_callback, for_each_edge): New parameter.
10063 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10065 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10066 out code checking if runtime alias check is possible to below ...
10067 Call the new function.
10068 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10069 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10071 2017-06-07 Marek Polacek <polacek@redhat.com>
10074 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10075 TYPE_OVERFLOW_WRAPS check.
10077 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10079 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10080 if versioning is required.
10081 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10082 peeling with the check for versioning.
10084 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10086 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10087 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10088 Set true to new parameter if new ssa variable is defined.
10089 (vect_gen_vector_loop_niters): Refactor. Set range information
10090 for the new vector loop bound variable.
10091 (vect_do_peeling): Ditto.
10093 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10095 * tree-affine.c (ssa.h): Include header file.
10096 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10097 has wrapping overflow behavior.
10099 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10101 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10103 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10105 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10106 (tree_to_aff_combination): ... here.
10108 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10110 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10111 reg_pressure model function.
10112 (ivopts_global_cost_for_size): Delete.
10113 (determine_set_costs, iv_ca_recount_cost): Call new model function
10114 ivopts_estimate_reg_pressure.
10116 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10118 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10119 expensive than udiv. Remove floating point cases from mod.
10121 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10123 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10124 Increase idiv cost.
10126 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10128 * config/aarch64/aarch64.md
10129 (copysignsf3): Fix mask generation.
10131 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10133 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10135 (class dump_manager): Add register_dumps method.
10136 * dumpfile.c: Include langhooks.h.
10137 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10138 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10139 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10140 (dump_manager::dump_register): Start with 512 entries instead of 32.
10141 (dump_manager::register_dumps): New method.
10142 * toplev.c (general_init): Instead of invoking register_dumps
10143 langhook, invoke register_dumps method on the dump manager.
10144 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10147 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10149 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10150 Say that # requires an associated define_split to exist, and that
10151 the define_split must be suitable for use after register allocation.
10153 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10155 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10156 (compute_outgoing_frequencies): Also initialize zero counts.
10157 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10158 around loops; preserve more of profile when nothing changes.
10160 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10162 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10164 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10165 * config/arm/arm-cpu-cdata.h: Regenerate.
10166 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10167 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10168 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10169 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10170 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10172 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10174 * config/arm/t-rmprofile: Likewise.
10175 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10177 2017-06-06 David S. Miller <davem@davemloft.net>
10180 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10181 blockage if function uses alloca.
10183 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10185 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10186 New "uid" fields to hold pretty-print IDs of group and ref.
10187 Memory references are now identified as <group_id>:<ref_id>
10188 instead of using [random] addresses.
10189 (dump_mem_details): Simplify, no functional change.
10190 (dump_mem_ref): Simplify and make output more concise.
10191 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10192 (find_or_create_group): Initialize group uid.
10193 (record_ref): Initialize ref uid. Improve debug output.
10194 (prune_group_by_reuse, should_issue_prefetch_p,)
10195 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10196 (mark_nontemporal_store, determine_loop_nest_reuse):
10197 Improve debug output.
10199 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10201 * dbgcnt.def (prefetch): New debug counter.
10202 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10203 (schedule_prefetches): Stop issueing prefetches if debug counter
10206 2017-06-06 Tom de Vries <tom@codesourcery.com>
10208 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10209 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10211 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10213 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10214 Use aarch64_reg_or_zero predicate for operand 4.
10215 (aarch64_compare_and_swap<mode> define_insn_and_split):
10216 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10217 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10219 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10221 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10222 (arm_compute_save_core_reg_mask): This.
10223 (thumb1_compute_save_reg_mask): Rename into ...
10224 (thumb1_compute_save_core_reg_mask): This.
10225 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10226 (arm_compute_frame_layout): Likewise.
10228 2017-06-06 Richard Biener <rguenther@suse.de>
10230 PR tree-optimization/80974
10231 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10232 keep or clear leaders SSA info.
10234 2017-06-06 Tom de Vries <tom@codesourcery.com>
10236 * config/nvptx/nvptx.c (split_mode_p): New function.
10237 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10239 2017-06-06 Tom de Vries <tom@codesourcery.com>
10241 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10243 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10246 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10249 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10251 * shrink-wrap.c (handle_simple_exit): Update profile.
10252 (try_shrink_wrapping): Upate profile.
10254 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10256 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10257 (tree_guess_outgoing_edge_probabilities): New.
10258 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10259 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10261 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10263 * ipa-split.c (split_function): Initialize return bb profile.
10265 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10267 * profile.c (compute_branch_probabilities): Also initialize
10268 EXIT_BLOCK profile.
10270 2017-06-06 Richard Biener <rguenther@suse.de>
10272 PR tree-optimization/80928
10273 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10274 (vect_analyze_loop_operations): Properly guard analysis for
10276 (vect_transform_loop): Likewise.
10277 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10278 (vect_model_induction_cost): Do not cost for pure SLP.
10279 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10280 of induction in inner loop vectorization.
10281 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10282 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10283 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10285 (vect_analyze_slp_cost_1): Cost induction.
10286 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10287 (vect_get_slp_vect_defs): Likewise.
10288 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10289 (vect_transform_stmt): Handle SLP reductions.
10290 * tree-vectorizer.h (vectorizable_induction): Adjust.
10292 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10294 * config/rs6000/rs6000.c (make_resolver_func): Update
10295 init_lowered_empty_function call.
10297 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10299 * doc/invoke.texi: Document the -fprofile-abs-path option.
10300 * common.opt (fprofile-abs-path): New option.
10301 * gcov-io.h (gcov_write_filename): Declare.
10302 * gcov-io.c (gcov_write_filename): New function.
10303 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10304 * profile.c (output_location): Likewise.
10306 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10308 * shring-wrap.c: Revert accidental commit.
10310 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10312 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10314 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10316 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10318 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10319 profile in callgraph edge.
10320 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10321 (apply_scale): Likewise.
10322 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10323 Also scale profile when inlining function with zero profile.
10324 (initialize_cfun): Update exit block profile even when it is zero.
10325 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10326 when profile is read.
10328 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10330 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10331 (CLONE_*): New constants to define the processors we can generate
10332 code for with the target_clone attribute.
10333 (rs6000_clone_map): New array to identify which clone processors
10334 the current program is running on.
10335 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10336 target_clone attribute.
10337 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10338 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10339 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10340 (cpu_expand_builtin): Add support for target_clone attribute.
10341 (rs6000_valid_attribute_p): Allow "default" attribute.
10342 (get_decl_name): New debug function to simplify printing the
10343 current function name in debugging statements.
10344 (rs6000_clone_priority): New functions to support the target_clone
10345 attribute, and be able to generate code to switch between ISA 2.05
10346 through ISA 3.0 (power6 through power9).
10347 (rs6000_compare_version_priority): Likewise.
10348 (rs6000_get_function_versions_dispatcher): Likewise.
10349 (make_resolver_func): Likewise.
10350 (add_condition_to_bb): Likewise.
10351 (dispatch_function_versions): Likewise.
10352 (rs6000_generate_version_dispatcher_body): Likewise.
10353 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10354 (fusion_gpr_load_p): Fix a spacing issue.
10355 * doc/extend.texi (Common Function Attributes): Document that the
10356 PowerPC supports the target_clone attribute.
10358 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10360 * config/arm/arm.h: explain F symbol found in description of ARM
10361 register allocation in its legend.
10363 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10365 * config/mips/frame-header-opt.c: Include profile-count.h.
10366 * config/riscv/riscv.c: Include profile-count.h
10368 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10370 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10372 (sm_set_flag_if_changed): Add bbs field.
10373 (execute_sm_if_changed_flag_set): Pass BBS.
10374 (execute_sm): Update.
10376 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10378 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10381 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10383 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10384 (peephole2): New peephole2 to emit the above.
10385 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10387 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10389 * config/aarch64/aarch64.c (define_peephole2 above
10390 *sub_<shift>_<mode>): New peephole.
10392 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10394 * config/i386/i386.c (make_resolver_func): Update.
10395 * Makefile.in: Add profile-count.h and profile-count.o
10396 * auto-profile.c (afdo_indirect_call): Update to new API.
10397 (afdo_set_bb_count): Update.
10398 (afdo_propagate_edge): Update.
10399 (afdo_propagate_circuit): Update.
10400 (afdo_calculate_branch_prob): Update.
10401 (afdo_annotate_cfg): Update.
10402 * basic-block.h: Include profile-count.h
10403 (struct edge_def): Turn count to profile_count.
10404 (struct basic_block_def): Likewie.
10405 (REG_BR_PROB_BASE): Move to profile-count.h
10406 (RDIV): Move to profile-count.h
10407 * bb-reorder.c (max_entry_count): Turn to profile_count.
10408 (find_traces): Update.
10409 (rotate_loop):Update.
10410 (connect_traces):Update.
10411 (sanitize_hot_paths):Update.
10412 * bt-load.c (migrate_btr_defs): Update.
10413 * cfg.c (RDIV): Remove.
10414 (init_flow): Use alloc_block.
10415 (alloc_block): Uninitialize count.
10416 (unchecked_make_edge): Uninitialize count.
10417 (check_bb_profile): Update.
10418 (dump_edge_info): Update.
10419 (dump_bb_info): Update.
10420 (update_bb_profile_for_threading): Update.
10421 (scale_bbs_frequencies_int): Update.
10422 (scale_bbs_frequencies_gcov_type): Update.
10423 (scale_bbs_frequencies_profile_count): New.
10424 * cfg.h (update_bb_profile_for_threading): Update.
10425 (scale_bbs_frequencies_profile_count): Declare.
10426 * cfgbuild.c (compute_outgoing_frequencies): Update.
10427 (find_many_sub_basic_blocks): Update.
10428 * cfgcleanup.c (try_forward_edges): Update.
10429 (try_crossjump_to_edge): Update.
10430 * cfgexpand.c (expand_gimple_tailcall): Update.
10431 (construct_exit_block): Update.
10432 * cfghooks.c (verify_flow_info): Update.
10433 (dump_bb_for_graph): Update.
10434 (split_edge): Update.
10435 (make_forwarder_block): Update.
10436 (duplicate_block): Update.
10437 (account_profile_record): Update.
10438 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10439 (get_estimated_loop_iterations): Update.
10440 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10441 (single_likely_exit): Update.
10442 * cfgloopmanip.c (scale_loop_profile): Update.
10444 (set_zero_probability): Update.
10445 (lv_adjust_loop_entry_edge): Update.
10446 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10447 (purge_dead_edges): Update.
10448 (rtl_account_profile_record): Update.
10449 * cgraph.c (cgraph_node::create): Uninitialize count.
10450 (symbol_table::create_edge): Uninitialize count.
10451 (cgraph_update_edges_for_call_stmt_node): Update.
10452 (cgraph_edge::dump_edge_flags): Update.
10453 (cgraph_node::dump): Update.
10454 (cgraph_edge::maybe_hot_p): Update.
10455 * cgraph.h: Include profile-count.h
10456 (create_clone), create_edge, create_indirect_edge): Update.
10457 (cgraph_node): Turn count to profile_count.
10458 (cgraph_edge0: Likewise.
10459 (make_speculative, clone): Update.
10460 (create_edge): Update.
10461 (init_lowered_empty_function): Update.
10462 * cgraphclones.c (cgraph_edge::clone): Update.
10463 (duplicate_thunk_for_node): Update.
10464 (cgraph_node::create_clone): Update.
10465 * cgraphunit.c (cgraph_node::analyze): Update.
10466 (cgraph_node::expand_thunk): Update.
10467 * final.c (dump_basic_block_info): Update.
10468 * gimple-streamer-in.c (input_bb): Update.
10469 * gimple-streamer-out.c (output_bb): Update.
10470 * graphite.c (print_global_statistics): Update.
10471 (print_graphite_scop_statistics): Update.
10472 * hsa-brig.c: Include basic-block.h.
10473 * hsa-dump.c: Include basic-block.h.
10474 * hsa-gen.c (T sum_slice): Update.
10475 (convert_switch_statements):Update.
10476 * hsa-regalloc.c: Include basic-block.h.
10477 * ipa-chkp.c (chkp_produce_thunks): Update.
10478 * ipa-cp.c (struct caller_statistics): Update.
10479 (init_caller_stats): Update.
10480 (gather_caller_stats): Update.
10481 (ipcp_cloning_candidate_p): Update.
10482 (good_cloning_opportunity_p): Update.
10483 (get_info_about_necessary_edges): Update.
10484 (dump_profile_updates): Update.
10485 (update_profiling_info): Update.
10486 (update_specialized_profile): Update.
10487 (perhaps_add_new_callers): Update.
10488 (decide_about_value): Update.
10489 (ipa_cp_c_finalize): Update.
10490 * ipa-devirt.c (struct odr_type_warn_count): Update.
10491 (struct decl_warn_count): Update.
10492 (struct final_warning_record): Update.
10493 (possible_polymorphic_call_targets): Update.
10494 (ipa_devirt): Update.
10495 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10496 * ipa-icf.c (sem_function::merge): Update.
10497 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10498 * ipa-inline.c (compute_uninlined_call_time): Update.
10499 (compute_inlined_call_time): Update.
10500 (want_inline_small_function_p): Update.
10501 (want_inline_self_recursive_call_p): Update.
10502 (edge_badness): Update.
10503 (lookup_recursive_calls): Update.
10504 (recursive_inlining): Update.
10505 (inline_small_functions): Update.
10506 (dump_overall_stats): Update.
10507 (dump_inline_stats): Update.
10508 * ipa-profile.c (ipa_profile_generate_summary): Update.
10509 (ipa_propagate_frequency): Update.
10510 (ipa_profile): Update.
10511 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10512 * ipa-utils.c (ipa_merge_profiles): Update.
10513 * loop-doloop.c (doloop_modify): Update.
10514 * loop-unroll.c (report_unroll): Update.
10515 (unroll_loop_runtime_iterations): Update.
10516 * lto-cgraph.c (lto_output_edge): Update.
10517 (lto_output_node): Update.
10518 (input_node): Update.
10519 (input_edge): Update.
10520 (merge_profile_summaries): Update.
10521 * lto-streamer-in.c (input_cfg): Update.
10522 * lto-streamer-out.c (output_cfg): Update.
10523 * mcf.c (create_fixup_graph): Update.
10524 (adjust_cfg_counts): Update.
10525 (sum_edge_counts): Update.
10526 * modulo-sched.c (sms_schedule): Update.
10527 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10528 * predict.c (maybe_hot_count_p): Update.
10529 (probably_never_executed): Update.
10530 (dump_prediction): Update.
10531 (combine_predictions_for_bb): Update.
10532 (propagate_freq): Update.
10533 (handle_missing_profiles): Update.
10534 (counts_to_freqs): Update.
10535 (rebuild_frequencies): Update.
10536 (force_edge_cold): Update.
10537 * predict.h: Include profile-count.h
10538 (maybe_hot_count_p, counts_to_freqs): UPdate.
10539 * print-rtl-function.c: Do not include cfg.h
10540 * print-rtl.c: Include basic-block.h
10541 * profile-count.c: New file.
10542 * profile-count.h: New file.
10543 * profile.c (is_edge_inconsistent): Update.
10544 (correct_negative_edge_counts): Update.
10545 (is_inconsistent): Update.
10546 (set_bb_counts): Update.
10547 (read_profile_edge_counts): Update.
10548 (compute_frequency_overlap): Update.
10549 (compute_branch_probabilities): Update; Initialize and deinitialize
10551 (branch_prob): Update.
10552 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10553 (edge_gcov_count): New.
10554 (bb_gcov_count): New.
10555 * shrink-wrap.c (try_shrink_wrapping): Update.
10556 * tracer.c (better_p): Update.
10557 * trans-mem.c (expand_transaction): Update.
10558 (ipa_tm_insert_irr_call): Update.
10559 (ipa_tm_insert_gettmclone_call): Update.
10560 * tree-call-cdce.c: Update.
10561 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10562 (gimple_duplicate_sese_tail): Update.
10563 (gimple_account_profile_record): Update.
10564 (execute_fixup_cfg): Update.
10565 * tree-inline.c (copy_bb): Update.
10566 (copy_edges_for_bb): Update.
10567 (initialize_cfun): Update.
10568 (freqs_to_counts): Update.
10569 (copy_cfg_body): Update.
10570 (expand_call_inline): Update.
10571 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10572 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10573 (try_unroll_loop_completely): Update.
10574 (try_peel_loop): Update.
10575 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10576 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10577 * tree-ssa-loop-split.c (connect_loops): Update.
10578 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10579 * tree-ssa-reassoc.c (branch_fixup): Update.
10580 * tree-ssa-tail-merge.c (replace_block_by): Update.
10581 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10582 (compute_path_counts): Update.
10583 (update_profile): Update.
10584 (recompute_probabilities): Update.
10585 (update_joiner_offpath_counts): Update.
10586 (estimated_freqs_path): Update.
10587 (freqs_to_counts_path): Update.
10588 (clear_counts_path): Update.
10589 (ssa_fix_duplicate_block_edges): Update.
10590 (duplicate_thread_path): Update.
10591 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10592 (struct switch_conv_info): Update.
10593 * tree-tailcall.c (decrease_profile): Update.
10594 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10595 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10596 * value-prof.c (check_counter): Update.
10597 (gimple_divmod_fixed_value): Update.
10598 (gimple_mod_pow2): Update.
10599 (gimple_mod_subtract): Update.
10600 (gimple_ic_transform): Update.
10601 (gimple_stringop_fixed_value): Update.
10602 * value-prof.h (gimple_ic): Update.
10604 2017-06-02 Carl Love <cel@us.ibm.com>
10606 * config/rs6000/rs6000-c: Add support for built-in functions
10607 vector double vec_doublee (vector signed int);
10608 vector double vec_doublee (vector unsigned int);
10609 vector double vec_doublee (vector float);
10610 vector double vec_doubleh (vector signed int);
10611 vector double vec_doubleh (vector unsigned int);
10612 vector double vec_doubleh (vector float);
10613 vector double vec_doublel (vector signed int);
10614 vector double vec_doublel (vector unsigned int);
10615 vector double vec_doublel (vector float);
10616 vector double vec_doubleo (vector signed int);
10617 vector double vec_doubleo (vector unsigned int);
10618 vector double vec_doubleo (vector float);.
10619 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10620 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10622 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10623 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10624 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10626 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10627 vec_doublel, vec_doubleh.
10628 * doc/extend.texi: Update the built-in documentation file for the
10629 new built-in functions.
10631 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10634 * ipa-inline-analysis.c (free_growth_caches): Set
10635 edge_removal_hook_holder to NULL after removing it.
10637 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10639 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10640 comparision with zero.
10642 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10643 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10644 for early expansion of vec_min and vec_max builtins.
10645 (builtin_function_type): Add min/max unsigned variants to those
10646 identified as having unsigned arguments.
10648 2017-06-02 Olivier Hainque <hainque@adacore.com>
10650 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10652 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10654 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10655 Use VALL_F16 iterator rather than VALL.
10657 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10659 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10660 Emit CBNZ inside loop when doing a strong exchange and comparing
10661 against zero. Generate the CC flags after the loop.
10663 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10665 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10666 (dl_section_ref): New.
10667 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10668 On AIX, append an expression to subtract the size of the
10669 section length to dl_section_ref.
10671 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10674 for early expansion of vector absolute builtins.
10676 2017-06-02 Richard Biener <rguenther@suse.de>
10678 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10679 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10681 2017-06-02 Richard Biener <rguenther@suse.de>
10683 PR tree-optimization/80948
10684 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10685 stmt order as well.
10687 2017-06-02 Richard Biener <rguenther@suse.de>
10689 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
10691 * tree-vect-stmts.c (process_use): Do not mark backedge defs
10692 for inductions as relevant.
10694 2017-06-02 Richard Biener <rguenther@suse.de>
10696 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
10697 (vectorizable_induction): ... this. Remove dead code.
10699 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
10701 * builtins. (expand_builtin_alloca): Remove second parameter and
10702 infer its value from the first parameter instead.
10703 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
10705 2017-06-02 Jakub Jelinek <jakub@redhat.com>
10707 PR rtl-optimization/80903
10708 * loop-doloop.c (add_test): Unshare sequence.
10710 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
10712 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
10714 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
10716 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
10718 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
10719 xlogue_layout::get_instance, logue_layout::xlogue_layout,
10720 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
10721 (xlogue_layout::get_stub_rtx): Make static.
10722 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
10723 (xlogue_layout::compute_stub_managed_regs): Rename to...
10724 (xlogue_layout::count_stub_managed_regs): ...this.
10725 (xlogue_layout::is_stub_managed_reg): New function.
10726 (xlogue_layout::m_stub_names): Rename to...
10727 (xlogue_layout::s_stub_names): ...this, make static.
10728 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
10729 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
10730 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
10731 xlogue_layout::s_stub_names): Instantiate statics.
10732 (stub_managed_regs): Remove.
10733 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
10734 (disable_call_ms2sysv_xlogues): Rename to...
10735 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
10736 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
10738 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
10739 change after reload_completed.
10740 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
10742 (ix86_expand_prologue): Likewise.
10743 (ix86_expand_epilogue): Likewise.
10744 (ix86_expand_split_stack_prologue): Likewise.
10745 (ix86_compute_frame_layout): Remove frame parameter ...
10746 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
10747 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
10749 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
10750 (ix86_frame): Move from here ...
10751 * config/i386/i386.h (ix86_frame): ... to here.
10752 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
10753 complete ix86_frame data structure instead. Remove some_ld_name.
10755 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10757 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
10758 symbols that hold a DECL_VALUE_EXPR.
10760 2017-06-01 Martin Jambor <mjambor@suse.cz>
10762 PR tree-optimization/80898
10763 * tree-sra.c (process_subtree_disqualification): Removed.
10764 (disqualify_candidate): Do not acll
10765 process_subtree_disqualification.
10766 (subtree_mark_written_and_enqueue): New function.
10767 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
10768 RHS has been disqualified and re-queue LHS if necessary. Apart
10769 from that, ignore disqualified RHS.
10771 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10773 * config/s390/s390.c (s390_emit_epilogue): Disable early return
10774 address fetch for z10 or later.
10776 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10778 * config/arc/arc.md (tst_movb): Add guard when splitting.
10780 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10782 * config/arc/arc.c (arc_can_eliminate): Test against
10783 arc_frame_pointer_needed.
10785 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10787 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
10788 to prevent store reordering.
10789 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
10790 (type): Add block type.
10791 (stack_tie): Define special instruction to be used in
10794 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10796 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
10797 constraint. It is not valid for the pattern.
10798 (noncommutative_binary_comparison): Likewise.
10800 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10802 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
10805 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10807 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
10808 be used by the reg-alloc.
10810 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10812 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
10813 reg-alloc when having mul64 or mul32x16 instructions.
10814 (mulsidi3): Likewise.
10815 (umulsidi3): Likewise.
10816 (mulsi32x16): New pattern.
10817 (mulsi64): Likewise.
10818 (mulsidi64): Likewise.
10819 (umulsidi64): Likewise.
10820 (MUL32x16_REG): Define.
10821 (mul64_600): Use MUL32x16_REG.
10822 (mac64_600): Likewise.
10823 (umul64_600): Likewise.
10824 (umac64_600): Likewise.
10826 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10828 * config/arc/arc.md (mulsi3_700): Make it commutative.
10830 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
10832 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
10834 (*sign_extendsidi2_insn): Likewise for movstosw.
10836 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10838 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
10839 the type of the input discriminant value. Convert the
10840 discriminant value of signedness vary.
10842 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
10844 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
10845 Add to -Wall section.
10847 2017-06-01 Richard Biener <rguenther@suse.de>
10849 PR middle-end/66313
10850 * fold-const.c (fold_plusminus_mult_expr): If the factored
10851 factor may be zero use a wrapping type for the inner operation.
10852 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
10853 and handle moved defs.
10854 (process_assignment): Properly guard the unary op case. Return a
10855 tri-state indicating that moving the stmt before the call may allow
10856 to continue. Pass through to_move.
10857 (find_tail_calls): Handle moving unrelated defs before
10860 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10863 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
10864 splitter result in the canonical way.
10866 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
10868 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
10869 also for 32bit target. Update insn attributes.
10870 (zero-extendsidi2 splitter): Allow all registers for operand 1.
10872 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
10874 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
10875 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
10876 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
10877 (_mm_maskz_min_ss): New intrinsics.
10879 2017-05-31 Martin Liska <mliska@suse.cz>
10881 * tree-vect-loop.c (vect_create_epilog_for_reduction):
10882 Change comment style to one we normally use.
10883 (vectorizable_reduction): Likewise.
10884 (vectorizable_induction): Likewise.
10885 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10886 (vectorizable_call): Likewise.
10887 (vectorizable_simd_clone_call): Likewise.
10888 (vectorizable_conversion): Likewise.
10889 (vectorizable_assignment): Likewise.
10890 (vectorizable_shift): Likewise.
10891 (vectorizable_operation): Likewise.
10892 (vectorizable_store): Likewise.
10893 (vectorizable_load): Likewise.
10894 * tree-vectorizer.h: Likewise.
10896 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
10898 * passes.c (emergency_dump_function): New.
10899 * tree-pass.h (emergency_dump_function): Declare.
10900 * plugin.c (plugins_internal_error_function): Remove.
10901 * plugin.h (plugins_internal_error_function): Remove declaration.
10902 * toplev.c (internal_error_function): New static function. Use it...
10903 (general_init): ...here.
10905 2017-05-31 Graham Markall <graham.markall@embecosm.com>
10907 * config/arc/arc.c (arc_print_operand): Handle constant operands.
10908 (arc_rtx_costs): Add costs for new patterns.
10909 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
10910 * config/arc/predicates.md: Add _1_2_3_operand predicate.
10912 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10914 * tree-ssa-strlen.c (get_next_strinfo): New function.
10915 (get_stridx_plus_constant): Use it.
10916 (zero_length_string): Likewise.
10917 (adjust_related_strinfos): Likewise.
10918 (adjust_last_stmt): Likewise.
10920 2017-05-31 Richard Biener <rguenther@suse.de>
10923 * config/i386/i386.c (ix86_expand_builtin): Remove assert
10924 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
10926 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10928 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
10929 loop_vinfo argument and use of dependence distance vectors.
10930 Check instead whether the two references differ only in their
10931 initial value and assume that they have the same alignment if the
10932 difference is a multiple of the vector alignment.
10933 (vect_analyze_data_refs_alignment): Update call accordingly.
10935 2017-05-31 Martin Liska <mliska@suse.cz>
10938 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
10940 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10942 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
10943 (create_intersect_range_checks): Move from ...
10944 * tree-data-ref.c (create_intersect_range_checks_index)
10945 (create_intersect_range_checks): ... to here.
10946 (create_runtime_alias_checks): New function factored from ...
10947 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
10948 here. Call above function.
10949 * tree-data-ref.h (create_runtime_alias_checks): New function.
10951 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10953 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
10954 segment length for dr_b and compute it in wide_int.
10956 2017-05-31 Richard Biener <rguenther@suse.de>
10958 PR tree-optimization/80906
10959 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
10960 and pass through iv_map.
10961 (copy_bb_and_scalar_dependences): Adjust.
10962 (translate_pending_phi_nodes): Likewise.
10963 (copy_loop_close_phi_args): Handle code-generating IVs instead
10966 2017-05-30 David Malcolm <dmalcolm@redhat.com>
10968 * diagnostic-color.c (color_dict): Add "type-diff".
10969 (parse_gcc_colors): Update comment.
10970 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10971 -fdiagnostics-show-template-tree and -fno-elide-type.
10972 (GCC_COLORS): Add type-diff to example.
10974 (-fdiagnostics-show-template-tree): New.
10975 (-fno-elide-type): New.
10976 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
10977 the pp_format_decoder callback. Call any m_format_postprocessor's
10979 (pretty_printer::pretty_printer): Initialize
10980 m_format_postprocessor.
10981 (pretty_printer::~pretty_printer): Delete any
10982 m_format_postprocessor.
10983 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
10984 (class format_postprocessor): New class.
10985 (struct pretty_printer::format_decoder): Document the new parameters.
10986 (struct pretty_printer::m_format_postprocessor): New field.
10987 * tree-diagnostic.c (default_tree_printer): Update for new
10988 bool and const char ** params.
10989 * tree-diagnostic.h (default_tree_printer): Likewise.
10991 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
10993 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
10994 (lwa_operand): Delete rs6000_gen_cell_microcode test.
10995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10996 rs6000_gen_cell_microcode code.
10997 (rs6000_final_prescan_insn): Delete.
10998 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
10999 "warn-cell-microcode" entries.
11000 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11001 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11002 throughout. Change cc_reg_not_micro_cr0_operand to
11003 cc_reg_not_cr0_operand throughout.
11004 (*extendhi<mode>2_noload): Delete.
11005 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11006 (mwarn-cell-microcode): Delete.
11007 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11008 -mgen-cell-microcode and -mwarn-cell-microcode.
11010 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11013 * config/i386/constraints.md (Yd): New constraint.
11015 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11016 and (?Yd, r) alternatives. Update insn attributes.
11017 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11018 and (?*Yd, r) alternatives. Update insn attributes.
11019 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11021 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11023 * gimplify.c (gimplify_modify_expr): Don't create a
11024 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11027 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11029 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11031 2017-05-30 Richard Biener <rguenther@suse.de>
11033 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11034 and reduc_def fields.
11035 (STMT_VINFO_REDUC_TYPE): New define.
11036 (STMT_VINFO_REDUC_DEF): Likewise.
11037 (vect_force_simple_reduction): Adjust prototype.
11038 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11039 (vect_is_simple_reduction): Remove check_reduction argument.
11040 (vect_force_simple_reduction): Adjust and set
11041 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11042 (vectorizable_reduction): Do not re-do reduction analysis
11043 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11044 * tree-parloops.c (gather_scalar_reductions): Adjust.
11046 2017-05-30 Richard Biener <rguenther@suse.de>
11048 PR middle-end/80901
11049 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11052 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11054 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11055 Introduce unknown_misalignment parameter and remove vf.
11056 (vect_peeling_hash_get_lowest_cost):
11057 Pass unknown_misalignment parameter.
11058 (vect_enhance_data_refs_alignment):
11059 Fix unsupportable data ref treatment.
11061 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11063 * tree-vect-data-refs.c (vect_get_data_access_cost):
11064 Workaround for SLP handling.
11065 (vect_enhance_data_refs_alignment):
11066 Compute costs for doing no peeling at all, compare to the best
11067 peeling costs so far and avoid peeling if cheaper.
11069 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11071 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11072 Return peeling info and set costs to zero for unlimited cost
11074 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11075 with unknown misalignment. Compute and costs for unknown
11076 misalignment, compare them to the costs for known misalignment
11077 and choose the cheapest for peeling.
11079 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11081 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11082 (vect_get_peeling_costs_all_drs): Create function.
11083 (vect_peeling_hash_get_lowest_cost):
11084 Use vect_get_peeling_costs_all_drs.
11085 (vect_peeling_supportable): Create function.
11086 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11088 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11090 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11091 DR_HAS_NEGATIVE_STEP.
11092 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11093 (vect_enhance_data_refs_alignment): Use.
11094 (vect_duplicate_ssa_name_ptr_info): Use.
11095 * tree-vectorizer.h (dr_misalignment): Use.
11096 (known_alignment_for_access_p): Use.
11098 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11101 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11103 (has_section_name): New function.
11105 2017-05-30 Martin Liska <mliska@suse.cz>
11108 * auto-profile.c (get_function_decl_from_block): Fix
11111 2017-05-30 Richard Biener <rguenther@suse.de>
11113 PR middle-end/80876
11114 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11116 2017-05-30 Martin Liska <mliska@suse.cz>
11118 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11119 * dumpfile.h (struct dump_file_info): Remove ctors.
11121 2017-05-30 Martin Liska <mliska@suse.cz>
11123 * predict.def: Fix GNU coding style.
11125 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11127 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11128 Mark 'to' argument with ATTRIBUTE_UNUSED.
11130 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11132 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11133 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11134 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11137 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11139 * doc/install.texi (Options specification): Restore entry of
11140 --enable-sjlj-exceptions.
11142 2017-05-27 Michael Eager <eager@eagercon.com>
11145 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11147 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11149 * config/microblaze/microblaze.h
11150 (FIXED_REGISTERS): Update in macro.
11151 (CALL_USED_REGISTERS): Update in macro.
11153 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11155 * doc/install.texi: Add links to macOS binary distributions.
11157 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11161 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11163 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11165 2017-05-26 Martin Liska <mliska@suse.cz>
11167 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11169 2017-05-26 Martin Liska <mliska@suse.cz>
11171 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11172 always leading ';; '.
11173 (dump_bb_info): Likewise.
11174 (brief_dump_cfg): Likewise.
11175 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11176 * dumpfile.c: Remove usage of TDF_VERBOSE.
11177 * dumpfile.h (enum dump_kind): Likewise.
11178 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11179 * print-tree.c (debug_verbose): Remove.
11180 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11181 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11182 * tree-diagnostic.c (default_tree_printer): Replace
11183 TDF_DIAGNOSTIC with TDF_SLIM.
11185 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11187 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11188 in parameter loop, rather than loop_vinfo.
11189 (create_intersect_range_checks): Ditto.
11190 (vect_create_cond_for_alias_checks): Update call to above functions.
11192 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11194 PR tree-optimization/80815
11195 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11196 for merging runtime alias checks. Handle negative DR_STEPs.
11198 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11200 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11202 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11203 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11204 out code pruning runtime alias checks.
11205 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11206 factored out from above.
11207 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11209 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11211 (prune_runtime_alias_test_list): New decalaration.
11213 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11215 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11216 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11217 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11218 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11219 (operator==, comp_dr_with_seg_len_pair): Ditto.
11220 (vect_prune_runtime_alias_test_list): Ditto.
11222 2017-05-26 Martin Liska <mliska@suse.cz>
11225 * params.def: Bound partial-inlining-entry-probability param.
11227 2017-05-26 Marek Polacek <polacek@redhat.com>
11230 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11233 2017-05-26 Richard Biener <rguenther@suse.de>
11235 PR tree-optimization/80842
11236 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11239 2017-05-26 Richard Biener <rguenther@suse.de>
11241 PR tree-optimization/80844
11242 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11244 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11246 * doc/md.texi (Machine Constraints): Update x86 family
11247 machine constraints section to match 'config/i386/constraints.md'.
11249 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11251 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11253 2017-05-25 Nathan Sidwell <nathan@acm.org>
11255 * doc/invoke.texi (--enable-languages): Update documentation.
11257 2017-05-25 Martin Liska <mliska@suse.cz>
11259 * dumpfile.c: Add TDF_FOLDING.
11260 * dumpfile.h (enum dump_kind): Likewise.
11261 * genmatch.c (dt_simplify::gen_1): Use it.
11263 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11265 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11267 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11269 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11270 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11272 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11274 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11275 * match.pd (X == C): Rewrite it here.
11276 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11277 with_certain_nonzero_bits2): New predicates.
11278 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11280 2017-05-24 Nathan Sidwell <nathan@acm.org>
11282 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11285 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11288 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11290 * config/powerpcspe: New port. Files are copied from the rs6000
11291 port, with "rs6000" in filenames replaced by "powerpcspe".
11293 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11295 PR rtl-optimization/80754
11296 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11298 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11300 * config/sparc/sparc.md (length): Return the correct value for -mflat
11301 sibcalls to match output_sibcall.
11303 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11307 * config/rs6000/rs6000.c (struct machine_function): Add new field
11309 (rs6000_get_separate_components): Init that field, use it.
11310 (rs6000_components_for_bb): Use the field.
11312 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11314 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11316 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11318 PR middle-end/80823
11319 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11321 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11324 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11325 address against address_operand predicate.
11326 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11328 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11330 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11331 parameters passed indirectly.
11333 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11335 * config/i386/i386.md (*movdi_internal): Remove SSE4
11336 alternative 18 (?r, *v). Update insn attributes.
11337 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11338 Update insn attributes.
11339 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11340 Update insn attributes.
11341 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11342 alternative 1 (r, v). Remove isa attribute.
11343 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11344 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11345 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11347 2017-05-23 Tom de Vries <tom@codesourcery.com>
11349 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11352 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11354 * cgraphunit.c (symbol_table::process_new_functions): Update.
11355 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11356 (inline_generate_summary): Rename to ...
11357 (ipa_fn_summary_generate): ... this one.
11358 (inline_read_summary): Rename to ...
11359 (ipa_fn_summary_read): ... this one.
11360 (inline_write_summary): Rename to ...
11361 (ipa_fn_summary_write): ... this one.
11362 (inline_free_summary): Rename to ...
11363 (ipa_free_fn_summary): ... this one.
11364 (pass_data_local_fn_summary, pass_local_fn_summary,
11365 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11366 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11367 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11368 make_pass_ipa_fn_summary): New.
11369 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11370 inline_write_summary, inline_free_summary): Remove.
11371 (ipa_free_fn_summary) : New.
11372 * ipa-inline.c (ipa_inline): Update.
11373 (pass_ipa_inline): Do not generate summaries.
11374 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11376 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11377 and add pass_ipa_fn_summary.
11378 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11380 (make_pass_inline_parameters): Remove.
11382 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11384 * omp-low.c (struct omp_context): Remove "default_kind" member.
11387 * omp-offload.c (execute_oacc_device_lower): Remove the
11388 parallelism dimensions function attributes for unparallelized
11389 OpenACC kernels constructs.
11391 2017-05-23 Martin Liska <mliska@suse.cz>
11393 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11395 (cgraph_edge::make_speculative): Likewise.
11396 (cgraph_edge::resolve_speculation): Likewise.
11397 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11398 (cgraph_node::dump): Likewise.
11399 * cgraph.h: Likewise.
11400 * cgraphunit.c (analyze_functions): Likewise.
11401 (symbol_table::compile): Likewise.
11402 * ipa-cp.c (print_all_lattices): Likewise.
11403 (determine_versionability): Likewise.
11404 (initialize_node_lattices): Likewise.
11405 (ipcp_verify_propagated_values): Likewise.
11406 (estimate_local_effects): Likewise.
11407 (update_profiling_info): Likewise.
11408 (create_specialized_node): Likewise.
11409 (perhaps_add_new_callers): Likewise.
11410 (decide_about_value): Likewise.
11411 (decide_whether_version_node): Likewise.
11412 (identify_dead_nodes): Likewise.
11413 (ipcp_store_bits_results): Likewise.
11414 * ipa-devirt.c (dump_targets): Likewise.
11415 (ipa_devirt): Likewise.
11416 * ipa-icf.c (sem_item::dump): Likewise.
11417 (sem_function::equals): Likewise.
11418 (sem_variable::equals): Likewise.
11419 (sem_item_optimizer::read_section): Likewise.
11420 (sem_item_optimizer::execute): Likewise.
11421 (congruence_class::dump): Likewise.
11422 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11423 (dump_inline_summary): Likewise.
11424 (estimate_node_size_and_time): Likewise.
11425 (inline_analyze_function): Likewise.
11426 * ipa-inline-transform.c (inline_call): Likewise.
11427 * ipa-inline.c (report_inline_failed_reason): Likewise.
11428 (want_early_inline_function_p): Likewise.
11429 (edge_badness): Likewise.
11430 (update_edge_key): Likewise.
11431 (inline_small_functions): Likewise.
11432 * ipa-profile.c (ipa_profile): Likewise.
11433 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11434 (ipa_make_edge_direct_to_target): Likewise.
11435 (remove_described_reference): Likewise.
11436 (ipa_impossible_devirt_target): Likewise.
11437 (propagate_controlled_uses): Likewise.
11438 (ipa_print_node_params): Likewise.
11439 (ipcp_transform_function): Likewise.
11440 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11441 (propagate_pure_const): Likewise.
11442 * ipa-reference.c (generate_summary): Likewise.
11443 (read_write_all_from_decl): Likewise.
11444 (propagate): Likewise.
11445 (ipa_reference_read_optimization_summary): Likewise.
11446 * ipa-utils.c (ipa_merge_profiles): Likewise.
11447 * ipa.c (walk_polymorphic_call_targets): Likewise.
11448 (symbol_table::remove_unreachable_nodes): Likewise.
11449 (ipa_single_use): Likewise.
11450 * passes.c (execute_todo): Likewise.
11451 * predict.c (drop_profile): Likewise.
11452 * symtab.c (symtab_node::get_dump_name): New function.
11453 (symtab_node::dump_name): Likewise.
11454 (symtab_node::dump_asm_name): Likewise.
11455 (symtab_node::dump_references): Likewise.
11456 (symtab_node::dump_referring): Likewise.
11457 (symtab_node::dump_base): Likewise.
11458 (symtab_node::debug_symtab): Likewise.
11459 * tree-sra.c (convert_callers_for_node): Likewise.
11460 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11461 * value-prof.c (init_node_map): Likewise.
11463 2017-05-23 Martin Liska <mliska@suse.cz>
11465 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11466 and symtab_node::debug_symtab to symbol_table::debug.
11467 * cgraphunit.c (analyze_functions): Use the renamed function.
11468 (symbol_table::compile): Likewise.
11469 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11470 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11471 * passes.c (execute_todo): Likewise.
11472 * symtab.c (symbol_table::dump): New function.
11473 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11475 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11477 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11478 that nonconst implies exec.
11480 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11482 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11483 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11484 (inline_edge_summary_vec): Turn into ...
11485 (ipa_call_summaries): ... this one.
11486 (redirect_to_unreachable, edge_set_predicate,
11487 evaluate_properties_for_edge, inline_summary_alloc,
11488 reset_ipa_call_summary, reset_inline_summary,
11489 inline_summary_t::duplicate): Update.
11490 (inline_edge_duplication_hook): Turn to ...
11491 (ipa_call_summary_t::duplicate): ... this one.
11492 (inline_edge_removal_hook): Turn to ...
11493 (ipa_call_summary_t::remove): ... this one.
11494 (dump_inline_edge_summary): Turn to ...
11495 (dump_ipa_call_summary): ... this one.
11496 (estimate_function_body_sizes): Update.
11497 (inline_update_callee_summaries): Update.
11498 (remap_edge_change_prob): Update.
11499 (remap_edge_summaries): Update.
11500 (inline_merge_summary): Update.
11501 (do_estimate_edge_time): Update.
11502 (inline_generate_summary): Update.
11503 (inline_read_section): Update.
11504 (inline_read_summary): Update.
11505 (inline_free_summary): Update.
11506 * ipa-inline.c (can_inline_edge_p): Update.
11507 (compute_inlined_call_time): Update.
11508 (want_inline_small_function_p): Update.
11509 (edge_badness): Update.
11510 (early_inliner): Update.
11511 * ipa-inline.h (inline_edge_summary): Turn to ...
11512 (ipa_call_summary): ... this one.
11513 (ipa_call_summary_t): New class.
11514 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11515 (ipa_call_summaries): New.
11516 (inline_edge_summary): Remove.
11517 (estimate_edge_growth): Update.
11518 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11519 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11520 * ipa-split.c (execute_split_functions): Update.
11521 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11523 2017-05-23 Tom de Vries <tom@codesourcery.com>
11525 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11526 attributes): Document rdrand effective target.
11528 2017-05-23 Tom de Vries <tom@codesourcery.com>
11530 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11531 attributes): Sort alphabetically.
11533 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11535 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11537 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11540 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11541 V2DF/V2DI splat into two separate patterns, one that handles
11542 registers, and the other that only handles memory. Drop support
11543 for splatting from a GPR on ISA 2.07 and then splitting the
11544 splat into direct move and splat.
11545 (vsx_splat_<mode>_reg): Likewise.
11546 (vsx_splat_<mode>_mem): Likewise.
11548 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11550 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11552 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11554 PR middle-end/80809
11555 * omp-low.c (finish_taskreg_remap): New function.
11556 (finish_taskreg_scan): If unit size of ctx->record_type
11557 is non-constant, unshare the size expression and replace
11558 decls in it with possible outer var refs.
11560 PR middle-end/80809
11561 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11562 GOVD_SHARED rather than GOVD_PRIVATE with it.
11563 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11564 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11566 PR middle-end/80853
11567 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11568 as last argument to build_outer_var_ref for pointer bases of array
11569 section reductions.
11571 2017-05-19 Martin Sebor <msebor@redhat.com>
11573 * print-tree.c (print_node): Print DECL_READ_P flag.
11575 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11577 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11578 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11579 * cgraph.c: Likewise.
11580 * cgraphunit.c: Likewise.
11581 * gengtype.c: Likewise.
11582 * ipa-cp.c: Likewise.
11583 * ipa-devirt.c: Likewise.
11584 * ipa-icf.c: Likewise.
11585 * ipa-predicate.c: Likewise.
11586 * ipa-profile.c: Likewise.
11587 * ipa-prop.c: Likewise.
11588 * ipa-split.c: Likewise.
11590 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11591 edge_predicate_pool, dump_inline_hints,
11592 inline_summary::account_size_time, redirect_to_unreachable,
11593 edge_set_predicate, set_hint_predicate,
11594 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11595 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11596 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11597 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11598 ipa_call_summary_t::remove, initialize_growth_caches,
11599 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11600 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11601 mark_modified, unmodified_parm_1, unmodified_parm,
11602 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11603 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11604 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11605 will_be_nonconstant_predicate, record_modified_bb_info,
11606 get_minimal_bb, record_modified, param_change_prob,
11607 phi_result_unknown_predicate, predicate_for_phi_result,
11608 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11609 estimate_function_body_sizes, compute_inline_parameters,
11610 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11611 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11612 inline_update_callee_summaries, remap_edge_change_prob,
11613 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11614 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11615 inline_analyze_function, inline_summary_t::insert,
11616 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11617 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11618 inline_free_summary): Move to ipa-fnsummary.h
11619 (predicate_t): Remove.
11620 * ipa-fnsummary.c: New file.
11621 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11622 (enum inline_hints_vals, inline_hints, agg_position_info,
11623 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11624 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11625 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11626 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11627 inline_read_summary, inline_write_summary, inline_free_summary,
11628 inline_analyze_function, initialize_inline_failed,
11629 inline_merge_summary, inline_update_overall_summary,
11630 compute_inline_parameters): Move to ipa-fnsummary.h
11631 * ipa-fnsummary.h: New file.
11632 * ipa-inline-transform.h: Include ipa-inline.h.
11633 * ipa-inline.c: LIkewise.
11635 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11637 * ipa-inline.c (edge_badness): Use inlined_time instead of
11638 inline_summaries->get.
11640 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11642 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11644 2017-05-22 Nathan Sidwell <nathan@acm.org>
11646 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11647 (fdump-lang): Document 'raw' option.
11648 * dumpfile.h (TDI_tu): Delete.
11649 * dumpfile.c (dump_files): Remove translation-unit.
11650 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11652 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11654 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11655 command option from $(AWK) call.
11656 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11657 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11658 [FORMAT]: Remove handling of variable.
11659 * config/avr/t-multilib: Regenerate.
11661 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11663 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11665 (dump_inline_summary): Do not print self_time.
11666 (estimate_function_body_sizes): Do not set self_time.
11667 (compute_inline_parameters): Likewise.
11668 (inline_read_section, inline_write_summary): Do not stream self_time.
11669 * ipa-inline.h (inline_summary): Drop self_time.
11671 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11673 * ipa-inline-analysis.c (account_size_time): Rename to ...
11674 (inline_summary::account_size_time): ... this one.
11675 (reset_ipa_call_summary): Turn to ...
11676 (ipa_call_summary::reset): ... this one.
11677 (reset_inline_summary): Turn to ...
11678 (inline_summary::reset): ... this one.
11679 (inline_summary_t::remove): Update.
11680 (inline_summary_t::duplicate): Update.
11681 (ipa_call_summary_t::remove): Update.
11682 (dump_inline_summary): Update.
11683 (estimate_function_body_sizes): Update.
11684 (compute_inline_parameters): Update.
11685 (estimate_node_size_and_time): Update.
11686 (inline_merge_summary): Update.
11687 (inline_update_overall_summary): Update.
11688 (inline_read_section): Update.
11689 (inline_write_summary): Update.
11690 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
11691 add account_size_time and reset member functions.
11692 (ipa_call_summary): Add reset function.
11693 * ipa-predicate.h (predicate::operator &): Constify.
11695 2017-05-22 Richard Biener <rguenther@suse.de>
11697 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
11699 2017-05-19 Jason Merrill <jason@redhat.com>
11701 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
11703 2017-05-19 Marek Polacek <polacek@redhat.com>
11706 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
11707 TYPE_OVERFLOW_WRAPS checks.
11709 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
11711 * tree-core.h (enum omp_clause_default_kind): Add
11712 "OMP_CLAUSE_DEFAULT_PRESENT".
11713 * tree-pretty-print.c (dump_omp_clause): Handle it.
11714 * gimplify.c (enum gimplify_omp_var_data): Add
11715 "GOVD_MAP_FORCE_PRESENT".
11716 (gimplify_adjust_omp_clauses_1): Map it to
11717 "GOMP_MAP_FORCE_PRESENT".
11718 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
11720 * gimplify.c (oacc_default_clause): Clarify.
11722 2017-05-19 Nathan Sidwell <nathan@acm.org>
11724 LANG_HOOK_REGISTER_DUMPS
11725 * toplev.c (general_init): Call register dump lang hook.
11726 * doc/invoke.texi: Document -fdump-lang option family.
11727 * dumpfile.c (dump_files): Remove class dump here.
11728 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
11729 * dumpfile.h (tree_dump_index): Remove TDI_class.
11730 * langhooks-def.h (lhd_register_dumps): Declare.
11731 (LANG_HOOKS_REGISTER_DUMPS): Define.
11732 (LANG_HOOKS_INITIALIZER): Add it.
11733 * langhooks.c (lhd_register_dumps): Define.
11734 * langhooks.h (struct lang_hooks): Add register_dumps.
11736 2017-05-19 Nathan Sidwell <nathan@acm.org>
11738 * context.h (context::set_passes): New.
11739 * context.c (context::context): Do not create pass manager.
11740 * toplev.c (general_init): Create pass manager here.
11742 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
11744 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
11745 use this splitter if two add or or instructions would also work for
11746 the constant we want to generate.
11748 2017-05-19 Richard Biener <rguenther@suse.de>
11751 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
11752 predicate evaluation.
11754 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11756 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
11758 * ipa-inline.c (want_inline_small_function_p): Do not cast to
11761 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11763 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11764 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11765 (inline_edge_summary_vec): Turn into ...
11766 (ipa_call_summaries): ... this one.
11767 (redirect_to_unreachable, edge_set_predicate,
11768 evaluate_properties_for_edge, inline_summary_alloc,
11769 reset_ipa_call_summary, reset_inline_summary,
11770 inline_summary_t::duplicate): Update.
11771 (inline_edge_duplication_hook): Turn to ...
11772 (ipa_call_summary_t::duplicate): ... this one.
11773 (inline_edge_removal_hook): Turn to ...
11774 (ipa_call_summary_t::remove): ... this one.
11775 (dump_inline_edge_summary): Turn to ...
11776 (dump_ipa_call_summary): ... this one.
11777 (estimate_function_body_sizes): Update.
11778 (inline_update_callee_summaries): Update.
11779 (remap_edge_change_prob): Update.
11780 (remap_edge_summaries): Update.
11781 (inline_merge_summary): Update.
11782 (do_estimate_edge_time): Update.
11783 (inline_generate_summary): Update.
11784 (inline_read_section): Update.
11785 (inline_read_summary): Update.
11786 (inline_free_summary): Update.
11787 * ipa-inline.c (can_inline_edge_p): Update.
11788 (compute_inlined_call_time): Update.
11789 (want_inline_small_function_p): Update.
11790 (edge_badness): Update.
11791 (early_inliner): Update.
11792 * ipa-inline.h (inline_edge_summary): Turn to ...
11793 (ipa_call_summary): ... this one.
11794 (ipa_call_summary_t): New class.
11795 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11796 (ipa_call_summaries): New.
11797 (inline_edge_summary): Remove.
11798 (estimate_edge_growth): Update.
11799 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11800 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11801 * ipa-split.c (execute_split_functions): Update.
11802 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11804 2017-05-19 Richard Biener <rguenther@suse.de>
11806 PR middle-end/80764
11807 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
11809 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
11811 * config/rs6000/rs6000.c (struct machine_function): Add field
11812 fpr_is_wrapped_separately.
11813 (rs6000_get_separate_components): Use 64 components. Handle the
11814 new FPR components.
11815 (rs6000_components_for_bb): Handle the FPR components.
11816 (rs6000_emit_prologue_components): Handle the FPR components.
11817 (rs6000_emit_epilogue_components): Handle the FPR components.
11818 (rs6000_set_handled_components): Handle the FPR components.
11819 (rs6000_emit_prologue): Don't output prologue code for those FPRs
11820 that are already separately shrink-wrapped.
11821 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
11822 that are already separately shrink-wrapped.
11824 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11827 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
11830 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
11831 (define_peephole2 for Altivec d-form load): Add peepholes to catch
11832 cases where the register allocator uses a move and an offsettable
11833 memory operation to/from a FPR register on ISA 2.06/2.07.
11834 (define_peephole2 for Altivec d-form store): Likewise.
11836 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
11839 * config/i386/mmx.md (*mov<mode>_internal): Enable
11840 alternatives 11, 12, 13 and 14 also for 32bit targets.
11841 Remove alternatives 15, 16, 17 and 18.
11842 * config/i386/sse.md (vec_concatv2di): Change
11843 alternative (!x, *y) to (x, ?!*Yn).
11845 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
11847 * dumpfile.h (enum dump_kind): Remove stray comma.
11849 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11851 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
11852 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
11853 predicate::num_conditions
11854 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
11855 (CHANGED): turn into predicate::changed.
11856 (agg_position_info): Move to ipa-predicate.h
11857 (add_condition, predicate::add_clause, predicate::operator &=,
11858 predicate::or_with, predicate::evaluate, predicate::probability,
11859 dump_condition, dump_clause, predicate::dump,
11860 predicate::remap_after_duplication, predicate::remap_after_inlining,
11861 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
11862 (evaluate_conditions_for_known_args): Update.
11863 (set_cond_stmt_execution_predicate): Update.
11864 * ipa-inline.h: Include ipa-predicate.h
11865 (condition, inline_param_summary, conditions, agg_position_info,
11866 predicate): Move to ipa-predicate.h
11867 * ipa-predicate.c: New file.
11868 * ipa-predicate.h: New file.
11870 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
11872 * final.c (leaf_function_p): Check we are not in a sequence.
11874 2017-05-18 Martin Liska <mliska@suse.cz>
11876 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
11877 * dumpfile.c (dump_register): Use new enum dump_kind.
11878 (get_dump_file_name): Likewise.
11879 (dump_enable_all): Likewise.
11880 (dump_switch_p_1): Likewise.
11881 (enable_rtl_dump_file): Remove usage of TDF_RTL.
11882 * dumpfile.h (enum dump_kind): New enum type.
11883 (struct dump_file_info): Create constructor and
11884 format fields and comments.
11885 * passes.c (pass_manager::register_one_dump_file):
11887 * statistics.c (statistics_early_init): Likewise.
11888 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
11889 TDF_TREE with TDF_SLIM.
11890 (gather_memory_references_ref): Likewise.
11892 2017-05-18 Martin Liska <mliska@suse.cz>
11894 * vec.h (struct vnull): Use it.
11896 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11898 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
11899 (true_predicate, false_predicate, true_predicate_p,
11900 false_predicate_p): Remove.
11901 (single_cond_predicate, not_inlined_predicate): Turn to member function
11903 (add_condition): Update.
11904 (add_clause): Turn to...
11905 (predicate::add_clause): ... this one; update; allow passing NULL
11907 (and_predicates): Turn to ...
11908 (predicate::operator &=): ... this one.
11909 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
11910 (or_predicates): Turn to ...
11911 (predicate::or_with): ... this one.
11912 (evaluate_predicate): Turn to ...
11913 (predicate::evaluate): ... this one.
11914 (predicate_probability): Turn to ...
11915 (predicate::probability): ... this one.
11916 (dump_condition): Update.
11917 (dump_predicate): Turn to ...
11918 (predicate::dump): ... this one.
11919 (account_size_time): Update.
11920 (edge_set_predicate): Update.
11921 (set_hint_predicate): UPdate.
11922 (evaluate_conditions_for_known_args): Update.
11923 (evaluate_properties_for_edge): Update.
11924 (remap_predicate_after_duplication): Turn to...
11925 (predicate::remap_after_duplication): ... this one.
11926 (remap_hint_predicate_after_duplication): Update.
11927 (inline_summary_t::duplicate): UPdate.
11928 (dump_inline_edge_summary): Update.
11929 (dump_inline_summary): Update.
11930 (set_cond_stmt_execution_predicate): Update.
11931 (set_switch_stmt_execution_predicate): Update.
11932 (compute_bb_predicates): Update.
11933 (will_be_nonconstant_expr_predicate): Update.
11934 (will_be_nonconstant_predicate): Update.
11935 (phi_result_unknown_predicate): Update.
11936 (predicate_for_phi_result): Update.
11937 (array_index_predicate): Update.
11938 (estimate_function_body_sizes): Update.
11939 (estimate_node_size_and_time): Update.
11940 (estimate_ipcp_clone_size_and_time): Update.
11941 (remap_predicate): Rename to ...
11942 (predicate::remap_after_inlining): ... this one.
11943 (remap_hint_predicate): Update.
11944 (inline_merge_summary): Update.
11945 (inline_update_overall_summary): Update.
11946 (estimate_size_after_inlining): Update.
11947 (read_predicate): Rename to ...
11948 (predicate::stream_in): ... this one.
11949 (read_inline_edge_summary): Update.
11950 (write_predicate): Rename to ...
11951 (predicate::stream_out): ... this one.
11952 (write_inline_edge_summary): Update.
11953 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
11954 (clause_t): Turn to uint32_t
11955 (predicate): Turn to class; implement constructor and operators
11957 (size_time_entry): Update.
11958 (inline_summary): Update.
11959 (inline_edge_summary): Update.
11961 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
11963 * fold-const.c (fold_binary_loc): Move transformation...
11964 * match.pd (C - X CMP X): ... here.
11966 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
11968 * config/sparc/sparc.c (sparc_option_override): Set function
11969 alignment for -mcpu=niagara7 to 64 to match the I$ line.
11970 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
11972 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
11974 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
11976 2017-05-18 Marek Polacek <polacek@redhat.com>
11979 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
11980 (pass_ubsan::execute): Call gimple_assign_single_p instead of
11981 gimple_assign_load_p.
11983 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
11985 PR middle-end/80692
11986 * real.c (do_compare): Give decimal_do_compare preference over
11987 comparing just the signs.
11989 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
11991 * doc/md.texi (Canonicalization of Instructions): Describe the
11992 canonical form of instructions that inherently set a condition
11995 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
11997 PR middle-end/80775
11998 * tree-cfg.c: Move deletion of unreachable case statements to after
11999 the merging of consecutive case labels.
12001 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12003 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12004 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12005 restoring of callee-saved registers.
12007 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12009 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12010 * config/visium/visium.c (single_set_and_flags): Likewise.
12011 * config/visium/visium.md (Substitutions): Likewise.
12013 2017-05-17 Martin Liska <mliska@suse.cz>
12015 * cfg.c: Introduce dump_flags_t type and
12016 use it instead of int type.
12018 * cfghooks.c: Likewise.
12019 * cfghooks.h (struct cfg_hooks): Likewise.
12020 * cfgrtl.c: Likewise.
12021 * cfgrtl.h: Likewise.
12022 * cgraph.c (cgraph_node::get_body): Likewise.
12023 * coretypes.h: Likewise.
12024 * domwalk.c: Likewise.
12025 * domwalk.h: Likewise.
12026 * dumpfile.c (struct dump_option_value_info): Likewise.
12027 (dump_enable_all): Likewise.
12028 (dump_switch_p_1): Likewise.
12029 (opt_info_switch_p): Likewise.
12030 * dumpfile.h (enum tree_dump_index): Likewise.
12031 (struct dump_file_info): Likewise.
12032 * genemit.c: Likewise.
12033 * generic-match-head.c: Likewise.
12034 * gengtype.c (open_base_files): Likewise.
12035 * gimple-pretty-print.c: Likewise.
12036 * gimple-pretty-print.h: Likewise.
12037 * graph.c (print_graph_cfg): Likewise.
12038 * graphite-scop-detection.c (dot_all_sese): Likewise.
12039 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12040 * loop-unroll.c (report_unroll): Likewise.
12041 * passes.c (pass_manager::register_one_dump_file): Likewise.
12042 * print-tree.c: Likewise.
12043 * statistics.c: Likewise.
12044 * tree-cfg.c: Likewise.
12045 * tree-cfg.h: Likewise.
12046 * tree-dfa.c: Likewise.
12047 * tree-dfa.h: Likewise.
12048 * tree-dump.c (dump_function): Likewise.
12049 * tree-dump.h (struct dump_info): Likewise.
12050 * tree-pretty-print.c: Likewise.
12051 * tree-pretty-print.h: Likewise.
12052 * tree-ssa-live.c: Likewise.
12053 * tree-ssa-live.h: Likewise.
12054 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12055 * tree-vect-loop.c: Likewise.
12056 * tree-vect-slp.c: Likewise.
12058 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12059 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12061 PR tree-optimization/80457
12062 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12063 of all arguments to a statement as scalar_to_vec operations.
12064 (vectorizable_call): Adjust call to vect_model_simple_cost for
12066 (vectorizable_conversion): Likewise.
12067 (vectorizable_assignment): Likewise.
12068 (vectorizable_shift): Likewise.
12069 (vectorizable_operation): Likewise.
12070 (vectorizable_comparison): Likewise.
12071 (vect_is_simple_cond): Record the def types for operands.
12072 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12073 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12074 for statement argument count.
12076 2017-05-16 Carl Love <cel@us.ibm.com>
12078 * config/rs6000/rs6000-c: Add support for built-in functions
12079 vector unsigned long long vec_bperm (vector unsigned long long,
12080 vector unsigned char)
12081 vector signed long long vec_mule (vector signed int,
12083 vector unsigned long long vec_mule (vector unsigned int,
12084 vector unsigned int)
12085 vector signed long long vec_mulo (vector signed int,
12087 vector unsigned long long vec_mulo (vector unsigned int,
12088 vector unsigned int)
12089 vector signed char vec_sldw (vector signed char,
12090 vector signed char,
12092 vector unsigned char vec_sldw (vector unsigned char,
12093 vector unsigned char,
12095 vector signed short vec_sldw (vector signed short,
12096 vector signed short,
12098 vector unsigned short vec_sldw (vector unsigned short,
12099 vector unsigned short,
12101 vector signed int vec_sldw (vector signed int,
12104 vector unsigned int vec_sldw (vector unsigned int,
12105 vector unsigned int,
12107 vector signed long long vec_sldw (vector signed long long,
12108 vector signed long long,
12110 vector unsigned long long vec_sldw (vector unsigned long long,
12111 vector unsigned long long,
12113 * config/rs6000/rs6000-c: Add support for built-in functions
12114 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12115 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12116 * doc/extend.texi: Update the built-in documentation for the
12117 new built-in functions.
12119 2017-05-16 Marek Polacek <polacek@redhat.com>
12123 * tree.c (save_expr): Don't fold the expression.
12125 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12127 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12128 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12129 and (?*y,m). Update insn attributes.
12131 2017-05-16 Martin Liska <mliska@suse.cz>
12133 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12134 flags argument of print_gimple_stmt, print_gimple_expr,
12135 print_generic_stmt and print_generic_expr.
12136 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12137 * coretypes.h: Likewise.
12138 * except.c (dump_eh_tree): Likewise.
12139 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12140 * gimple-pretty-print.h: Likewise.
12141 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12142 (backprop::push_to_worklist): Likewise.
12143 (backprop::pop_from_worklist): Likewise.
12144 (backprop::process_use): Likewise.
12145 (backprop::intersect_uses): Likewise.
12146 (note_replacement): Likewise.
12147 * gimple-ssa-store-merging.c
12148 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12149 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12150 (pass_store_merging::execute): Likewise.
12151 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12152 (ssa_base_cand_dump_callback): Likewise.
12153 (dump_incr_vec): Likewise.
12154 (replace_refs): Likewise.
12155 (replace_mult_candidate): Likewise.
12156 (create_add_on_incoming_edge): Likewise.
12157 (create_phi_basis): Likewise.
12158 (insert_initializers): Likewise.
12159 (all_phi_incrs_profitable): Likewise.
12160 (introduce_cast_before_cand): Likewise.
12161 (replace_one_candidate): Likewise.
12162 * gimplify.c (gimplify_expr): Likewise.
12163 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12164 (set_rename): Likewise.
12165 (rename_uses): Likewise.
12166 (copy_loop_phi_nodes): Likewise.
12167 (add_close_phis_to_merge_points): Likewise.
12168 (copy_loop_close_phi_args): Likewise.
12169 (copy_cond_phi_args): Likewise.
12170 (graphite_copy_stmts_from_block): Likewise.
12171 (translate_pending_phi_nodes): Likewise.
12172 * graphite-poly.c (print_pdr): Likewise.
12173 (dump_gbb_cases): Likewise.
12174 (dump_gbb_conditions): Likewise.
12175 (print_scop_params): Likewise.
12176 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12177 (build_cross_bb_scalars_use): Likewise.
12178 (gather_bbs::before_dom_children): Likewise.
12179 * hsa-dump.c (dump_hsa_immed): Likewise.
12180 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12181 (get_replacement_map): Likewise.
12182 * ipa-inline-analysis.c (dump_condition): Likewise.
12183 (estimate_function_body_sizes): Likewise.
12184 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12185 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12186 * ipa-prop.c (ipa_dump_param): Likewise.
12187 (ipa_print_node_jump_functions_for_edge): Likewise.
12188 (ipa_modify_call_arguments): Likewise.
12189 (ipa_modify_expr): Likewise.
12190 (ipa_dump_param_adjustments): Likewise.
12191 (ipa_dump_agg_replacement_values): Likewise.
12192 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12193 * ipa-pure-const.c (check_stmt): Likewise.
12194 (pass_nothrow::execute): Likewise.
12195 * ipa-split.c (execute_split_functions): Likewise.
12196 * omp-offload.c (dump_oacc_loop_part): Likewise.
12197 (dump_oacc_loop): Likewise.
12198 * trans-mem.c (tm_log_emit): Likewise.
12199 (tm_memopt_accumulate_memops): Likewise.
12200 (dump_tm_memopt_set): Likewise.
12201 (dump_tm_memopt_transform): Likewise.
12202 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12203 (print_loop): Likewise.
12204 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12205 (chkp_gather_checks_info): Likewise.
12206 (chkp_get_check_result): Likewise.
12207 (chkp_remove_check_if_pass): Likewise.
12208 (chkp_use_outer_bounds_if_possible): Likewise.
12209 (chkp_reduce_bounds_lifetime): Likewise.
12210 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12211 (chkp_mark_completed_bounds): Likewise.
12212 (chkp_register_incomplete_bounds): Likewise.
12213 (chkp_mark_invalid_bounds): Likewise.
12214 (chkp_maybe_copy_and_register_bounds): Likewise.
12215 (chkp_build_returned_bound): Likewise.
12216 (chkp_get_bound_for_parm): Likewise.
12217 (chkp_build_bndldx): Likewise.
12218 (chkp_get_bounds_by_definition): Likewise.
12219 (chkp_generate_extern_var_bounds): Likewise.
12220 (chkp_get_bounds_for_decl_addr): Likewise.
12221 * tree-chrec.c (chrec_apply): Likewise.
12222 * tree-data-ref.c (dump_data_reference): Likewise.
12223 (dump_subscript): Likewise.
12224 (dump_data_dependence_relation): Likewise.
12225 (analyze_overlapping_iterations): Likewise.
12226 * tree-inline.c (expand_call_inline): Likewise.
12227 (tree_function_versioning): Likewise.
12228 * tree-into-ssa.c (dump_defs_stack): Likewise.
12229 (dump_currdefs): Likewise.
12230 (dump_names_replaced_by): Likewise.
12231 (dump_update_ssa): Likewise.
12232 (update_ssa): Likewise.
12233 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12234 * tree-parloops.c (build_new_reduction): Likewise.
12235 (try_create_reduction_list): Likewise.
12236 (ref_conflicts_with_region): Likewise.
12237 (oacc_entry_exit_ok_1): Likewise.
12238 (oacc_entry_exit_single_gang): Likewise.
12239 * tree-pretty-print.h: Likewise.
12240 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12241 (get_scalar_evolution): Likewise.
12242 (add_to_evolution): Likewise.
12243 (get_loop_exit_condition): Likewise.
12244 (analyze_evolution_in_loop): Likewise.
12245 (analyze_initial_condition): Likewise.
12246 (analyze_scalar_evolution): Likewise.
12247 (instantiate_scev): Likewise.
12248 (number_of_latch_executions): Likewise.
12249 (gather_chrec_stats): Likewise.
12250 (final_value_replacement_loop): Likewise.
12251 (scev_const_prop): Likewise.
12252 * tree-sra.c (dump_access): Likewise.
12253 (disqualify_candidate): Likewise.
12254 (create_access): Likewise.
12255 (reject): Likewise.
12256 (maybe_add_sra_candidate): Likewise.
12257 (create_access_replacement): Likewise.
12258 (analyze_access_subtree): Likewise.
12259 (analyze_all_variable_accesses): Likewise.
12260 (sra_modify_assign): Likewise.
12261 (initialize_constant_pool_replacements): Likewise.
12262 (find_param_candidates): Likewise.
12263 (decide_one_param_reduction): Likewise.
12264 (replace_removed_params_ssa_names): Likewise.
12265 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12266 * tree-ssa-copy.c (dump_copy_of): Likewise.
12267 (copy_prop_visit_cond_stmt): Likewise.
12268 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12269 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12270 (record_equivalences_from_stmt): Likewise.
12271 * tree-ssa-dse.c (compute_trims): Likewise.
12272 (delete_dead_call): Likewise.
12273 (delete_dead_assignment): Likewise.
12274 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12275 (forward_propagate_into_cond): Likewise.
12276 (pass_forwprop::execute): Likewise.
12277 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12278 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12280 (move_computations_worker): Likewise.
12281 (execute_sm): Likewise.
12282 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12283 (remove_exits_and_undefined_stmts): Likewise.
12284 (remove_redundant_iv_tests): Likewise.
12285 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12286 (adjust_iv_update_pos): Likewise.
12287 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12288 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12289 (value_replacement): Likewise.
12290 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12291 * tree-ssa-pre.c (print_pre_expr): Likewise.
12292 (get_representative_for): Likewise.
12293 (create_expression_by_pieces): Likewise.
12294 (insert_into_preds_of_block): Likewise.
12295 (eliminate_insert): Likewise.
12296 (eliminate_dom_walker::before_dom_children): Likewise.
12297 (eliminate): Likewise.
12298 (remove_dead_inserted_code): Likewise.
12299 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12300 * tree-ssa-reassoc.c (get_rank): Likewise.
12301 (eliminate_duplicate_pair): Likewise.
12302 (eliminate_plus_minus_pair): Likewise.
12303 (eliminate_not_pairs): Likewise.
12304 (undistribute_ops_list): Likewise.
12305 (eliminate_redundant_comparison): Likewise.
12306 (update_range_test): Likewise.
12307 (optimize_range_tests_var_bound): Likewise.
12308 (optimize_vec_cond_expr): Likewise.
12309 (rewrite_expr_tree): Likewise.
12310 (rewrite_expr_tree_parallel): Likewise.
12311 (linearize_expr): Likewise.
12312 (break_up_subtract): Likewise.
12313 (linearize_expr_tree): Likewise.
12314 (attempt_builtin_powi): Likewise.
12315 (attempt_builtin_copysign): Likewise.
12316 (transform_stmt_to_copy): Likewise.
12317 (transform_stmt_to_multiply): Likewise.
12318 (dump_ops_vector): Likewise.
12319 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12320 (print_scc): Likewise.
12321 (set_ssa_val_to): Likewise.
12322 (visit_reference_op_store): Likewise.
12323 (visit_use): Likewise.
12324 (sccvn_dom_walker::before_dom_children): Likewise.
12325 (run_scc_vn): Likewise.
12326 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12328 (expr_hash_elt::print): Likewise.
12329 (const_and_copies::pop_to_marker): Likewise.
12330 (const_and_copies::record_const_or_copy_raw): Likewise.
12331 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12332 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12333 (dump_predicates): Likewise.
12334 (find_uninit_use): Likewise.
12335 (warn_uninitialized_phi): Likewise.
12336 (pass_late_warn_uninitialized::execute): Likewise.
12337 * tree-ssa.c (verify_vssa): Likewise.
12338 (verify_ssa): Likewise.
12339 (maybe_optimize_var): Likewise.
12340 * tree-vrp.c (dump_value_range): Likewise.
12341 (dump_all_value_ranges): Likewise.
12342 (dump_asserts_for): Likewise.
12343 (register_edge_assert_for_2): Likewise.
12344 (vrp_visit_cond_stmt): Likewise.
12345 (vrp_visit_switch_stmt): Likewise.
12346 (vrp_visit_stmt): Likewise.
12347 (vrp_visit_phi_node): Likewise.
12348 (simplify_cond_using_ranges_1): Likewise.
12349 (fold_predicate_in): Likewise.
12350 (evrp_dom_walker::before_dom_children): Likewise.
12351 (evrp_dom_walker::push_value_range): Likewise.
12352 (evrp_dom_walker::pop_value_range): Likewise.
12353 (execute_early_vrp): Likewise.
12355 2017-05-16 Richard Biener <rguenther@suse.de>
12357 * dwarf2out.c (loc_list_from_tree_1): Do not create
12358 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12360 2017-05-16 Richard Biener <rguenther@suse.de>
12362 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12364 (note_variable_value_in_expr): If we resolved the decl ref
12365 do not push to the stack.
12367 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12369 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12370 operations in fast-math mode.
12371 (vaddq_f16): Likewise.
12372 (vmul_f16): Likewise.
12373 (vmulq_f16): Likewise.
12374 (vsub_f16): Likewise.
12375 (vsubq_f16): Likewise.
12376 * config/arm/neon.md (add<mode>3): New.
12378 (fma:<VH:mode>3): New. Also remove outdated comment.
12381 2017-05-16 Martin Liska <mliska@suse.cz>
12385 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12386 (odr_types_equivalent_p): Likewise.
12388 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12390 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12392 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12395 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12396 non-interunit SSE move alternatives with '?'.
12397 (zero-extendsidi peephole2): New peephole to skip intermediate
12398 general register in SSE zero-extend sequence.
12400 2017-05-15 Jeff Law <law@redhat.com>
12402 * reorg.c (relax_delay_slots): Create a new variable to hold
12403 the temporary target rather than clobbering TARGET_LABEL.
12405 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12406 missing argument to extract_bit_field call.
12407 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12409 2017-05-15 Martin Liska <mliska@suse.cz>
12412 * gcc.c (process_command): Do not allow empty argument of -o option.
12414 2017-05-15 Renlin Li <renlin.li@arm.com>
12416 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12417 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12418 * config/aarch64/constraints.md (Usf): Add long call check.
12419 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12420 (call_value): Likewise.
12421 (sibcall): Likewise.
12422 (sibcall_value): Likewise.
12424 (call_value_insn): New.
12425 (sibcall_insn): Update rtx pattern.
12426 (sibcall_value_insn): Likewise.
12427 (call_internal): Remove.
12428 (call_value_internal): Likewise.
12429 (sibcall_internal): Likewise.
12430 (sibcall_value_internal): Likewise.
12431 (call_reg): Likewise.
12432 (call_symbol): Likewise.
12433 (call_value_reg): Likewise.
12434 (call_value_symbol): Likewise.
12436 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12439 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12441 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12443 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12444 compatible with CCGOCmode and with CCZmode.
12446 2017-05-14 Martin Sebor <msebor@redhat.com>
12448 PR middle-end/77671
12449 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12450 (gimple_fold_builtin_snprintf): Same.
12451 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12452 (gimple_fold_builtin_snprintf): Same.
12453 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12454 of character types.
12455 (is_call_safe): New function.
12456 (try_substitute_return_value): Call it.
12457 (try_simplify_call): New function.
12458 (pass_sprintf_length::handle_gimple_call): Call it.
12460 2017-05-14 Martin Sebor <msebor@redhat.com>
12462 PR middle-end/80669
12463 * builtins.c (expand_builtin_stpncpy): Simplify.
12465 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12467 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12468 * config/i386/i386.h
12469 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12470 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12471 (struct machine_function): Add new members call_ms2sysv,
12472 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12473 (struct machine_frame_state): New fields sp_realigned and
12474 sp_realigned_offset.
12475 * config/i386/i386.c
12476 (enum xlogue_stub): New enum.
12477 (enum xlogue_stub_sets): New enum.
12478 (class xlogue_layout): New class.
12479 (struct ix86_frame): New fields stack_realign_allocate_offset,
12480 stack_realign_offset and outlined_save_offset. Modify comments to
12481 detail stack layout when using out-of-line stubs.
12482 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12483 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12484 -mcall-ms2sysv-xlogues.
12485 (stub_managed_regs): New static variable.
12486 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12487 registers managed by out-of-line stub.
12488 (disable_call_ms2sysv_xlogues): New function.
12489 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12490 m->call_ms2sysv when appropriate and compute frame layout for
12492 (sp_valid_at, fp_valid_at): New inline functions.
12493 (choose_basereg): New function.
12494 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12496 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12497 Use align parameter of choose_baseaddr to generated aligned SSE movs
12499 (pro_epilogue_adjust_stack): Modify to track
12500 machine_frame_state::sp_realigned.
12501 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12502 (ix86_nsaved_sseregs): Likewise.
12503 (ix86_emit_save_regs): Likewise.
12504 (ix86_emit_save_regs_using_mov): Likewise.
12505 (ix86_emit_save_sse_regs_using_mov): Likewise.
12506 (get_scratch_register_on_entry): Likewise.
12507 (gen_frame_set): New function.
12508 (gen_frame_load): Likewise.
12509 (gen_frame_store): Likewise.
12510 (emit_outlined_ms2sysv_save): Likewise.
12511 (emit_outlined_ms2sysv_restore): Likewise.
12512 (ix86_expand_prologue): Modify stack re-alignment code and call
12513 emit_outlined_ms2sysv_save when appropriate.
12514 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12515 parameter rtx_insn *insn, which allows the function to be used to only
12516 generate the notes.
12517 (ix86_expand_epilogue): Modify validity checks of frame and stack
12518 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12519 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12520 * config/i386/predicates.md
12521 (save_multiple): New predicate.
12522 (restore_multiple): Likewise.
12523 * config/i386/sse.md
12524 (save_multiple<mode>): New pattern.
12525 (save_multiple_realign<mode>): Likewise.
12526 (restore_multiple<mode>): Likewise.
12527 (restore_multiple_and_return<mode>): Likewise.
12528 (restore_multiple_leave_return<mode>): Likewise.
12529 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12531 2017-05-14 Julia Koval <julia.koval@intel.com>
12533 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12534 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12535 (__builtin_ia32_xsetbv): New builtins.
12536 * config/i386/i386.c (ix86_expand_special_args_builtin):
12538 (ix86_expand_builtin): Special expand for new intrinsics.
12539 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12540 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12541 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12543 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12545 * cfganal.c (inverted_post_order_compute): Change argument type
12547 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12548 * df-core.c (rest_of_handle_df_initialize): Adjust.
12549 (rest_of_handle_df_finish): Likewise.
12550 (df_analyze_1): Likewise.
12551 (df_analyze): Likewise.
12552 (loop_inverted_post_order_compute): Change argument to be a vec *.
12553 (df_analyze_loop): Adjust.
12554 (df_get_n_blocks): Likewise.
12555 (df_get_postorder): Likewise.
12556 * df.h (struct df_d): Change field to be a vec.
12557 * lcm.c (compute_laterin): Adjust.
12558 (compute_available): Likewise.
12559 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12560 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12561 * tree-ssa-pre.c (compute_antic): Likewise.
12563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12565 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12566 (depth_first_search::depth_first_search): Change structure init
12567 function to this constructor.
12568 (depth_first_search::add_bb): Rename function to this member.
12569 (depth_first_search::execute): Likewise.
12570 (flow_dfs_compute_reverse_finish): Adjust.
12572 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12574 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12575 (longest_simple_path): Likewise.
12576 * shrink-wrap.c (spread_components): Likewise.
12577 (disqualify_problematic_components): Likewise.
12578 (emit_common_heads_for_components): Likewise.
12579 (emit_common_tails_for_components): Likewise.
12580 (insert_prologue_epilogue_for_components): Likewise.
12582 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12584 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12587 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12589 * df-core.c (df_set_blocks): Start using auto_bitmap.
12590 (df_compact_blocks): Likewise.
12591 * df-problems.c (df_rd_confluence_n): Likewise.
12592 * df-scan.c (df_insn_rescan_all): Likewise.
12593 (df_process_deferred_rescans): Likewise.
12594 (df_update_entry_block_defs): Likewise.
12595 (df_update_exit_block_uses): Likewise.
12596 (df_entry_block_bitmap_verify): Likewise.
12597 (df_exit_block_bitmap_verify): Likewise.
12598 (df_scan_verify): Likewise.
12599 * lra-constraints.c (lra_constraints): Likewise.
12600 (undo_optional_reloads): Likewise.
12601 (lra_undo_inheritance): Likewise.
12602 * lra-remat.c (calculate_gen_cands): Likewise.
12603 (do_remat): Likewise.
12604 * lra-spills.c (assign_spill_hard_regs): Likewise.
12605 (spill_pseudos): Likewise.
12606 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12607 (bitmap_set_subtract_values): Likewise.
12609 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12611 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12612 management with auto_bitmap.
12613 (fix_inter_tick): Likewise.
12614 (fix_recovery_deps): Likewise.
12615 * ira.c (add_store_equivs): Likewise.
12616 (find_moveable_pseudos): Likewise.
12617 (split_live_ranges_for_shrink_wrap): Likewise.
12618 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12619 (rtx_reuse_manager::seen_def_p): Likewise.
12620 (rtx_reuse_manager::set_seen_def): Likewise.
12621 * print-rtl.h (class rtx_reuse_manager): Likewise.
12623 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12625 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12627 (migrate_btr_def): Likewise.
12628 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12629 * df-core.c (loop_post_order_compute): Likewise.
12630 (loop_inverted_post_order_compute): Likewise.
12631 * hsa-common.h: Likewise.
12632 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12633 * init-regs.c (initialize_uninitialized_regs): Likewise.
12634 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12635 (inline_small_functions): Likewise.
12636 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12637 * ira.c (combine_and_move_insns): Likewise.
12638 (build_insn_chain): Likewise.
12639 * loop-invariant.c (find_invariants): Likewise.
12640 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12641 * predict.c (tree_predict_by_opcode): Likewise.
12642 (predict_paths_leading_to): Likewise.
12643 (predict_paths_leading_to_edge): Likewise.
12644 (estimate_loops_at_level): Likewise.
12645 (estimate_loops): Likewise.
12646 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12647 (spread_components): Likewise.
12648 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12649 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12650 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12651 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12652 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12653 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12654 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12655 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12656 (mark_threaded_blocks): Likewise.
12657 (thread_through_all_blocks): Likewise.
12658 * tree-ssa.c (verify_ssa): Likewise.
12659 (execute_update_addresses_taken): Likewise.
12660 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12662 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12664 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12666 (post_order_compute): Likewise.
12667 (inverted_post_order_compute): Likewise.
12668 (pre_and_rev_post_order_compute_fn): Likewise.
12670 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12672 * genrecog.c (int_set::int_set): Explicitly construct our
12673 auto_vec base class.
12674 * vec.h (auto_vec::auto_vec): New constructor.
12676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12678 * bitmap.h (class auto_bitmap): New constructor taking
12679 bitmap_obstack * argument.
12681 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12683 * bitmap.h (class auto_bitmap): Change type of m_bits to
12684 bitmap_head, and adjust ctor / dtor and member operators.
12686 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12688 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12689 when returned register mode doesn't match original mode.
12691 2017-05-12 Jeff Law <law@redhat.com>
12692 Jakub Jelinek <jakub@redhat.com>
12694 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
12695 we look for cc setter after the compare-elim changes.
12696 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
12697 within the vector to match what compare-elim now expects.
12698 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
12699 (xorsi3_flags, one_cmplsi2_flags): Likewise.
12701 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
12702 after the compare-elim changes.
12703 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
12704 the vector to match what compare-elim now expects.
12705 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
12706 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
12707 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
12708 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
12709 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
12711 * config/visium/visium.c (single_set_and_flags): Fix where
12712 we look for cc setter after the compare-elim changes.
12713 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
12714 with the vector to match what compare-elim now expects.
12715 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
12716 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
12717 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
12718 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
12719 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
12720 (neg<mode>2_insn_set_overflow): Likewise.
12722 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
12724 PR middle-end/79794
12725 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
12726 maybe_expand_insn call, set ops[0].target. If still set after call,
12727 set alt_rtl. Add extra arg to recursive calls.
12728 (extract_bit_field): Add alt_rtl argument. Pass to
12730 * expmed.h (extract_bit_field): Fix prototype.
12731 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
12732 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
12733 to extract_bit_field_calls.
12734 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
12735 Pass alt_rtl to extract_bit_field calls.
12736 * calls.c (store_unaligned_arguments_into_psuedos)
12737 load_register_parameters): Pass extra NULL to extract_bit_field calls.
12738 * optabs.c (maybe_legitimize_operand): Clear op->target when call
12740 * optabs.h (struct expand_operand): Add target bitfield.
12742 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12744 * compare-elim.c (try_eliminate_compare): Canonicalize
12745 operation with embedded compare to
12746 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
12747 (set (reg) (operation)].
12749 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
12751 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12754 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
12755 cost of adding a carry flag for ADC instruction.
12756 [case MINUS]: Ignore the cost of subtracting a carry flag
12757 for SBB instruction.
12759 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
12761 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
12763 * config/rs6000/bmiintrin.h: New file.
12764 * config/rs6000/bmi2intrin.h: New file.
12765 * config/rs6000/x86intrin.h: New file.
12767 2017-05-12 Jeff Law <law@redhat.com>
12769 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
12772 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
12774 PR middle-end/80707
12775 * tree-cfg.c: Remove cfg edges of unreachable case statements.
12777 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12779 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12780 early expansion of vector divide builtins.
12781 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
12782 builtins identified as having unsigned arguments.
12784 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12786 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
12787 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
12788 expansion of vector logical operations (and, andc, or, xor,
12791 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12793 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
12794 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
12796 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12798 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12799 early GIMPLE expansion of vector multiplies.
12801 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12803 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
12804 TARGET_HAVE_MOVT conditional.
12805 (movt splitter): Likewise.
12807 2017-05-12 Richard Biener <rguenther@suse.de>
12809 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
12810 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12811 Fold all stmts not inplace.
12813 2017-05-12 Richard Biener <rguenther@suse.de>
12815 PR tree-optimization/80713
12816 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
12817 inserted_exprs bit for not removed stmts.
12819 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
12821 PR middle-end/69921
12822 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
12823 parallelized" attribute for parallelized OpenACC kernels.
12824 * omp-offload.c (execute_oacc_device_lower): Use it.
12826 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
12827 Set "oacc kernels" attribute.
12828 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
12829 parameter. Adjust all users.
12830 (oacc_fn_attrib_kernels_p): Remove function.
12831 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
12832 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
12833 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
12834 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
12835 assert "oacc kernels" attribute is set.
12837 2017-05-11 Carl Love <cel@us.ibm.com>
12839 * config/rs6000/rs6000-c: Add support for built-in functions
12840 vector unsigned char vec_popcnt (vector signed char)
12841 vector unsigned char vec_popcnt (vector unsigned char)
12842 vector unsigned short vec_popcnt (vector signed short)
12843 vector unsigned short vec_popcnt (vector unsigned short)
12844 vector unsigned int vec_popcnt (vector signed int)
12845 vector unsigned int vec_popcnt (vector unsigned int)
12846 vector unsigned long long vec_popcnt (vector signed long long)
12847 vector unsigned long long vec_popcnt (vector unsigned long long)
12848 vector signed long long vec_slo (vector signed long long,
12849 vector signed char)
12850 vector signed long long vec_slo (vector signed long long,
12851 vector unsigned char)
12852 vector unsigned long long vec_slo (vector unsigned long long,
12853 vector signed char)
12854 vector unsigned long long vec_slo (vector unsigned long long,
12855 vector unsigned char)
12856 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
12857 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
12858 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
12859 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
12860 * doc/extend.texi: Update the built-in documentation file for the
12861 new built-in functions.
12863 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12865 * attribs.h (sorted_attr_string): Move machine independent
12866 functions for target clone support from the i386 port to common
12867 code. Rename ix86_function_versions to common_function_versions.
12868 Rename make_name to make_unique_name.
12869 (common_function_versions): Likewise.
12870 (make_unique_name): Likewise.
12871 (make_dispatcher_decl): Likewise.
12872 (is_function_default_version): Likewise.
12873 * attribs.c (attr_strcmp): Likewise.
12874 (sorted_attr_string): Likewise.
12875 (common_function_versions): Likewise.
12876 (make_unique_name): Likewise.
12877 (make_dispatcher_decl): Likewise.
12878 (is_function_default_version): Likewise.
12879 * config/i386/i386.c (attr_strcmp): Likewise.
12880 (sorted_attr_string): Likewise.
12881 (ix86_function_versions): Likewise.
12882 (make_name): Likewise.
12883 (make_dispatcher_decl): Likewise.
12884 (is_function_default_version): Likewise.
12885 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
12887 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12890 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12891 Account for direct move costs for vec_construct of integer
12894 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
12897 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
12898 (UNSPEC_STX_ATOMIC): Ditto.
12899 (loaddi_via_sse): New insn.
12900 (storedi_via_sse): Ditto.
12901 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
12902 Update corresponding peephole2 patterns.
12903 (atomic_storedi_fpu): Ditto.
12905 2017-05-11 Julia Koval <julia.koval@intel.com>
12907 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
12908 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
12910 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
12911 (__builtin_ia32_rsqrt14ss_mask): New builtins.
12912 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
12914 2017-05-11 Nathan Sidwell <nathan@acm.org>
12916 * graphite-poly.c: Include dumpfile.h.
12918 * dumpfle.h (dump_function): Declare here ...
12919 * tree-dump.h (dump_function): ... not here.
12920 * dumpfile.c: #include tree-cfg.h.
12921 (dump_function): Move here from ...
12922 * tree-dump.c (dump_function): ... here.
12923 * gimplify.c: #include splay-tree.h, not tree-dump.h.
12924 * graphite-poly.c: Don't include tree-dump.h.
12925 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
12926 * print-tree.c: Likewise.
12927 * stor-layout.c: Likewise.
12928 * tree-nested.c: Likewise.
12930 * dumpfile.c (dump_start): Use TDF_FLAGS.
12931 (dump_enable_all): Fix TDF_KIND check thinko.
12933 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12935 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12936 array entries to represent two legal parameterizations of the
12937 overloaded __builtin_cmpb function, as represented by the
12938 P6_OV_BUILTIN_CMPB constant.
12939 (altivec_resolve_overloaded_builtin): Add special case handling
12940 for the __builtin_cmpb function, as represented by the
12941 P6_OV_BUILTIN_CMPB constant.
12942 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
12943 (BU_P6_64BIT_2): New macro.
12944 (BU_P6_OVERLOAD_2): New macro
12945 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
12946 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
12947 (CMPB): Add overload support to represent both 32-bit and 64-bit
12948 compare-bytes function.
12949 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12950 support for TARGET_CMPB.
12951 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
12952 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12953 documentation of the __builtin_cmpb overloaded built-in function.
12955 2017-05-11 Richard Biener <rguenther@suse.de>
12957 PR tree-optimization/80705
12958 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
12959 bases are not vectorizable.
12961 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12963 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
12964 when counting register pressure.
12966 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12968 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
12969 (iv_ca_more_deps): Renamed to ...
12970 (iv_ca_compare_deps): ... this.
12971 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
12973 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12975 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
12977 (determine_group_iv_costs): ... here.
12978 (find_inv_vars_cb): Record inv var if it's not recorded before.
12980 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12982 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
12983 (get_shiftadd_cost): Ditto.
12985 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12987 * tree-ssa-address.c: Include header file.
12988 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
12990 (add_to_parts): Refactor.
12991 (addr_to_parts): New parameter. Update use of move_hint_to_base.
12992 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
12995 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12997 PR tree-optimization/53090
12998 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13000 (extract_cond_operands): Detect condition with IV on both sides
13001 and return COMP_IV_EXPR_2.
13002 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13003 (rewrite_use_compare): Simplify by removing call to function
13004 extract_cond_operands.
13006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13008 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13009 (extract_cond_operands): Detect condition comparing against non-
13010 invariant bound and return appropriate enum value.
13011 (find_interesting_uses_cond): Update use of extract_cond_operands.
13012 Handle its return value accordingly.
13013 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13015 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13017 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13018 nonlinear iv_use computation in loop invariant sensitive way.
13020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13022 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13023 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13027 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13028 (dump_cand): Support iv_cand.inv_exprs.
13029 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13031 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13036 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13038 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13039 as local function. Include necessary header files.
13040 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13042 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13044 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13046 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13048 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13049 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13050 RSHIFT_EXPR and BIT_NOT_EXPR.
13052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13054 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13055 (adjust_setup_cost): New parameter supporting round up adjustment.
13056 (struct address_cost_data): Delete.
13057 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13058 (split_address_cost, ptr_difference_cost): Delete.
13059 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13060 (struct ainc_cost_data): New struct.
13061 (get_address_cost_ainc): New function.
13062 (get_address_cost, get_computation_cost): Reimplement.
13063 (determine_group_iv_cost_address): Record inv_expr for all uses of
13065 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13066 (iv_ca_has_deps): Reimplemented to ...
13067 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13069 (iv_ca_extend): Call iv_ca_more_deps.
13071 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13073 * tree-ssa-address.c (struct mem_address): Move to header file.
13074 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13075 * tree-ssa-address.h (struct mem_address): Move from C file.
13076 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13078 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13080 * tree-affine.h (aff_combination_type): New interface.
13081 (aff_combination_zero_p): Remove static.
13082 (aff_combination_const_p): New interface.
13083 (aff_combination_singleton_var_p): New interfaces.
13085 2017-05-11 Richard Biener <rguenther@suse.de>
13087 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13088 Skip unreachable blocks and destinations.
13089 (eliminate): Move stmt removal and fixup ...
13090 (fini_eliminate): ... here. Skip inserted exprs.
13091 (pass_pre::execute): Move fini_pre after fini_eliminate.
13092 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13093 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13094 PRE to get rid of dead code that has invalid SSA form and
13095 split critical edges again.
13097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13099 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13101 2017-05-11 Richard Biener <rguenther@suse.de>
13103 * passes.c (execute_function_todo): Verify loops if they are
13104 said to be up-to-date.
13105 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13106 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13108 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13111 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13112 handle calling assemble_external ourself.
13115 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13116 modes with zero size. Enhance comment.
13118 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13120 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13121 built-ins for vec_xl and vec_xst with short and char pointer
13124 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13126 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13127 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13128 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13129 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13130 (_mm_maskz_min_round_ss): New intrinsics.
13131 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13132 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13133 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13134 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13135 (__builtin_ia32_minss_mask_round): New builtins.
13136 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13138 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13140 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13141 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13143 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13146 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13148 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13149 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13150 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13151 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13152 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13153 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13154 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13155 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13156 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13157 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13158 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13159 (__builtin_ia32_mulss_mask_round): New builtins.
13160 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13161 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13162 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13164 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13165 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13167 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13170 2017-05-10 Julia Koval <julia.koval@intel.com>
13172 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13173 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13174 (_mm256_setr_m128i): New intrinsics.
13176 2017-05-10 Julia Koval <julia.koval@intel.com>
13178 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13179 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13180 (_mm_maskz_rcp14_ss): New intrinsics.
13181 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13182 (__builtin_ia32_rcp14ss_mask): New builtins.
13183 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13185 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13187 PR tree-optimization/51513
13188 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13189 (assert_unreachable_fallthru_edge_p): Use it.
13190 (group_case_labels_stmt): Likewise.
13191 * tree-cfg.h: Prototype it.
13192 * stmt.c: Include cfghooks.h and tree-cfg.h.
13193 (emit_case_dispatch_table) <gap_label>: New local variable.
13194 Use it to fill dispatch table gaps.
13195 Test for default_label before updating probabilities.
13196 (expand_case) <default_label>: Remove unneeded initialization.
13197 Test for unreachable default case statement and remove its edge.
13198 Set default_label accordingly.
13199 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13201 2017-05-10 Carl Love <cel@us.ibm.com>
13203 * config/rs6000/rs6000-c: Add support for built-in functions
13204 vector signed char vec_neg (vector signed char)
13205 vector signed short int vec_neg (vector short int)
13206 vector signed int vec_neg (vector signed int)
13207 vector signed long long vec_neg (vector signed long long)
13208 vector float vec_neg (vector float)
13209 vector double vec_neg (vector double)
13210 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13212 * config/rs6000/altivec.h: Add define for vec_neg
13213 * doc/extend.texi: Update the built-in documentation for the
13214 new built-in functions.
13216 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13218 PR tree-optimization/77644
13219 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13221 2017-05-10 Nathan Sidwell <nathan@acm.org>
13223 * dumpfile.h (TDI_lang_all): New.
13224 (TDF_KIND): New. Renumber others
13225 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13227 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13229 (get_dump_file_name): Adjust suffix generation.
13230 (dump_enable_all): Use TDF_KIND.
13231 * doc/invoke.texi (-fdump-lang-all): Document.
13233 * dumpfile.h: Tabify.
13235 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13238 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13239 Move member access before delete.
13241 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13243 * tree-inline.c (expand_call_inline): Split block at stmt
13246 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13249 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13250 are now unused after splitting mov{sf,sd}_hardfloat.
13251 (f32_lr2): Likewise.
13252 (f32_lm): Likewise.
13253 (f32_lm2): Likewise.
13254 (f32_li): Likewise.
13255 (f32_li2): Likewise.
13256 (f32_lv): Likewise.
13257 (f32_sr): Likewise.
13258 (f32_sr2): Likewise.
13259 (f32_sm): Likewise.
13260 (f32_sm2): Likewise.
13261 (f32_si): Likewise.
13262 (f32_si2): Likewise.
13263 (f32_sv): Likewise.
13264 (f32_dm): Likewise.
13265 (f32_vsx): Likewise.
13266 (f32_av): Likewise.
13267 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13268 For movsf, order stores so the VSX stores occur before the GPR
13269 store which encourages the register allocator to use a traditional
13270 FPR instead of a GPR. For movsd, order the stores so that the GPR
13271 store comes before the VSX stores to allow the power6 to work.
13272 This is due to the power6 not having a 32-bit integer store
13273 instruction from a FPR.
13274 (movsf_hardfloat): Likewise.
13275 (movsd_hardfloat): Likewise.
13277 2017-05-09 Martin Sebor <msebor@redhat.com>
13279 PR translation/80280
13280 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13283 PR translation/80280
13284 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13285 data member added in r247778.
13286 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13288 2017-05-09 Nathan Sidwell <nathan@acm.org>
13290 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13292 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13295 2017-05-09 Marek Polacek <polacek@redhat.com>
13297 * doc/invoke.texi: Fix typo.
13299 2017-05-09 Richard Biener <rguenther@suse.de>
13301 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13302 (vrp_val_is_min): Likewise.
13303 (set_value_range_to_value): Likewise.
13304 (set_value_range_to_nonnegative): Likewise.
13305 (gimple_assign_nonzero_p): Likewise.
13306 (gimple_stmt_nonzero_p): Likewise.
13307 (vrp_int_const_binop): Likewise. Remove unreachable case.
13308 (adjust_range_with_scev): Adjust comments.
13309 (compare_range_with_value): Likewise.
13310 (extract_range_from_phi_node): Likewise.
13311 (test_for_singularity): Likewise.
13313 2017-05-09 Richard Biener <rguenther@suse.de>
13315 * tree-vrp.c (get_single_symbol): Add assert that we don't
13316 get overflowed constants as invariant part.
13317 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13318 checks. Use wi::cmp instead of recursing for integer constants.
13319 (compare_values): Just ignore whether we assumed undefined
13320 overflow instead of failing the compare.
13321 (extract_range_for_var_from_comparison_expr): Add comment before the
13322 TREE_NO_WARNING sets.
13323 (test_for_singularity): Likewise.
13324 (extract_range_from_comparison): Do not disable optimization
13325 when we assumed undefined overflow.
13326 (extract_range_basic): Remove init of unused var.
13328 2017-05-09 Richard Biener <rguenther@suse.de>
13330 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13331 (extract_range_from_multiplicative_op_1): Adjust.
13332 (extract_range_from_binary_expr_1): Use int_const_binop.
13334 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13337 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13338 rs6000_store_data_bypass_p in seven define_bypass directives and
13339 in several comments.
13340 * config/rs6000/rs6000-protos.h: Add prototype for
13341 rs6000_store_data_bypass_p function.
13342 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13343 function implements slightly different (rs6000-specific) semantics
13344 than store_data_bypass_p, returning false rather than aborting
13345 with assertion error when arguments do not satisfy the
13346 requirements of store data bypass.
13347 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13348 rs6000_store_data_bypass_p.
13350 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13352 * config/xtensa/xtensa-protos.h
13353 (xtensa_initial_elimination_offset): New declaration.
13354 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13355 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13356 macro definition, add case for FRAME_POINTER_REGNUM when
13357 FRAME_GROWS_DOWNWARD.
13358 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13359 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13360 xtensa_initial_elimination_offset.
13362 2017-05-08 Nathan Sidwell <nathan@acm.org>
13364 * doc/invoke.texi: Alphabetize -fdump options.
13366 2017-05-08 Martin Sebor <msebor@redhat.com>
13368 PR translation/80280
13369 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13371 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13373 * target.def (compute_frame_layout): New optional target hook.
13374 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13375 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13376 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13378 * reload1.c (verify_initial_elim_offsets): Likewise.
13379 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13380 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13381 (arm_get_frame_offsets): Split up into this ...
13382 (arm_compute_frame_layout): ... and this function.
13384 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13386 * config/aarch64/constraints.md (Usa): New constraint.
13387 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13389 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13391 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13392 with_multilib_list after it has been checked.
13394 2017-05-08 Richard Biener <rguenther@suse.de>
13396 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13397 (bitmap_set_subtract_values): Likewise.
13399 2017-05-08 Richard Biener <rguenther@suse.de>
13401 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13402 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13404 (gimple_stmt_nonzero_warnv_p): Rename to ...
13405 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13407 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13408 (extract_range_basic): Adjust, do not disable propagation on
13409 strict overflow sensitive simplification.
13410 (vrp_visit_cond_stmt): Likewise.
13412 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13414 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13415 body size unconditionally.
13417 2017-05-07 Jeff Law <law@redhat.com>
13420 2017-05-06 Jeff Law <law@redhat.com>
13421 PR tree-optimization/78496
13422 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13425 PR tree-optimization/78496
13426 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13427 (simplify_stmt_using_ranges): Call it.
13428 (vrp_dom_walker::before_dom_children): Extract equivalences
13429 from an ASSERT_EXPR with an equality comparison against a
13432 2017-05-06 Jeff Law <law@redhat.com>
13434 PR tree-optimization/78496
13435 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13438 PR tree-optimization/78496
13439 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13440 (simplify_stmt_using_ranges): Call it.
13441 (vrp_dom_walker::before_dom_children): Extract equivalences
13442 from an ASSERT_EXPR with an equality comparison against a
13445 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13447 * lra-constraints.c (lra_copy_reg_equiv): New function.
13448 (split_reg): Use it to copy equivalence information from the
13449 original register to the spill register.
13451 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13453 PR rtl-optimization/75964
13454 * simplify-rtx.c (simplify_const_relational_operation): Remove
13455 invalid handling of comparisons of integer ABS.
13457 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13459 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13460 initialize to zero.
13461 (init_regs): Remove declaration.
13462 (function_arg_advance_32): Initialize error_p as boolean variable.
13464 2017-05-05 Nathan Sidwell <nathan@acm.org>
13466 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13467 lines. Use for (;;).
13469 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13471 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13472 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13473 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13474 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13475 VF=2 that require versioning.
13477 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13479 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13482 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13484 * diagnostic.h (diagnostic_override_option_index): Convert from
13485 macro to inline function.
13487 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13489 * diagnostic.c (last_module_changed_p): New function.
13490 (set_last_module): New function.
13491 (diagnostic_report_current_module): Convert macro usage to
13492 the above functions.
13493 * diagnostic.h (diagnostic_context::last_module): Strengthen
13494 from const line_map * to const line_map_ordinary *.
13495 (diagnostic_last_module_changed): Delete macro.
13496 (diagnostic_set_last_module): Delete macro.
13498 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13500 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13501 with diagnostic_report_diagnostic.
13502 (diagnostic_n_impl_richloc): Likewise.
13503 * diagnostic.h (report_diagnostic): Delete macro.
13504 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13505 with diagnostic_report_diagnostic.
13506 * substring-locations.c (format_warning_va): Likewise.
13508 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13510 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13511 save/restor of format_spec. Move option-printing code to...
13512 (print_option_information): ...this new function, and
13513 reimplement by simply printing to the pretty_printer,
13514 rather than appending to the format string.
13516 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13518 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13519 handling logic into...
13520 (update_effective_level_from_pragmas): ...this new function.
13522 2017-05-04 Andrew Waterman <andrew@sifive.com>
13524 * config/riscv/riscv.opt (mstrict-align): New option.
13525 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13526 (SLOW_UNALIGNED_ACCESS): Define.
13527 (riscv_slow_unaligned_access): Declare.
13528 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13530 (riscv_slow_unaligned_access): New variable.
13531 (rocket_tune_info): Set slow_unaligned_access to true.
13532 (optimize_size_tune_info): Set slow_unaligned_access to false.
13533 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13534 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13535 (riscv_option_override): Set riscv_slow_unaligned_access.
13536 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13538 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13540 * config/riscv/riscv.md: Unify indentation.
13542 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13547 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13549 (extendsi<mode>2): Add support for doing sign extension via
13550 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13551 don't have ISA 3.0 instructions.
13552 (extendsi<mode>2 splitter): Likewise.
13553 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13554 generate the normal insns since SImode can now go in vector
13555 registers. Disallow the special UNSPECs needed for previous
13556 machines to hide SImode being used. Add new insns
13557 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13558 (fix_trunc<mode>si2_stfiwx): Likewise.
13559 (fix_trunc<mode>si2_internal): Likewise.
13560 (fixuns_trunc<mode>si2): Likewise.
13561 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13562 (fctiw<u>z_<mode>_smallint): Likewise.
13563 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13564 of floating point to 32-bit integer from doing a direct move to
13565 the GPR registers to do a store.
13566 (fctiwz_<mode>): Break long line.
13568 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13570 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13571 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13572 (addr_list, addr_offset_valid_p): New.
13573 (split_address_groups): Check offset validity with above function.
13574 (gt-tree-ssa-loop-ivopts.h): Include header file.
13576 2017-05-05 Nathan Sidwell <nathan@acm.org>
13578 * config.gcc (arm*-*-*): Add missing 'fi'.
13580 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13582 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13583 included in -fopt-info does not matter.
13584 * doc/optinfo.texi (-fopt-info): Fix description of default
13585 behavour. Explicitly say order of options included in -fopt-info
13588 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13590 * config.gcc: Allow combinations of aprofile and rmprofile values for
13591 --with-multilib-list.
13592 * config/arm/t-multilib: New file.
13593 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13594 variables. Remove setting of ISA and floating-point ABI in
13595 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13596 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13597 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13598 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13600 * config/arm/t-rmprofile: Likewise except for the matches changes.
13601 * doc/install.texi (--with-multilib-list): Document the combination of
13602 aprofile and rmprofile values and warn about pitfalls in doing that.
13604 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13606 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13607 (movdi_aarch64): Likewise.
13609 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13611 PR tree-optimization/80632
13612 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13614 (build_arrays): Initialize it for virtual phis.
13615 (fix_phi_nodes): Use it for virtual phis.
13617 PR tree-optimization/80558
13618 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13619 [x, y] op z into [x op, y op z] for op & or | if conditions
13622 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13623 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13626 * config/arm/arm.md (use_literal_pool): Remove.
13627 (64-bit immediate split): No longer takes cost into consideration
13628 if arm_disable_literal_pool is enabled.
13629 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13630 used when arm_disable_literal_pool is enabled.
13631 (arm_max_const_double_inline_cost): Remove use of
13632 arm_disable_literal_pool.
13633 (push_minipool_fix): Add assert.
13634 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13635 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13636 (no_literal_pool_sf_immediate): New.
13638 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13640 PR tree-optimization/80613
13641 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13642 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13644 2017-05-05 Richard Biener <rguenther@suse.de>
13646 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13648 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13650 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13651 of this flag from insn conditions due to removal from r247495.
13653 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13655 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13657 (arm_early_store_addr_dep_ptr): Likewise.
13658 * config/arm/aarch-common-protos.h
13659 (arm_early_load_addr_dep_ptr): Add prototype.
13660 (arm_early_store_addr_dep_ptr): Likewise.
13661 * config/arm/cortex-a53.md: Add new bypasses.
13663 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13665 * tree.c (next_type_uid): Change type to unsigned.
13666 (type_hash_canon): Decrement back next_type_uid if
13667 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13668 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13671 2017-05-04 Martin Sebor <msebor@redhat.com>
13673 * builtins.c: Fix a trivial typo in a comment.
13675 PR middle-end/79234
13676 * builtins.c (check_sizes): Adjust to handle reading past the end.
13677 Avoid printing excessive upper bound of ranges. Use %E to print
13678 tree nodes instead of converting them to %wu.
13679 (expand_builtin_memchr): New function.
13680 (compute_dest_size): Rename...
13681 (compute_objsize): ...to this.
13682 (expand_builtin_memcpy): Adjust.
13683 (expand_builtin_mempcpy): Adjust.
13684 (expand_builtin_strcat): Adjust.
13685 (expand_builtin_strcpy): Adjust.
13686 (check_strncat_sizes): Adjust.
13687 (expand_builtin_strncat): Adjust.
13688 (expand_builtin_strncpy): Adjust and simplify.
13689 (expand_builtin_memset): Adjust.
13690 (expand_builtin_bzero): Adjust.
13691 (expand_builtin_memcmp): Adjust.
13692 (expand_builtin): Handle memcmp.
13693 (maybe_emit_chk_warning): Check strncat just once.
13695 2017-05-04 Martin Sebor <msebor@redhat.com>
13697 PR preprocessor/79214
13698 PR middle-end/79222
13699 PR middle-end/79223
13700 * builtins.c (check_sizes): Add inlining context and issue
13701 warnings even when -Wno-system-headers is set.
13702 (check_strncat_sizes): Same.
13703 (expand_builtin_strncat): Same.
13704 (expand_builtin_memmove): New function.
13705 (expand_builtin_stpncpy): Same.
13706 (expand_builtin): Handle memmove and stpncpy.
13708 2017-05-04 Bin Cheng <bin.cheng@arm.com>
13710 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
13711 which is not used any more.
13713 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13715 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
13717 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13719 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
13720 (cortexa53_tunings): Likewise.
13721 (cortexa57_tunings): Likewise.
13722 (cortexa72_tunings): Likewise.
13723 (cortexa73_tunings): Likewise.
13725 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13727 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
13728 Set loop alignment to 8.
13730 2017-05-04 Martin Sebor <msebor@redhat.com>
13732 PR translation/80280
13733 * builtins.c (expand_builtin_object_size): Add missing quoting to
13734 %D and like directives.
13735 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
13736 (hsa_type_for_tree_type): Same.
13737 (verify_function_arguments): Same.
13738 * symtab.c (symbol_table::change_decl_assembler_name): Same.
13739 * varasm.c (get_section): Same.
13742 2017-05-04 Martin Sebor <msebor@redhat.com>
13744 PR translation/80280
13745 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
13747 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13749 * config/aarch64/aarch64.c (generic_addrcost_table):
13750 Change HI/TI mode setting.
13752 2017-05-04 Martin Jambor <mjambor@suse.cz>
13754 PR tree-optimization/80622
13755 * tree-sra.c (comes_initialized_p): New function.
13756 (build_accesses_from_assign): Only set write lazily when
13757 comes_initialized_p is false.
13758 (analyze_access_subtree): Use comes_initialized_p.
13759 (propagate_subaccesses_across_link): Assert !comes_initialized_p
13760 instead of testing for PARM_DECL.
13762 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13764 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
13765 constraint on operand 0 to allow more general addressing modes.
13766 Adjust output template.
13767 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
13769 * config/aarch64/aarch64-protos.h
13770 (aarch64_address_valid_for_prefetch_p): Declare prototype.
13771 * config/aarch64/constraints.md (Dp): New address constraint.
13772 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
13775 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
13777 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
13778 update use of estimate_ipcp_clone_size_and_time.
13779 (estimate_local_effects): Update use of
13780 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
13781 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
13782 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
13783 Return nonspecialized time.
13785 2017-05-04 Richard Biener <rguenther@suse.de>
13787 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
13788 for the last VUSE which def dominates the PHI. Directly call
13790 (get_continuation_for_phi_1): Remove.
13792 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
13794 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
13795 to explain the use of truncating division. Cap the number of
13796 iterations to the maximum given by nb_iterations_upper_bound,
13799 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13801 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
13802 * configure: Regenerate.
13803 * config.in: Regenerate.
13804 * config/i386/driver-mingw32.c: new file.
13805 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
13806 * config.host: Link driver-mingw32.o on MinGW host.
13807 * doc/install.texi: Document new --enable-mingw-wildcard configure
13810 2017-05-04 Marek Polacek <polacek@redhat.com>
13812 PR tree-optimization/80612
13813 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
13815 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13816 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
13818 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
13819 (movt splitter): Likewise.
13820 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
13821 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
13822 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
13823 block for Thumb-1 with MOVT.
13824 (thumb2_legitimate_address_p): Move code block ...
13825 (can_avoid_literal_pool_for_label_p): ... into this new function.
13826 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
13828 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
13829 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
13830 "M-profile targets with the MOVT instruction".
13832 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13834 * config/arm/arm-builtins.c (arm_init_builtins): Rename
13835 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
13836 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
13838 2017-05-04 Martin Liska <mliska@suse.cz>
13840 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
13841 variable cond_code.
13843 2017-05-04 Richard Biener <rguenther@suse.de>
13845 * tree.c (array_at_struct_end_p): Handle arrays at struct
13846 end with flexarrays more conservatively. Refactor and treat
13847 arrays of arrays or aggregates more strict. Fix
13848 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
13849 * tree.c (array_at_struct_end_p): Adjust prototype.
13850 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13851 * gimple-fold.c (get_range_strlen): Likewise.
13852 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
13854 2017-05-04 Richard Biener <rguenther@suse.de>
13856 PR tree-optimization/31130
13857 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
13859 (supports_overflow_infinity): Likewise.
13860 (is_negative_overflow_infinity): Likewise.
13861 (is_positive_overflow_infinity): Likewise.
13862 (is_overflow_infinity): Likewise.
13863 (stmt_overflow_infinity): Likewise.
13864 (overflow_infinity_range_p): Likewise.
13865 (usable_range_p): Remove as always returning true.
13866 (make_overflow_infinity): Remove.
13867 (negative_overflow_infinity): Likewise.
13868 (positive_overflow_infinity): Likewise.
13869 (avoid_overflow_infinity): Likewise.
13870 (set_value_range): Adjust accordingly.
13871 (set_value_range_to_nonnegative): Likewise, remove now unused
13872 overflow_infinity arg.
13873 (vrp_operand_equal_p): Adjust.
13874 (update_value_range): Likewise.
13875 (range_int_cst_singleton_p): Likewise.
13876 (operand_less_p): Likewise.
13877 (compare_values_warnv): Likewise.
13878 (extract_range_for_var_from_comparison_expr): Likewise.
13879 (vrp_int_const_binop): Likewise.
13880 (zero_nonzero_bits_from_vr): Likewise.
13881 (extract_range_from_multiplicative_op_1): Likewise.
13882 (extract_range_from_binary_expr_1): Likewise.
13883 (extract_range_from_unary_expr): Likewise.
13884 (extract_range_from_comparison): Likewise.
13885 (extract_range_basic): Likewise.
13886 (adjust_range_with_scev): Likewise.
13887 (compare_ranges): Likewise.
13888 (compare_range_with_value): Likewise.
13889 (dump_value_range): Likewise.
13890 (test_for_singularity): Likewise, remove strict_overflow_p parameter
13892 (simplify_cond_using_ranges): Adjust.
13894 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13896 * brig-builtins.def: Added a builtin for class_f64.
13897 * builtin-types.def: Added a builtin type needed by class_f64.
13899 2017-05-03 Jason Merrill <jason@redhat.com>
13901 * timevar.def: Add TV_CONSTEXPR.
13903 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13905 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
13907 2017-05-03 Martin Jambor <mjambor@suse.cz>
13909 * ipa-prop.c (ipa_update_after_lto_read): Removed.
13910 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
13911 * ipa-cp.c (ipcp_propagate_stage): Do not call
13912 ipa_update_after_lto_read.
13913 * ipa-inline.c (ipa_inline): Likewise.
13915 2017-05-03 Martin Jambor <mjambor@suse.cz>
13917 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
13918 tag. Added a default constructor and a destructor.
13919 (ipa_edge_args_sum_t): New class;
13920 (ipa_edge_args_sum): Declare.
13921 (ipa_edge_args_vector): Remove declaration.
13922 (IPA_EDGE_REF): Use ipa_edge_args_sum.
13923 (ipa_free_edge_args_substructures): Remove declaration.
13924 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
13925 (ipa_edge_args_info_available_for_edge_p): Likewise.
13926 * ipa-prop.c (ipa_edge_args_vector): Removed.
13927 (edge_removal_hook_holder): Likewise.
13928 (edge_duplication_hook_holder): Likewise.
13929 (ipa_edge_args_sum): New variable.
13930 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
13931 ipa_edge_args_vector.
13932 (ipa_free_edge_args_substructures): Likewise.
13933 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
13934 ipa_edge_args_vector.
13935 (ipa_edge_removal_hook): Turned into method
13936 ipa_edge_args_sum_t::remove.
13937 (ipa_edge_duplication_hook): Turned into method
13938 ipa_edge_args_sum_t::duplicate.
13939 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
13940 registering edge hooks.
13941 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
13942 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
13943 ipa_edge_args_sum instead of ipa_edge_args_vector.
13944 * ipa-profile.c (ipa_profile): Likewise.
13946 2017-05-03 Martin Jambor <mjambor@suse.cz>
13948 * symbol-summary.h (function_summary): New method exists.
13949 (function_summary::symtab_removal): Deallocate through release.
13950 (call_summary): New class.
13951 (gt_ggc_mx): New overload.
13952 (gt_pch_nx): Likewise.
13953 (gt_pch_nx): Likewise.
13955 2017-05-03 Jeff Law <law@redhat.com>
13957 PR tree-optimization/78496
13958 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
13959 from simplify_cond_using_ranges. Split off code to walk
13960 backwards through casts into ...
13961 (simplify_cond_using_ranges_2): New function.
13962 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
13963 (execute_vrp): After identifying jump threads, call
13964 simplify_cond_using_ranges_2.
13966 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
13969 * ipa-inline.h (inline_summary): Add ctor.
13970 (create_ggc): Do not use ggc_cleared_alloc.
13972 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
13973 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13975 * gcc.c (handle_braces): Support escaping in switch matching
13977 * doc/invoke.texi (Spec Files): Document it.
13978 Remove superfluous @code markup in items.
13980 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13982 * diagnostic-show-locus.c (struct column_range): New struct.
13983 (get_affected_columns): New function.
13984 (get_printed_columns): New function.
13985 (struct correction): New struct.
13986 (correction::ensure_capacity): New function.
13987 (correction::ensure_terminated): New function.
13988 (struct line_corrections): New struct.
13989 (line_corrections::~line_corrections): New dtor.
13990 (line_corrections::add_hint): New function.
13991 (layout::print_trailing_fixits): Reimplement in terms of the new
13993 (selftest::test_overlapped_fixit_printing): New function.
13994 (selftest::diagnostic_show_locus_c_tests): Call it.
13996 2017-05-03 Nathan Sidwell <nathan@acm.org>
13998 Canonicalize canonical type hashing
13999 * tree.h (type_hash_canon_hash): Declare.
14000 * tree.c (type_hash_list, attribute_hash_list): Move into
14001 type_hash_canon_hash.
14002 (build_type_attribute_qual_variant): Break out hash code calc into
14003 type_hash_canon_hash.
14004 (type_hash_canon_hash): New. Generic type hash computation.
14005 (build_range_type_1, build_array_type_1, build_function_type,
14006 build_method_type_directly, build_offset_type, build_complex_type,
14007 make_vector_type): Call it.
14009 2017-05-03 Richard Biener <rguenther@suse.de>
14011 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14012 When all DRs have unknown misaligned do not always peel
14013 when there is a store but apply the same costing model as if
14014 there were only loads.
14016 2017-05-03 Richard Biener <rguenther@suse.de>
14019 PR tree-optimization/80492
14020 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14021 compare_base_decls returning dont-know properly.
14023 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14025 * config/arm/iterators.md (CCSI): New mode iterator.
14026 (arch): New mode attribute.
14027 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14028 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14029 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14030 code iterator for success result mode.
14031 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14032 the corresponding new insn generators.
14034 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14037 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14038 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14040 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14042 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14043 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14044 (DDR_A): Wrap DDR argument in brackets.
14045 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14046 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14047 (DDR_REVERSED_P): Likewise.
14049 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14051 PR tree-optimization/79472
14052 * tree-switch-conversion.c (struct switch_conv_info): Add
14053 contiguous_range and default_case_nonstandard fields.
14054 (collect_switch_conv_info): Compute contiguous_range and
14055 default_case_nonstandard fields, don't clear final_bb if
14056 contiguous_range and only the default case doesn't have the required
14058 (check_all_empty_except_final): Set default_case_nonstandard instead
14059 of failing if contiguous_range and the default case doesn't have empty
14061 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14062 and only the default case doesn't have the required constants. Skip
14064 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14065 if default_case_nonstandard.
14066 (build_constructors): Build constant 1 just once. Assert that default
14067 values aren't inserted in between cases if contiguous_range. Skip
14069 (build_arrays): Skip virtual phis.
14070 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14071 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14072 Handle virtual phis.
14073 (gen_inbound_check): Handle default_case_nonstandard case.
14074 (process_switch): Adjust check_final_bb caller. Call
14075 gather_default_values with the first non-default case instead of
14076 default case if default_case_nonstandard.
14078 2017-05-02 Nathan Sidwell <nathan@acm.org>
14080 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14081 check. Fix formatting.
14083 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14085 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14086 errors when comparing specialized and unspecialized times.
14088 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14090 * diagnostic-show-locus.c
14091 (layout::should_print_annotation_line_p): Make private.
14092 (layout::print_annotation_line): Make private.
14093 (layout::annotation_line_showed_range_p): Make private.
14094 (layout::show_ruler): Make private.
14095 (layout::print_source_line): Make private. Pass in line and
14096 line_width, rather than calling location_get_source_line. Drop
14098 (layout::print_leading_fixits): New method.
14099 (layout::print_any_fixits): Rename to...
14100 (layout::print_trailing_fixits): ...this, and make private.
14101 Don't print newline fixits.
14102 (diagnostic_show_locus): Move logic for printing one row into...
14103 (layout::print_line): ...this new function. Move the
14104 location_get_source_line call and error-handling from
14105 print_source_line to here. Call print_leading_fixits, and rename
14106 print_any_fixits to print_trailing_fixits.
14107 (selftest::test_fixit_insert_containing_newline): Update now that
14108 newlines are partially supported.
14109 (selftest::test_fixit_insert_containing_newline_2): New test.
14110 (selftest::test_fixit_replace_containing_newline): Update comments.
14111 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14112 * edit-context.c (class added_line): New class.
14113 (class edited_line): Describe newline handling in comment.
14114 (edited_line::actually_edited_p): New method.
14115 (edited_line::print_content): Delete redundant decl.
14116 (edited_line::m_predecessors): New field.
14117 (edited_file::print_content): Call edited_line::print_content.
14118 (edited_file::print_diff): Update to support newlines.
14119 (edited_file::print_diff_hunk): Likewise.
14120 (edited_file::print_run_of_changed_lines): New function.
14121 (edited_file::print_diff_line): Convert to...
14122 (print_diff_line): ...this.
14123 (edited_file::get_effective_line_count): New function.
14124 (edited_line::edited_line): Initialize new field m_predecessors.
14125 (edited_line::~edited_line): Clean up m_predecessors.
14126 (edited_line::apply_fixit): Handle newlines.
14127 (edited_line::get_effective_line_count): New function.
14128 (edited_line::print_content): New function.
14129 (edited_line::print_diff_lines): New function.
14130 (selftest::test_applying_fixits_insert_containing_newline): New
14132 (selftest::test_applying_fixits_replace_containing_newline): New
14134 (selftest::insert_line): New function.
14135 (selftest::test_applying_fixits_multiple_lines): Add example of
14137 (selftest::edit_context_c_tests): Call the new tests.
14139 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14141 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14142 parameter cand. Update dump information.
14143 (get_computation_cost): Update uses.
14145 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14147 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14148 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14149 (get_computation_at, rewrite_use_address): Update use of
14150 get_computation_aff.
14152 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14154 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14155 (get_computation): Delete.
14156 (get_computation_cost): Implement like get_computation_cost_at.
14157 Use get_computation_at.
14158 (get_computation_cost_at): Delete.
14159 (rewrite_use_nonlinear_expr): Use get_computation_at.
14160 (rewrite_use_compare, remove_unused_ivs): Ditto.
14162 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14164 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14168 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14169 (ivopts_global_cost_for_size): Rename parameter and update uses.
14170 (iv_ca_recount_cost): Update uses.
14171 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14172 candidates seperately in n_invs and n_cands.
14173 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14175 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14177 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14178 (find_inv_vars_cb): New.
14179 (find_depends): Renamed to ...
14180 (find_inv_vars): ... this.
14181 (add_candidate_1, force_var_cost): Call find_inv_vars.
14182 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14184 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14186 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14187 inv_vars. Add inv_exprs.
14188 (struct iv_cand): Rename depends_on to inv_vars.
14189 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14190 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14191 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14192 (dump_cand): Dump inv_vars.
14193 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14194 (record_invariant, find_depends, add_candidate_1): Ditto.
14195 (set_group_iv_cost, force_var_cost): Ditto.
14196 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14197 (get_computation_cost_at, get_computation_cost): Ditto.
14198 (determine_group_iv_cost_generic): Ditto.
14199 (determine_group_iv_cost_address): Ditto.
14200 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14201 (determine_group_iv_costs): Ditto.
14202 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14203 (iv_ca_set_remove_invariants): Renamed to ...
14204 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14205 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14206 (iv_ca_set_add_invariants): Renamed to ...
14207 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14208 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14209 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14210 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14211 (create_new_ivs): Remove useless dump.
14213 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14215 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14217 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14218 (iv_ca_set_no_cp, create_new_iv): Ditto.
14220 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14222 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14224 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14226 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14227 function tree_check2.
14229 2017-05-02 Martin Liska <mliska@suse.cz>
14231 * doc/gcov.texi: Add missing preposition.
14232 * gcov.c (function_info::function_info): Properly fill up
14233 all member variables.
14235 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14237 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14239 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14241 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14243 2017-05-02 Martin Liska <mliska@suse.cz>
14246 * lto-streamer-in.c (lto_read_tree_1): Remove
14247 LTO_STREAMER_DEBUG.
14248 * lto-streamer.c (struct tree_hash_entry): Likewise.
14249 (struct tree_entry_hasher): Likewise.
14250 (tree_entry_hasher::hash): Likewise.
14251 (tree_entry_hasher::equal): Likewise.
14252 (lto_streamer_init): Likewise.
14253 (lto_orig_address_map): Likewise.
14254 (lto_orig_address_get): Likewise.
14255 (lto_orig_address_remove): Likewise.
14256 * lto-streamer.h: Likewise.
14257 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14258 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14260 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14262 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14263 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14264 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14265 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14266 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14267 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14268 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14269 (mm_maskz_sub_ss): New intrinsics.
14270 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14271 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14272 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14273 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14274 (__builtin_ia32_subss_mask_round): New builtins.
14275 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14276 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14277 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14279 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14280 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14282 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14285 2017-05-02 Martin Jambor <mjambor@suse.cz>
14287 PR tree-optimization/78687
14288 * tree-sra.c (access): New field parent.
14289 (process_subtree_disqualification): New function.
14290 (disqualify_candidate): Call it.
14291 (build_accesses_from_assign): Reset write flag if creating an
14293 (build_access_subtree): Fill in parent field and also prpagate
14294 down grp_write flag.
14295 (create_artificial_child_access): New parameter set_grp_write, set
14296 grp_write to its value.
14297 (propagate_subaccesses_across_link): Also propagate grp_write flag
14299 (propagate_all_subaccesses): Push the closest parent back to work
14300 queue if add_access_to_work_queue returned true.
14302 2017-05-02 Richard Biener <rguenther@suse.de>
14304 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14305 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14306 -fstrict-overflow documentation.
14307 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14308 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14309 flag_strict_overflow.
14310 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14311 * lto-opts.c (lto_write_options): Do not stream it.
14312 * lto-wrapper.c (merge_and_complain): Do not handle it.
14313 * opts.c (default_options_table): Do not set -fstrict-overflow.
14314 (finish_options): Likewise do not clear it when sanitizing.
14315 * simplify-rtx.c (simplify_const_relational_operation): Do not
14316 test flag_strict_overflow.
14318 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14320 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14321 using enabled attribute.
14322 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14323 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14324 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14325 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14326 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14327 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14328 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14329 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14330 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14331 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14333 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14335 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14337 2017-05-02 Richard Biener <rguenther@suse.de>
14339 PR tree-optimization/80591
14341 2017-04-10 Richard Biener <rguenther@suse.de>
14343 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14346 2017-05-02 Richard Biener <rguenther@suse.de>
14348 PR tree-optimization/80549
14349 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14350 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14351 headers if they do not have a preheader.
14353 2017-05-02 Martin Liska <mliska@suse.cz>
14356 * common.opt: Fix typo.
14357 * doc/invoke.texi: Likewise.
14359 2017-05-01 Jan Beulich <jbeulich@suse.com>
14361 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14362 swapping, add (x,x,m,x,n) alternative.
14364 2017-05-01 Nathan Sidwell <nathan@acm.org>
14366 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14367 unnecessary unadjusted_alignment check.
14369 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14372 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14374 * gimplify.c (gimplify_cilk_detach): New function.
14375 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14376 * tree-core.h: Document EXPR_CILK_SPAWN.
14377 * tree.h (EXPR_CILK_SPAWN): Define.
14379 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14381 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14382 to use new fixit_hint representation, using the "replace" logic.
14383 (get_line_span_for_fixit_hint): Likewise.
14384 (layout::print_any_fixits): Likewise.
14385 (selftest::test_one_liner_many_fixits): Rename to...
14386 (selftest::test_one_liner_many_fixits_1): ...this, and update
14387 comment and expected output to reflect that the multiple fix-it
14388 hints are now consolidated into one insertion.
14389 (selftest::test_one_liner_many_fixits_2): New test.
14390 (selftest::test_diagnostic_show_locus_one_liner): Update for
14392 (selftest::test_fixit_consolidation): Update for fix-it API
14394 * diagnostic.c (print_parseable_fixits): Likewise.
14395 * edit-context.c (edited_line::m_line_events): Convert from
14396 auto_vec <line_event *> to auto_vec <line_event>.
14397 (class line_event): Convert from abstract base class to a concrete
14398 class, taking over the role of replace_event.
14399 (class insert_event): Delete.
14400 (class replace_event): Rename to class line_event. Convert to
14402 (edit_context::add_fixits): Reimplement.
14403 (edit_context::apply_insert): Delete.
14404 (edit_context::apply_replace): Rename to...
14405 (edit_context::apply_fixit): ...this. Convert to half-open range.
14406 (edited_file::apply_insert): Delete.
14407 (edited_file::apply_replace): Rename to...
14408 (edited_file::apply_fixit): ...this.
14409 (edited_line::~edited_line): Drop deletion of events.
14410 (edited_line::apply_insert): Delete.
14411 (edited_line::apply_replace): Rename to...
14412 (edited_line::apply_fixit): ...this. Convert to half-open range.
14413 Update for change to type of m_line_events.
14414 * edit-context.h (edit_context::apply_insert): Delete.
14415 (edit_context::apply_replace): Rename to...
14416 (edit_context::apply_fixit): ...this.
14418 2017-05-01 Martin Sebor <msebor@redhat.com>
14420 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14423 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14426 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14427 __get_cpuid_max returns 0.
14428 (__get_cpuid_count): Ditto.
14430 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14432 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14433 replacement expression is another instance of one of its arguments.
14435 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14438 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14439 check for stack push/pop autoinc.
14440 * config/i386/i386.c (ix86_agi_dependent): Return false
14441 if the only reason why modified_in_p returned true is that
14442 addr is SP based and set_insn is a push or pop.
14444 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14446 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14449 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14452 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14453 (account_size_time): Use two predicates - exec_pred and
14455 (evaluate_conditions_for_known_args): Compute both clause and
14457 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14458 (inline_summary_t::duplicate): Update.
14459 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14461 (compute_inline_parameters): Likewise.
14462 (estimate_edge_size_and_time): Update caluclation of time.
14463 (estimate_node_size_and_time): Compute both time and nonspecialized
14465 (estimate_ipcp_clone_size_and_time): Update.
14466 (inline_merge_summary): Update.
14467 (do_estimate_edge_time): Update.
14468 (do_estimate_edge_size): Update.
14469 (do_estimate_edge_hints): Update.
14470 (inline_read_section, inline_write_summary): Stream both new predicates.
14471 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14473 (compute_inlined_call_time): Cleanup.
14474 (big_speedup_p): Update.
14475 (edge_badness): Update.
14476 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14477 (size_time_entry): Replace predicate by exec_predicate and
14478 nonconst_predicate.
14479 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14480 (estimate_edge_time): Return also nonspec_time.
14481 (reset_edge_growth_cache): Update.
14483 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14485 PR rtl-optimization/80491
14486 * ifcvt.c (noce_process_if_block): When looking for x setter
14487 with missing else_bb, don't check only the insn right before
14488 cond_earliest, but look for the last insn that x is modified in
14489 within the same bb.
14491 PR rtl-optimization/80491
14492 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14494 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14496 PR tree-optimization/80487
14497 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14499 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14501 PR tree-optimization/79697
14502 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14503 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14504 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14506 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14507 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14509 2017-04-28 Martin Sebor <msebor@redhat.com>
14511 PR tree-optimization/80523
14512 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14513 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14515 (maybe_warn, format_directive, parse_directive): Use new functions.
14516 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14518 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14520 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14522 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14524 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14525 target_header_dir): Set correctly.
14526 * configure: Regenerated.
14527 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14528 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14529 instead of SYSTEM_HEADER_DIR.
14531 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14533 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14534 (estimate_local_effects): Likewise.
14535 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14536 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14537 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14538 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14539 do_estimate_edge_time, estimate_edge_time): Likewise.
14540 * ipa-inline-analysis.c (estimate_node_size_and_time,
14541 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14542 (estimate_time_after_inlining): Remove.
14544 2017-04-28 Martin Liska <mliska@suse.cz>
14546 * doc/gcov.texi: Enhance documentation of gcov.
14548 2017-04-28 Martin Liska <mliska@suse.cz>
14550 * doc/gcov.texi: Sort options in alphabetic order.
14551 * doc/gcov-dump.texi: Likewise.
14552 * doc/gcov-tool.texi: Likewise.
14553 * gcov.c (print_usage): Likewise.
14554 * gcov-dump.c (print_usage): Likewise.
14555 * gcov-tool.c (print_merge_usage_message): Likewise.
14556 (print_rewrite_usage_message): Likewise.
14557 (print_overlap_usage_message): Likewise.
14559 2017-04-28 Martin Liska <mliska@suse.cz>
14561 PR gcov-profile/53915
14562 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14564 2017-04-28 Martin Liska <mliska@suse.cz>
14566 PR gcov-profile/79891
14567 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14568 is marked by compiler as living on a line.
14569 (get_cycles_count): Remove usage of the union.
14570 (output_intermediate_file): Likewise.
14571 (find_source): Fix GNU coding style.
14572 (accumulate_line_counts): Remove old non-all block mode.
14573 (output_lines): Remove usage of the union.
14574 * profile.c (output_location): Include all BBs, even if
14575 belonging to a same line (and file) as a previous BB.
14577 2017-04-28 Martin Liska <mliska@suse.cz>
14579 * gcov.c (process_args): Handle new argument 'w'.
14580 (read_graph_file): Assign ID to BBs.
14581 (output_branch_count): Display BB # if verbose flag is set.
14582 (output_lines): Likewise for arcs.
14583 (print_usage): Add '--verbose' option help.
14584 * doc/gcov.texi: Document --verbose (-w) option.
14586 2017-04-28 Martin Liska <mliska@suse.cz>
14588 * gcov.c (struct block_location_info): New struct.
14589 (process_file): Fill up the new structure.
14590 (read_graph_file): Replace usage of encoding by the newly added
14592 (add_line_counts): Likewise.
14593 (accumulate_line_counts): Remove usage of the union.
14594 (function_info::function_info): New function.
14595 (function_info::~function_info): Likewise.
14596 (process_file): Call delete instead of release_function.
14597 (release_function): Release the function.
14598 (release_structures): Call delete instead of release_function.
14599 (solve_flow_graph): Replace usage of num_blocks.
14600 (find_exception_blocks): Likewise.
14601 (output_lines): Fix GNU coding style.
14603 2017-04-28 Martin Liska <mliska@suse.cz>
14606 * coverage.c (coverage_remove_note_file): New function.
14607 * coverage.h: Declare the function.
14608 * toplev.c (finalize): Clean if an error has been seen.
14610 2017-04-28 Martin Liska <mliska@suse.cz>
14612 PR gcov-profile/80031
14613 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14614 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14615 * gcov.c (read_graph_file): Read just number of blocks.
14616 * profile.c (branch_prob): Do not stream 0 flags per a basic
14619 2017-04-28 Martin Liska <mliska@suse.cz>
14621 * gcov-dump.c (tag_*): Add new argument to declarations.
14622 (dump_gcov_file): Likewise.
14623 (tag_blocks): Add and use new argument depth.
14624 (tag_arcs): Likewise.
14625 (tag_lines): Likewise.
14626 (tag_counters): Likewise.
14627 (tag_summary): Likewise.
14628 (dump_working_sets): Use depth to do a proper indentation.
14630 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14633 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14634 * symtab.c (symtab_node::debug_symtab): Move definition here.
14636 2017-04-28 Richard Biener <rguenther@suse.de>
14638 * lto-streamer.h (LTO_major_version): Bump to 7.
14640 2017-04-28 Richard Biener <rguenther@suse.de>
14642 * tree-vrp.c (assert_info): New struct.
14643 (add_assert_info): New helper.
14644 (register_edge_assert_for_2): Refactor to add asserts to a vector
14646 (register_edge_assert_for_1): Likewise.
14647 (register_edge_assert_for): Likewise.
14648 (finish_register_edge_assert_for): New helper actually registering
14649 asserts where live on edge.
14650 (find_conditional_asserts): Adjust.
14651 (find_switch_asserts): Likewise.
14652 (evrp_dom_walker::try_find_new_range): Generalize.
14653 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14655 2017-04-27 Marek Polacek <polacek@redhat.com>
14658 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14659 arg10 and arg11 to itype.
14661 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14663 * doc/extend.texi (Object Size Checking): Improve grammar.
14665 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14668 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14669 that the logic for permitting reciprocal estimates matches that
14672 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14675 * tree.c (type_cache_hasher::equal): Only compare
14676 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14677 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14678 non-aggregate element types.
14679 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14680 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14682 2017-04-27 Richard Biener <rguenther@suse.de>
14684 PR middle-end/80533
14685 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14686 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14687 keeping a reference to a trailing array.
14689 2017-04-27 Richard Biener <rguenther@suse.de>
14691 PR middle-end/80539
14692 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
14693 being in loop-closed SSA form conservatively.
14694 (chrec_fold_multiply_poly_poly): Likewise.
14696 2017-04-27 Tamar Christina <tamar.christina@arm.com>
14698 PR middle-end/79665
14699 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
14700 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
14702 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14705 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
14706 (aarch64_function_arg_alignment): Return unsigned int again, but still
14707 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
14708 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14709 Don't emit -Wpsabi note.
14710 (aarch64_function_arg_boundary): Likewise.
14711 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14714 2017-04-26 Nathan Sidwell <nathan@acm.org>
14716 * tree.h (crc32_unsigned_n): Declare.
14717 (crc32_unsigned, crc32_unsigned): Make inline.
14718 * tree.c (crc32_unsigned_bits): Replace with ...
14719 (crc32_unsigned_n): ... this.
14720 (crc32_unsigned, crc32_byte): Remove.
14721 (crc32_string): Remove unnecessary braces.
14723 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14725 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
14726 * ipa-inline-analysis.c (MAX_TIME): Remove.
14727 (account_size_time): Use sreal for time.
14728 (dump_inline_summary): Update.
14729 (estimate_function_body_sizes): Update.
14730 (estimate_edge_size_and_time): Update.
14731 (estimate_calls_size_and_time): Update.
14732 (estimate_node_size_and_time): Update.
14733 (inline_merge_summary): Update.
14734 (inline_update_overall_summary): Update.
14735 (estimate_time_after_inlining): Update.
14736 (inline_read_section): Update.
14737 (inline_write_summary): Update.
14738 * ipa-inline.c (compute_uninlined_call_time): Update.
14739 (compute_inlined_call_time): Update.
14740 (recursive_inlining): Update.
14741 (inline_small_functions): Update.
14742 (dump_overall_stats): Update.
14743 * ipa-inline.h: Include sreal.h.
14744 (size_time_entry): Turn time to sreal.
14745 (inline_summary): Turn self_time nad time to sreal.
14747 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14749 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
14751 (sreal::stream_out, sreal::stream_in): New.
14752 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
14754 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14756 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
14759 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
14762 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
14764 (dimode_scalar_chain::compute_convert_gain): Ditto.
14765 (dimode_scalar_chain::make_vector_copies): Ditto.
14766 (dimode_scalar_chain::convert_reg): Ditto.
14767 (dimode_scalar_chain::convert_insn): Ditto.
14768 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
14769 (VI248_AVX512BW_1): New mode iterator.
14770 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
14771 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
14774 2017-04-25 Martin Sebor <msebor@redhat.com>
14776 PR tree-optimization/80497
14777 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
14778 constants are representable in HOST_WIDE_INT.
14779 (parse_directive): Ditto.
14781 2017-04-25 Martin Sebor <msebor@redhat.com>
14784 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
14785 (new_zero_array): Adjust signature.
14786 (dom_info::dom_init): Used unsigned rather that size_t.
14787 (dom_info::dom_info): Same.
14789 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14790 Jakub Jelinek <jakub@redhat.com>
14793 * config/arm/arm.c: Include gimple.h.
14794 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14795 returns negative, increment ncrn only if it returned positive.
14796 (arm_needs_doubleword_align): Return int instead of bool,
14797 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
14798 members, but if there is any such non-FIELD_DECL
14799 > PARM_BOUNDARY aligned decl, return -1 instead of false.
14800 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14801 returns negative, increment nregs only if it returned positive.
14802 (arm_setup_incoming_varargs): Likewise.
14803 (arm_function_arg_boundary): Emit -Wpsabi note if
14804 arm_needs_doubleword_align returns negative, return
14805 DOUBLEWORD_ALIGNMENT only if it returned positive.
14807 2017-04-25 Marek Polacek <polacek@redhat.com>
14810 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
14811 first argument to type.
14813 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
14816 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
14817 type checks to test for compatibility instead of equality.
14819 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14820 Jakub Jelinek <jakub@redhat.com>
14823 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
14825 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
14826 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
14827 the alignment computation, but return their maximum in warn_alignment.
14828 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14829 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
14831 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
14832 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14835 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14837 * config/arc/simdext.md (dmpyh): Fix typo.
14839 2017-04-25 Richard Biener <rguenther@suse.de>
14841 PR tree-optimization/80492
14842 * alias.c (compare_base_decls): Handle registers with asm
14843 specification conservatively.
14844 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14845 compare_base_decls returning dont-know properly.
14847 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14849 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
14850 (legitimate_offset_address_p): New function.
14851 (arc_legitimate_address_p): Use above function.
14853 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14855 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
14857 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14859 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
14860 ACCH registers whenever they are available.
14862 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14864 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
14865 double regs fix when not used.
14867 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14869 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
14871 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
14872 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
14874 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14876 * config/arc/arc.c (arc_output_addsi): Check for h-register class
14877 when emitting short ADD instructions.
14879 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14881 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
14883 (cmpsi_cc_c_insn): Likewise.
14884 (cbranchsi4_scratch): Compute proper instruction length using
14885 compact_hreg_operand.
14886 * config/arc/predicates.md (compact_hreg_operand): New predicate.
14888 2017-04-25 Richard Biener <rguenther@suse.de>
14890 PR middle-end/80509
14891 * passes.c (pass_manager::pass_manager): Initialize
14892 m_name_to_pass_map.
14894 2017-04-25 Richard Biener <rguenther@suse.de>
14896 PR tree-optimization/79201
14897 * tree-ssa-sink.c (statement_sink_location): Handle calls.
14899 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14902 * config/s390/vector.md: Split MEM->GPR vector moves for
14903 non-s_operand addresses.
14905 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14908 * config/s390/predicates.md (reload_const_wide_int_operand): New
14910 * config/s390/s390.md ("movti"): Remove d/P alternative.
14911 ("movti_bigconst"): New pattern definition.
14913 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14916 * s390-protos.h (s390_expand_cs_hqi): Removed.
14917 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
14918 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
14919 modes as well as CCZ1mode and CCZmode.
14920 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
14921 signature of s390_emit_compare_and_swap.
14922 (s390_expand_cs_hqi): Likewise, make static.
14923 (s390_expand_cs_tdsi): Generate an explicit compare before trying
14924 compare-and-swap, in some cases.
14925 (s390_expand_cs): Wrapper function.
14926 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
14928 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
14929 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
14930 patterns for small and large integers. Forbid symref memory operands.
14931 Move expander to s390.c. Require cc register.
14932 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
14933 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
14934 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
14935 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
14936 symref memory operands. Remove CC mode and call s390_match_ccmode
14938 ("atomic_exchange<mode>"): Allow and implement all integer modes.
14940 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14942 * config/s390/s390.md (define_peephole2): New peephole to help
14943 combining the load-and-test pattern with volatile memory.
14945 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14947 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
14948 with CCZmode for TARGET_Z196.
14950 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14952 PR rtl-optimization/80501
14953 * combine.c (make_compound_operation_int): Set subreg_code to SET
14954 even for AND with mask of the sign bit of mode.
14956 PR rtl-optimization/80500
14957 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
14958 sum's initial value.
14960 2017-04-25 Julian Brown <julian@codesourcery.com>
14961 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14963 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
14965 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
14967 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
14969 2017-04-25 Julian Brown <julian@codesourcery.com>
14970 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14972 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
14973 (thunderx2t99_sha): New Reservation.
14975 2017-04-25 Julian Brown <julian@codesourcery.com>
14976 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14978 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
14979 type for 1-element load.
14981 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
14983 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
14985 2017-04-24 Martin Jambor <mjambor@suse.cz>
14987 PR tree-optimization/80293
14988 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
14989 char arrays not totally scalarizable if it is false.
14990 (analyze_all_variable_accesses): Pass correct value in the new
14991 parameter. Add a statistics counter.
14993 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
14995 PR middle-end/79931
14996 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
14998 2017-04-24 Richard Biener <rguenther@suse.de>
15000 PR tree-optimization/80494
15001 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15002 out for complex types.
15004 2017-04-24 Richard Biener <rguenther@suse.de>
15006 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15007 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15008 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15009 (DFS): Adjust and never fail.
15010 (sccvn_dom_walker::fail): Remove.
15011 (sccvn_dom_walker::before_dom_children): Adjust.
15012 (run_scc_vn): Likewise and never fail.
15013 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15014 (pass_fre::execute): Likewise.
15016 2017-04-24 Richard Biener <rguenther@suse.de>
15018 PR tree-optimization/79725
15019 * tree-ssa-sink.c (statement_sink_location): Return whether
15020 failure reason was zero uses. Move that check later.
15021 (sink_code_in_bb): Deal with zero uses by removing the stmt
15024 2017-04-24 Richard Biener <rguenther@suse.de>
15027 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15028 pointer-based references.
15030 2017-04-24 Richard Biener <rguenther@suse.de>
15033 * pass_manager.h (pass_manager::operator new): Remove.
15034 (pass_manager::operator delete): Likewise.
15035 * passes.c (pass_manager::operator new): Remove.
15036 (pass_manager::operator delete): Likewise.
15037 (pass_manager::pass_manager): Zero individual pass members.
15039 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15042 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15043 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15044 Check "XEXP (src, 1)" operand here.
15045 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15046 Check "XEXP (src, 1)" operand here.
15047 (dimode_scalar_chain::make_vector_copies): Detect count register
15048 of a shift instruction. Zero extend count register from QImode
15049 to DImode to satisfy vector shift pattern count operand predicate.
15050 Substitute vector shift count operand with a DImode copy.
15051 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15054 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15056 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15057 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15058 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15059 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15060 (UNSPEC_NOREX_MEM): Remove definition.
15062 2017-04-21 Richard Biener <rguenther@suse.de>
15064 PR tree-optimization/79547
15065 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15066 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15067 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15068 without any constraints.
15070 2017-04-21 Richard Biener <rguenther@suse.de>
15072 PR tree-optimization/78847
15073 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15075 2017-04-21 Richard Biener <rguenther@suse.de>
15077 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15078 (build_distinct_type_copy): Likewise.
15079 (build_variant_type_copy): Likewise.
15080 * tree.c (build_qualified_type): Pass down mem-stat info.
15081 (build_distinct_type_copy): Likewise.
15082 (build_variant_type_copy): Likewise.
15084 2017-04-21 Richard Biener <rguenther@suse.de>
15086 PR tree-optimization/80237
15087 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15089 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15090 for a simplified result.
15092 2016-04-21 Richard Biener <rguenther@suse.de>
15094 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15095 sth as strict as a simple_iv but a chrec without symbols and an
15096 operand defined in the loop we are peeling (and not some subloop).
15097 (propagate_constants_for_unrolling): Propagate all constants.
15099 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15102 * config/i386/i386.c (print_reg): Remove assert for disalowed
15103 regno values, call output_operand_lossage instead.
15105 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15108 * config/i386/constraints.md (Yc): New register constraint.
15109 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15110 Use Yc constraint for alternative 2 of operand 0. Remove
15111 preferred_for_speed attribute.
15113 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15115 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15116 lastprivate clauses in SIMT case.
15118 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15120 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15122 2017-04-20 Richard Biener <rguenther@suse.de>
15124 PR tree-optimization/57796
15125 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15126 as N scalar stores.
15127 (vect_model_load_cost): Cost gathers as N scalar loads.
15129 2017-04-20 Richard Biener <rguenther@suse.de>
15131 * ggc-page.c (ggc_allocated_p): Rename to ...
15132 (safe_lookup_page_table_entry): ... this and return the lookup
15134 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15136 2017-04-20 Richard Biener <rguenther@suse.de>
15138 PR tree-optimization/80453
15139 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15140 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15141 from the conditions.
15142 (vn_phi_eq): Pass them down.
15143 (vn_phi_lookup): Record them.
15144 (vn_phi_insert): Likewise.
15146 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15148 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15149 uninitialized variable warning to avoid buffer overrun.
15151 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15154 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15155 is suppressed for '{ 0 }' in C.
15157 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15159 * BASE-VER: Set to 8.0.0.
15161 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15163 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15164 priority .init_array and .fini_array section with SECTION_NOTYPE
15167 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15169 PR middle-end/80423
15170 * tree.h (build_array_type): Add typeless_storage default argument.
15171 * tree.c (type_cache_hasher::equal): Also compare
15172 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15173 (build_array_type): Add typeless_storage argument, set
15174 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15176 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15177 (build_array_type): Likewise. Add typeless_storage argument.
15179 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15180 Jakub Jelinek <jakub@redhat.com>
15182 PR tree-optimization/80426
15183 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15184 operation on symbolic operands, also compute the overflow for the
15185 invariant part when the operation degenerates into a negation.
15187 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15190 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15191 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15194 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15196 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15199 * config/avr/avr.c (tree.h): Include it.
15200 (cgraph.h): Include it.
15201 (avr_encode_section_info): Don't warn for uninitialized progmem
15202 variable if it's just an alias.
15204 2017-04-19 Richard Biener <rguenther@suse.de>
15207 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15208 when needed by AutoPGO.
15210 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15213 * doc/lto.texi: Remove an extra 'that'.
15215 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15217 PR rtl-optimization/80429
15218 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15219 are only used in debug insns.
15221 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15222 Vladimir Makarov <vmakarov@redhat.com>
15224 * config/sparc/predicates.md (input_operand): Add comment. Return
15225 true for any memory operand when LRA is in progress.
15226 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15228 2017-04-18 Jeff Law <law@redhat.com>
15231 * mips.md ({return,simple_return}_internal): Do not overwrite
15234 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15236 PR tree-optimization/80443
15237 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15238 instead of adding 1, subtract -1 and similarly instead of subtracting
15241 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15243 PR rtl-optimization/80357
15244 * haifa-sched.c (tmp_bitmap): New variable.
15245 (model_recompute): Handle duplicate use records.
15246 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15247 (free_global_sched_pressure_data): Free it.
15249 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15252 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15253 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15254 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15255 instead of SYSTEM_HEADER_DIR.
15257 2017-04-18 Jeff Law <law@redhat.com>
15259 PR middle-end/80422
15260 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15261 predecessors after walking up the insn chain.
15263 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15266 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15267 sizetype type into debug info.
15269 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15272 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15273 unneeded test for TARGET_UPPER_REGS_SF.
15274 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15276 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15279 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15280 instead of gsi_after_labels.
15282 2017-04-18 Jeff Law <law@redhat.com>
15284 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15288 2017-04-13 Jeff Law <law@redhat.com>
15289 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15290 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15292 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15295 * config/avr/avr.c (intl.h): Include it.
15296 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15298 2017-04-18 Martin Liska <mliska@suse.cz>
15300 PR gcov-profile/78783
15301 * gcov-tool.c (gcov_output_files): Validate that destination
15302 file is either removed by the tool or by a user.
15304 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15305 Guy Benyei <guybe@mellanox.com>
15307 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15308 block, and do not negate it, the stored id is already negative.
15310 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15312 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15314 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15317 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15318 masks of options that should be turned off if the VSX vector
15319 options are turned off.
15320 (OTHER_P8_VECTOR_MASKS): Likewise.
15321 (OTHER_VSX_VECTOR_MASKS): Likewise.
15322 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15323 rs6000_disable_incompatible_switches to validate no type switches
15325 (rs6000_incompatible_switch): New function to disallow turning on
15326 other vector options if -mno-vsx, -mno-power8-vector, or
15327 -mno-power9-vector are specified.
15329 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15331 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15333 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15335 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15336 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15337 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15338 (ARG_POINTER_CFA_OFFSET): Likewise.
15340 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15342 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15343 conditions to take advantage of various optimizations.
15345 2017-04-13 Jeff Law <law@redhat.com>
15347 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15348 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15349 (zero_extendsidi2_dext): Likewise.
15351 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15354 * fold-const.c (fold_ternary_loc): Revert
15355 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15358 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15360 PR rtl-optimization/80343
15361 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15362 new scratch pseudo.
15364 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15367 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15368 to ubsan_encode_value.
15370 2017-04-13 Jeff Law <law@redhat.com>
15372 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15373 appearing in DEBUG_INSNs.
15375 2017-04-13 Martin Liska <mliska@suse.cz>
15377 PR gcov-profile/80413
15378 * gcov-io.c (gcov_write_string): Copy to buffer just when
15379 allocated size is greater than zero.
15381 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15384 * dwarf2out.c (decls_for_scope): Ignore declarations of
15385 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15387 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15390 * ipa-visibility.c (non_local_p): Fix typos.
15391 (localize_node): When localizing symbol in same comdat group,
15392 dissolve the group only when we know external symbols are going
15394 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15396 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15398 PR tree-optimization/79390
15399 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15400 order does not result in usable sequence, retry with reversed operand
15406 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15407 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15408 op0 instead of fold_convert_loc (loc, type, arg0).
15410 2017-04-12 Jeff Law <law@redhat.com>
15412 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15413 has a delay slot in the generated code.
15415 * config/cris/cris.md (cris_preferred_reload_class): Return
15416 GENNONACR_REGS rather than GENERAL_REGS.
15418 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15421 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15422 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15423 signedness of the result type.
15425 2017-04-12 Richard Biener <rguenther@suse.de>
15426 Jeff Law <law@redhat.com>
15428 PR tree-optimization/80359
15429 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15430 trim stores to TARGET_MEM_REFs.
15432 2017-04-12 Richard Biener <rguenther@suse.de>
15434 PR tree-optimization/79390
15435 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15436 threading case even more.
15438 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15441 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15442 for quad_address_p for TImode, instead of just not indexed_address.
15444 2017-04-12 Richard Biener <rguenther@suse.de>
15445 Bernd Edlinger <bernd.edlinger@hotmail.de>
15447 PR middle-end/79671
15448 * alias.c (component_uses_parent_alias_set_from): Handle
15449 TYPE_TYPELESS_STORAGE.
15450 (get_alias_set): Likewise.
15451 * tree-core.h (tree_type_common): Add typeless_storage flag.
15452 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15453 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15454 for types containing members with TYPE_TYPELESS_STORAGE.
15455 (place_field): Likewise.
15456 (layout_type): Likewise for ARRAY_TYPE.
15457 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15458 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15459 TYPE_TYPELESS_STORAGE.
15460 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15462 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15465 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15466 first argument to type.
15468 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15472 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15473 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15474 (rs6000_expand_binop_builtin): Likewise.
15475 (rs6000_expand_ternop_builtin): Likewise; also add missing
15476 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15477 vshasigma built-ins.
15478 * doc/extend.texi: Document that vec_xxpermdi's third argument
15479 must be a constant.
15481 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15483 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15484 Use shift_const cost parameter when calculating gain of STV shifts.
15486 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15488 PR rtl-optimization/70478
15489 * lra-constraints.c (process_alt_operands): Check memory for
15490 disfavoring memory insn operand.
15492 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15494 PR middle-end/80100
15495 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15496 left shift in unsigned HOST_WIDE_INT type.
15498 PR rtl-optimization/80385
15499 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15500 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15503 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15504 if they have any depend clauses.
15506 2017-04-11 Martin Liska <mliska@suse.cz>
15509 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15510 * ipa-split.c (split_function): Create a local comdat symbol
15511 if caller is in a comdat group.
15513 2017-04-11 Martin Liska <mliska@suse.cz>
15516 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15519 2017-04-11 Martin Sebor <msebor@redhat.com>
15521 PR middle-end/80364
15522 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15523 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15525 (directive::set_width, directive::set_precision, format_character):
15527 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15530 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15533 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15534 conflict, set target->arch_name instead of target->cpu_name.
15536 2017-04-11 Richard Biener <rguenther@suse.de>
15538 PR tree-optimization/80374
15539 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15540 build_zero_cst, remove fold_convertible_p check again.
15542 2017-04-11 Martin Liska <mliska@suse.cz>
15545 * ubsan.c (instrument_object_size): Do not instrument register
15548 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15551 * config/i386/i386-builtin-types.def
15552 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15553 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15554 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15555 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15556 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15557 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15558 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15559 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15560 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15561 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15562 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15563 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15564 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15565 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15566 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15567 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15568 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15569 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15570 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15571 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15572 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15573 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15574 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15575 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15576 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15577 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15578 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15579 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15580 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15581 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15582 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15583 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15584 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15585 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15586 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15587 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15588 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15589 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15590 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15591 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15592 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15593 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15594 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15595 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15596 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15597 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15599 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15600 flag to second_arg_count, handle 4 argument function type _COUNT
15601 aliases, handle second_arg_count on second argument rather than last.
15603 2017-04-10 Jeff Law <law@redhat.com>
15605 PR tree-optimization/80374
15606 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15607 record anything if we can not convert integer_zero_node to the
15610 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15613 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15614 Enhance special handling given to the TARGET_P9_MINMAX option in
15615 relation to certain other options.
15617 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15619 PR tree-optimization/80153
15620 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15621 remove POINTER_PLUS_EXPR's base part directly, rather than through
15624 2017-04-10 Richard Biener <rguenther@suse.de>
15625 Bin Cheng <bin.cheng@arm.com>
15627 PR tree-optimization/80153
15628 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15629 the first element of pointer type aff_tree. Build result expr in
15631 (add_elt_to_tree): Convert to type unconditionally. Remove other
15632 fold_convert calls.
15633 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15634 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15636 2017-04-10 Richard Biener <rguenther@suse.de>
15638 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15641 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15643 PR rtl-optimization/70478
15644 * lra-constraints.c (curr_small_class_check): New.
15645 (update_and_check_small_class_inputs): New.
15646 (process_alt_operands): Update curr_small_class_check. Disfavor
15647 alternative insn memory operands. Check available regs for small
15650 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15653 * config/mips/mips.opt (-mvirt): Update description.
15654 * doc/invoke.texi (-mvirt): Likewise.
15656 2017-04-10 Richard Biener <rguenther@suse.de>
15658 PR middle-end/80362
15659 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15660 looking for NEGATE_EXPR in -A / -B to A / B folding.
15662 2017-04-10 Martin Liska <mliska@suse.cz>
15664 PR gcov-profile/80224
15665 * gcov.c (print_usage): Fix usage string.
15666 (get_gcov_intermediate_filename): Remove.
15667 (output_gcov_file): Use both for normal and intermediate format.
15668 (generate_results): Do not initialize special file for
15669 intermediate format.
15671 2017-04-10 Richard Biener <rguenther@suse.de>
15673 PR tree-optimization/80304
15674 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15677 2017-04-10 Nathan Sidwell <nathan@acm.org>
15680 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15681 (rs6000_init_builtins): Use it.
15683 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15685 * config/arm/arm.md (<mrc>): Add mode to SET source.
15686 (<mrrc>): Likewise.
15688 2017-04-10 Richard Biener <rguenther@suse.de>
15690 PR middle-end/80344
15691 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
15693 2017-04-10 Jakub Jelinek <jakub@redhat.com>
15696 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
15697 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
15698 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
15699 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
15700 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
15701 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
15702 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
15703 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
15704 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
15705 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
15706 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
15707 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
15708 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
15709 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
15710 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
15711 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
15712 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
15713 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
15714 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
15715 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
15716 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
15717 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
15718 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
15720 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
15722 PR rtl-optimization/70478
15723 * lra-constraints.c: Reverse the last patch.
15725 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
15727 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
15728 Add comment for WCHAR_T.
15730 2017-04-08 Martin Liska <mliska@suse.cz>
15733 2017-04-07 Martin Liska <mliska@suse.cz>
15736 * ipa-split.c (split_function): Add function part to a same comdat
15739 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15742 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
15744 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
15746 * rs6000/rs6000.c (vec_load_pendulum): Rename...
15747 (vec_pairing): ...to this.
15748 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
15749 (rs6000_sched_init): Adjust for name change.
15750 (struct rs6000_sched_context): Likewise.
15751 (rs6000_init_sched_context): Likewise.
15752 (rs6000_set_sched_context): Likewise.
15754 2017-04-07 Jakub Jelinek <jakub@redhat.com>
15760 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
15762 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
15763 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
15764 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
15766 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
15768 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
15770 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15772 PR rtl-optimization/70703
15773 * ira-color.c (update_conflict_hard_regno_costs): Use
15774 int64_t instead of HOST_WIDE_INT.
15776 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15778 PR rtl-optimization/70478
15779 * lra-constraints.c (process_alt_operands): Disfavor alternative
15780 insn memory operands.
15782 2017-04-07 Jeff Law <law@redhat.com>
15784 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
15785 CALL and NOTE_INSN_CALL_ARG_LOCATION.
15787 2017-04-07 Martin Liska <mliska@suse.cz>
15790 * config/aarch64/aarch64.c (aarch64_process_target_attr):
15791 Show error message instead of an ICE.
15793 2017-04-07 Martin Liska <mliska@suse.cz>
15796 * ipa-split.c (split_function): Add function part to a same comdat
15799 2017-04-07 Richard Biener <rguenther@suse.de>
15801 PR middle-end/80341
15802 * tree.c (get_unwidened): Also handle ! for_type case for
15804 * convert.c (do_narrow): Split out from ...
15805 (convert_to_integer_1): ... here. Do not pass final truncation
15806 type to get_unwidened for TRUNC_DIV_EXPR.
15808 2017-04-07 Richard Biener <rguenther@suse.de>
15810 * tree-affine.c (wide_int_ext_for_comb): Take type rather
15812 (aff_combination_const): Adjust.
15813 (aff_combination_scale): Likewise.
15814 (aff_combination_add_elt): Likewise.
15815 (aff_combination_add_cst): Likewise.
15816 (aff_combination_convert): Likewise.
15817 (add_elt_to_tree): Likewise. Remove unused argument.
15818 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
15820 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
15822 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
15824 * config/arm/arm.c (arm_default_short_enums): Use
15825 ARM_DEFAULT_SHORT_ENUMS.
15826 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
15828 2017-04-06 Jakub Jelinek <jakub@redhat.com>
15831 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
15832 members with redundant out-of-class redeclaration.
15834 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15837 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
15838 * config/i386/i386.md (*zero_extendsidi2):
15839 Add (?*x,*x) and (?*v,*v) alternatives.
15841 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15844 * config/i386/i386.c (ix86_expand_builtin)
15845 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
15846 mode from insn data. Convert operands to insn operand mode.
15847 Copy operands that don't satisfy insn predicate to a register.
15849 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
15851 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
15854 2017-04-06 Richard Biener <rguenther@suse.de>
15856 PR tree-optimization/80334
15857 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
15858 preserve alignment of accesses.
15860 2017-04-06 Richard Biener <rguenther@suse.de>
15862 PR tree-optimization/80262
15863 * tree-sra.c (build_ref_for_offset): Preserve address-space
15865 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
15866 Drop useless address-space information on MEM_REF offsets.
15868 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
15870 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
15872 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15874 PR rtl-optimization/70703
15875 * ira-color.c (update_conflict_hard_regno_costs): Use
15876 HOST_WIDE_INT instead of long.
15878 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
15881 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
15882 not defined for x86_64 target. Add -mmmx target option when __SSE2__
15884 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
15885 for x86_64 target. Handle -m3dnowa option.
15887 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15889 PR rtl-optimization/70703
15890 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
15891 (update_conflict_hard_regno_costs): Use long instead of unsigned
15892 arithmetic for cost calculation.
15894 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15895 Bernd Edlinger <bernd.edlinger@hotmail.de>
15898 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
15901 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
15904 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
15905 ptr_mode with Pmode throughout.
15906 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
15907 into probe_stack_range and use DImode.
15909 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
15912 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
15913 call_eh_return is true.
15915 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15917 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
15918 Initialize last_match_fntype_index.
15920 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15923 * tree-nvr.c: Include internal-fn.h.
15924 (pass_return_slot::execute): Ignore internal calls without
15927 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15928 Richard Biener <rguenther@suse.de>
15931 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
15932 captures used multiple times, except for the last use.
15933 * generic-match-head.c: Include gimplify.h.
15935 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15937 PR tree-optimization/79390
15938 * target.h (struct noce_if_info): Declare.
15939 * targhooks.h (default_noce_conversion_profitable_p): Declare.
15940 * target.def (noce_conversion_profitable_p): New target hook.
15941 * ifcvt.h (struct noce_if_info): New type, moved from ...
15942 * ifcvt.c (struct noce_if_info): ... here.
15943 (noce_conversion_profitable_p): Renamed to ...
15944 (default_noce_conversion_profitable_p): ... this. No longer
15946 (noce_try_store_flag_constants, noce_try_addcc,
15947 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
15948 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
15949 instead of noce_conversion_profitable_p.
15950 * config/i386/i386.c: Include ifcvt.h.
15951 (ix86_option_override_internal): Don't override
15952 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
15953 (ix86_noce_conversion_profitable_p): New function.
15954 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
15955 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
15956 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
15957 * doc/tm.texi: Regenerated.
15959 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15961 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
15964 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15967 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
15968 instructions for small multiply cores.
15970 2017-04-04 Jeff Law <law@redhat.com>
15972 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
15974 (mips_expand_vec_perm_const): Initialize elements in orig_perm
15975 that are not set by the loop over the elements.
15977 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15980 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
15981 int mode, convert_modes it to mode as unsigned, otherwise use
15982 lowpart_subreg to mode rather than SImode.
15983 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
15984 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
15985 Use DImode instead of SImode for the shift count operand.
15986 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
15989 2017-04-04 Richard Biener <rguenther@suse.de>
15991 PR middle-end/80281
15992 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
15993 arithmetic done for the negate or the plus. Simplify.
15994 (A - (-B) -> A + B): Likewise.
15995 * fold-const.c (split_tree): Make sure to not negate pointers.
15997 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
15999 PR rtl-optimization/60818
16000 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16001 a compare of comparisons with the thing compared if this results
16002 in a different machine mode.
16004 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16006 * alias.c (base_alias_check): Fix typo in comment.
16007 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16008 * cgraphunit.c (symbol_table::compile): Likewise.
16009 * collect2.c (maybe_run_lto_and_relink): Likewise.
16010 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16011 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16012 * config/avr/avr.c (avr_map_op_t): Likewise.
16013 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16014 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16015 * config/epiphany/epiphany.md (movcc): Likewise.
16016 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16017 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16019 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16020 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16021 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16022 * config/sh/sh.c (sh_rtx_costs): Likewise.
16023 * fold-const.c (fold_truth_andor): Likewise.
16024 * genautomata.c (collapse_flag): Likewise.
16025 * gengtype.h (struct type::u::s): Likewise.
16026 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16027 * input.c (FORMAT_AMOUNT): Likewise.
16028 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16029 (known_aggs_to_agg_replacement_list): Likewise.
16030 * ipa-inline-analysis.c: Likewise.
16031 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16032 * ipa-polymorphic-call.c
16033 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16034 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16035 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16037 * modulo-sched.c (apply_reg_moves): Likewise.
16038 * omp-expand.c (build_omp_regions_1): Likewise.
16039 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16040 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16041 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16042 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16043 * value-prof.c: Likewise.
16044 * var-tracking.c (val_reset): Likewise.
16046 2017-04-03 Richard Biener <rguenther@suse.de>
16048 PR tree-optimization/80275
16049 * fold-const.c (split_address_to_core_and_offset): Handle
16052 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16054 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16055 descriptors is at least equal to that of functions.
16057 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16059 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16061 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16064 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16065 (mov<IMOD4:mode>): New expander.
16066 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16068 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16070 PR rtl-optimization/79405
16071 * fwprop.c (propagations_left): New variable.
16072 (forward_propagate_into): Decrement it.
16073 (fwprop_init): Initialize it.
16074 (fw_prop): If the variable has reached zero, stop propagating.
16075 (fwprop_addr): Ditto.
16077 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16080 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16081 a FUNCTION_DECL, pass it as decl instead of origin to
16084 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16086 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16089 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16092 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16093 TARGET_VSX_SMALL_INTEGER.
16095 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16097 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16098 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16100 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16102 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16103 extraction from odd-numbered MSA register.
16105 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16107 PR middle-end/80173
16108 * expmed.c (store_bit_field_1): Don't attempt to create
16109 a word subreg out of hard registers wider than word if they
16110 have HARD_REGNO_NREGS of 1 for their mode.
16112 PR middle-end/80163
16113 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16114 conversions to integer types wider than word and pointer.
16117 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16118 (rtx_equal_for_cselib_p): Pass 0 to it.
16119 * cselib.c (cselib_hasher::equal): Likewise.
16120 (rtx_equal_for_cselib_1): Add depth argument. If depth
16121 is 128, don't look up VALUE locs and punt. Increment
16122 depth in recursive calls when walking VALUE locs.
16124 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16126 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16127 (make_gcov_file_name): Use the canonical path name for generating
16129 (read_line): Fix handling of files with ascii null bytes.
16131 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16133 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16134 to initialise a vector register instead
16135 of using a const_int.
16137 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16139 PR translation/80189
16140 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16141 diagnostic messages.
16143 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16146 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16147 (dfp_diex_<mode>): Update mode of operand 1.
16148 * doc/extend.texi (dxex, dxexq): Document change to return type.
16149 (diex, diexq): Document change to argument type.
16151 2017-03-30 Martin Jambor <mjambor@suse.cz>
16154 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16155 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16156 it reflects the signature changes performed at the callee side.
16157 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16158 to cgraph_build_function_type_skip_args.
16159 (build_function_decl_skip_args): Adjust call to the above function.
16161 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16164 * config/i386/sse.md
16165 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16166 register as dest whenever it is a MEM not rtx_equal_p to the
16167 corresponding dup operand, and when forcing into reg move the
16168 reg into the memory afterwards.
16169 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16170 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16171 for the force_reg mode.
16172 (avx512vl_vextractf128<mode>): Use register as dest either
16173 always when a MEM, or when it is a MEM not rtx_equal_p to the
16174 corresponding dup operand, or even not when it is a CONST_VECTOR
16175 depending on the mode and lo vs. hi.
16176 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16178 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16179 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16180 Likewise. Require that operands[2] is even.
16181 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16182 Remove extraneous parens. Require that operands[2] is a multiple
16184 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16185 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16186 disallow memory then.
16188 2017-03-30 Richard Biener <rguenther@suse.de>
16190 PR tree-optimization/77498
16191 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16192 to non-constants over backedges.
16194 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16196 PR rtl-optimization/80233
16197 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16198 as last_combined_insn. Do not test for BARRIER_P separately.
16200 2017-03-29 Andreas Schwab <schwab@suse.de>
16203 * calls.c (prepare_call_address): Convert funexp to Pmode before
16204 copying to temp reg.
16206 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16208 PR tree-optimization/80158
16209 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16210 Handle possible future case of more than one alternate
16212 (replace_rhs_if_not_dup): Likewise.
16213 (replace_one_candidate): Likewise.
16215 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16217 PR rtl-optimization/80193
16218 * ira.c (ira): Do not check allocation for LRA.
16220 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16222 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16223 (nvptx_output_simt_exit): Declare.
16224 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16225 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16226 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16227 (nvptx_declare_function_name): ...here. Emit declaration of local
16228 memory space buffer for omp_simt_enter insn.
16229 (nvptx_output_unisimt_switch): New.
16230 (nvptx_output_softstack_switch): New.
16231 (nvptx_output_simt_enter): New.
16232 (nvptx_output_simt_exit): New.
16233 * config/nvptx/nvptx.h (struct machine_function): New fields
16234 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16235 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16236 (UNSPECV_SIMT_EXIT): Ditto.
16237 (omp_simt_enter_insn): New insn.
16238 (omp_simt_enter): New expansion.
16239 (omp_simt_exit): New insn.
16240 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16242 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16243 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16244 (expand_GOMP_SIMT_EXIT): New.
16245 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16246 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16247 (GOMP_SIMT_EXIT): Ditto.
16248 * target-insns.def (omp_simt_enter): New insn.
16249 (omp_simt_exit): Ditto.
16250 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16252 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16253 (lower_rec_input_clauses): Likewise.
16254 (lower_lastprivate_clauses): Handle SIMT privatization.
16256 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16257 (ompdevlow_adjust_simt_enter): New.
16258 (find_simtpriv_var_op): New.
16259 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16260 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16262 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16263 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16264 (copy_decl_for_dup_finish): Ditto.
16266 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16268 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16271 * config/i386/i386.c (ix86_option_override_internal): Always
16272 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16274 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16276 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16278 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16280 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16281 mark new edge's irreducible flag accordign to it.
16282 (vect_do_peeling): Check loop preheader edge's irreducible flag
16283 and pass it to function slpeel_add_loop_guard.
16285 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16287 PR tree-optimization/80218
16288 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16289 Update block frequencies and counts.
16291 2017-03-28 Richard Biener <rguenther@suse.de>
16293 PR tree-optimization/78644
16294 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16295 of a simplification result we may not use it at all.
16297 2017-03-28 Richard Biener <rguenther@suse.de>
16300 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16301 without arguments, generate default definition of a SSA name.
16303 2017-03-28 Richard Biener <rguenther@suse.de>
16305 PR middle-end/80222
16306 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16307 TYPE_REF_CAN_ALIAS_ALL references.
16308 * fold-const.c (fold_indirect_ref_1): Likewise.
16310 2017-03-28 Martin Liska <mliska@suse.cz>
16313 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16314 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16316 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16317 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16319 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16320 (EXTRA_SPECS): Define.
16321 (SUBTARGET_EXTRA_SPECS): Likewise.
16322 (SUBTARGET_CPP_SPEC): Likewise.
16323 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16324 SUBTARGET_EXTRA_SPECS.
16325 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16327 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16329 * config/arc/simdext.md (vst64_insn): Update pattern.
16330 (vld32wh_insn): Likewise.
16331 (vld32wl_insn): Likewise.
16332 (vld64_insn): Likewise.
16333 (vld32_insn): Likewise.
16335 2017-03-28 Marek Polacek <polacek@redhat.com>
16338 * fold-const.c (fold_comparison): Use protected_set_expr_location
16339 instead of SET_EXPR_LOCATION.
16341 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16343 * tree.c (add_expr): Avoid name lookup warning.
16345 2017-03-27 Jeff Law <law@redhat.com>
16347 PR tree-optimization/80216
16348 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16349 function name. Limit recursion depth.
16350 (record_temporary_equivalences): Corresponding changes.
16352 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16354 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16357 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16360 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16362 * cfgcleanup.c (reg_note_cfa_p): New array.
16363 (insns_have_identical_cfa_notes): New function.
16364 (old_insns_match_p): Don't cross-jump in between /f
16365 and non-/f instructions. If both i1 and i2 are frame related,
16366 verify all CFA notes, their order and content.
16368 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16371 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16372 HImode and SImode with zero extend to DImode to one insn.
16373 (bswap<mode>2_extenddi): Likewise.
16374 (bswapsi2_extenddi): Likewise.
16375 (bswaphi2_extendsi): Likewise.
16376 (bswaphi2): Combine bswap HImode and SImode into one insn.
16377 Separate memory insns from swapping register.
16378 (bswapsi2): Likewise.
16379 (bswap<mode>2): Likewise.
16380 (bswaphi2_internal): Delete, no longer used.
16381 (bswapsi2_internal): Likewise.
16382 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16383 store, and gpr<-gpr swap insns.
16384 (bswap<mode>2_store): Likewise.
16385 (bswaphi2_reg): Register only splitter, combine with the splitter.
16386 (bswaphi2 splitter): Likewise.
16387 (bswapsi2_reg): Likewise.
16388 (bswapsi2 splitter): Likewise.
16389 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16390 the insns into load, store, and register/register insns.
16391 (bswapdi2_ldbrx): Likewise.
16392 (bswapdi2_load): Likewise.
16393 (bswapdi2_store): Likewise.
16394 (bswapdi2_reg): Likewise.
16396 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16398 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16399 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16401 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16404 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16407 special handling for target option conflicts between dform
16408 options (-mpower9-dform, -mpower9-dform-vector,
16409 -mpower9-dform-scalar) and -mno-direct-move.
16411 2017-03-27 Richard Biener <rguenther@suse.de>
16413 PR tree-optimization/80181
16414 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16416 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16418 * config/arc/predicates.md (move_double_src_operand): Replace the
16419 call to move_double_src_operand with a call to address_operand.
16421 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16423 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16424 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16425 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16427 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16429 * config/arc/predicates.md (long_immediate_loadstore_operand):
16430 Consider scaled addresses cases.
16432 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16434 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16435 restored when in interrupt.
16436 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16437 doesn't have delay slot.
16439 2017-03-27 Richard Biener <rguenther@suse.de>
16442 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16443 inlined thunk clones.
16445 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16448 * asan.c (instrument_derefs): Copy over last operand from
16449 original COMPONENT_REF to the new COMPONENT_REF with
16450 DECL_BIT_FIELD_REPRESENTATIVE.
16451 * ubsan.c (instrument_object_size): Likewise.
16453 2017-03-27 Richard Biener <rguenther@suse.de>
16455 PR tree-optimization/80170
16456 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16457 sure DR/SCEV didnt fold in constants we do not see when looking
16458 at the reference base alignment.
16460 2017-03-27 Richard Biener <rguenther@suse.de>
16462 PR middle-end/80171
16463 * gimple-fold.c (fold_ctor_reference): Properly guard against
16464 NULL return value from canonicalize_constructor_val.
16466 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16469 * config/i386/i386.c (ix86_expand_builtin)
16470 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16471 flags reg setting and flags reg using instructions.
16472 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16473 clobbering instructions to zero extend op2.
16475 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16477 * doc/install.texi (Configuration) <--with-aix-soname>:
16478 Update link to AIX ld.
16480 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16482 PR rtl-optimization/80160
16483 PR rtl-optimization/80159
16484 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16485 reg_alternate_class into account.
16487 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16490 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16491 to consider in curr_insn_transform.
16493 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16495 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16496 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16497 and emit_mode_inner.
16499 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16501 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16502 argument to the overloaded builtin variants. Use the new flag to
16503 deprecate certain builtin variants.
16504 * config/s390/s390-builtin-types.def: Add new builtin types.
16505 * config/s390/s390-builtins.h: Support new flags field for
16506 overloaded builtins.
16507 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16508 (s390_macro_to_expand): Enable vector float data type.
16509 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16510 builtins by incrementing the __VEC__ version number.
16511 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16513 (s390_resolve_overloaded_builtin): Emit error messages depending
16514 on the builtin flags.
16515 * config/s390/s390.c (s390_expand_builtin): Support additional
16516 flags argument. Change error message to match the messages
16517 emitted in s390-c.c.
16518 * config/s390/s390.md: New UNSPEC_* constants.
16519 (op_type): Add new instruction types.
16520 * config/s390/vecintrin.h: Add new builtins and test data class
16522 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16523 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16524 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16525 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16526 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16527 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16529 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16530 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16531 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16532 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16534 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16535 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16536 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16537 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16539 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16540 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16541 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16542 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16543 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16544 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16545 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16547 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16548 ("vec_scatter_element<V_HW_4:mode>_DI")
16549 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16550 ("vec_fpint<mode>", "vflls")
16551 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16552 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16553 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16554 ("*vec_cmphe<mode>_cc"): ... these.
16556 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16557 mode constant instead of magic value.
16559 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16561 * config/s390/s390.c (s390_expand_vec_compare): Support other
16562 vector floating point modes than just V2DF.
16563 (s390_expand_vcond): Likewise.
16564 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16565 (s390_cannot_change_mode_class): Prevent mode changes between TF
16566 and V1TF in vector registers.
16567 * config/s390/s390.md (DF, SF): New mode attributes.
16568 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16569 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16570 SFmode support for VRs.
16571 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16573 (VFT, VF_HW): New mode iterators.
16574 (vw, sdx): New mode attributes.
16575 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16576 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16577 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16578 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16579 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16580 also the new vector floating point modes. Renaming to ...
16582 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16583 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16584 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16585 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16586 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16587 ("vec_unordered<mode>"): ... these.
16589 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16590 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16591 ("*vec_extendv2df"): New insn definitions.
16593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16595 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16596 ("mulditi3_2", "*muldi3_sign"): New patterns.
16597 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16598 rename the pattern definition.
16600 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16602 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16604 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16606 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16608 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16609 instruction if possible.
16610 * config/s390/vector.md (vec_halfnumelts): New mode
16612 ("*vec_vllezlf<mode>"): New pattern.
16614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16616 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16617 ("popcountv4si2", "popcountv2di2"): Rename to ...
16618 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16619 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16621 ("popcount<mode>2_vxe"): New pattern.
16623 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16625 * common/config/s390/s390-common.c (processor_flags_table): Add
16627 * config.gcc: Add arch12.
16628 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16629 Default to arch12 for unknown CPU model numbers.
16630 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16631 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16632 PROCESSOR_max sanity check.
16633 * config/s390/s390-opts.h (enum processor_type): Add
16635 * config/s390/s390.c (processor_table): Add arch12.
16636 (s390_expand_builtin): Add check for B_VXE flag.
16637 (s390_issue_rate): Add PROCESSOR_ARCH12.
16638 (s390_get_sched_attrmask): Likewise.
16639 (s390_get_unit_mask): Likewise.
16640 (s390_sched_score): Enable z13 scheduling for arch12.
16641 (s390_sched_reorder): Likewise.
16642 (s390_sched_variable_issue): Likewise.
16643 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16645 (s390_tune_attr): Use z13 scheduling also for arch12.
16646 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16647 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16648 (TARGET_VXE_P): New macros.
16649 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16650 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16651 * config/s390/s390.opt: Add arch12 as processor_type.
16653 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16655 * config/s390/s390.md
16656 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16657 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16658 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16660 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16661 Rename expanders to ...
16663 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16664 ("fixuns_truncdddi2_emu"): ... these.
16666 ("fixuns_trunc<mode>si2_emu"): New expander.
16668 ("*fixuns_truncdfdi2_z13"): Rename to ...
16669 ("*fixuns_truncdfdi2_vx"): ... this.
16671 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16673 * config/s390/2964.md: Remove the single element vector compare
16674 instructions which are no longer used.
16675 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16677 (s390_canonicalize_comparison): Remove handling of DFmode
16679 (s390_expand_vec_compare_scalar): Remove function.
16680 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16681 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16683 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16687 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16689 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
16690 will anyway by matched by mov<mode>_64dfp.
16692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16694 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
16695 vlef/vstef. Add missing operand to vleif.
16697 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16699 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
16700 pair for all vector types with 64 bit elements.
16701 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
16702 * config/s390/vector.md (V_HW_64): ... here.
16703 (V_128_NOSINGLE): New mode iterator.
16704 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
16705 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
16706 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
16707 ("*vec_load_pairv2di"): Change to ...
16708 ("*vec_load_pair<mode>"): ... this one.
16710 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16712 * config/s390/constraints.md: Add comments.
16713 (jKK): Reject element sizes > 8 bytes.
16714 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
16716 * config/s390/s390.md: Add the s_operand checks formerly in
16717 s390_split_ok_p to various splitters where they are still
16719 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
16720 for 128 bit vectors. Plus two splitters.
16722 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16724 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
16727 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16730 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
16731 error if the boundary argument is not constant.
16733 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16735 PR rtl-optimization/80112
16736 * loop-doloop.c (doloop_condition_get): Don't check condition
16737 if cmp isn't SET with IF_THEN_ELSE src.
16739 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16741 PR tree-optimization/80158
16742 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
16743 replacing a candidate statement, also replace it for the
16744 candidate's alternate interpretation.
16745 (replace_rhs_if_not_dup): Likewise.
16746 (replace_one_candidate): Likewise.
16748 2017-03-24 Richard Biener <rguenther@suse.de>
16750 PR tree-optimization/80167
16751 * graphite-isl-ast-to-gimple.c
16752 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
16754 (translate_isl_ast_to_gimple::get_rename): Likewise.
16756 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16758 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
16759 handling of certain combinations of target options, including the
16760 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
16761 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
16763 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16766 * config/arm/arm.md (*load_multiple): Add reload_completed to
16767 matching condition.
16769 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16770 Richard Biener <rguenth@suse.de>
16772 PR tree-optimization/79908
16773 PR tree-optimization/80136
16774 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16775 been cast away, gimplify_and_add suffices.
16777 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
16779 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
16781 2017-03-23 Richard Biener <rguenther@suse.de>
16783 PR tree-optimization/80032
16784 * gimplify.c (gimple_push_cleanup): Forced unconditional
16785 cleanups still have to go to the conditional_cleanups
16788 2017-03-22 Jakub Jelinek <jakub@redhat.com>
16790 PR tree-optimization/80072
16791 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
16793 (next_operand_entry_id): Change type to unsigned int.
16794 (sort_by_operand_rank): Make sure to return the right return value
16795 even if unsigned fields are bigger than INT_MAX.
16796 (struct oecount): Change cnt and id type to unsigned int.
16797 (oecount_hasher::equal): Formatting fix.
16798 (oecount_cmp): Make sure to return the right return value
16799 even if unsigned fields are bigger than INT_MAX.
16800 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
16803 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
16804 TREE_READONLY on result if writing it more than once.
16807 * doc/invoke.texi (-fsanitize=thread): Document that with
16808 -fnon-call-exceptions atomics are not able to throw
16812 * tsan.c: Include tree-eh.h.
16813 (instrument_builtin_call): Call maybe_clean_eh_stmt or
16814 maybe_clean_or_replace_eh_stmt where needed.
16815 (instrument_memory_accesses): Add cfg_changed argument.
16816 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
16817 if it returned true.
16818 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
16820 PR rtl-optimization/63191
16821 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
16822 wrapper function, moved the whole old content into ...
16823 (ix86_delegitimize_address_1): ... this. New inline function.
16824 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
16825 true as last argument instead of ix86_delegitimize_address.
16827 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16829 * config/aarch64/aarch64.c (generic_branch_cost): Copy
16830 cortexa57_branch_cost.
16832 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16834 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
16836 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16839 * doc/md.texi (Constraints): Document wA constraint.
16840 * config/rs6000/constraints.md (wA): New.
16841 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
16842 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
16843 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
16844 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
16846 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
16849 * gimplify.c (is_oacc_declared): New function.
16850 (oacc_default_clause): Use it to set default flags for acc declared
16851 variables inside parallel regions.
16852 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
16853 declared variables.
16854 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
16855 declare attribute to any decl as necessary.
16857 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16860 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
16861 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
16862 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
16863 (arm_arch_lpae): This.
16864 * config/arm/arm.c (arm_arch7ve): Rename into ...
16865 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
16866 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
16869 2017-03-22 Martin Liska <mliska@suse.cz>
16872 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
16873 error message instead of an ICE.
16875 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16877 * doc/extend.texi (6.11 Additional Floating Types): Revise.
16879 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
16881 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
16883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16886 2017-03-21 Martin Sebor <msebor@redhat.com>
16888 * doc/extend.texi: Use "cannot" instead of "can't."
16889 * doc/hostconfig.texi: Same.
16890 * doc/install.texi: Same.
16891 * doc/invoke.texi: Same.
16892 * doc/loop.texi: Same.
16893 * doc/md.texi: Same.
16894 * doc/objc.texi: Same.
16895 * doc/rtl.texi: Same.
16896 * doc/tm.texi: Same.
16897 * doc/tm.texi.in: Same.
16898 * doc/trouble.texi: Same.
16900 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
16903 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
16904 (collect_checksum_attributes): Set it.
16905 (die_checksum_ordered): Use it.
16907 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16909 PR tree-optimization/79908
16910 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
16911 change: For a VA_ARG whose LHS has been cast away, use
16912 force_gimple_operand to construct the side effects.
16914 2017-03-21 David Malcolm <dmalcolm@redhat.com>
16916 PR translation/80001
16917 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
16918 more amenable to translation.
16919 (oacc_loop_auto_partitions): Likewise.
16921 2017-03-21 Marek Polacek <polacek@redhat.com>
16922 Martin Sebor <msebor@redhat.com>
16924 PR tree-optimization/80109
16925 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
16926 on INTEGRAL_TYPE_P.
16928 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16929 Segher Boessenkool <segher@kernel.crashing.org>
16932 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
16933 check reg_used_between_p between insn and one of succ or succ2
16934 depending on if succ is artificial insn not inserted into insn
16937 2017-03-21 Martin Liska <mliska@suse.cz>
16939 PR gcov-profile/80081
16940 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
16941 * doc/gcc.texi: Include gcov-dump stuff.
16942 * doc/gcov-dump.texi: New file.
16944 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
16946 PR rtl-optimization/79150
16947 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
16948 conditional jump, if the jump is the last insn of the loop.
16950 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16951 Richard Biener <rguenth@suse.de>
16953 PR tree-optimization/79908
16954 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16955 been cast away, use force_gimple_operand to construct the side
16958 2017-03-21 Martin Liska <mliska@suse.cz>
16960 PR libfortran/79956
16961 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
16964 2017-03-21 Brad Spengler <spender@grsecurity.net>
16967 * plugin.c (htab_hash_plugin): New function.
16968 (add_new_plugin): Use it and adjust.
16969 (parse_plugin_arg_opt): Adjust.
16970 (init_one_plugin): Likewise.
16972 2017-03-21 Richard Biener <rguenther@suse.de>
16974 PR tree-optimization/80032
16975 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
16976 if set force the cleanup to happen unconditionally.
16977 (gimplify_target_expr): Push inserted clobbers with force_uncond
16978 to avoid them being removed by control-dependent DCE.
16980 2017-03-21 Richard Biener <rguenther@suse.de>
16982 PR tree-optimization/80122
16983 * tree-inline.c (copy_bb): Do not expans va-arg packs or
16984 va_arg_pack_len when the inlined call stmt requires pack
16986 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
16988 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16991 * tsan.c (instrument_builtin_call): If the memory model argument
16992 is not a constant, assume it is valid.
16995 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
16998 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17000 PR rtl-optimization/79910
17001 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17002 if its dest is used by an insn before I2 (other than the combined
17003 insns themselves, which are properly handled already).
17005 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17008 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17010 * combine.c (record_used_regs): New static function.
17011 (try_combine): Handle situations where there is an additional
17012 instruction between I2 and I3 which needs to have a LOG_LINK
17016 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17018 * combine.c (try_combine): Delete redundant i1 test. Call
17019 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17021 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17024 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17025 alternatives 13/14.
17027 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17029 PR tree-optimization/80054
17030 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17031 the optimization if a PHI or any of its arguments is not dominated
17032 by the candidate's basis. Use gphi* rather than gimple* as
17034 (replace_profitable_candidates): Clean up a gimple* variable that
17035 should be a gphi* variable.
17037 2017-03-20 Martin Sebor <msebor@redhat.com>
17040 * doc/extend.texi (attribute constructor): Document present limitation.
17042 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17045 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17046 __POWER9_VECTOR__ #ifdef control, change template definition to
17047 use Power9-specific built-in function.
17048 (vec_any_eq): Likewise.
17049 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17050 to control outcomes from this test.
17051 (vector_ae_<mode>p): For VEC_F modes, likewise.
17053 2017-03-20 Ian Lance Taylor <iant@google.com>
17055 * config/i386/i386.c (ix86_function_regparm): Save an extra
17056 register for -fsplit-stack with DECL_STATIC_CHAIN.
17058 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17061 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17062 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17064 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17066 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17068 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17071 2017-03-20 Marek Polacek <polacek@redhat.com>
17074 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17076 2017-03-20 Richard Biener <rguenther@suse.de>
17078 PR tree-optimization/80113
17079 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17080 allocate extra SSA name for PHI def.
17081 (add_close_phis_to_outer_loops): Likewise.
17082 (add_close_phis_to_merge_points): Likewise.
17083 (copy_loop_close_phi_args): Likewise.
17084 (copy_cond_phi_nodes): Likewise.
17086 2017-03-20 Martin Liska <mliska@suse.cz>
17088 PR middle-end/79753
17089 * tree-chkp.c (chkp_build_returned_bound): Do not build
17090 returned bounds for a LHS that's not a BOUNDED_P type.
17092 2017-03-20 Martin Liska <mliska@suse.cz>
17096 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17097 COMPLEX_CST and VECTOR_CST.
17099 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17102 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17103 target operand. A new splitter adds the clobber statement in case
17104 the target operand is dead anyway.
17106 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17108 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17109 to age-old versions of binutils and glibc.
17111 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17113 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17115 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17117 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17119 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17121 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17122 requirement for binutils 2.13.
17124 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17126 * combine.c (try_combine): Delete redundant i1 test. Call
17127 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17129 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17131 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17132 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17134 <riscv64-*-elf>: Re-arrange section
17135 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17136 <riscv32-*-linux>: Likewise.
17137 <riscv64-*-elf>: Likewise
17138 <riscv64-*-linux>: Likewise.
17140 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17143 * aarch64.opt(verbose-cost-dump): Fix typo.
17145 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17148 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17149 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17151 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17153 * reload.c (find_reloads): When reloading a nonoffsettable address,
17154 use RELOAD_OTHER for it and its address reloads.
17156 PR rtl-optimization/79910
17157 * combine.c (record_used_regs): New static function.
17158 (try_combine): Handle situations where there is an additional
17159 instruction between I2 and I3 which needs to have a LOG_LINK
17162 2017-03-17 Jeff Law <law@redhat.com>
17164 PR tree-optimization/71437
17165 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17166 conditional in the hash table first.
17167 (vrp_dom_walker::before_dom_children): Extract condition from
17168 ASSERT_EXPR. Record condition, its inverion and any implied
17169 conditions as well.
17171 2017-03-17 Marek Polacek <polacek@redhat.com>
17172 Markus Trippelsdorf <markus@trippelsdorf.de>
17174 PR tree-optimization/80079
17175 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17178 2017-03-17 Richard Biener <rguenther@suse.de>
17180 PR middle-end/80075
17181 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17182 Properly verify the LHS before the RHS possibly claims to be
17184 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17187 2017-03-17 Martin Jambor <mjambor@suse.cz>
17189 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17190 (List of -O2 options): Likewise.
17191 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17194 2017-03-17 Tom de Vries <tom@codesourcery.com>
17196 * gcov-dump.c (print_usage): Print bug_report_url.
17198 2017-03-17 Richard Biener <rguenther@suse.de>
17200 PR middle-end/80050
17201 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17202 (parser::peek): Likewise.
17204 2017-03-17 Richard Biener <rguenther@suse.de>
17206 PR tree-optimization/80048
17207 * sese.c (free_sese_info): Properly release rename_map and
17208 copied_bb_map elements.
17210 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17212 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17213 Add linked-list forward and backlinks. Insert on
17214 construction, remove on destruction.
17215 (class pass_store_merging): Add m_stores_head field.
17216 (pass_store_merging::terminate_and_process_all_chains):
17217 Iterate over m_stores_head list.
17218 (pass_store_merging::terminate_all_aliasing_chains):
17220 (pass_store_merging::execute): Check for debug stmts first.
17221 Push new chains onto the m_stores_head stack.
17223 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17226 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17227 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17228 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17230 2017-03-16 Jeff Law <law@redhat.com>
17232 PR tree-optimization/71437
17233 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17234 member function. Implementation moved into after_dom_children
17235 member function and into the threader's thread_outgoing_edges
17237 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17238 some code into new thread_outgoing_edges.
17239 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17240 definition. Simplify marker handling (do it here). Assume we always
17241 have the available expression and the const/copies tables.
17242 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17244 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17245 * tree-vrp.c (equiv_stack): No longer file scoped.
17246 (vrp_dom_walker): New class.
17247 (vrp_dom_walker::before_dom_children): New member function.
17248 (vrp_dom_walker::after_dom_children): Likewise.
17249 (identify_jump_threads): Setup domwalker. Use it rather than
17250 walking edges in a random order by hand. Simplify setup/finalization.
17251 (finalize_jump_threads): Remove.
17252 (vrp_finalize): Do not call identify_jump_threads here.
17253 (execute_vrp): Do it here instead and call thread_through_all_blocks
17256 PR tree-optimization/71437
17257 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17259 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17261 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17262 (dom_opt_dom_walker::thread_across_edge): Remove
17263 handle_dominating_asserts argument. All callers changed.
17264 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17265 changes. Remove calls to lhs_of_dominating_assert. Other
17266 uses of handle_dominating_asserts turn into unconditional code
17267 (simplify_control_stmt_condition_1): Likewise.
17268 (simplify_control_stmt_condition): Likewise.
17269 (thread_through_normal_block, thread_across_edge): Likewise.
17270 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17271 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17272 object if it is not an SSA_NAME.
17273 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17274 before calling into the VRP specific simplifiers.
17275 (identify_jump_threads): Remove handle_dominating_asserts
17278 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17281 * tree-diagnostic.c (default_tree_printer): No longer static.
17282 * tree-diagnostic.h (default_tree_printer): New prototype.
17284 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17286 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17287 Change ins into fmov.
17289 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17291 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17292 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17293 Use h_con constraint for operand 1.
17294 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17295 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17297 2017-03-15 Jeff Law <law@redhat.com>
17299 PR tree-optimization/71437
17300 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17301 (record_temporary_equivalences): Use it.
17303 PR tree-optimization/71437
17304 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17305 tree-ssa-scopedtables.
17306 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17307 (record_conditions, record_cond, vuse_eq): Likewise.
17308 (record_edge_info): Adjust to API tweak of record_conditions.
17309 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17310 (record_temporary_equivalences, optimize_stmt): Likewise.
17311 (eliminate_redundant_computations): Likewise.
17312 (record_equivalences_from_stmt): Likewise.
17313 * tree-ssa-scopedtables.c: Include options.h and params.h.
17314 (vuse_eq): New function, moved from tree-ssa-dom.c
17315 (build_and_record_new_cond): Likewise.
17316 (record_conditions): Likewise. Accept vector of conditions rather
17317 than edge_equivalence structure for first argument.
17318 for the first argument.
17319 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17320 from tree-ssa-dom.c.
17321 (avail_exprs_stack::record_cond): Likewise.
17322 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17323 from tree-ssa-dom.c.
17324 (avail_exprs_stack): Add new member functions lookup_avail_expr
17326 (record_conditions): Declare.
17328 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17331 * lra-constraints.c (process_alt_operands): Increase reject for
17332 reloading an input/output operand.
17334 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17337 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17338 insns to convert from signed/unsigned char/short to IEEE 128-bit
17340 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17342 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17345 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17346 subreg of inner mode for values already in registers.
17348 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17350 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17351 iteration reg is used after the loop.
17353 2017-03-14 Martin Sebor <msebor@redhat.com>
17355 PR tree-optimization/79800
17356 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17357 precision in negative-positive range.
17358 (format_floating): Call non-const overload with adjusted precision.
17360 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17363 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17366 2017-03-14 Martin Sebor <msebor@redhat.com>
17368 PR middle-end/80020
17369 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17370 * builtins.def (aligned_alloc): Use it.
17373 * Makefile.in (GTFILES): Add calls.c.
17374 * calls.c: Include "gt-calls.h".
17376 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17378 PR rtl-optimization/79728
17379 * regs.h (struct target_regs): New field
17380 x_contains_allocatable_regs_of_mode.
17381 (contains_allocatable_regs_of_mode): New macro.
17382 * reginfo.c (init_reg_sets_1): Initialize it, and change
17383 contains_reg_of_mode so it includes global regs as well.
17384 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17385 rather than contains_regs_of_mode.
17387 2017-03-14 Martin Liska <mliska@suse.cz>
17389 * doc/invoke.texi: Document options that can't be combined with
17390 -fcheck-pointer-bounds.
17392 2017-03-14 Martin Liska <mliska@suse.cz>
17394 PR middle-end/79831
17395 * doc/invoke.texi (-Wchkp): Document the option.
17397 2017-03-14 Martin Liska <mliska@suse.cz>
17399 * Makefile.in: Install gcov-dump.
17401 2017-03-14 Martin Liska <mliska@suse.cz>
17403 * multiple_target.c (expand_target_clones): Bail out for
17404 an invalid attribute.
17406 2017-03-14 Richard Biener <rguenther@suse.de>
17408 * alias.c (struct alias_set_entry): Pack properly.
17409 * cfgloop.h (struct loop): Likewise.
17410 * cse.c (struct set): Likewise.
17411 * ipa-utils.c (struct searchc_env): Likewise.
17412 * loop-invariant.c (struct invariant): Likewise.
17413 * lra-remat.c (struct cand): Likewise.
17414 * recog.c (struct change_t): Likewise.
17415 * rtl.h (struct address_info): Likewise.
17416 * symbol-summary.h (function_summary): Likewise.
17417 * tree-loop-distribution.c (struct partition): Likewise.
17418 * tree-object-size.c (struct object_size_info): Likewise.
17419 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17420 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17421 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17422 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17423 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17424 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17425 (struct _stmt_vec_info): Likewise.
17427 2017-03-14 Martin Liska <mliska@suse.cz>
17430 * multiple_target.c (create_dispatcher_calls): Check that
17431 a target can create a function dispatcher.
17433 2017-03-14 Martin Liska <mliska@suse.cz>
17436 * multiple_target.c (expand_target_clones): Drop local.local
17437 flag for default implementation.
17439 2017-03-14 Richard Biener <rguenther@suse.de>
17441 PR tree-optimization/80030
17442 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17444 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17446 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17447 gcc_fallthrough() instead of __attribute__((fallthrough));
17449 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17451 * doc/gcc.texi: Remove "up" link to (DIR).
17452 * doc/gccint.texi: Ditto.
17454 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17456 * doc/install.texi (Specific) <avr>: Remove reference to
17459 2017-03-13 Jeff Law <law@redhat.com>
17461 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17462 attribute rather than comments.
17464 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17465 match_scratch operand is highest.
17467 2017-03-13 Martin Liska <mliska@suse.cz>
17469 PR middle-end/78339
17470 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17471 is a CHKP clone, use original declaration.
17473 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17475 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17476 (arc_conditional_register_usage): Use a different allocation order
17477 when optimizing for size.
17478 * common/config/arc/arc-common.c (arc_option_optimization_table):
17479 Section anchors default on when optimizing for size.
17481 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17483 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17485 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17487 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17488 * config/arc/arc.md (cpu_facility): Add cd variant.
17489 (*movqi_insn): Add code density variant.
17490 (*movhi_insn): Likewise.
17491 (*movqi_insn): Likewise.
17492 (*addsi3_mixed): Likewise.
17493 (subsi3_insn): Likewise.
17495 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17497 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17499 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17501 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17502 expressions with MINUS and UNARY ops.
17504 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17507 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17509 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17510 between vec_select and vector argument.
17511 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17512 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17513 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17514 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17515 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17516 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17518 2017-03-13 Richard Biener <rguenther@suse.de>
17521 * params.def (vect-max-peeling-for-alignment): Fix typo.
17523 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17525 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17526 issue that only occurred with binutils below 2.18.
17528 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17530 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17531 refer to binutils 2.11/2.12 minimum.
17533 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17535 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17536 ftp.kernel.org and simplify binutils requirement.
17538 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17540 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17542 (Optimize Options): Ditto. Also remove redundancy.
17544 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17546 PR translation/79848
17547 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17549 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17550 to G_ to avoid double translation.
17552 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17554 PR translation/79923
17555 * auto-profile.c (get_combined_location): Convert leading
17556 character of diagnostics to lower case and remove trailing period.
17557 (read_profile): Likewise for various diagnostics.
17558 * config/arm/arm.c (arm_option_override): Remove trailing period
17559 from various diagnostics.
17560 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17561 (msp430_expand_delay_cycles): Likewise.
17563 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17566 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17567 full command-line argument, rather than just "str".
17568 (aarch64_validate_march): Likewise.
17569 (aarch64_validate_mtune): Likewise.
17571 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17573 PR rtl-optimization/78911
17574 * lra-assigns.c (must_not_spill_p): New function.
17575 (spill_for): Use it.
17577 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17579 PR tree-optimization/79981
17580 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17581 ATOMIC_COMPARE_EXCHANGE ifn result.
17582 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17583 IFN_ATOMIC_COMPARE_EXCHANGE.
17585 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17588 * opts.c (parse_sanitizer_options): Add missing question mark to
17589 "did you mean" message.
17591 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17593 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17595 (VMULEUH_UNS): Likewise.
17596 (VMULOUB_UNS): Likewise.
17597 (VMULOUH_UNS): Likewise.
17598 * config/rs6000/rs6000.c (builtin_function_type): Remove
17599 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17601 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17604 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17605 x with result of extra_parsing_for_operand_code_0.
17606 (function_reader::extra_parsing_for_operand_code_0): Convert
17607 return type from void to rtx, returning x. When reading
17608 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17609 larger size containing struct block_symbol.
17611 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17613 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17614 -mfloat128-hardware without -m64.
17616 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17619 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17620 entries to the case statement that marks unsigned arguments to
17621 overloaded functions.
17623 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17625 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17626 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17628 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17631 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17632 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17634 2017-03-10 Martin Liska <mliska@suse.cz>
17638 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17639 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17642 2017-03-10 Olivier Hainque <hainque@adacore.com>
17644 * tree-switch-conversion (array_value_type): Start by resetting
17645 candidate type to it's main variant.
17647 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17649 PR rtl-optimization/79909
17650 * combine.c (try_combine): Use simplify_replace_rtx on individual
17651 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17652 of the whole CALL_INSN_FUNCTION_USAGE.
17654 PR tree-optimization/79972
17655 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17656 get_range_info on SSA_NAMEs. Formatting fixes.
17658 2017-03-10 Richard Biener <rguenther@suse.de>
17659 Jakub Jelinek <jakub@redhat.com>
17661 PR tree-optimization/77975
17662 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17663 edge to be constant.
17664 (get_val_for): For constant x return it. Formatting fix.
17665 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17666 would use the same bases as the current one.
17668 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17670 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17671 instead of vec_select for V1TImode.
17672 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17674 (VSX_LE_128): Add V1TI to this mode iterator.
17675 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17676 (*vsx_le_perm_store_<mode>): Likewise.
17677 (pre-reload splitter for VSX stores): Likewise.
17678 (post-reload splitter for VSX stores): Likewise.
17679 (*vsx_xxpermdi2_le_<mode>): Likewise.
17680 (*vsx_lxvd2x2_le_<mode>): Likewise.
17681 (*vsx_stxvd2x2_le_<mode>): Likewise.
17683 2017-03-09 Michael Eager <eager@eagercon.com>
17685 Correct failures with --enable-checking=yes,rtl.
17687 * config/microblaze/microblaze.c (microblaze_expand_shift):
17688 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17689 test for const0_rtx.
17690 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
17691 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
17693 2017-03-09 Richard Biener <rguenther@suse.de>
17695 PR tree-optimization/79977
17696 * graphite-scop-detection.c (scop_detection::merge_sese):
17697 Handle the case of extra exits to blocks dominating the entry.
17699 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
17701 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17704 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
17706 PR rtl-optimization/79949
17707 * lra-constraints.c (process_alt_operands): Check memory when
17708 trying to predict a cycle. Print about the overall increase.
17710 2017-03-09 Richard Biener <rguenther@suse.de>
17712 PR middle-end/79971
17713 * gimple-expr.c (useless_type_conversion_p): Preserve
17714 TYPE_SATURATING for fixed-point types.
17716 2017-03-09 Richard Biener <rguenther@suse.de>
17719 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
17720 alignment of BLKmode params.
17722 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17725 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
17726 (VALL_NO_V2Q): Likewise.
17728 * config/aarch64/aarch64-simd.md
17729 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
17731 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
17732 VALL_NO_V2Q mode iterator.
17733 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
17735 2017-03-09 Martin Liska <mliska@suse.cz>
17737 PR tree-optimization/79631
17738 * tree-chkp-opt.c (chkp_is_constant_addr): Call
17739 tree_int_cst_sign_bit just for INTEGER constants.
17741 2017-03-09 Martin Liska <mliska@suse.cz>
17745 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
17748 2017-03-09 Marek Polacek <polacek@redhat.com>
17751 * tree.c (inchash::add_expr): Handle TREE_VEC.
17753 2017-03-09 Martin Liska <mliska@suse.cz>
17756 (chkp_narrow_size_and_offset): New function.
17757 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
17758 (void chkp_parse_bit_field_ref): New function.
17759 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
17760 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
17762 2017-03-09 Martin Liska <mliska@suse.cz>
17765 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
17766 (chkp_find_bounds_1): Remove gcc_unreachable.
17768 2017-03-09 Jakub Jelinek <jakub@redhat.com>
17771 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
17772 BUILT_IN_SYNC*, determine the access type from the size suffix and
17773 always build a MEM_REF with that type. Handle forgotten
17774 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
17777 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
17778 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
17779 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
17780 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
17781 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
17782 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
17783 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
17784 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
17785 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
17786 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
17787 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
17788 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
17789 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
17790 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
17791 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
17792 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
17793 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
17794 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
17795 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
17796 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
17797 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
17798 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
17799 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
17800 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
17801 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
17802 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
17803 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
17804 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
17805 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
17806 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
17807 definitions outside of __OPTIMIZE__ guarded section.
17810 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
17811 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
17812 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
17813 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
17816 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17818 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
17819 ("vfenez<mode>"): Add missing constraints.
17821 2017-03-08 Martin Sebor <msebor@redhat.com>
17824 * config/nds32/nds32.c (nds32_option_override):
17825 Fix misspelled diagnostic.
17827 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17830 * gimplify.c (gimplify_omp_for): Replace index var in outer
17831 taskloop statement with an artificial variable and add
17832 OMP_CLAUSE_PRIVATE clause for it.
17834 2017-03-08 Richard Biener <rguenther@suse.de>
17836 PR tree-optimization/79955
17837 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
17838 for accesses that are completely outside of the variable.
17840 2017-03-08 Andrew Haley <aph@redhat.com>
17842 PR tree-optimization/79943
17843 * tree-ssa-loop-split.c (compute_new_first_bound): When
17844 calculating the new upper bound, (END-BEG) should be added, not
17847 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17849 * config/avr/avr.md (setmemhi): Make sure match_dup
17850 operand number comes before match_scratch.
17852 2017-03-08 Richard Biener <rguenther@suse.de>
17854 PR tree-optimization/79920
17855 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
17856 with ncopies == 1 to ...
17857 (vect_transform_slp_perm_load): ... here. Properly compute
17858 all element loads by iterating VF times over the group. Do
17859 not handle ncopies (computed in a broken way) in
17860 vect_create_mask_and_perm.
17862 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17865 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
17866 is a uniform vector, use uniform_vector_p return value instead of
17867 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
17869 2017-03-07 Marek Polacek <polacek@redhat.com>
17871 PR middle-end/79809
17872 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
17873 (alloca_call_type): Likewise.
17875 2017-03-07 Martin Liska <mliska@suse.cz>
17877 * gcov.c (process_args): Put comment to correct location.
17879 2017-03-07 Martin Liska <mliska@suse.cz>
17881 PR middle-end/68270
17882 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
17883 Use array_at_struct_end_p instead of DECL_CHAIN (field).
17884 (chkp_narrow_bounds_for_field): Likewise.
17885 (chkp_parse_array_and_component_ref): Pass one more argument to
17888 2017-03-07 Richard Biener <rguenther@suse.de>
17890 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
17893 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17895 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
17896 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
17898 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17901 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
17902 to end of description.
17903 (PARAM_MAX_STORES_TO_MERGE): Likewise.
17905 2017-03-07 Jakub Jelinek <jakub@redhat.com>
17907 PR rtl-optimization/79901
17908 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
17910 (*avx512f_<code><mode>3<mask_name>): ... this.
17911 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
17912 iterator instead of VI8_AVX2_AVX512BW.
17914 PR rtl-optimization/79901
17915 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
17916 min/max expander, expand it using expand_vec_cond_expr.
17919 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
17922 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17925 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
17926 to void * for PCH reasons.
17927 * dwarf2out.c (output_loc_operands, output_die): Cast
17928 v.val_vec.array to unsigned char *.
17930 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
17933 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
17936 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
17938 PR rtl-optimization/79571
17939 * lra-constraints.c (process_alt_operands): Calculate static
17940 reject and subtract it from overall when only addresses will be
17943 2017-03-06 Julia Koval <julia.koval@intel.com>
17946 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
17947 incoming stack boundary to 128 for 64-bit targets.
17949 2017-03-06 Richard Biener <rguenther@suse.de>
17951 PR tree-optimization/79894
17952 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
17953 to NULL after folding it.
17955 2017-03-06 Richard Biener <rguenther@suse.de>
17957 PR tree-optimization/79824
17958 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
17959 check disabling peeling for gaps.
17961 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
17963 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17964 attributes): Document gettimeofday.
17966 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17968 * config/s390/s390.c (s390_option_override_internal): Set
17969 PARAM_MIN_VECT_LOOP_BOUND
17971 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17973 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
17974 * config/s390/s390.md: Likewise.
17976 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17979 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
17980 (<avx2_avx512>_perm<mode>): Rename to ...
17981 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
17983 (<avx512>_perm<mode>_mask): Rename to ...
17984 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
17986 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
17987 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
17988 instead of VI8F_256_512.
17989 (avx512f_perm<mode>): New define_expand.
17990 (avx512f_perm<mode>_mask): Likewise.
17991 (avx512f_perm<mode>_1<mask_name>): New define_insn.
17992 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
17994 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17996 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
17997 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
17999 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18001 2017-03-06 Martin Liska <mliska@suse.cz>
18004 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18005 when having a SSA NAME w/o VAR_DECL assigned to it.
18007 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18009 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18010 msa_dpsub_<su>_d): Fix MODE for vec_select.
18012 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18014 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18016 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18018 2017-03-06 Richard Biener <rguenther@suse.de>
18020 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18021 * plugin.c (register_plugin_info): Likewise.
18022 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18024 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18026 * config/i386/sse.md (sse_storehps, sse_storelps,
18027 avx_<castmode><avxsizesuffix>_<castmode>,
18028 avx512f_<castmode><avxsizesuffix>_<castmode>,
18029 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18030 in condition that at least one operand is not a MEM.
18032 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18034 PR middle-end/79805
18035 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18036 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18038 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18039 gimple_call_nothrow_p flag based on whether original builtin can throw.
18040 If it can, emit following stmts on the fallthrough edge.
18041 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18042 don't create new bb if inserting just debug stmts on the edge, try to
18043 insert them on the fallthru bb or just reset debug stmts.
18045 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18048 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18049 restore recog_data (including the operand rtxes inside it) around
18050 the call to get_insn_template.
18052 2017-03-03 Martin Sebor <msebor@redhat.com>
18054 PR tree-optimization/79699
18055 * context.c (context::~context): Free MPFR caches to avoid
18056 a memory leak on program exit.
18058 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18060 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18061 Use wide_int::ulow () instead of .elt (0).
18063 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18065 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18066 (*pushxf): Limit oF constraint to 32bit targets and add oC
18067 constraint for 64bit targets.
18068 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18069 (*pushdf): Change rmF constraint to rmC.
18071 2017-03-03 Martin Liska <mliska@suse.cz>
18073 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18074 Remove unused variable.
18076 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18079 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18080 is a memory operand, increase num_memory.
18081 (ix86_expand_args_builtin): Likewise.
18083 2017-03-03 Jan Hubicka <jh@suse.cz>
18086 * ipa-devirt.c (maybe_record_node): Properly handle
18087 __cxa_pure_virtual visibility.
18089 2017-03-03 Martin Liska <mliska@suse.cz>
18091 PR tree-optimization/79803
18092 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18094 (pass_loop_prefetch::execute): Disabled optimization if an
18095 assumption about L1 cache size is not met.
18097 2017-03-03 Martin Liska <mliska@suse.cz>
18099 PR rtl-optimization/79574
18100 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18101 (hash_scan_set): Likewise.
18102 (dump_hash_table): Likewise.
18103 (hoist_code): Likewise.
18105 2017-03-03 Richard Biener <rguenther@suse.de>
18107 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18108 in place of uhigh ().
18109 (fixed_convert_from_real): Likewise.
18111 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18114 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18116 2017-03-03 Richard Biener <rguenther@suse.de>
18118 PR middle-end/79818
18119 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18120 TYPE_OVERFLOW_UNDEFINED check.
18122 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18124 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18126 (vector_ae_<mode>_p): Likewise.
18127 (vector_nez_<mode>_p): Likewise.
18128 (vector_ne_v2di_p): Likewise.
18129 (vector_ae_v2di_p): Likewise.
18130 (vector_ne_<mode>_p): Likewise.
18131 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18133 (vsx_tsqrt<mode>2_fe): Likewise.
18135 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18138 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18140 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18142 PR rtl-optimization/79780
18143 * cprop.c (one_cprop_pass): When second and further conditional trap
18144 in a single basic block is turned into an unconditional trap, turn it
18145 into a deleted note to avoid RTL verification failures.
18147 2017-03-02 Richard Biener <rguenther@suse.de>
18149 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18151 2017-03-02 Richard Biener <rguenther@suse.de>
18153 PR tree-optimization/79345
18155 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18156 param and abort the walk, returning -1 if it is hit.
18157 (walk_aliased_vdefs): Take a limit param and pass it on.
18158 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18159 defaulting to 0 and return a signed int.
18160 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18161 (check_defs): New helper.
18162 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18163 about uninitialized memory.
18164 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18165 bogus uninitialized warning.
18166 (fixed_convert_from_real): Likewise.
18168 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18170 PR tree-optimization/66768
18171 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18172 iv_use if base object can't be determined.
18174 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18176 PR tree-optimization/79345
18177 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18178 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18179 (get_pattern_stats): Initialize it.
18180 * genemit.c (gen_expand): Verify match_scratch numbers come after
18181 match_operand/match_dup numbers.
18182 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18183 match_scratch numbers.
18184 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18186 * config/s390/s390.md (trunctdsd2): Likewise.
18188 2017-03-02 Richard Biener <rguenther@suse.de>
18190 * wide-int.h (wide_int_storage::operator=): Implement in terms
18193 2017-03-02 Richard Biener <rguenther@suse.de>
18195 PR tree-optimization/79777
18196 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18197 the to insert expression to sth existing.
18199 2017-03-01 Martin Sebor <msebor@redhat.com>
18201 PR middle-end/79692
18202 * gimple-ssa-sprintf.c
18203 (directive::known_width_and_precision): New function.
18204 (format_integer): Use it.
18205 (get_mpfr_format_length): Consider the full range of precision
18206 when computing %g output with the # flag. Set the likely byte
18207 count to 3 rather than 1 when precision is indeterminate.
18208 (format_floating): Correct the lower bound of precision.
18210 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18212 * doc/invoke.texi: Document default code model for 64-bit Linux.
18214 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18217 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18218 udiv rather than div since input pattern is unsigned.
18220 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18222 * config/i386/i386.c (print_reg): Warn for values of
18223 unsupported size in integer register.
18225 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18228 * config/rs6000/predicates.md (current_file_function_operand): Do
18229 not allow self calls to be local if the function is replaceable.
18231 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18234 * config/rs6000/altivec.h (vec_ctz and others): Change the
18235 preprocessor macro that controls conditional compilation from
18236 _ARCH_PWR9 to __POWER9_VECTOR__.
18237 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18238 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18239 control (instead of _ARCH_PWR9 control) so that template
18240 definition uses power9-specific function.
18241 (vec_any_eq): Likewise.
18242 (vec_all_ne): Change macro definition to use a power9-specific
18243 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18244 _ARCH_PWR9 control).
18245 (vec_any_eq) Likewise.
18246 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18247 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18248 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18249 support for xvcmpnedp instruction.
18250 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18251 macro expansion so that Power9 implementation of vec_all_ne does
18252 not use the AltiVec predicate framework.
18253 (VCMPNEH_P): Likewise.
18254 (VCMPNEW_P): Likewise.
18255 (VCMPNED_P): Likewise.
18256 (VCMPNEFP_P): Likewise.
18257 (VCMPNEDP_P): Likewise.
18258 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18259 implementation of vec_any_eq to not use AltiVec predicate
18261 (VCMPAEH_P): Likewise.
18262 (VCMPAEW_P): Likewise.
18263 (VCMPAED_P): Likewise.
18264 (VCMPAEFP_P): Likewise.
18265 (VCMPAEDP_P): Likewise.
18266 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18267 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18268 not use the AltiVec predicate framework.
18269 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18270 of vec_any_eq to not use AltiVec predicate framework.
18271 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18272 support for predefined __POWER9_VECTOR__ macro to indicate that
18273 Power9 instruction selection is enabled.
18274 (altivec_overloaded_builtins): Remove extraneous
18275 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18276 function argument types RS6000_BTI_bool_V16QI and
18277 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18278 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18279 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18280 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18281 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18282 Power9 for implementations of vec_cmpne. Change the signature for
18283 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18284 (representing vec_all_ne) to remove the previously described first
18285 argument of type RS6000_BTI_INTSI, as this was an artifact of
18286 reliance on the AltiVec predicate framework, which is no longer
18287 used in the implementation of these functions. Add
18288 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18289 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18290 since, unlike the AltiVec predicate framework implementation, we
18291 do not share function descriptors between vec_alle and vec_anyeq.
18292 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18293 set of modes that receive special treatment even when
18294 TARGET_P9_VECTOR is true. The special treatment emits code that
18295 does not depend on Power9 instructions.
18296 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18297 define_expand to not rely on AltiVec predicate framework.
18298 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18300 (vector_ne_v2di_p): Change this define_expand to not rely on
18301 AltiVec predicate framework.
18302 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18304 (vector_ne_<mode>_p): Change this define_expand to not rely on
18305 AltiVec predicate framework.
18306 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18308 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18309 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18310 define_insn pattern.
18311 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18312 define_insn pattern because the xvcmpne<VSs>. instruction is not
18314 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18315 instruction is not supported.
18317 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18319 * config/nvptx/nvptx.c: Include intl.h.
18321 2017-03-01 Martin Jambor <mjambor@suse.cz>
18324 * ipa-prop.h (ipa_bits): Removed field known.
18325 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18326 to pointers. Adjusted their comments to warn about their sharing.
18327 (ipcp_transformation_summary): Change bits to a vector of pointers.
18328 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18329 (ipa_get_ipa_bits_for_value): Declare.
18330 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18331 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18332 (ipa_bits_hash_table): Likewise.
18333 (ipa_vr_ggc_hash_traits): Likewise.
18334 (ipa_vr_hash_table): Likewise.
18335 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18336 being pointers and vr_known being removed.
18337 (ipa_set_jf_unknown): Likewise.
18338 (ipa_get_ipa_bits_for_value): New function.
18339 (ipa_set_jfunc_bits): Likewise.
18340 (ipa_get_value_range): New overloaded functions.
18341 (ipa_set_jfunc_vr): Likewise.
18342 (ipa_compute_jump_functions_for_edge): Use the above functions to
18343 construct bits and vr parts of jump functions.
18344 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18345 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18347 (ipcp_grow_transformations_if_necessary): Also allocate
18348 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18350 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18351 them. Fix too long lines.
18352 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18353 vr_known being removed.
18354 (ipa_read_jump_function): Use new setter functions to construct bits
18355 and vr parts of jump functions or set them to NULL.
18356 (write_ipcp_transformation_info): Adjust for bits being pointers.
18357 (read_ipcp_transformation_info): Likewise.
18358 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18360 Include gt-ipa-prop.h.
18361 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18363 (ipcp_store_bits_results): Likewise.
18364 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18365 Do not write to existing jump functions but use a temporary instead.
18367 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18370 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18371 attempt to use its first operand as BIT_FIELD_REF base.
18373 2017-03-01 Richard Biener <rguenther@suse.de>
18375 PR middle-end/79721
18376 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18377 interpolating formula in wrapping arithmetic.
18378 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18380 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18382 PR tree-optimization/79734
18383 * tree-vect-generic.c (expand_vector_condition): Optimize
18384 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18385 Handle VEC_COND_EXPR where comparison has different inner width from
18386 type's inner width.
18388 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18390 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18391 markup, and similar issues. Remove @opindex entries for things
18392 that aren't options. Add missing -mmpy-option entries.
18394 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18396 PR tree-optimization/79737
18397 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18398 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18399 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18400 instead of byte_size. Formatting fix.
18401 (shift_bytes_in_array_right): Formatting fix.
18403 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18406 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18407 condition on optimize for the leaf function test.
18409 2017-02-28 Martin Liska <mliska@suse.cz>
18412 * read-rtl-function.c (function_reader::handle_unknown_directive):
18413 Bail out when one uses -flto.
18415 2017-02-28 Martin Liska <mliska@suse.cz>
18417 * common.opt: Replace space with tabular for options of <number>
18419 * config/i386/i386.opt: Show <number> value for
18420 -mlarge-data-threshold.
18421 * opts.c (print_filtered_help): Do not display number in hexadecimal
18424 2017-02-28 Martin Liska <mliska@suse.cz>
18426 * common.opt: Fix --help=option -Q for options which are of
18429 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18431 * config/i386/i386.c (print_reg): Error out for values
18432 of 8-bit size in invalid integer register.
18434 2017-02-28 Martin Sebor <msebor@redhat.com>
18436 PR tree-optimization/79691
18437 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18439 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18442 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18443 gcc_unreachable with output_operand_lossage.
18445 2017-02-28 Richard Biener <rguenther@suse.de>
18447 PR tree-optimization/79740
18448 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18450 (visit_nary_op): Insert the nary into the hashtable if we
18451 pattern-matched sth.
18452 * tree-ssa-pre.c (eliminate_insert): Robustify.
18454 2017-02-28 Richard Biener <rguenther@suse.de>
18456 PR middle-end/79731
18457 * fold-const.c (decode_field_reference): Reject out-of-bound
18460 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18462 * config/i386/i386.c: Include intl.h.
18463 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18464 instead of just cond ? "..." : "...".
18465 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18466 * coverage.c (read_counts_file): Likewise.
18467 * omp-offload.c: Include intl.h.
18468 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18469 of just cond ? "..." : "...".
18470 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18471 of just cond ? "..." : "...".
18473 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18476 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18478 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18479 'tune for' CPU name.
18480 * config/arm/arm-cpu-data.h: Regenerated.
18482 2017-02-28 Richard Biener <rguenther@suse.de>
18484 PR tree-optimization/79732
18485 * tree-inline.c (expand_call_inline): Do not shadow var.
18487 2017-02-28 Richard Biener <rguenther@suse.de>
18489 PR tree-optimization/79723
18490 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18491 address-space properly.
18493 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18495 * doc/optinfo.texi (Optimization groups): Fix option used for
18497 * doc/invoke.texi (-fopt-info): Document "omp".
18498 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18499 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18500 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18501 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18502 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18504 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18506 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18507 OPTGROUP_OMP with "omp".
18509 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18512 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18513 for arithmetic shift of unsigned V2DI.
18515 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18517 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18518 arc/linux.h headers.
18519 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18520 (LINK_SPEC): Likewise.
18521 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18522 (EXTRA_SPECS): Likewise.
18523 (STARTFILE_SPEC): Likewise.
18524 (ENDFILE_SPEC): Likewise.
18525 (LIB_SPEC): Likewise.
18526 (TARGET_SDATA_DEFAULT): Likewise.
18527 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18528 (MULTILIB_DEFAULTS): Likewise.
18529 (DWARF2_UNWIND_INFO): Likewise.
18530 * config/arc/big.h: New file.
18531 * config/arc/elf.h: Likewise.
18532 * config/arc/linux.h: Likewise.
18533 * config/arc/t-uClibc: Remove.
18535 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18537 PR tree-optimization/77536
18538 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18539 (tree_transform_and_unroll_loop): Use above function to compute the
18540 estimated niter of unrolled loop and use it when scaling profile.
18541 Also use count info rather than frequency if it's non-zero.
18542 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18543 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18544 (vect_transform_loop): Call above function.
18546 2017-02-27 Richard Biener <rguenther@suse.de>
18548 PR tree-optimization/45397
18549 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18550 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18551 (visit_nary_op): Add pattern matching for CSEing sign-changed
18552 or truncated operations with wider ones.
18554 2017-02-27 Richard Biener <rguenther@suse.de>
18556 PR tree-optimization/79690
18557 * tree-vect-stmts.c (vectorizable_store): Use vector type
18558 built from the DR with address-space.
18560 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18562 * doc/invoke.texi (Optimize Options): Refine the description
18563 of asan-use-after-return.
18565 2017-02-25 Alan Modra <amodra@gmail.com>
18567 PR rtl-optimization/79584
18568 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18569 base, not ad->base_term, the reg within base. Remove assertion
18570 that ad->base == ad->base_term. Replace gen_int_mode using
18571 bogus mode with const0_rtx.
18573 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18575 PR middle-end/79396
18576 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18577 FMA_EXPR like tcc_binary or tcc_unary.
18579 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18582 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18584 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18585 (output_loc_operands): Handle DW_OP_call_ref and
18586 DW_OP_GNU_variable_value.
18587 (struct variable_value_struct): New type.
18588 (struct variable_value_hasher): Likewise.
18589 (variable_value_hash): New variable.
18590 (string_types): Remove.
18591 (copy_loc_descr): New function.
18592 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18593 (prepend_loc_descr_to_each): New function.
18594 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18595 instead of add_loc_descr_to_each if the first argument is single
18596 location list and the second has multiple.
18597 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18598 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18599 when looking for variable value which doesn't have other location info.
18600 (loc_list_from_tree): Formatting fix.
18601 (gen_array_type_die): Simplify DW_AT_string_length handling.
18602 (adjust_string_types): Remove.
18603 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18604 string_types. Call resolve_variable_values.
18605 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18606 (resolve_addr_in_expr): Likewise. Add A argument.
18607 (copy_deref_exprloc): Remove deref argument. Adjust for the
18608 original expression being DW_OP_GNU_variable_value with optionally
18609 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18610 optionally after it.
18611 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18612 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18613 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18614 (variable_value_hasher::hash, variable_value_hasher::equal): New
18616 (resolve_variable_value_in_expr, resolve_variable_value,
18617 resolve_variable_values, note_variable_value_in_expr,
18618 note_variable_value): New functions.
18619 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18622 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18625 * opts.h (handle_generated_option): Add GENERATED_P argument.
18626 * opts-common.c (handle_option): Adjust function comment.
18627 (handle_generated_option): Add GENERATED_P argument, pass it to
18629 (control_warning_option): Pass false to handle_generated_option
18631 * opts.c (maybe_default_option): Pass true to handle_generated_option
18633 * optc-gen.awk: Likewise.
18635 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18637 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18638 a REG, look at the REG it is a SUBREG of.
18639 (splitter for cmpeqsi_t): Ditto.
18641 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18643 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18644 the special USEs with the pattern of the insn, not the insn itself.
18646 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18649 * doc/invoke.texi: Document -mload-store-pairs.
18651 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18652 Sandra Loosemore <sandra@codesourcery.com>
18654 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18655 argument isn't a CONST_INT.
18656 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18657 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18658 (nios2_validate_compare): Bypass alternate compare logic if *op2
18659 is not a CONST_INT.
18660 (ldstwm_operation_p): Return false if first_base is not a REG or
18661 if first_offset is not a CONST_INT.
18663 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18665 * config/cris/cris.md: Use correct operand in a define_peephole2.
18667 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18669 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18671 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18673 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18674 this_insn if it is an INSN or JUMP_INSN.
18675 (force_offsettable): Look at base, not at addr.
18676 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18677 on things that aren't necessarily CONST_INTs.
18679 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18681 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18682 -mfpmath=sse is the default also for x86-32 targets with SSE2
18683 instruction set when @option{-ffast-math} is enabled
18685 2017-02-24 Jeff Law <law@redhat.com>
18687 PR rtl-optimizatoin/79286
18688 * ira.c (update_equiv_regs): Drop may_trap_p exception to
18691 2017-02-24 Richard Biener <rguenther@suse.de>
18693 PR tree-optimization/79389
18694 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
18697 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
18699 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
18700 function comment to reflect reality.
18701 (loop_exits_before_overflow): Fix typo in function description.
18703 2017-02-24 Richard Biener <rguenther@suse.de>
18705 PR tree-optimization/79389
18706 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
18707 properly that a threading opportunity exists. Detect conditional
18708 copy/constant propagation opportunities.
18710 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
18712 * config/visium/visium.md (type): Add trap.
18713 (b): New mode attribute.
18714 (*btst): Rename into...
18715 (*btst<mode>): ...this and adjust.
18716 (*cbranchsi4_btst_insn): Rename into...
18717 (*cbranch<mode>4_btst_insn): ...this and adjust.
18718 (trap): New define_insn.
18720 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18722 PR tree-optimization/79389
18723 * ifcvt.c (struct noce_if_info): Add rev_cond field.
18724 (noce_reversed_cond_code): New function.
18725 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
18726 reversed_comparison_code. Formatting fix.
18727 (noce_try_store_flag): Test rev_cond != NULL in addition to
18728 reversed_comparison_code.
18729 (noce_try_store_flag_constants): Likewise.
18730 (noce_try_store_flag_mask): Likewise.
18731 (noce_try_addcc): Use rev_cond if non-NULL instead of
18732 reversed_comparison_code.
18733 (noce_try_cmove_arith): Likewise. Formatting fixes.
18734 (noce_try_minmax, noce_try_abs): Clear rev_cond.
18735 (noce_find_if_block): Initialize rev_cond.
18736 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
18737 instead of false as last argument never attempt to reverse it
18740 2017-02-23 Bin Cheng <bin.cheng@arm.com>
18742 PR tree-optimization/79663
18743 * tree-predcom.c (combine_chains): Process refs in reverse order
18744 only for ZERO length chains, and add explaining comment.
18746 2017-02-23 Jeff Law <law@redhat.com>
18748 PR tree-optimization/79578
18749 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
18750 in call to operand_equal_p.
18752 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
18755 * config/i386/cpuid.h: Fix another undefined behavior.
18757 2017-02-23 Richard Biener <rguenther@suse.de>
18759 PR tree-optimization/79683
18760 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
18761 vector types for data-refs.
18763 2017-02-23 Martin Liska <mliska@suse.cz>
18765 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
18767 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18769 PR middle-end/79665
18770 * internal-fn.c (get_range_pos_neg): Moved to ...
18771 * tree.c (get_range_pos_neg): ... here. No longer static.
18772 * tree.h (get_range_pos_neg): New prototype.
18773 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
18774 are known to be in between 0 and signed maximum inclusive, try to
18775 expand both unsigned and signed divmod and use the cheaper one from
18778 2017-02-22 Jeff Law <law@redhat.com>
18780 PR tree-optimization/79578
18781 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
18782 to compare base operands.
18784 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
18787 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
18788 gpc_reg_operand instead of fpr_reg_operand.
18790 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
18792 * config/mips/mips.c (mips_return_in_memory): Force FP
18793 vector types to be returned in memory for o32 ABI.
18795 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18797 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
18798 instead of DW_TAG_member for static data member declarations and don't
18799 set no_linkage_name for static inline data members.
18800 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
18803 2017-02-22 Martin Liska <mliska@suse.cz>
18805 * doc/invoke.texi: Replace inequality signs with square brackets
18808 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18810 PR tree-optimization/68644
18811 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
18813 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
18816 * lra-constraints.c (simplify_operand_subreg): Handle
18817 WORD_REGISTER_OPERATIONS targets.
18819 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18822 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
18823 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
18824 elimination by swapping fld*.
18826 2017-02-22 Richard Biener <rguenther@suse.de>
18828 PR tree-optimization/79673
18829 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
18830 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
18831 irrelevant address-space qualifiers and avoiding a
18832 ADDR_SPACE_CONVERT_EXPR from fold_convert.
18834 2017-02-22 Richard Biener <rguenther@suse.de>
18836 PR tree-optimization/79666
18837 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
18838 to not symbolically negate if that may introduce undefined
18841 2017-02-22 Martin Liska <mliska@suse.cz>
18844 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
18845 * data-streamer-out.c (streamer_write_gcov_count_stream):
18847 * value-prof.c (stream_out_histogram_value): Make assert more
18848 precise based on type of counter.
18850 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
18853 * config/i386/i386.md (standard_x87sse_constant_load splitter):
18854 Use nonimmediate_operand instead of memory_operand for operand 1.
18855 (float-extend standard_x87sse_constant_load splitter): Ditto.
18857 2017-02-21 Jeff Law <law@redhat.com>
18859 PR tree-optimization/79621
18860 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
18861 blocks with edges to themselves.
18863 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18866 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
18867 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
18868 Use gimple_call_builtin_p.
18871 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
18872 on temporarily removed DEBUG_INSNs.
18874 PR tree-optimization/79649
18875 * tree-loop-distribution.c (classify_partition): Give up on
18876 non-generic address space loads/stores.
18878 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
18880 * doc/loop.texi (Loop manipulation): Remove nonexistent
18881 tree_ssa_loop_version from the documentation.
18882 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
18884 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18887 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
18888 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
18889 * config/rs6000/rs6000.c: Include except.h.
18890 (rs6000_expand_split_stack_prologue): Call
18891 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
18893 2017-02-21 Martin Jambor <mjambor@suse.cz>
18896 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
18897 have been analyzed.
18899 2017-02-21 Martin Jambor <mjambor@suse.cz>
18901 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
18902 for backward compatibility only.
18903 * doc/invoke.texi (Option Summary): Remove all references to
18904 -fipa-cp-alignment.
18906 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
18910 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 2017-02-21 Martin Liska <mliska@suse.cz>
18917 * config/i386/i386.opt: Replace -masm-dialect with -masm.
18919 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
18921 PR translation/79638
18922 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
18924 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
18927 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
18928 (arm_function_ok_for_sibcall): Return false for an indirect call by
18929 descriptor if all the argument registers are used.
18930 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
18931 alignment of the function.
18933 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18935 PR tree-optimization/61441
18936 * simplify-rtx.c (simplify_const_unary_operation): For
18937 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
18938 the sNaN unmodified.
18940 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18942 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18943 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18944 instead of SYSTEM_HEADER_DIR.
18946 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
18947 Martin Liška <mliska@suse.cz>
18949 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
18950 Fix typos and grammar, use active voice, and clarify.
18952 2017-02-20 Marek Polacek <polacek@redhat.com>
18954 PR middle-end/79537
18955 * gimplify.c (gimplify_expr): Handle unused *&&L;.
18958 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
18960 2017-02-20 Jakub Jelinek <jakub@redhat.com>
18963 * config/i386/i386.c (ix86_expand_builtin): Handle
18964 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
18965 ix86_builtins_isa[fcode].isa as a requirement of those
18966 flags and any other flag in the bitmask.
18967 (ix86_init_mmx_sse_builtins): Use 0 instead of
18968 ~OPTION_MASK_ISA_64BIT as mask.
18969 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
18970 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
18971 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
18972 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
18974 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18977 * lra-constraints.c (split_reg): Check requested split mode
18978 is supported by the register.
18980 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18982 * lra-constraints.c (simplify_operand_subreg): Remove early
18985 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18988 * lra-constraints.c (curr_insn_transform): Tighten condition
18989 for converting SUBREG reloads from OP_OUT to OP_INOUT.
18991 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18994 * lra-constraints.c (curr_insn_transform): Handle
18995 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18997 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19000 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19002 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19004 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19007 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19010 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19012 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19015 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19017 * final.c (last_columnnum, override_columnnum): New variables.
19018 (final_start_function): Set last_columnnum, pass it to begin_prologue
19019 hook and pass 0 to dwarf2out_begin_prologue.
19020 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19021 to source_line debug hook.
19022 (notice_source_line): Compute last_columnnum and for debug_column_info
19023 return true on column changes.
19024 * debug.h (struct gcc_debug_hooks): Add column argument to
19025 source_line and begin_prologue hooks.
19026 (debug_nothing_int_charstar_int_bool): Remove prototype.
19027 (debug_nothing_int_int_charstar,
19028 debug_nothing_int_int_charstar_int_bool): New prototypes.
19029 (dwarf2out_begin_prologue): Add column argument.
19030 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19031 begin_prologue hooks.
19032 (debug_nothing_int_charstar_int_bool): Remove.
19033 (debug_nothing_int_int_charstar,
19034 debug_nothing_int_int_charstar_int_bool): New functions.
19035 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19036 through to dwarf2out_source_line.
19037 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19038 (dwarf2out_source_line): Add column argument, emit it if requested.
19039 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19041 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19042 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19043 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19044 through to dwarf2out_begin_prologue.
19045 (vmsdbgout_source_line): Add column argument, pass it through to
19046 dwarf2out_source_line.
19047 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19048 dbxout_source_line caller.
19049 (dbxout_source_line): Add column argument.
19051 * common.opt (gno-column-info, gcolumn-info): New options.
19052 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19053 (check_die): Also test for multiple DW_AT_decl_column attributes.
19054 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19055 DW_AT_decl_column if requested.
19056 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19058 (gen_variable_die): Likewise.
19059 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19060 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19063 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19064 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19065 (ix86_handle_option): Handle OPT_m3dnowa.
19066 * doc/invoke.texi (-m3dnowa): Document.
19067 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19068 -m3dnowa instead of -m3dnow -march=athlon.
19071 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19072 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19074 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19077 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19078 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19079 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19080 generator for vsx_xxpermdi_<mode>_be.
19081 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19082 force big-endian semantics.
19083 (vsx_xxpermdi_<mode>_be): New define_expand with same
19084 implementation as previous version of vsx_xxpermdi_<mode>.
19086 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19088 PR tree-optimization/79327
19089 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19090 variable, its initialization and use.
19092 2017-02-17 Julia Koval <julia.koval@intel.com>
19094 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19095 (OPTION_MASK_ISA_PKU_UNSET): New.
19096 (ix86_handle_option): Handle -mrdpid.
19097 * config/i386/cpuid.h (bit_RDPID): New.
19098 * config/i386/driver-i386.c (host_detect_local_cpu):
19099 Detect RDPID feature.
19100 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19101 * config/i386/i386-c.c (ix86_target_macros_internal):
19103 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19104 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19105 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19106 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19107 * config/i386/i386.md (define_insn "rdpid"): New.
19108 * config/i386/i386.opt Add -mrdpid.
19109 * config/i386/immintrin.h (_rdpid_u32): New.
19111 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19113 PR rtl-optimization/79541
19114 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19115 instead of transforming it into USE.
19117 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19119 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19120 If HONOR_SNANS (SFmode) force the input to a register.
19121 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19122 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19123 an frsp or similar insn.
19125 2017-02-17 Martin Liska <mliska@suse.cz>
19127 PR rtl-optimization/79577
19128 * params.def (selsched-max-sched-times): Increase minimum to 1.
19130 2017-02-17 Martin Liska <mliska@suse.cz>
19132 PR rtl-optimization/79574
19133 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19135 2017-02-17 Martin Liska <mliska@suse.cz>
19137 PR tree-optimization/79529
19138 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19139 ssa_defined_default_def_p to handle cases which are implicitly
19141 * tree-ssa.c (ssa_defined_default_def_p): New function.
19142 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19143 which are implicitly defined.
19144 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19146 2017-02-17 Richard Biener <rguenther@suse.de>
19148 PR middle-end/79576
19149 * params.def (max-ssa-name-query-depth): Limit to 10.
19151 2017-02-17 Richard Biener <rguenther@suse.de>
19153 PR tree-optimization/79552
19154 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19157 2017-02-17 Richard Biener <rguenther@suse.de>
19160 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19162 2017-02-17 Marek Polacek <polacek@redhat.com>
19164 PR middle-end/79536
19165 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19166 (fold_negate_expr): New wrapper.
19168 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19170 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19171 Correct terminology and de-emphasize pre-standard behavior.
19173 2017-02-16 Alan Modra <amodra@gmail.com>
19175 PR rtl-optimization/79286
19176 * ira.c (def_dominates_uses): New function.
19177 (update_equiv_regs): Don't create an equivalence for insns that
19178 may trap where the register def does not dominate the use.
19180 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19182 PR rtl-optimization/78127
19183 * lra.c (lra): Call lra_eliminate before finish the loop after
19186 2017-02-16 Richard Biener <rguenther@suse.de>
19188 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19190 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19191 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19192 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19193 (isl_val_int_from_wi): New function.
19194 (extract_affine_gmp): Rename to ...
19195 (extract_affine_wi): ... this, take a widest_int.
19196 (extract_affine_int): Just wrap extract_affine_wi.
19197 (add_param_constraints): Use isl_val_int_from_wi.
19198 (add_loop_constraints): Likewise, and extract_affine_wi.
19200 2017-02-15 Jeff Law <law@redhat.com>
19202 PR middle-end/79521
19203 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19204 ira_init_register_move_cost_if_necessary.
19206 2017-02-15 Martin Sebor <msebor@redhat.com>
19208 PR middle-end/32003
19209 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19210 removed in a prior commit.
19212 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19214 PR tree-optimization/79347
19215 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19216 counters during peeling.
19218 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19220 * Makefile.in (site.exp): Remove "set ISLVER".
19222 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19225 * real.c (real_from_integer): Call real_convert even for decimal.
19227 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19230 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19232 2017-02-14 Andrew Pinski <apinski@cavium.com>
19234 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19235 cores and change the partno/implementer to be correct.
19236 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19237 the 'B" as the implementer.
19238 * config/aarch64/aarch64-tune.md: Regenerate.
19240 2017-02-14 Carl Love <cel@us.ibm.com>
19242 * config/rs6000/rs6000.c: Add case statement entry to make the
19243 xvcvuxdsp built-in argument unsigned.
19244 * config/rs6000/vsx.md: Fix the source and return operand types so they
19245 match the instruction definitions from the ISA document. Fix typo
19246 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19249 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19252 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19253 member early_clobber_alts.
19254 * lra-lives.c (reg_early_clobber_p): New.
19255 (process_bb_lives): Use it.
19256 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19257 (debug_operand_data): Initialize early_clobber_alts.
19258 (setup_operand_alternative): Set up early_clobber_alts.
19259 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19260 alternatives to new_insn_reg.
19261 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19263 (lra_update_insn_regno_info): Pass the new arg.
19265 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19267 PR middle-end/79505
19268 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19269 (new_oacc_loop_raw): Don't clear already cleared fields.
19272 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19273 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19274 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19276 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19279 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19281 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19284 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19285 the extra instruction to the right place to store 128-bit constant
19288 2017-02-14 Martin Sebor <msebor@redhat.com>
19290 PR middle-end/79448
19291 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19292 warning for strings of unknown length.
19294 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19296 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19298 2017-02-14 Jeff Law <law@redhat.com>
19301 * ira-costs.c (scan_one_insn): Initialize register move costs
19302 for pseudos seen in USE/CLOBBER insns.
19304 PR tree-optimization/79095
19305 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19306 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19307 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19308 if the operands are known to be not equal, then the resulting range
19310 (intersect_ranges): If the new range is ~[0,0] and the old range is
19311 wide, then prefer ~[0,0].
19312 * tree-vrp.c (overflow_comparison_p_1): New function.
19313 (overflow_comparison_p): New function.
19314 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19315 if NAME is used in an overflow test.
19316 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19317 overflow check that can be expressed as an equality test, then adjust
19318 ops to be that equality test.
19320 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19322 * config/s390/s390-builtin-types.def: Remove flags argument.
19323 * config/s390/s390.c (s390_init_builtins): Likewise.
19325 2017-02-14 Martin Liska <mliska@suse.cz>
19327 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19328 vector. Fix trailing white spaces.
19330 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19332 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19335 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19337 PR rtl-optimization/68664
19338 * config/arm/arm.c (arm_sched_can_speculate_insn):
19339 New function. Declare prototype.
19340 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19342 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19344 PR rtl-optimization/68664
19345 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19347 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19349 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19351 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19352 max skip bytes for function, loop and jump.
19354 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19356 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19357 ABS_EXPR for gimple dump.
19359 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19362 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19364 PR tree-optimization/79408
19365 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19366 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19367 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19368 also if rhs1 is INTEGER_CST.
19370 2017-02-14 Richard Biener <rguenther@suse.de>
19372 PR middle-end/79432
19373 * tree-into-ssa.c (insert_phi_nodes): When the function can
19374 have abnormal edges rewrite SSA names with broken use-def
19375 dominance out of SSA and register them for PHI insertion.
19377 2017-02-13 Martin Sebor <msebor@redhat.com>
19379 PR middle-end/79496
19380 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19381 clearing info.nowrite flag when snprintf size argument is a range.
19383 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19385 * cprop.c (cprop_jump): Add missing space in string literal.
19386 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19387 (get_constraint_for_component_ref): Likewise.
19388 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19389 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19390 * lra-constraints.c (process_alt_operands): Likewise.
19391 * ipa-inline.c (inline_small_functions): Likewise.
19392 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19393 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19394 * trans-mem.c (diagnose_tm_1_op): Likewise.
19395 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19396 (grid_parallel_clauses_gridifiable): Likewise.
19398 * config/nvptx/mkoffload.c (process): Add space in between
19401 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19402 "MOD4_SSE_REGS" and "ALL_REGS".
19404 * spellcheck.c (test_data): Add , in between "foo" and "food".
19406 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19409 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19410 boundary crossing check and subsequent code generation agree.
19412 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19414 * config/aarch64/aarch64.c (has_memory_op): Delete.
19415 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19418 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19420 PR rtl-optimization/79388
19421 PR rtl-optimization/79450
19422 * combine.c (distribute_notes): When removing TEM_INSN for which
19423 corresponding dest has last value recorded, invalidate that last
19426 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19428 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19429 of explicit '@'. Add missing assembly comment marker on branch costs
19432 2017-02-13 Nathan Sidwell <nathan@acm.org>
19434 * gengtype-lex.l (<in_struct>): Add '/'.
19436 2017-02-13 Martin Liska <mliska@suse.cz>
19439 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19441 2017-02-13 Richard Biener <rguenther@suse.de>
19443 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19445 * configure: Re-generate.
19446 * config.in: Likewise.
19447 * graphite-dependences.c: Simplify as if
19448 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19449 * graphite-isl-ast-to-gimple.c: Likewise.
19450 * graphite-optimize-isl.c: Likewise.
19451 * graphite-poly.c: Likewise.
19452 * graphite-sese-to-poly.c: Likewise.
19453 * graphite.h: Likewise.
19454 * toplev.c: Include isl/version.h and use isl_version () for
19455 printing the ISL version.
19456 * doc/install.texi: Update ISL requirement.
19458 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19460 * doc/standards.texi (Standards): Update reference to
19463 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19465 * doc/extend.texi (Named Address Spaces): sourceware.org now
19467 * doc/install.texi (Binaries): Ditto.
19470 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19472 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19473 standard terminology "stringize"/"stringizing" throughout.
19474 * doc/cppinternals.texi: Likewise.
19476 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19478 * doc/extend.texi: Fix some spelling mistakes and typos.
19479 * doc/invoke.texi: Likewise.
19481 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19484 * params.def (inline-min-speedup) Change from 10 to 8.
19486 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19488 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19491 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19494 * ipa-inline-analysis.c (get_minimal_bb): New function.
19495 (record_modified): Use it.
19496 (remap_edge_change_prob): Handle also ancestor functions.
19498 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19500 * doc/contrib.texi (Contributors): Remove broken link into
19501 the Mauve CVS repository.
19503 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19505 PR middle-end/79454
19506 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19507 result computation whenever lhs doesn't have vector mode, not
19508 just when it has BLKmode.
19510 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19512 * doc/makefile.texi (profiledbootstrap): Refer to the
19513 installation instructions only in textual form.
19515 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19518 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19520 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19522 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19523 (Specific): Update mingw-w64 reference.
19525 (Specific): Remove broken link to Renesas RX processor.
19527 2017-02-10 Richard Biener <rguenther@suse.de>
19529 * toplev.c (process_options): Do not mention obsolete graphite
19530 options when printing sorry message about missing graphite support.
19531 Mention -floop-nest-optimize.
19533 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19535 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19536 (vtst_p16): Likewise.
19537 (vtstq_p8): Likewise.
19538 (vtstq_p16): Likewise.
19540 (vtstq_p64): Likewise.
19541 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19542 (vset_lane_p64): New.
19543 (vsetq_lane_p64): New.
19545 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19547 PR tree-optimization/79411
19548 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19549 stmt operands are SSA_NAMEs used in abnormal phis.
19550 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19553 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19556 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19559 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19561 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19563 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19565 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19569 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19570 not arbitrary TREE_CONSTANT.
19573 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19574 "omp declare target link" attribute unless is_global_var.
19575 * omp-offload.c (find_link_var_op): Likewise.
19577 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19578 Chung-Lin Tang <cltang@codesourcery.com>
19580 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19582 (gimplify_adjust_omp_clauses): Don't delete TILE.
19583 (gimplify_omp_for): Deal with TILE.
19584 * internal-fn.c (expand_GOACC_TILE): New function.
19585 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19587 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19588 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19590 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19591 avoid DIV for outermost collapse var.
19592 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19593 Remove out of date comments, fix whitespace.
19594 * omp-general.c (omp_extract_for_data): Deal with tiling.
19595 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19596 adjust OLF_DIM_BASE value.
19597 (struct omp_for_data): Add tiling field.
19598 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19599 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19600 for auto loops. Remove default auto determining, moved to
19601 oacc_loop_fixed_partitions.
19602 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19603 stmts, add e_mask field.
19604 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19605 (oacc_thread_numbers): Use oacc_dim_call.
19606 (oacc_xform_tile): New.
19607 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19608 (finish_oacc_loop): Adjust for ifns vector.
19609 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19610 add case for GOACC_TILE fns.
19611 (oacc_loop_xform_loop): Delete.
19612 (oacc_loop_process): Iterate over call list directly, and add
19613 handling for GOACC_TILE fns.
19614 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19616 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19617 vector partitioning to outer loops. Assign 2 partitions to loops
19618 when available. Add TILE handling.
19619 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19620 (execite_oacc_device_lower): Process GOACC_TILE fns,
19621 ignore unknown specs.
19622 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19623 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19624 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19626 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19628 * configure.ac (ACX_BUGURL): Update.
19629 * configure: Regenerate.
19631 2017-02-09 Richard Biener <rguenther@suse.de>
19633 PR tree-optimization/69823
19634 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19635 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19637 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19639 * config/arc/arc-c.def: Add __NPS400__ definition.
19640 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19641 (TARGET_NPS400): Define.
19643 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19645 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19647 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19648 pointer, arch_info.
19649 (arc_cpu_types): Fill the arch_info field with a pointer into the
19650 arc_arch_types table.
19651 (arc_selected_cpu): Declare.
19652 * config/arc/arc.c (arc_selected_cpu): Make global.
19653 (arc_selected_arch): Delete.
19654 (arc_base_cpu): Delete.
19655 (arc_override_options): Remove references to deleted variables,
19656 update access to arch information.
19657 (ARC_OPT): Update access to arch information.
19658 (ARC_OPTX): Likewise.
19659 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19660 (TARGET_ARC600): Update access to arch information.
19661 (TARGET_ARC601): Likewise.
19662 (TARGET_ARC700): Likewise.
19663 (TARGET_EM): Likewise.
19664 (TARGET_HS): Likewise.
19665 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19668 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19671 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19672 condition/operands for integer GE/LE/GEU/LEU operations.
19674 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19676 PR translation/79397
19677 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19680 2017-02-08 Martin Jambor <mjambor@suse.cz>
19683 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19684 whether allocation happened.
19685 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19686 nothing was allocated.
19688 2017-02-08 Jakub Jelinek <jakub@redhat.com>
19690 PR tree-optimization/79408
19691 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
19692 constant, but SSA_NAME with a known integer range, use the minimum
19693 of that range instead of op1 to determine if modulo can be replaced
19694 with its first operand.
19696 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19698 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
19700 2017-02-08 Richard Biener <rguenther@suse.de>
19702 PR tree-optimization/71824
19703 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19704 Check all loops contained in the merged region.
19706 2017-02-07 Andrew Pinski <apinski@cavium.com>
19708 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
19710 2017-02-07 Andrew Pinski <apinski@cavium.com>
19712 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
19713 (thunderxt88): Likewise.
19714 (thunderxt81): Disable LSE and change v8.1 to v8.
19715 (thunderxt83): Likewise.
19717 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19718 Richard Biener <rguenther@suse.de>
19720 PR middle-end/79399
19721 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
19722 type from int to size_t.
19723 * ira-costs.c (struct_costs_size): Change type from int to size_t.
19725 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19727 PR rtl-optimization/79386
19728 * cprop.c (bypass_conditional_jumps): Initialize
19729 bypass_last_basic_block already before splitting bbs after
19730 unconditional traps...
19731 (bypass_conditional_jumps): ... rather than here.
19734 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
19735 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
19736 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
19737 fix -masm=intel patterns.
19739 2017-02-07 Richard Biener <rguenther@suse.de>
19741 PR tree-optimization/79256
19742 PR middle-end/79278
19743 * builtins.c (get_object_alignment_2): Use min_align_of_type
19744 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
19745 and ADJUST_FIELD_ALIGN.
19747 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
19749 * doc/tm.texi: Regenerate.
19750 * stor-layout.c (layout_decl): Adjust.
19751 (update_alignment_for_field): Likewise.
19752 (place_field): Likewise.
19753 (min_align_of_type): Likewise.
19754 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
19755 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
19756 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
19757 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
19758 * config/frv/frv.c (frv_adjust_field_align): Likewise.
19759 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
19760 * config/i386/i386.c (x86_field_alignment): Likewise.
19761 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
19762 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
19763 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
19764 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
19765 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
19766 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
19770 2017-01-30 Richard Biener <rguenther@suse.de>
19772 PR tree-optimization/79256
19773 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
19774 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
19777 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
19779 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
19780 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
19781 builtins to SImode and emit a zero-extend, if necessary.
19783 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19785 * docs/invoke.texi (RISC-V Options): Alphabetize.
19787 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19789 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
19792 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19794 * config/riscv/riscv.c: New file.
19795 * common/config/riscv/riscv-common.c: Likewise.
19796 * config.gcc: Likewise.
19797 * config/riscv/constraints.md: Likewise.
19798 * config/riscv/elf.h: Likewise.
19799 * config/riscv/generic.md: Likewise.
19800 * config/riscv/linux.h: Likewise.
19801 * config/riscv/multilib-generator: Likewise.
19802 * config/riscv/peephole.md: Likewise.
19803 * config/riscv/pic.md: Likewise.
19804 * config/riscv/predicates.md: Likewise.
19805 * config/riscv/riscv-builtins.c: Likewise.
19806 * config/riscv/riscv-c.c: Likewise.
19807 * config/riscv/riscv-ftypes.def: Likewise.
19808 * config/riscv/riscv-modes.def: Likewise.
19809 * config/riscv/riscv-opts.h: Likewise.
19810 * config/riscv/riscv-protos.h: Likewise.
19811 * config/riscv/riscv.h: Likewise.
19812 * config/riscv/riscv.md: Likewise.
19813 * config/riscv/riscv.opt: Likewise.
19814 * config/riscv/sync.md: Likewise.
19815 * config/riscv/t-elf-multilib: Likewise.
19816 * config/riscv/t-linux: Likewise.
19817 * config/riscv/t-linux-multilib: Likewise.
19818 * config/riscv/t-riscv: Likewise.
19819 * configure.ac: Likewise.
19820 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
19821 Waterman as RISC-V maintainers.
19822 * doc/install.texi: Add RISC-V entries.
19823 * doc/invoke.texi: Add RISC-V options section.
19824 * doc/md.texi: Add RISC-V constraints section.
19825 * configure: Regenerated.
19827 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19830 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
19831 false values to be constant vectors with all 0 or all 1 bits set.
19832 (vcondu<mode><mode>): Likewise.
19833 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
19835 (fpmask_comparison_operator): Update comment.
19836 (vecint_comparison_operator): New predicate.
19837 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
19838 vector conditionals when the true and false values are constant
19839 vectors with all 0 bits or all 1 bits set.
19841 2017-02-06 Martin Sebor <msebor@redhat.com>
19843 PR tree-optimization/79376
19844 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
19846 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
19848 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
19849 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
19850 to simplify split condition.
19852 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19854 * omp-expand.c (oxpand_omp_atomic_fetch_op,
19855 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
19858 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
19860 PR rtl-optimization/68664
19861 * target.def (can_speculate_insn): New hook.
19862 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
19863 * doc/tm.texi: Regenerate.
19864 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
19865 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
19866 (rs6000_sched_can_speculate_insn): New function.
19868 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19870 PR tree-optimization/79284
19871 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
19872 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
19873 vectorizable_mask_load_store, vectorizable_operation,
19874 vect_is_simple_cond, get_same_sized_vectype): Use it instead
19875 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
19876 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
19877 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
19878 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
19879 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19880 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
19881 is_gimple_assign (stmt). Replace another such test with
19882 is_gimple_assign (stmt).
19884 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19887 * config/avr/avr.c (rtl-iter.h): Include it.
19888 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
19889 (avr_legitimate_combined_insn): ...and implementation.
19891 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
19893 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
19894 * config/s390/s390.c (s390_const_operand_ok)
19895 (s390_canonicalize_comparison, s390_extract_part)
19896 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
19897 (s390_contiguous_bitmask_p, s390_rtx_costs)
19898 (legitimize_pic_address): Likewise.
19899 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
19900 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
19901 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
19902 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
19903 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
19905 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19907 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
19908 REGNO($0) == REGNO($1).
19910 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19912 * config/s390/linux.h(SIZE_TYPE): Add comment.
19914 2017-02-06 Julian Brown <julian@codesourcery.com>
19915 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19916 Virendra Pathak <virendra.pathak@broadcom.com>
19918 * config/aarch64/aarch64-cores.def: Change the scheduler
19920 * config/aarch64/aarch64.md: Include thunderx2t99.md.
19921 * config/aarch64/thunderx2t99.md: New file.
19923 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19925 * doc/standards.texi (Go Language): Update link to language
19928 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19930 * tree-eh.c (lower_resx): Sanitize profile.
19931 (cleanup_empty_eh_move_lp): Likewise.
19933 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19936 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
19938 * cfgloopmanip.h (loop_version): Update prototype.
19939 * modulo-sched.c (sms_schedule): Update call of loop_version.
19940 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
19941 * tree-parloops.c (gen_parallel_loop): Likewise.
19942 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
19943 * tree-ssa-loop-split.c (split_loop): Likewise.
19944 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
19945 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19947 2017-02-05 Martin Liska <mliska@suse.cz>
19950 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
19952 (print_operand_address): Initialize a struct to zero.
19954 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19956 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
19957 garbage collector only in textual form.
19959 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19961 * doc/extend.texi (x86 specific memory model extensions for
19962 transactional memory): Simplify a phrase.
19964 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
19967 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
19968 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
19969 (atomic_storedi_1): Likewise.
19971 2017-02-04 Jakub Jelinek <jakub@redhat.com>
19973 PR tree-optimization/79338
19974 * tree-parloops.c (gather_scalar_reductions): Don't call
19975 vect_analyze_loop_form for loop->inner before destroying loop's
19978 2017-02-03 Martin Sebor <msebor@redhat.com>
19980 PR tree-optimization/79327
19981 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
19982 when precision has resulted in leading zeros.
19983 (format_integer): Adjust the likely counter to assume an unknown
19984 argument that may be zero is non-zero.
19986 2017-02-03 Jason Merrill <jason@redhat.com>
19989 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
19990 avoid copying non-taken branch.
19992 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19994 PR tree-optimization/79340
19995 * tree-vect-loop.c (vectorizable_reduction): Release
19996 vec_defs elements after safe_splicing them into other vectors.
19999 PR tree-optimization/79327
20000 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20001 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20003 (format_integer): Use wide_int_to_tree instead of build_int_cst
20004 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20005 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20006 of shortest and longest sequence.
20008 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20010 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20011 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20013 2017-02-03 Walter Lee <walt@tilera.com>
20016 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20017 after initial stackframe link reg save.
20018 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20020 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20023 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20024 wu for stxssp alternative.
20026 2017-02-03 Martin Sebor <msebor@redhat.com>
20028 PR tree-optimization/79352
20029 * gimple-fold.c (get_range_strlen): Add argument.
20030 (get_range_strlen): Change return type to bool.
20031 (get_maxval_strlen): Pass in a dummy argument.
20032 * gimple-fold.h (get_range_strlen): Change return type to bool.
20033 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20034 * tree.h (array_at_struct_end_p): Add argument.
20035 * tree.c (array_at_struct_end_p): Handle it.
20037 2017-02-03 Martin Liska <mliska@suse.cz>
20040 * multiple_target.c (create_dispatcher_calls): Redirect edge
20041 from a caller of a dispatcher.
20042 (expand_target_clones): Make the clones local.
20043 (ipa_target_clone): Do both target clones and resolvers.
20044 (ipa_dispatcher_calls): Remove the pass.
20045 (pass_dispatcher_calls::gate): Likewise.
20046 (make_pass_dispatcher_calls): Likewise.
20047 * passes.def (pass_target_clone): Put as very first IPA early
20050 2017-02-03 Martin Liska <mliska@suse.cz>
20052 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20053 in case of a function with ifunc attribute.
20055 2017-02-03 Martin Liska <mliska@suse.cz>
20057 * cgraph.c (cgraph_node::dump): Dump function version info.
20058 * symtab.c (symtab_node::dump_base): Add missing new line.
20060 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20062 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20063 (ifcombine_ifandif): Use it.
20065 2017-02-03 Martin Liska <mliska@suse.cz>
20067 * doc/invoke.texi: Document default value for
20068 use-after-scope-direct-emission-threshold.
20070 2017-02-03 Martin Liska <mliska@suse.cz>
20072 PR tree-optimization/79339
20073 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20074 (format_floating): Likewise.
20076 2017-02-03 Martin Liska <mliska@suse.cz>
20079 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20081 (ipa_node_params_t::remove): Likewise.
20082 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20083 initialization from removed ipa_node_params_t::insert.
20084 (ipa_node_params::~ipa_node_params): Move from removed
20085 ipa_node_params_t::release.
20086 * symbol-summary.h (symbol_summary::m_released): New member.
20087 Do not release a summary twice. Do not allow to call finalizer
20088 for types of a summary that live in GGC memory.
20090 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20092 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20095 2017-02-02 Martin Sebor <msebor@redhat.com>
20097 PR middle-end/79275
20098 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20099 (format_string): Tighten up the range of output for non-constant
20100 strings and correct the expected range for wide non-constant strings.
20102 2017-02-02 Martin Sebor <msebor@redhat.com>
20104 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20106 PR middle-end/32003
20107 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20109 (-fdump-tree-@var): Add to index and document how to come up
20110 with pass-specific option and dump file names.
20111 (-fdump-passes): Clarify where to look for output.
20113 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20115 PR middle-end/77445
20116 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20117 statistics of the analyzed path; allow threading for speed when
20118 any of BBs along the path are optimized for speed.
20120 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20122 PR middle-end/78468
20123 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20124 settings of the virtual registers.
20127 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20129 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20130 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20133 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20135 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20136 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20138 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20140 * config/s390/s390.md: Add missing comments with the expanded
20142 * config/s390/vector.md: Likewise.
20143 * config/s390/vx-builtins.md: Likewise.
20145 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20148 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20149 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20150 conditions on a single line.
20152 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20154 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20155 __S390_VX__ to __VX__.
20157 2017-02-01 Andrew Pinski <apinski@cavium.com>
20159 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20160 stmt_info to record_stmt_cost.
20161 (vect_get_known_peeling_cost): Pass stmt_info if known to
20163 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20164 cpu_vector_cost field into
20165 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20166 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20167 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20168 splitting of scalar_stmt_cost and vec_stmt_cost.
20169 (thunderx_vector_cost): Likewise.
20170 (cortexa57_vector_cost): LIkewise.
20171 (exynosm1_vector_cost): Likewise.
20172 (xgene1_vector_cost): Likewise.
20173 (thunderx2t99_vector_cost): Improve after the splitting of the two
20175 (aarch64_builtin_vectorization_cost): Update for the splitting of
20176 scalar_stmt_cost and vec_stmt_cost.
20178 2017-02-01 Torvald Riegel <triegel@redhat.com>
20179 Richard Henderson <rth@redhat.com>
20181 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20182 conditional on existance of a fast atomic load.
20183 * optabs-query.c (can_atomic_load_p): New function.
20184 * optabs-query.h (can_atomic_load_p): Declare it.
20185 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20186 no fast atomic load is available for the particular size of access.
20187 (expand_atomic_compare_and_swap): Likewise.
20188 (expand_atomic_load): Likewise.
20189 (expand_atomic_store): Likewise.
20190 (expand_atomic_fetch_op): Likewise.
20191 * testsuite/lib/target-supports.exp
20192 (check_effective_target_sync_int_128): Remove x86 because it provides
20193 no fast atomic load.
20194 (check_effective_target_sync_int_128_runtime): Likewise.
20196 2017-02-01 Richard Biener <rguenther@suse.de>
20198 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20199 (graphite_transform_loops): Provide opt-info for optimized nests.
20200 * tree-parloop.c (parallelize_loops): Provide opt-info for
20201 parallelized loops.
20203 2017-02-01 Richard Biener <rguenther@suse.de>
20205 PR middle-end/79315
20206 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20207 was not set before.
20209 2017-02-01 Richard Biener <rguenther@suse.de>
20211 PR tree-optimization/71824
20212 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20213 Verify the loops are valid in the merged SESE region.
20214 (scop_detection::can_represent_loop_1): Check analyzing the
20215 evolution of the number of iterations in the region succeeds.
20217 2017-01-31 Ian Lance Taylor <iant@golang.org>
20219 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20220 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20222 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20224 PR preprocessor/79210
20225 * input.c (get_substring_ranges_for_loc): Replace line_width
20226 assertion with error-handling.
20228 2017-01-31 Richard Biener <rguenther@suse.de>
20230 PR tree-optimization/77318
20231 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20232 (create_pw_aff_from_tree): Take loop parameter.
20233 (add_condition_to_pbb): Pass loop of the condition to
20234 create_pw_aff_from_tree.
20236 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20238 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20239 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20241 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20245 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20247 (convert_int_to_float128): Likewise.
20248 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20249 (convert_int_to_float128): Likewise.
20250 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20251 (UNSPEC_IEEE128_CONVERT): Likewise.
20252 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20253 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20254 Use local variables for IBM extended format.
20255 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20256 (fix_trunc<mode>si2_fprs): Likewise.
20257 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20258 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20259 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20260 to know that we can now have integers of all sizes in vector
20262 (fix<uns>_<mode>di2_hw): Likewise.
20263 (float<uns>_<mode>si2_hw): Likewise.
20264 (fix_<mode>si2_hw): Likewise.
20265 (fixuns_<mode>si2_hw): Likewise.
20266 (float<uns>_<mode>di2_hw): Likewise.
20267 (float_<mode>di2_hw): Likewise.
20268 (float_<mode>si2_hw): Likewise.
20269 (floatuns_<mode>di2_hw): Likewise.
20270 (floatuns_<mode>si2_hw): Likewise.
20271 (xscvqp<su>wz_<mode>): Delete, no longer used.
20272 (xscvqp<su>dz_<mode>): Likewise.
20273 (xscv<su>dqp_<mode>): Likewise.
20274 (ieee128_mfvsrd_64bit): Likewise.
20275 (ieee128_mfvsrd_32bit): Likewise.
20276 (ieee128_mfvsrwz): Likewise.
20277 (ieee128_mtvsrw): Likewise.
20278 (ieee128_mtvsrd_64bit): Likewise.
20279 (ieee128_mtvsrd_32bit): Likewise.
20281 2017-01-31 Martin Liska <mliska@suse.cz>
20284 * ipa-prop.c (ipa_free_all_node_params): Call release method
20285 instead of ~sumbol_summary to not to trigger double times
20288 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20290 PR tree-optimization/71691
20291 * bitmap.h (class auto_bitmap): New.
20292 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20293 is_maybe_undefined instead of ssa_undefined_value_p.
20295 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20297 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20298 __S390_ARCH_LEVEL__ to __ARCH__.
20300 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20302 PR tree-optimization/79267
20303 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20304 if should_remove_lhs_p is true.
20306 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20309 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20310 (add_alignment_attribute): New.
20311 (base_type_die): Add alignment attribute.
20312 (subrange_type_die): Likewise.
20313 (modified_type_die): Likewise.
20314 (gen_array_type_die): Likewise.
20315 (gen_descr_array_type_die: Likewise.
20316 (gen_enumeration_type_die): Likewise.
20317 (gen_subprogram_die): Likewise.
20318 (gen_variable_die): Likewise.
20319 (gen_field_die): Likewise.
20320 (gen_ptr_to_mbr_type_die): Likewise.
20321 (gen_struct_or_union_type_die): Likewise.
20322 (gen_subroutine_type_die): Likewise.
20323 (gen_typedef_die): Likewise.
20324 (base_type_cmp): Compare alignment attribute.
20326 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20329 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20330 (setb_unsigned) New pattern for setb with CCUNS.
20331 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20332 subfc./subfe sequence to avoid overflow problems. Generate a
20333 shorter sequence with cmpld/setb for power9.
20334 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20335 for generating subfc. instruction.
20336 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20337 now uses this instruction.
20339 2017-01-30 Ian Lance Taylor <iant@google.com>
20342 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20343 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20345 2017-01-30 Martin Sebor <msebor@redhat.com>
20347 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20348 Move constant to the right of a relational operator.
20349 (get_mpfr_format_length, format_character, format_string): Ditto.
20350 (should_warn_p, maybe_warn): Same.
20352 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20354 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20357 * asan.c (get_translation_unit_decl): Remove function.
20358 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20360 2017-01-30 Martin Liska <mliska@suse.cz>
20362 PR gcov-profile/79259
20363 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20364 -fprofile-generate.
20366 2017-01-30 Martin Liska <mliska@suse.cz>
20369 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20370 Initialize variables with NULL value.
20372 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20375 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20377 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20379 2017-01-30 Richard Biener <rguenther@suse.de>
20381 PR tree-optimization/79276
20382 * tree-vrp.c (process_assert_insertions): Properly adjust common
20383 when removing a duplicate.
20385 * gcc.dg/torture/pr79276.c: New testcase.
20387 2017-01-30 Richard Biener <rguenther@suse.de>
20389 PR tree-optimization/79256
20390 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20391 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20393 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20395 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20398 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20399 ("*r<noxa>sbg_<mode>_sll_bitmask")
20400 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20401 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20402 Use contiguous_bitmask_nowrap_operand.
20404 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20407 * config/rs6000/altivec.h (vec_xl): Revise #define.
20408 (vec_xst): Likewise.
20410 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20412 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20414 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20416 PR rtl-optimization/79194
20417 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20418 traps before call to bypass_conditional_jumps.
20420 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20422 PR tree-optimization/71374
20423 * lra-constraints.c (check_conflict_input_operands): New.
20424 (match_reload): Use it.
20426 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20429 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20430 account to calculate conflict_set.
20432 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20434 PR rtl-optimization/78559
20435 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20436 other_insn in combine.
20438 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20440 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20441 uint16_type_node for BT_UINT16.
20443 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20445 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20446 "RTL Tests" to menu.
20447 (GIMPLE Tests): New node.
20448 (RTL Tests): New node.
20450 2017-01-27 Richard Biener <rguenther@suse.de>
20452 PR tree-optimization/79245
20453 * tree-loop-distribution.c (distribute_loop): Apply cost
20454 modeling also to detected patterns.
20456 2017-01-27 Richard Biener <rguenther@suse.de>
20458 PR tree-optimization/71433
20459 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20460 (compare_assert_loc): New function.
20461 (process_assert_insertions): Sort and optimize assert locations
20462 to remove duplicates and push down identical assertions on
20463 edges to their destination block.
20465 2017-01-27 Richard Biener <rguenther@suse.de>
20467 PR tree-optimization/79244
20468 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20469 out SSA names even if abnormal.
20471 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20473 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20474 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20475 instead of MPFR_RNDN.
20477 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20480 * arm.c (arm_option_override): Don't call build_target_option_node
20481 until after doing all option overrides.
20482 (arm_valid_target_attribute_tree): Likewise.
20484 2017-01-27 Martin Liska <mliska@suse.cz>
20486 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20487 for {cd}tors and C++ {cd}tors.
20489 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20491 * config/s390/s390.md ("*setmem_long_and")
20492 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20494 2017-01-26 Martin Sebor <msebor@redhat.com>
20496 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20499 2017-01-26 Martin Sebor <msebor@redhat.com>
20501 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20502 HAVE_DFmode before using XFmode or DFmode.
20503 (parse_directive): Avoid using the z length modifier to avoid
20504 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20506 PR middle-end/78703
20507 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20508 to accept adjustment as an array.
20509 (get_int_range): New function.
20510 (struct directive): Make width and prec arrays.
20511 (directive::set_width, directive::set_precision): Call get_int_range.
20512 (format_integer, format_floating): Handle width and precision ranges.
20513 (format_string, parse_directive): Same.
20515 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20518 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20519 comdat_type_p set, just clone them, but keep the children in the
20523 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20524 which have direct callers with -fvar-tracking-assignments enabled
20526 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20527 inside of type units.
20529 2017-01-26 Martin Sebor <msebor@redhat.com>
20531 PR middle-end/78703
20532 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20534 (struct format_result): Replace number_chars, number_chars_min,
20535 and number_chars_max with a single member of struct result_range.
20537 (format_result::operator+=): Adjust.
20538 (struct fmtresult): Remove bounded. Handle likely and unlikely
20540 (fmtresult::adjust_for_width_or_precision): New function.
20541 (fmtresult:type_max_digits): New function.
20542 (bytes_remaining): Handle likely and unlikely counters.
20543 (min_bytes_remaining): Remove.
20544 (format_percent): Simplify.
20545 (format_integer, format_floating): Set likely and unlikely counters.
20546 (get_string_length, format_character, format_string): Same.
20547 (format_plain, should_warn_p): New function.
20548 (maybe_warn): Call should_warn_p. Update diagnostic messages
20549 and handle those for all directives, including plain strings.
20550 (format_directive): Handle likely and unlikely counters.
20551 Remove unnecessary quoting from diagnostics. Add an informational
20553 (add_bytes): Remove.
20554 (pass_sprintf_length::compute_format_length): Simplify.
20555 (try_substitute_return_value): Handle likely and unlikely counters.
20557 2017-01-26 Carl Love <cel@us.ibm.com>
20559 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20560 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20562 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20565 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20566 endianess for subregs into account.
20567 * lra-constraints.c (lra_constraints): Do risky transformations
20568 always on the first iteration.
20569 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20570 last_call_used_reg_set.
20571 (process_bb_lives): Define and use last_call_used_reg_set.
20572 * lra.c (lra): Always continue after lra_constraints on the first
20575 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20577 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20579 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20581 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20583 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20584 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20585 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20586 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20587 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20588 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20589 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20590 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20591 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20593 2017-01-26 Marek Polacek <polacek@redhat.com>
20596 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20597 for the third operand.
20599 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20601 PR middle-end/79236
20602 * omp-low.c (struct omp_context): Add simt_stmt field.
20603 (scan_omp_for): Return omp_context *.
20604 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20605 context to the _simt_ SIMD stmt.
20606 (lower_omp_for): For combined SIMD with sibling _simt_
20607 SIMD, make sure to use the same decls in _looptemp_
20608 clauses as in the sibling.
20610 2017-01-26 David Sherwood <david.sherwood@arm.com>
20612 PR middle-end/79212
20613 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20616 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20619 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20620 emit fld b; fld a; if possible.
20622 * brig-builtins.def: Update copyright years.
20623 * config/arm/arm_acle_builtins.def: Update copyright years.
20625 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20628 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20629 constraint instead of o for the stxsd instruction.
20631 2017-01-25 Carl Love <cel@us.ibm.com>
20633 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20634 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20636 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20638 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20640 2017-01-25 Richard Biener <rguenther@suse.de>
20642 PR tree-optimization/69264
20643 * target.def (vector_alignment_reachable): Improve documentation.
20644 * doc/tm.texi: Regenerate.
20645 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20647 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20648 earlier changes with respect to TYPE_USER_ALIGN.
20649 (vector_alignment_reachable_p): Likewise. Improve dumping.
20651 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20654 * config/arm/arm.md (xordi3): Force constant operand into a register
20657 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20659 * doc/invoke.texi (-fstore-merging): Correct default optimization
20660 levels at which it is enabled.
20661 (-O): Move -fstore-merging from list to...
20664 2017-01-25 Richard Biener <rguenther@suse.de>
20667 * omp-expand.c: Include debug.h.
20668 (expand_omp_taskreg): Make sure to generate early debug before
20669 outlining anything from a function.
20670 (expand_omp_target): Likewise.
20671 (grid_expand_target_grid_body): Likewise.
20673 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20676 * asan.c (get_translation_unit_decl): New function.
20677 (asan_add_global): Extract modules file name from globals
20678 TRANSLATION_UNIT_DECL name.
20680 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20683 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20684 for long calls with APCS frame and VFP.
20686 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20688 * cfg.c (original_copy_tables_initialized_p): New function.
20689 * cfg.h (original_copy_tables_initialized_p): New decl.
20690 * cfgrtl.c (relink_block_chain): Guard the call to
20691 free_original_copy_tables with a call to
20692 original_copy_tables_initialized_p.
20693 * cgraph.h (symtab_node::native_rtl_p): New decl.
20694 * cgraphunit.c (symtab_node::native_rtl_p): New function.
20695 (symtab_node::needed_p): Don't assert for early assembly output
20696 for __RTL functions.
20697 (cgraph_node::finalize_function): Set "force_output" for __RTL
20699 (cgraph_node::analyze): Bail out early for __RTL functions.
20700 (analyze_functions): Update assertion to support __RTL functions.
20701 (cgraph_node::expand): Bail out early for __RTL functions.
20702 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
20704 * function.h (struct function): Update comment for field
20706 * gimple-expr.c: Include "tree-pass.h".
20707 (gimple_has_body_p): Return false for __RTL functions.
20708 * Makefile.in (OBJS): Add run-rtl-passes.o.
20709 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
20711 (gcc::pass_manager::get_clean_slate): New accessor.
20712 * passes.c: Include "insn-addr.h".
20713 (should_skip_pass_p): Add logging. Update logic for running
20714 "expand" to be compatible with both __GIMPLE and __RTL. Guard
20715 property-provider override so it is only done for gimple passes.
20717 (skip_pass): New function.
20718 (execute_one_pass): Call skip_pass when skipping passes.
20719 * read-md.c (md_reader::read_char): Support filtering
20720 the input to a subset of line numbers.
20721 (md_reader::md_reader): Initialize fields
20722 m_first_line and m_last_line.
20723 (md_reader::read_file_fragment): New function.
20724 * read-md.h (md_reader::read_file_fragment): New decl.
20725 (md_reader::m_first_line): New field.
20726 (md_reader::m_last_line): New field.
20727 * read-rtl-function.c (function_reader::create_function): Only
20728 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
20729 curr_properties. Set DECL_INITIAL to a dummy block.
20730 (read_rtl_function_body_from_file_range): New function.
20731 * read-rtl-function.h (read_rtl_function_body_from_file_range):
20733 * run-rtl-passes.c: New file.
20734 * run-rtl-passes.h: New file.
20736 2017-01-24 Jeff Law <law@redhat.com>
20738 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
20741 2017-01-24 Bin Cheng <bin.cheng@arm.com>
20743 PR tree-optimization/79159
20744 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
20745 (record_nonwrapping_iv): Improve boundary using above function if no
20746 value range information.
20748 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
20749 Martin Jambor <mjambor@suse.cz>
20751 * brig-builtins.def: New file.
20752 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
20753 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
20754 (DEF_HSAIL_SAT_BUILTIN): Likewise.
20755 (DEF_HSAIL_INTR_BUILTIN): Likewise.
20756 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
20757 * builtin-types.def (BT_INT8): New.
20758 (BT_INT16): Likewise.
20759 (BT_UINT8): Likewise.
20760 (BT_UINT16): Likewise.
20761 (BT_FN_ULONG): Likewise.
20762 (BT_FN_UINT_INT): Likewise.
20763 (BT_FN_UINT_ULONG): Likewise.
20764 (BT_FN_UINT_LONG): Likewise.
20765 (BT_FN_UINT_PTR): Likewise.
20766 (BT_FN_ULONG_PTR): Likewise.
20767 (BT_FN_INT8_FLOAT): Likewise.
20768 (BT_FN_INT16_FLOAT): Likewise.
20769 (BT_FN_UINT32_FLOAT): Likewise.
20770 (BT_FN_UINT16_FLOAT): Likewise.
20771 (BT_FN_UINT8_FLOAT): Likewise.
20772 (BT_FN_UINT64_FLOAT): Likewise.
20773 (BT_FN_UINT16_UINT32): Likewise.
20774 (BT_FN_UINT32_UINT16): Likewise.
20775 (BT_FN_UINT16_UINT16_UINT16): Likewise.
20776 (BT_FN_INT_PTR_INT): Likewise.
20777 (BT_FN_UINT_PTR_UINT): Likewise.
20778 (BT_FN_LONG_PTR_LONG): Likewise.
20779 (BT_FN_ULONG_PTR_ULONG): Likewise.
20780 (BT_FN_VOID_UINT64_UINT64): Likewise.
20781 (BT_FN_UINT8_UINT8_UINT8): Likewise.
20782 (BT_FN_INT8_INT8_INT8): Likewise.
20783 (BT_FN_INT16_INT16_INT16): Likewise.
20784 (BT_FN_INT_INT_INT): Likewise.
20785 (BT_FN_UINT_FLOAT_UINT): Likewise.
20786 (BT_FN_FLOAT_UINT_UINT): Likewise.
20787 (BT_FN_ULONG_UINT_UINT): Likewise.
20788 (BT_FN_ULONG_UINT_PTR): Likewise.
20789 (BT_FN_ULONG_ULONG_ULONG): Likewise.
20790 (BT_FN_UINT_UINT_UINT): Likewise.
20791 (BT_FN_VOID_UINT_PTR): Likewise.
20792 (BT_FN_UINT_UINT_PTR: Likewise.
20793 (BT_FN_UINT32_UINT64_PTR): Likewise.
20794 (BT_FN_INT_INT_UINT_UINT): Likewise.
20795 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
20796 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
20797 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
20798 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
20799 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
20800 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
20801 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
20802 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
20803 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
20804 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
20805 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
20806 * doc/frontends.texi: List BRIG FE.
20807 * doc/install.texi (Testing): Add BRIG tesring requirements.
20808 * doc/invoke.texi (Overall Options): Mention BRIG.
20809 * doc/standards.texi (Standards): Doucment BRIG HSA version.
20811 2017-01-24 Richard Biener <rguenther@suse.de>
20813 PR translation/79208
20814 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
20816 2017-01-24 Martin Jambor <mjambor@suse.cz>
20819 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
20820 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
20821 and known_contexts.
20823 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
20825 PR middle-end/79123
20826 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
20827 casts from signed to unsigned really don't have a range.
20829 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
20831 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
20832 GMP_RNDx for compatiblity.
20834 2017-01-24 Martin Liska <mliska@suse.cz>
20837 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
20838 that would prevent us to call alloca with -1 as argument.
20840 2017-01-24 Jakub Jelinek <jakub@redhat.com>
20842 * dwarf2out.c (output_compilation_unit_header, output_file_names):
20843 Avoid -Wformat-security warning.
20845 2017-01-23 Andrew Pinski <apinski@cavium.com>
20847 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
20850 2017-01-23 Martin Sebor <msebor@redhat.com>
20852 PR middle-end/78703
20853 * gimple-ssa-sprintf.c (warn_level): New global.
20854 (format_integer): Use it here and throughout the rest of the file.
20855 Use the same switch to compute sign as base.
20856 (maybe_warn): New function.
20857 (format_directive): Factor out warnings into maybe_warn.
20858 Add debugging output. Use warn_level.
20859 (add_bytes): Use warn_level.
20860 (pass_sprintf_length::compute_format_length): Add debugging output.
20861 (try_substitute_return_value): Same.
20862 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
20864 PR middle-end/78703
20865 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
20866 (struct fmtresult, format_integer, format_floating): Adjust.
20867 (fmtresult::fmtresult): Set max correctly in two argument ctor.
20868 (get_string_length, format_string,format_directive): Same.
20869 (pass_sprintf_length::compute_format_length): Same.
20870 (try_substitute_return_value): Simplify slightly.
20872 PR middle-end/78703
20873 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
20874 (fmtresult::operator+=): Outlined.
20875 (struct fmtresult): Add ctors.
20876 (struct conversion_spec): Rename...
20877 (struct directive): ...to this. Add and remove data members.
20878 (directive::set_width, directive::set_precision): New functions.
20879 (format_percent): Use fmtresult ctor.
20880 (get_width_and_precision): Remove.
20881 (format_integer): Make naming changes. Avoid computing width and
20883 (format_floating): Same. Adjust indentation.
20884 (format_character, format_none): New functions.
20885 (format_string): Moved character handling to format_character.
20886 (format_directive): Remove arguments, change return type.
20887 (parse_directive): New function.
20888 (pass_sprintf_length::compute_format_length): Move directive
20889 parsing to parse_directive.
20891 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20893 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
20894 (assign_assembler_name_if_needed): ... this.
20895 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
20896 (assign_assembler_name_if_needed): ... this.
20897 (free_lang_data_in_cgraph): Adjust callers.
20898 * cgraphunit.c (cgraph_node::analyze): Likewise.
20899 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
20902 2017-01-23 Richard Biener <rguenther@suse.de>
20904 PR tree-optimization/79088
20905 PR tree-optimization/79188
20906 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
20907 resetting loop bounds after last path deletion. Reset loop
20908 bounds of the target loop, make code match the comments.
20909 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
20910 Make sure loops need no fixups.
20912 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20914 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
20915 exponent support with double type for first argument.
20916 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
20917 type returned by __builtin_vec_extract_sig,
20918 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
20919 functions from "vector int" to "vector unsigned int" or from
20920 "vector long long int" to "vector unsigned long long int".
20921 Changed type returned by __builtin_vec_extract_exp,
20922 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
20923 functions from "vector int" to "vector unsigned int" or from
20924 "vector long long int" to "vector unsigned long long int".
20925 Changed return type of __builtin_vec_test_data_class,
20926 __builtin_vec_test_data_class_sp, and
20927 __builtin_vec_test_data_class_dp from "vector int" to
20928 "vector bool int" or from "vector long long int" to "vector bool
20929 long long int" and changed second argument type from "unsigned
20930 int" to "int". Added new overloaded function forms "vector float
20931 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
20932 "vector float __builtin_vec_insert_exp_sp (vector float, vector
20933 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
20934 double, vector unsigned long long int)" and "vector double
20935 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
20936 long int)". Changed return type of
20937 __builtin_scalar_test_data_class and
20938 __builtin_scalar_test_data_class_sp and
20939 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
20940 int" and changed second argument from "unsigned int" to "int".
20941 Changed type returned by __builtin_scalar_test_neg,
20942 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
20943 from "int" to "bool int". Added new overloaded function form
20944 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
20945 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
20946 exponent double-precision with floating point first argument.
20947 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
20948 documentation of scalar_test_data_class, scalar_test_neg,
20949 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
20950 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
20951 vec_test_data_class built-in functions to reflect refinements in
20952 their type signatures.
20954 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
20956 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
20958 (aarch64_elf_asm_destructor): Likewise.
20960 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
20962 PR rtl-optimization/78634
20963 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
20964 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
20965 * ifcvt.c (noce_try_cmove): Add missing cost check.
20967 PR rtl-optimization/71724
20968 * combine.c (if_then_else_cond): Look for situations where it is
20969 beneficial to undo the work of one of the recursive calls.
20971 2017-01-23 Bin Cheng <bin.cheng@arm.com>
20973 PR tree-optimization/70754
20974 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
20975 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
20976 combined stmt before it if not NULL.
20977 (combine_chains): Process refs reversely and compute dominance point
20980 2017-01-23 Martin Liska <mliska@suse.cz>
20982 PR tree-optimization/79196
20983 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
20984 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
20986 (strlen_optimize_stmt): Call the renamed function.
20988 2017-01-23 Michael Matz <matz@suse.de>
20990 PR tree-optimization/78384
20991 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
20993 2017-01-23 Richard Biener <rguenther@suse.de>
20995 PR tree-optimization/79186
20996 * tree-vrp.c (register_new_assert_for): Make sure we've seen
20997 both incoming edges before moving an assert.
20999 2017-01-23 Martin Jambor <mjambor@suse.cz>
21001 * ipa-prop.c (load_from_param_1): Removed.
21002 (load_from_unmodified_param): Bits from load_from_param_1 put back
21004 (load_from_param): Removed.
21005 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21006 with stmt. Reverted back to use of load_from_unmodified_param.
21008 2017-01-23 Martin Jambor <mjambor@suse.cz>
21011 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21012 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21013 field a pointer to garbage collected vector, mark lattices and
21014 ipcp_orig_node with GTY((skip)).
21015 (ipa_get_param_count): Adjust to descriptors being a pointer.
21016 (ipa_get_param): Likewise.
21017 (ipa_get_type): Likewise.
21018 (ipa_get_param_move_cost): Likewise.
21019 (ipa_set_param_used): Likewise.
21020 (ipa_get_controlled_uses): Likewise.
21021 (ipa_set_controlled_uses): Likewise.
21022 (ipa_is_param_used): Likewise.
21023 (ipa_node_params_t): Move into garbage collector. New methods insert
21025 (ipa_node_params_sum): Annotate wth GTY(()).
21026 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21028 (ipa_load_from_parm_agg): Adjust declaration.
21029 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21030 * ipa-profile.c (ipa_profile): Likewise.
21031 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21032 (ipa_populate_param_decls): Make descriptors parameter garbage
21034 (ipa_dump_param): Adjust to descriptors being a pointer.
21035 (ipa_alloc_node_params): Likewise.
21036 (ipa_initialize_node_params): Likewise.
21037 (load_from_param_1): Make descriptors parameter garbage collected.
21038 (load_from_unmodified_param): Likewise.
21039 (load_from_param): Likewise.
21040 (ipa_load_from_parm_agg): Likewise.
21041 (ipa_node_params::~ipa_node_params): Removed.
21042 (ipa_free_all_node_params): Remove call to delete operator.
21043 (ipa_node_params_t::insert): New.
21044 (ipa_node_params_t::remove): Likewise.
21045 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21046 copy known_csts and known_contexts vectors.
21047 (ipa_read_node_info): Adjust to descriptors being a pointer.
21048 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21050 (ipcp_transform_function): Make descriptors variable garbage
21053 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21055 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21056 * config/i386/avx512dqintrin.h: Ditto.
21057 * config/i386/avx512fintrin.h: Ditto.
21058 * config/i386/i386.c: Handle new builtins.
21059 * config/i386/i386-builtin.def: Add new builtins.
21060 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21061 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21063 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21064 Martin Liska <mliska@suse.cz>
21066 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21067 * asan.c (asan_expand_poison_ifn): Support stores and use
21068 appropriate ASAN report function.
21069 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21070 * internal-fn.def (ASAN_POISON_USE): Declare.
21071 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21072 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21073 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21074 ASAN_POISON calls w/o LHS.
21075 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21076 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21077 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21078 * gimplify.c (asan_poison_variables): Add attribute
21079 use_after_scope_memory to variables that really needs to live
21081 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21082 having the attribute.
21084 2017-01-23 Martin Liska <mliska@suse.cz>
21086 * asan.c (create_asan_shadow_var): New function.
21087 (asan_expand_poison_ifn): Likewise.
21088 * asan.h (asan_expand_poison_ifn): New declaration.
21089 * internal-fn.c (expand_ASAN_POISON): Likewise.
21090 * internal-fn.def (ASAN_POISON): New builtin.
21091 * sanopt.c (pass_sanopt::execute): Expand
21092 asan_expand_poison_ifn.
21093 * tree-inline.c (copy_decl_for_dup_finish): Make function
21095 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21096 * tree-ssa.c (is_asan_mark_p): New function.
21097 (execute_update_addresses_taken): Rewrite local variables
21098 (identified just by use-after-scope as addressable) into SSA.
21100 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21102 * doc/install.texi (Specific): opensource.apple.com uses https
21103 now. Remove trailing slash.
21105 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21107 * README.Portability: Remove note on an Irix compatibility issue.
21109 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21111 * gcov.c (INCLUDE_ALGORITHM): Define.
21112 (INCLUDE_VECTOR): Define.
21113 No longer include <vector> and <algorithm> directly.
21115 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21117 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21119 * doc/invoke.texi (Code Gen Options): Ditto.
21121 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21124 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21126 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21128 rtl-optimization/79125
21129 * cprop.c (local_cprop_pass): Handle cases where we make an
21130 unconditional trap.
21132 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21136 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21137 read from, for big endian.
21139 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21141 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21142 register pauth builtins for LP64 only.
21144 2017-01-20 Marek Polacek <polacek@redhat.com>
21147 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21150 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21152 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21154 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21155 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21156 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21158 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21161 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21162 in big-endian mode when they are not a single duplicated value.
21164 2017-01-20 Richard Biener <rguenther@suse.de>
21166 * BASE-VER: Bump to 7.0.1.
21168 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21170 * omp-low.c (omplow_simd_context): New struct. Use it...
21171 (lower_rec_simd_input_clauses): ...here and...
21172 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21173 references to idx, lane, max_vf, is_simt.
21175 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21177 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21180 2017-01-20 Martin Jambor <mjambor@suse.cz>
21182 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21183 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21185 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21186 (GTFILES): Rename hsa.c to hsa-common.c.
21187 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21188 * hsa-dump.c: Likewise.
21189 * hsa-gen.c: Likewise.
21190 * hsa-regalloc.c: Likewise.
21191 * ipa-hsa.c: Likewise.
21192 * omp-expand.c: Likewise.
21193 * omp-low.c: Likewise.
21194 * toplev.c: Likewise.
21196 2017-01-20 Marek Polacek <polacek@redhat.com>
21199 * doc/invoke.texi: Document -Wduplicated-branches.
21200 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21201 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21202 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21203 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21204 return 0 only when not OEP_LEXICOGRAPHIC.
21205 (fold_build_cleanup_point_expr): Use the expression
21206 location when building CLEANUP_POINT_EXPR.
21207 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21208 * tree.c (add_expr): Handle error_mark_node.
21210 2017-01-20 Martin Liska <mliska@suse.cz>
21213 * tree-profile.c (init_ic_make_global_vars): Do not call
21215 (gimple_init_gcov_profiler): Likewise.
21217 2017-01-20 Martin Liska <mliska@suse.cz>
21220 * cgraph.h (maybe_create_reference): Remove argument and
21222 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21224 * ipa-cp.c (create_specialized_node): Likewise.
21225 * symtab.c (symtab_node::maybe_create_reference): Handle
21226 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21228 2017-01-20 Martin Liska <mliska@suse.cz>
21230 * read-rtl-function.c (function_reader::create_function): Use
21231 build_decl instread of build_decl_stat.
21233 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21235 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21236 * config/i386/avx512dqintrin.h: Ditto.
21237 * config/i386/avx512fintrin.h: Ditto.
21238 * config/i386/i386-builtin-types.def: Add new types.
21239 * config/i386/i386.c: Handle new types.
21240 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21241 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21242 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21243 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21244 (__builtin_ia32_kshiftridi): New.
21245 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21247 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21251 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21252 define to rs6000_init_stack_protect_guard.
21253 (rs6000_init_stack_protect_guard): New function.
21255 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21256 Yunqiang Su <yunqiang.su@imgtec.com>
21258 * config.gcc (supported_defaults): Add madd4.
21259 (with_madd4): Add validation.
21260 (all_defaults): Add madd4.
21261 * config/mips/mips.opt (mmadd4): New option.
21262 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21264 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21266 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21267 (ISA_HAS_FUSED_MADD4): Likewise.
21268 * doc/invoke.texi (-mmadd4): Document the new option.
21269 * doc/install.texi (--with-madd4): Document the new option.
21271 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21273 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21274 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21275 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21276 (aarch64_init_pauth_hint_builtins): New.
21277 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21278 (aarch64_expand_builtin): Expand new builtins.
21280 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21282 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21283 * combine-stack-adj.c (no_unhandled_cfa): Handle
21284 REG_CFA_TOGGLE_RA_MANGLE.
21285 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21286 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21287 info for return address signing.
21288 (aarch64_expand_epilogue): Likewise.
21290 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21292 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21293 * config/aarch64/aarch64-protos.h
21294 (aarch64_return_address_signing_enabled): New declaration.
21295 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21297 (aarch64_expand_prologue): Sign return address before it's pushed onto
21299 (aarch64_expand_epilogue): Authenticate return address fetched from
21301 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21302 (aarch64_attributes): New function attributes for "sign-return-address".
21303 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21304 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21305 ("*do_return"): Generate combined instructions according to key index.
21306 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21307 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21309 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21310 * config/aarch64/aarch64.opt (msign-return-address=): New.
21311 * doc/extend.texi (AArch64 Function Attributes): Documents
21312 "sign-return-address=".
21313 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21315 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21317 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21318 overall option summary.
21320 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21322 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21323 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21324 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21325 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21327 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21329 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21330 -mpower9-minmax by default for -mcpu=power9.
21331 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21332 128-bit floating point.
21334 2017-01-20 Alan Modra <amodra@gmail.com>
21336 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21337 optimizing for size.
21339 2017-01-20 Alan Modra <amodra@gmail.com>
21342 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21343 for strcmp and strncmp from corresponding builtin decl.
21345 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21347 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21348 instead of i386/rtems-64.h.
21349 * config/i386/rtems-64.h: Remove.
21351 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21355 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21357 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21359 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21361 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21362 Change int to HOST_WIDE_INT.
21363 * config/aarch64/aarch64-protos.h
21364 (aarch64_simd_gen_const_vector_dup): Likewise.
21365 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21367 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21369 * langhooks-def.h (lhd_type_for_size): New decl.
21370 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21371 * langhooks.c (lhd_type_for_size): New function, taken from
21374 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21376 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21377 define_bypass for CR latency.
21378 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21379 (power9-alu2): Add define_bypass for CR latency.
21381 (power9-mul): Update insn latency.
21382 (power9-mul-compare): Update insn latency, bypass latency and remove
21385 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21387 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21389 * config/aarch64/aarch64.md
21390 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21391 aarch64_nopcrelative_literal_loads.
21392 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21394 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21396 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21397 TARGET_LOONGSON_3A.
21398 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21400 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21403 * config.gcc (supported_defaults): Add lxc1-sxc1.
21404 (with_lxc1_sxc1): Add validation.
21405 (all_defaults): Add lxc1-sxc1.
21406 * config/mips/mips.opt (mlxc1-sxc1): New option.
21407 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21409 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21410 __mips_no_lxc1_sxc1.
21411 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21412 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21413 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21415 2017-01-19 Richard Biener <rguenther@suse.de>
21417 PR tree-optimization/72488
21418 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21419 sure to restore SSA info.
21420 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21422 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21424 PR rtl-optimization/79121
21425 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21426 of the inner type when shifting an extended value.
21428 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21431 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21432 interposable aliases.
21434 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21437 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21438 Use the evmergelohi instruction.
21439 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21440 (mov_sitf_e500_subreg8_2_be): Likewise.
21441 (mov_sitf_e500_subreg12_2_le): Likewise.
21442 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21443 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21444 (mov_sitf_e500_subreg8_2_le): Likewise.
21445 (mov_sitf_e500_subreg12_2_be): Likewise.
21447 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21449 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21450 attribute from vecsimple to vecperm.
21451 (altivec_vbpermq2): Likewise.
21453 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21456 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21458 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21459 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21460 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21461 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21462 case where N arg is SIZE_MAX.
21463 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21464 (cmpstrsi): Add pattern.
21466 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21468 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21469 __builtin_vec_revb builtins.
21470 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21471 built-in functions to support generation of the ISA 3.0 XXBR<x>
21472 vector byte reverse instructions.
21473 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21474 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21475 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21476 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21477 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21478 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21479 (P9V_BUILTIN_VEC_REVB): Likewise.
21480 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21481 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21482 (p9_xxbrq_v16qi): Likewise.
21483 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21484 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21485 (p9_xxbrh_v8hi): Likewise.
21486 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21487 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21488 vec_revb built-in functions.
21490 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21492 PR rtl-optimization/78952
21493 * config/i386/i386.md (any_extract): New code iterator.
21494 (*insvqi_2): Use any_extract for source operand.
21495 (*insvqi_3): Use any_shiftrt for source operand.
21497 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21499 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21501 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21503 2017-01-18 Matthias Klose <doko@ubuntu.com>
21505 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21507 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21509 * config/rs6000/altivec.h (vec_bperm): Change #define.
21510 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21511 (altivec_vbpermq2): New define_insn.
21512 (altivec_vbpermd): Likewise.
21513 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21514 function interface.
21515 (VBPERMD): Likewise.
21516 (VBPERM): New polymorphic function interface.
21517 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21518 Add entries for P9V_BUILTIN_VEC_VBPERM.
21519 * doc/extend.texi: Add interfaces for vec_bperm.
21521 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21523 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21524 first letter of error messages.
21525 (s390_resolve_overloaded_builtin): Likewise.
21526 * config/s390/s390.c (s390_expand_builtin): Likewise.
21527 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21528 (s390_valid_target_attribute_inner_p): Likewise.
21529 * config/s390/s390.md ("tabort"): Likewise.
21531 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21533 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21534 (ISA_AVOID_DIV_HILO): New macro.
21535 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21536 (ISA_HAS_DDIV): Likewise.
21538 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21540 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21542 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21544 * doc/invoke.texi (fabi-version): Spelling fix.
21546 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21549 * doc/invoke.texi (fabi-version): Mention mangling fix for
21552 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21555 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21557 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21560 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21561 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21563 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21565 (stack_protect_test): Ditto.
21566 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21567 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21569 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21570 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21571 -mstack-protector-guard-offset=.
21572 (RS/6000 and PowerPC Options): Ditto.
21574 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21576 * config/i386/i386.h (MASK_CLASS_P): New define.
21577 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21578 there are no registers from different register sets also when
21579 mask registers are used. Update function comment.
21580 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21581 to (*k/*r) and (*k/*km) alternatives.
21583 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21585 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21586 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21587 (EH_RETURN_HANDLER_RTX): New define.
21588 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21589 Force frame pointer in EH return functions.
21590 (aarch64_expand_epilogue): Add barrier for eh_return.
21591 (aarch64_final_eh_return_addr): Remove.
21592 (aarch64_eh_return_handler_rtx): New function.
21593 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21595 (aarch64_eh_return_handler_rtx): New prototype.
21597 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21599 * config/rs6000/altivec.h (vec_rlmi): New #define.
21600 (vec_vrlnm): Likewise.
21601 (vec_rlnm): Likewise.
21602 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21603 (UNSPEC_VRLNM): Likewise.
21604 (VIlong): New mode iterator.
21605 (altivec_vrl<VI_char>mi): New define_insn.
21606 (altivec_vrl<VI_char>nm): Likewise.
21607 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21609 (VRLDNM): Likewise.
21610 (RLNM): New polymorphic function entry.
21611 (VRLWMI): New monomorphic function entry.
21612 (VRLDMI): Likewise.
21613 (RLMI): New polymorphic function entry.
21614 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21615 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21616 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21619 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21622 * dwarf2out.c (field_byte_offset): Restore the
21623 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21624 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21625 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21628 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21631 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21633 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21636 * dwarf2out.c (add_data_member_location_attribute): For constant
21637 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21638 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21639 DW_AT_byte_size attributes.
21641 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21643 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21644 after forcing to constant memory when the code model is medium.
21646 2017-01-17 Julia Koval <julia.koval@intel.com>
21649 * config/i386/avx512fintrin.h
21650 (_mm512_i32gather_ps): Change __addr type to void const*.
21651 (_mm512_mask_i32gather_ps): Ditto.
21652 (_mm512_i32gather_pd): Ditto.
21653 (_mm512_mask_i32gather_pd): Ditto.
21654 (_mm512_i64gather_ps): Ditto.
21655 (_mm512_mask_i64gather_ps): Ditto.
21656 (_mm512_i64gather_pd): Ditto.
21657 (_mm512_mask_i64gather_pd): Ditto.
21658 (_mm512_i32gather_epi32): Ditto.
21659 (_mm512_mask_i32gather_epi32): Ditto.
21660 (_mm512_i32gather_epi64): Ditto.
21661 (_mm512_mask_i32gather_epi64): Ditto.
21662 (_mm512_i64gather_epi32): Ditto.
21663 (_mm512_mask_i64gather_epi32): Ditto.
21664 (_mm512_i64gather_epi64): Ditto.
21665 (_mm512_mask_i64gather_epi64): Ditto.
21666 (_mm512_i32scatter_ps): Change __addr type to void*.
21667 (_mm512_mask_i32scatter_ps): Ditto.
21668 (_mm512_i32scatter_pd): Ditto.
21669 (_mm512_mask_i32scatter_pd): Ditto.
21670 (_mm512_i64scatter_ps): Ditto.
21671 (_mm512_mask_i64scatter_ps): Ditto.
21672 (_mm512_i64scatter_pd): Ditto.
21673 (_mm512_mask_i64scatter_pd): Ditto.
21674 (_mm512_i32scatter_epi32): Ditto.
21675 (_mm512_mask_i32scatter_epi32): Ditto.
21676 (_mm512_i32scatter_epi64): Ditto.
21677 (_mm512_mask_i32scatter_epi64): Ditto.
21678 (_mm512_i64scatter_epi32): Ditto.
21679 (_mm512_mask_i64scatter_epi32): Ditto.
21680 (_mm512_i64scatter_epi64): Ditto.
21681 (_mm512_mask_i64scatter_epi64): Ditto.
21682 * config/i386/avx512pfintrin.h
21683 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21684 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21685 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21686 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21687 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21688 (_mm512_prefetch_i32scatter_ps): Ditto.
21689 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
21690 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
21691 (_mm512_prefetch_i64scatter_pd): Ditto.
21692 (_mm512_prefetch_i64scatter_ps): Ditto.
21693 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
21694 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
21695 * config/i386/avx512vlintrin.h
21696 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
21697 (_mm_mmask_i32gather_ps): Ditto.
21698 (_mm256_mmask_i32gather_pd): Ditto.
21699 (_mm_mmask_i32gather_pd): Ditto.
21700 (_mm256_mmask_i64gather_ps): Ditto.
21701 (_mm_mmask_i64gather_ps): Ditto.
21702 (_mm256_mmask_i64gather_pd): Ditto.
21703 (_mm_mmask_i64gather_pd): Ditto.
21704 (_mm256_mmask_i32gather_epi32): Ditto.
21705 (_mm_mmask_i32gather_epi32): Ditto.
21706 (_mm256_mmask_i32gather_epi64): Ditto.
21707 (_mm_mmask_i32gather_epi64): Ditto.
21708 (_mm256_mmask_i64gather_epi32): Ditto.
21709 (_mm_mmask_i64gather_epi32): Ditto.
21710 (_mm256_mmask_i64gather_epi64): Ditto.
21711 (_mm_mmask_i64gather_epi64): Ditto.
21712 (_mm256_i32scatter_ps): Change __addr type to void*.
21713 (_mm256_mask_i32scatter_ps): Ditto.
21714 (_mm_i32scatter_ps): Ditto.
21715 (_mm_mask_i32scatter_ps): Ditto.
21716 (_mm256_i32scatter_pd): Ditto.
21717 (_mm256_mask_i32scatter_pd): Ditto.
21718 (_mm_i32scatter_pd): Ditto.
21719 (_mm_mask_i32scatter_pd): Ditto.
21720 (_mm256_i64scatter_ps): Ditto.
21721 (_mm256_mask_i64scatter_ps): Ditto.
21722 (_mm_i64scatter_ps): Ditto.
21723 (_mm_mask_i64scatter_ps): Ditto.
21724 (_mm256_i64scatter_pd): Ditto.
21725 (_mm256_mask_i64scatter_pd): Ditto.
21726 (_mm_i64scatter_pd): Ditto.
21727 (_mm_mask_i64scatter_pd): Ditto.
21728 (_mm256_i32scatter_epi32): Ditto.
21729 (_mm256_mask_i32scatter_epi32): Ditto.
21730 (_mm_i32scatter_epi32): Ditto.
21731 (_mm_mask_i32scatter_epi32): Ditto.
21732 (_mm256_i32scatter_epi64): Ditto.
21733 (_mm256_mask_i32scatter_epi64): Ditto.
21734 (_mm_i32scatter_epi64): Ditto.
21735 (_mm_mask_i32scatter_epi64): Ditto.
21736 (_mm256_i64scatter_epi32): Ditto.
21737 (_mm256_mask_i64scatter_epi32): Ditto.
21738 (_mm_i64scatter_epi32): Ditto.
21739 (_mm_mask_i64scatter_epi32): Ditto.
21740 (_mm256_i64scatter_epi64): Ditto.
21741 (_mm256_mask_i64scatter_epi64): Ditto.
21742 (_mm_i64scatter_epi64): Ditto.
21743 (_mm_mask_i64scatter_epi64): Ditto.
21744 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
21745 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
21746 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
21747 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
21748 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
21749 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
21750 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
21751 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
21752 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
21753 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
21754 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
21755 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
21756 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
21757 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
21758 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
21759 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
21760 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
21761 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
21762 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
21763 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
21764 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
21765 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
21766 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
21767 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
21768 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
21769 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
21770 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
21771 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
21772 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
21773 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
21774 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
21775 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
21776 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
21777 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
21778 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
21779 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
21780 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
21781 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
21782 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
21783 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
21784 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
21785 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
21786 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
21787 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
21788 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
21789 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
21790 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
21791 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
21792 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
21793 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
21794 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
21795 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
21796 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
21797 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
21798 definitions accordingly.
21800 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
21801 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21804 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
21807 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
21810 * ira-conflicts.c (ira_build_conflicts): Update total conflict
21811 hard regs for inner regno.
21813 2017-01-17 Martin Liska <mliska@suse.cz>
21816 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
21817 assumption and add comment.
21819 2017-01-17 Nathan Sidwell <nathan@acm.org>
21821 * ipa-visibility.c (localize_node): New function, broken out of ...
21822 (function_and_variable_visibility): ... here. Call it.
21824 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21826 PR middle-end/77445
21827 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
21828 correctly set frequency of oudgoing edge.
21829 (duplicate_thread_path): Fix profile updating.
21831 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21834 * configure.ac: Add GCC_BASE_VER.
21835 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
21836 version from BASE-VER file.
21837 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
21838 (gcc.o): Depend on $(BASEVER).
21839 * common.opt (dumpfullversion): New option.
21840 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
21841 * doc/invoke.texi: Document -dumpfullversion.
21842 * doc/install.texi: Document --with-gcc-major-version-only.
21843 * configure: Regenerated.
21845 2017-01-17 Richard Biener <rguenther@suse.de>
21847 PR tree-optimization/71433
21848 * tree-vrp.c (register_new_assert_for): Merge same asserts
21849 on all incoming edges.
21850 (process_assert_insertions_for): Handle insertions at the
21853 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
21855 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
21856 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
21858 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
21861 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
21864 2017-01-17 Alan Modra <amodra@gmail.com>
21867 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
21868 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
21869 symbolic stack limit when pic.
21871 2017-01-16 Martin Sebor <msebor@redhat.com>
21873 PR tree-optimization/78608
21874 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
21876 2017-01-16 Jeff Law <law@redhat.com>
21879 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
21880 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
21881 for several include directories that may be relative to sysroot.
21882 * config/i386/x-mingw32 (gplus_includedir): Define.
21883 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
21884 (native_system_includedir): Likewise.
21885 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
21886 override if TARGET_SYSTEM_ROOT is defined.
21887 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
21889 PR tree-optimization/79090
21890 PR tree-optimization/33562
21891 PR tree-optimization/61912
21892 PR tree-optimization/77485
21893 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
21894 and computed trims into the dump file.
21896 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21898 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
21900 2017-01-16 Jakub Jelinek <jakub@redhat.com>
21903 * gimplify.c (gimplify_init_constructor): If want_value and
21904 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
21908 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
21909 sequence. Formatting fixes.
21910 (doloop_optimize): Formatting fixes.
21913 * gcc.c (debug_level_greater_than_spec_func): New function.
21914 (static_spec_functions): Add debug-level-gt spec function.
21915 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
21917 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21918 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21919 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
21920 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
21921 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
21922 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
21924 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
21926 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
21927 QImode fixups to general and mask registers only.
21929 2017-01-16 Carl Love <cel@us.ibm.com>
21931 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
21932 for built-in functions
21933 vector signed char vec_nabs (vector signed char)
21934 vector signed short vec_nabs (vector signed short)
21935 vector signed int vec_nabs (vector signed int)
21936 vector signed long long vec_nabs (vector signed long long)
21937 vector float vec_nabs (vector float)
21938 vector double vec_nabs (vector double)
21939 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
21941 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
21942 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
21943 * doc/extend.texi: Update the documentation file for the new built-in
21946 2017-01-16 Martin Sebor <msebor@redhat.com>
21948 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
21951 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21953 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
21954 UNSPEC_VSX__XXSPLTD to require special splat handling.
21956 2017-01-16 David Malcolm <dmalcolm@redhat.com>
21959 * system.h: Poison strndup.
21961 2017-01-16 Alan Modra <amodra@gmail.com>
21964 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
21967 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
21969 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
21971 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
21973 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
21974 call recog here. Assert that INSN_CODE (insn) is non-negative.
21976 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21979 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
21981 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
21982 in the currently scheduled RTL fragment.
21984 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21986 PR rtl-optimization/78751
21987 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
21990 2017-01-14 Jeff Law <law@redhat.com>
21992 PR tree-optimization/79090
21993 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
21994 variable length stores.
21995 (compute_trims): Delete dead assignment to *trim_tail.
21996 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
21999 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22001 PR rtl-optimization/78626
22002 PR rtl-optimization/78727
22003 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22004 of a block, and split such blocks after everything else is finished.
22006 2017-01-14 Alan Modra <amodra@gmail.com>
22009 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22010 target legitimate_combined_insn.
22011 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22012 (rs6000_legitimate_combined_insn): New function.
22013 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22015 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22016 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22017 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22019 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22021 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22023 2017-01-13 Jeff Law <law@redhat.com>
22025 PR tree-optimization/33562
22026 PR tree-optimization/61912
22027 PR tree-optimization/77485
22028 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22030 (delete_dead_assignment): Likewise.
22031 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22032 statement to delete_dead_call and delete_dead_assignment.
22034 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22037 * substring-locations.c (format_warning_va): Strengthen case 1 so
22038 that both endpoints of the substring must be within the format
22039 range for just the substring to be printed.
22041 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22043 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22044 * config/i386/i386.c (ix86_target_string): Add missing options
22045 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22046 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22047 flags_other and ix86_target_other to flags2_other. Display unknown
22049 (ix86_valid_target_attribute_inner_p): Add missing options and
22050 reorder options by implied ISAs, as in ix86_target_string.
22052 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22054 * hash-table.h (hash_table::too_empty_p): New function.
22055 (hash_table::expand): Use it.
22056 (hash_table::traverse): Likewise.
22057 (hash_table::empty_slot): Use sizeof (value_type) instead of
22058 sizeof (PTR) to convert bytes to elements. Shrink the table
22059 if the current size is excessive for the current number of
22062 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22064 * ira-costs.c (record_reg_classes): Break from the inner loop
22065 early once alt_fail is known to be true. Update outer loop
22066 handling accordingly.
22068 2017-01-13 Jeff Law <law@redhat.com>
22070 * tree-ssa-dse.c (decrement_count): New function.
22071 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22072 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22073 when we know the partially dead statement is a mem* function.
22075 PR tree-optimization/61912
22076 PR tree-optimization/77485
22077 * tree-ssa-dse.c: Include expr.h.
22078 (maybe_trim_constructor_store): New function.
22079 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22081 PR tree-optimization/33562
22082 PR tree-optimization/61912
22083 PR tree-optimization/77485
22084 * doc/invoke.texi: Document new dse-max-object-size param.
22085 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22086 * tree-ssa-dse.c: Include params.h.
22087 (dse_store_status): New enum.
22088 (initialize_ao_ref_for_dse): New, partially extracted from
22090 (valid_ao_ref_for_dse, normalize_ref): New.
22091 (setup_live_bytes_from_ref, compute_trims): Likewise.
22092 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22093 (maybe_trim_partially_dead_store): Likewise.
22094 (maybe_trim_complex_store): Likewise.
22095 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22096 Track what bytes live from the original store. Return tri-state
22097 for dead, partially dead or live.
22098 (dse_dom_walker): Add constructor, destructor and new private members.
22099 (delete_dead_call, delete_dead_assignment): New extracted from
22101 (dse_optimize_stmt): Make a member of dse_dom_walker.
22102 Use initialize_ao_ref_for_dse.
22104 PR tree-optimization/33562
22105 PR tree-optimization/61912
22106 PR tree-optimization/77485
22107 * sbitmap.h (bitmap_count_bits): Prototype.
22108 (bitmap_clear_range, bitmap_set_range): Likewise.
22109 * sbitmap.c (bitmap_clear_range): New function.
22110 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22112 2017-01-13 Martin Liska <mliska@suse.cz>
22115 * function.c (set_cfun): Add new argument force.
22116 * function.h (set_cfun): Likewise.
22117 * ipa-inline-transform.c (inline_call): Use the function when
22118 strict alising from is dropped for function we inline to.
22120 2017-01-13 Richard Biener <rguenther@suse.de>
22122 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22123 for dumping GIMPLE INTEGER_CSTs.
22125 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22127 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22128 to 201112L since C++17.
22130 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22133 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22134 if -fsanitize=kernel-address is present.
22136 2017-01-13 Richard Biener <rguenther@suse.de>
22138 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22139 as _Literal ( type ) number in case usual suffixes do not
22140 preserve all information.
22142 2017-01-13 Richard Biener <rguenther@suse.de>
22144 PR tree-optimization/77283
22145 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22146 and ssa-iterators.h.
22147 (is_feasible_trace): Implement a cost model based on joiner
22150 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22153 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22154 char or short to __float128/_Float128 directly.
22156 2017-01-12 Martin Sebor <msebor@redhat.com>
22158 to -Wformat-overflow.
22159 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22160 (min_bytes_remaining): Same.
22161 (get_string_length): Same.
22162 (format_string): Same.
22163 (format_directive): Same.
22165 (pass_sprintf_length::handle_gimple_call): Same.
22167 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22169 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22170 info.nowrite calls with no lhs that can't throw. Return bool
22171 whether gsi_remove has been called or not.
22172 (pass_sprintf_length::handle_gimple_call): Return bool whether
22173 try_substitute_return_value called gsi_remove. Formatting fix.
22174 (pass_sprintf_length::execute): Don't use gsi_remove if
22175 handle_gimple_call returned true.
22178 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22179 be removed due to side-effects, don't remove following barrier nor
22180 turn the successor edge into fallthru edge.
22182 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22185 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22186 element-reversing loads and stores as not swappable.
22188 2017-01-12 Nathan Sidwell <nathan@acm.org>
22189 Nicolai Stange <nicstange@gmail.com>
22191 * combine.c (try_combine): Don't ignore result of overlap checking
22192 loop. Combine overlap & asm check into single loop.
22194 2017-01-12 Richard Biener <rguenther@suse.de>
22196 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22197 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22199 2017-01-12 Richard Biener <rguenther@suse.de>
22201 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22202 and TS_TARGET_OPTION directly derive from TS_BASE.
22203 * tree-core.h (tree_optimization_option): Derive from tree_base.
22204 (tree_target_option): Likewise.
22206 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22208 * config/i386/i386.c (memory_address_length): Increase len
22209 only when rip_relative_addr_p returns false.
22211 2017-01-11 Julia Koval <julia.koval@intel.com>
22213 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22214 (OPTION_MASK_ISA_SGX_SET): New.
22215 (ix86_handle_option): Handle OPT_msgx.
22216 * config.gcc: Added sgxintrin.h.
22217 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22218 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22219 * config/i386/i386.c (ix86_target_string): Add -msgx.
22221 (ix86_option_override_internal): Handle new options.
22222 (ix86_valid_target_attribute_inner_p): Add sgx.
22223 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22224 * config/i386/i386.opt: Add msgx.
22225 * config/i386/sgxintrin.h: New file.
22226 * config/i386/x86intrin.h: Add sgxintrin.h.
22228 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22231 * fold-const.c (maybe_nonzero_address): Return 1 for function
22233 (tree_single_nonzero_warnv_p): Don't handle function local objects
22237 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22240 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22243 * opts.c: Include "spellcheck.h"
22244 (struct string_fragment): New struct.
22245 (struct edit_distance_traits<const string_fragment &>): New
22247 (get_closest_sanitizer_option): New function.
22248 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22250 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22252 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22254 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22255 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22256 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22257 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22258 for initial die_offset if dwarf_split_debug_info.
22259 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22260 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22262 (output_skeleton_debug_sections): Formatting fix. Use
22263 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22264 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22266 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22268 * config/arm/cortex-a53.md: Add bypasses for
22269 cortex_a53_r2f_cvt.
22270 (cortex_a53_r2f): Only use for transfers.
22271 (cortex_a53_f2r): Likewise.
22272 (cortex_a53_r2f_cvt): Add reservation for conversions.
22273 (cortex_a53_f2r_cvt): Likewise.
22275 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22277 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22278 to all inlined functions, change static to extern.
22280 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22283 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22285 (arm_assemble_integer): Likewise.
22287 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22289 * config.gcc: Use new awk script to check CPU, FPU and architecture
22290 parameters for --with-... options.
22291 * config/arm/parsecpu.awk: New file
22292 * config/arm/arm-cpus.in: New file.
22293 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22295 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22297 * config/arm/t-arm: Update dependency rules.
22298 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22299 of processing .def files.
22300 * config/arm/genopt.sh: Deleted.
22301 * config/arm/gentune.sh: Deleted.
22302 * config/arm/arm-cores.def: Deleted.
22303 * config/arm/arm-arches.def: Deleted.
22304 * config/arm/arm-fpus.def: Deleted.
22305 * config/arm/arm-tune.md: Regenerated.
22306 * config/arm/arm-tables.opt: Regenerated.
22307 * config/arm/arm-cpu.h: New generated file.
22308 * config/arm/arm-cpu-data.h: New generated file.
22309 * config/arm/arm-cpu-cdata.h: New generated file.
22311 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22314 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22316 (input_varpool_node): Unpack dynamically_initialized bit.
22318 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22320 PR rtl-optimization/79032
22321 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22322 the alignment of the adjusted memory reference against that of MODE,
22323 instead of the alignment of the original memory reference.
22325 2017-01-11 Martin Jambor <mjambor@suse.cz>
22327 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22329 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22330 decorated functions.
22332 2017-01-11 Richard Biener <rguenther@suse.de>
22334 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22335 set range/nonnull info for PHI results. Do not set it on
22336 stmts marked for removal.
22338 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22340 * expr.c (store_field): In the bitfield case, fetch the return value
22341 from the registers before applying a single big-endian adjustment.
22342 Always do a final load for a BLKmode value not larger than a word.
22344 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22347 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22348 that we correctly handle column numbers greater than
22349 LINE_MAP_MAX_COLUMN_NUMBER.
22351 2017-01-10 Martin Sebor <msebor@redhat.com>
22353 PR middle-end/78245
22354 * gimple-ssa-sprintf.c (get_destination_size): Call
22355 {init,fini}object_sizes.
22356 * tree-object-size.c (addr_object_size): Adjust.
22357 (pass_through_call): Adjust.
22358 (pass_object_sizes::execute): Adjust.
22359 * tree-object-size.h (fini_object_sizes): Declare.
22361 2017-01-10 Martin Sebor <msebor@redhat.com>
22363 PR tree-optimization/78775
22364 * builtins.c (get_size_range): Move...
22365 * calls.c: ...to here.
22366 (alloc_max_size): Accept zero argument.
22367 (operand_signed_p): Remove.
22368 (maybe_warn_alloc_args_overflow): Call get_size_range.
22369 * calls.h (get_size_range): Declare.
22371 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22373 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22374 from TI's devices.csv file as of September 2016.
22375 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22377 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22379 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22380 * doc/invoke.texi: Likewise.
22381 * doc/md.texi: Likewise.
22382 * doc/objc.texi: Likewise.
22384 2017-01-10 Joshua Conner <joshconner@google.com>
22386 * config/arm/fuchsia-elf.h: New file.
22387 * config/fuchsia.h: New file.
22388 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22389 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22391 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22393 2016-01-10 Richard Biener <rguenther@suse.de>
22395 PR tree-optimization/79034
22396 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22397 Propagate out degenerate PHIs in the joiner.
22399 2017-01-10 Martin Liska <mliska@suse.cz>
22401 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22402 (sort_congruence_classes_by_decl_uid): Likewise.
22403 (sort_congruence_class_groups_by_decl_uid): Likewise.
22404 (sem_item_optimizer::merge_classes): Sort class, groups in these
22405 classes and members in the groups by DECL_UID of declarations.
22406 This would make merge operations stable.
22408 2017-01-10 Martin Liska <mliska@suse.cz>
22410 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22411 usage of m_classes_vec.
22412 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22413 (sem_item_optimizer::get_group_by_hash): Likewise.
22414 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22415 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22416 (sem_item_optimizer::verify_classes): Likewise.
22417 (sem_item_optimizer::process_cong_reduction): Likewise.
22418 (sem_item_optimizer::dump_cong_classes): Likewise.
22419 (sem_item_optimizer::merge_classes): Likewise.
22420 * ipa-icf.h (congruence_class_hash): Rename from
22421 congruence_class_group_hash. Remove declaration of m_classes_vec.
22423 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22425 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22426 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22427 * config.gcc: Add avx512vpopcntdqintrin.h.
22428 * config/i386/avx512vpopcntdqintrin.h: New.
22429 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22430 * config/i386/i386-builtin-types.def: Add new types.
22431 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22432 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22433 __builtin_ia32_vpopcountq_v8di_mask): New.
22434 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22435 __AVX512VPOPCNTDQ__.
22436 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22437 (PTA_AVX512VPOPCNTDQ): Define.
22438 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22439 TARGET_AVX512VPOPCNTDQ_P): Define.
22440 * config/i386/i386.opt: Add mavx512vpopcntdq.
22441 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22442 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22444 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22446 PR middle-end/77484
22447 * predict.def (PRED_CALL): Set to 67.
22449 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22451 * expr.c (store_field): In the bitfield case, if the value comes from
22452 a function call and is of an aggregate type returned in registers, do
22453 not modify the field mode; extract the value in all cases if the mode
22454 is BLKmode and the size is not larger than a word.
22456 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22459 * config/i386/cpuid.h: Fix undefined behavior.
22461 2017-01-04 Jeff Law <law@redhat.com>
22463 PR tree-optimization/79007
22464 PR tree-optimization/67955
22465 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22466 conservative for pt.null when flag_non_call_exceptions is on.
22468 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22470 PR translation/79019
22471 PR translation/79020
22472 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22473 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22474 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22476 * config/avr/avr.opt (maccumulate-args): Likewise.
22477 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22478 * common.opt (freport-bug): Likewise.
22479 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22480 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22481 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22482 translatable string.
22483 * config/i386/i386.c (function_value_32): Likewise.
22484 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22485 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22487 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22488 * common/config/msp430/msp430-common.c (msp430_handle_option):
22490 * symtab.c (symtab_node::verify_base): Likewise.
22491 * opts.c (set_debug_level): Likewise.
22492 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22493 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22494 missing whitespace to translatable strings.
22495 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22496 * config/sh/superh.h: Likewise.
22497 * config/i386/xopintrin.h: Likewise.
22498 * config/i386/znver1.md: Likewise.
22499 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22500 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22501 * double-int.h (struct double_int): Likewise.
22502 * double-int.c (div_and_round_double): Likewise.
22503 * wide-int.cc: Likewise.
22504 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22505 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22506 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22507 (crossjumps_occurred): ... this.
22508 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22511 PR tree-optimization/78899
22512 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22513 returning bool return struct loop *, NULL for failure and the new
22515 (versionable_outer_loop_p): Don't version outer loop if it has
22516 dont_vectorized bit set.
22517 (tree_if_conversion): When versioning outer loop, ensure
22518 tree_if_conversion is performed also on the inner loop of the
22519 non-vectorizable outer loop copy.
22520 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22521 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22522 prevent vectorization of it.
22523 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22524 the outer loop vectorization of the non-scalar version is attempted
22525 before vectorization of the inner loop in scalar version. If
22526 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22527 vectorization of its inner loop.
22528 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22529 has 2 inner loops, rename also on edges from bb whose single pred
22530 is outer_loop->header. Fix typo in function comment.
22532 2017-01-09 Martin Sebor <msebor@redhat.com>
22535 * asan.c (asan_emit_stack_protection): Increase local buffer size
22536 to avoid snprintf truncation warning.
22538 2017-01-09 Andrew Pinski <apinski@cavium.com>
22540 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22541 to reference thunderx2t99 for the tuning structure
22542 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22544 (thunderx2t99_extra_costs): This.
22545 * config/aarch64/aarch64-tune.md: Regenerate.
22546 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22547 (vulcan_addrcost_table): This.
22548 (vulcan_regmove_cost): Rename to ...
22549 (thunderx2t99_regmove_cost): This.
22550 (vulcan_vector_cost): Rename to ...
22551 (thunderx2t99_vector_cost): this.
22552 (vulcan_branch_cost): Rename to ...
22553 (thunderx2t99_branch_cost): This.
22554 (vulcan_tunings): Rename to ...
22555 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22556 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22558 2017-01-09 Martin Jambor <mjambor@suse.cz>
22562 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22563 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22564 (propagate_vr_accross_jump_function): Use the above function for all
22565 value range computations for pass-through jump functions and type
22566 converasion from explicit value range values.
22567 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22568 parameters from TYPE_ARG_TYPES.
22569 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22570 (ipa_write_node_info): Stream type of the actual argument.
22571 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22573 2017-01-09 Martin Liska <mliska@suse.cz>
22576 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22577 (lookup_compiler): Do not show error message with have_E.
22579 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22581 PR tree-optimization/78938
22582 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22583 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22584 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22585 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22588 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22590 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22593 2017-01-09 Richard Biener <rguenther@suse.de>
22595 PR tree-optimization/78997
22596 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22597 name condition properly.
22599 2017-01-09 Richard Biener <rguenther@suse.de>
22602 * dwarf2out.c (is_cxx): New overload with context.
22603 (is_naming_typedef_decl): Use it.
22605 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22607 * invoke.texi (Option Summary): Correct spacing in option lists
22608 and add line breaks to fix over-long lines.
22610 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22612 PR middle-end/17660
22614 * extend.texi (Common Variable Attributes): Add xref to GCC
22615 Internals manual to explain mode attribute keywords.
22617 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22620 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22621 and Preprocessor Options.
22622 (Options for Linking): Document -pthread here....
22623 (RS/6000 and PowerPC Options): ...not here.
22624 (Solaris 2 Options): ...or here.
22625 * doc/cppopts.texi: Document -pthread.
22627 2017-01-08 Martin Sebor <msebor@redhat.com>
22629 PR middle-end/77708
22630 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22631 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22632 New member functions.
22633 (format_directive): Used them.
22635 (pass_sprintf_length::handle_gimple_call): Same.
22636 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22637 to avoid truncation for any argument.
22638 (extract_affine_mul): Same.
22639 * tree.c (get_file_function_name): Same.
22641 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22643 PR middle-end/77484
22644 * predict.def (PRED_INDIR_CALL): Set to 86.
22646 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22648 PR preprocessor/54124
22649 * doc/cppopts.texi: Reformat -d subtable to list the full name
22650 of the options. Add cross-reference to the docs for the general
22651 compiler -d options.
22652 * doc/invoke.texi (Developer Options): Add cross-reference to the
22653 preprocessor-specific -d option documentation.
22655 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22657 PR preprocessor/13498
22658 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22659 redudant material, and reflect new command-line options.
22660 (System Headers): Likewise.
22662 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22664 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22665 -isystem, and -idirafter. Copy-edit.
22666 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22667 default for -ftrack-macro-expansion. Delete obsolete and
22668 badly-formatted implementation details about -fdebug-cpp output.
22669 * doc/cppwarnopts.texi: Copy-edit.
22671 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22674 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22675 that the transition from a max line width >= 1<<10 to narrower
22676 lines works correctly.
22678 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22680 * doc/options.texi (PerFunction): New.
22681 * opt-functions.awk (switch_flags): Map both Optimization and
22682 PerFunction to CL_OPTIMIZATION.
22683 * opth-gen.awk: Test for PerFunction flag along with
22685 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22686 it only when the latter is present. Skip those that don't in
22687 the hash function generator.
22688 * common.opt (fvar-tracking): Mark as PerFunction instead of
22690 (fvar-tracking-assignments): Likewise.
22691 (fvar-tracking-assignments-toggle): Likewise.
22692 (fvar-tracking-uninit): Likewise.
22694 2017-01-07 Jakub Jelinek <jakub@redhat.com>
22696 PR translation/79018
22697 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
22700 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
22703 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
22704 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
22705 TARGET_LONG_JUMP_TABLE_OFFSETS.
22706 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
22707 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
22708 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
22709 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
22710 * config/m68k/m68k.md (tablejump expander): Likewise.
22711 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
22712 TARGET_LONG_JUMP_TABLE_OFFSETS.
22713 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
22714 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
22716 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
22717 David Holsgrove <david.holsgrove@xilinx.com>
22719 * common/config/microblaze/microblaze-common.c
22720 (TARGET_EXCEPT_UNWIND_INFO): Remove.
22721 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
22723 * config/microblaze/microblaze.c (microblaze_must_save_register)
22724 (microblaze_expand_epilogue, microblaze_return_addr): Handle
22726 (microblaze_eh_return): New function.
22727 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
22728 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
22729 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
22730 * config/microblaze/microblaze.md (eh_return): New pattern.
22732 2017-01-06 Jakub Jelinek <jakub@redhat.com>
22734 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
22735 GCC_DIAGNOSTIC_STRINGIFY): Define.
22737 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
22739 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22741 * config/arm/arm.md (<mcrr>): New.
22743 * config/arm/arm.c (arm_arch5te): New.
22744 (arm_option_override): Set arm_arch5te.
22745 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
22747 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
22748 (arm_mcrr_qualifiers): ... this. New.
22749 (MRRC_QUALIFIERS): Define to...
22750 (arm_mrrc_qualifiers): ... this. New.
22751 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
22753 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
22754 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
22755 (MRRCI, mrrc, MRRC): New.
22756 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
22757 VUNSPEC_MRRC2): New.
22759 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22761 * config/arm/arm.md (<mcr>): New.
22763 * config/arm/arm.c (arm_coproc_builtin_available): Add
22764 support for mcr, mrc, mcr2 and mrc2.
22765 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
22766 (arm_mcr_qualifiers): ... this. New.
22767 (MRC_QUALIFIERS): Define to ...
22768 (arm_mrc_qualifiers): ... this. New.
22769 (MCR_QUALIFIERS): Define to ...
22770 (arm_mcr_qualifiers): ... this. New.
22771 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
22773 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
22774 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
22775 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
22776 VUNSPEC_MRC2): New.
22778 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22780 * config/arm/arm.md (*ldc): New.
22784 * config/arm/arm.c (arm_coproc_builtin_available): Add
22785 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
22786 (arm_coproc_ldc_stc_legitimate_address): New.
22787 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
22788 'qualifier_const_pointer'.
22789 (LDC_QUALIFIERS): Define to...
22790 (arm_ldc_qualifiers): ... this. New.
22791 (STC_QUALIFIERS): Define to...
22792 (arm_stc_qualifiers): ... this. New.
22793 * config/arm/arm-protos.h
22794 (arm_coproc_ldc_stc_legitimate_address): New.
22795 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
22796 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
22797 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
22798 stc2, stcl, stc2l): New.
22799 * config/arm/constraints.md (Uz): New.
22800 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
22801 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
22802 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
22803 VUNSPEC_STC2L): New.
22805 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22807 * config/arm/arm.md (<cdp>): New.
22808 * config/arm/arm.c (neon_const_bounds): Rename this ...
22809 (arm_const_bounds): ... this.
22810 (arm_coproc_builtin_available): New.
22811 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
22812 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
22813 (CDP_QUALIFIERS): Define to...
22814 (arm_cdp_qualifiers): ... this. New.
22816 (arm_expand_builtin_args): Add case for 6 arguments.
22817 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
22818 (arm_const_bounds): ... this.
22819 (arm_coproc_builtin_available): New.
22820 * config/arm/arm_acle.h (__arm_cdp): New.
22822 * config/arm/arm_acle_builtins.def (cdp): New.
22824 * config/arm/iterators.md (CDPI,CDP,cdp): New.
22825 * config/arm/neon.md: Rename all 'neon_const_bounds' to
22826 'arm_const_bounds'.
22827 * config/arm/types.md (coproc): New.
22828 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
22829 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
22830 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
22831 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
22833 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22835 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
22836 (UBINOP_QUALIFIERS): New.
22838 (acle_builtin_data): New. Change comment.
22839 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
22840 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
22841 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
22842 arm_acle_builtins.def.
22843 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
22844 (arm_init_acle_builtins): New.
22845 (CRC32_BUILTIN): Remove.
22846 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
22847 crc32cb, crc32ch and crc32cw.
22848 (arm_init_crc32_builtins): Remove.
22849 (arm_init_builtins): Use arm_init_acle_builtins rather
22850 than arm_init_crc32_builtins.
22851 (arm_expand_acle_builtin): New.
22852 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
22853 * config/arm/arm_acle_builtins.def: New.
22855 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22857 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
22858 (arm_builtin_datum): ... this.
22859 (arm_init_neon_builtin): Rename to ...
22860 (arm_init_builtin): ... this. Add a new parameters PREFIX
22861 and USE_SIG_IN_NAME.
22862 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
22863 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
22864 'arm_builtin_datum'.
22865 (arm_init_vfp_builtins): Likewise.
22866 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
22867 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
22868 (arm_expand_neon_args): Rename to ...
22869 (arm_expand_builtin_args): ... this. Rename builtin_arg
22870 enum values and differentiate between ARG_BUILTIN_MEMORY
22871 and ARG_BUILTIN_NEON_MEMORY.
22872 (arm_expand_neon_builtin_1): Rename to ...
22873 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
22874 values, arm_expand_builtin_args and add bool parameter NEON.
22875 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
22876 (arm_expand_vfp_builtin): Likewise.
22877 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
22879 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22881 PR middle-end/77484
22882 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
22883 * predict.c (tree_estimate_probability_bb): Reverse direction of
22884 polymorphic call predictor.
22886 2017-01-06 David Malcolm <dmalcolm@redhat.com>
22888 * passes.c (execute_one_pass): Split out pass-skipping logic into...
22889 (determine_pass_name_match): ...this new function and...
22890 (should_skip_pass_p): ...this new function.
22892 2017-01-06 Nathan Sidwell <nathan@acm.org>
22894 * ipa-visibility.c (function_and_variable_visibility): Reformat
22895 comments and long lines. Remove extrneous if.
22896 * symtab.c (symtab_node::make_decl_local): Fix code format.
22897 (symtab_node::set_section_for_node): Fix comment typo.
22899 2017-01-06 Martin Liska <mliska@suse.cz>
22902 * lra-constraints.c: Rename invariant to lra_invariant.
22903 * predict.c (set_even_probabilities): Initialize e to NULL.
22905 2017-01-05 Martin Sebor <msebor@redhat.com>
22907 PR tree-optimization/78910
22908 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
22909 (format_integer): Correct off-by-one error in the handling
22910 of precision with negative numbers in signed conversions..
22912 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
22914 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
22916 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22918 PR tree-optimization/71016
22919 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
22920 factor_out_conditional_conversion. Formatting fix.
22921 (factor_out_conditional_conversion): Add cond_stmt argument.
22922 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
22923 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
22926 2017-01-05 David Malcolm <dmalcolm@redhat.com>
22928 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
22929 read-rtl-function.o, and selftest-rtl.o.
22930 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
22931 (selftest::aarch64_test_loading_full_dump): New function.
22932 (selftest::aarch64_run_selftests): New function.
22933 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
22934 selftest::aarch64_run_selftests.
22935 * config/i386/i386.c
22936 (selftest::ix86_test_loading_dump_fragment_1): New function.
22937 (selftest::ix86_test_loading_call_insn): New function.
22938 (selftest::ix86_test_loading_full_dump): New function.
22939 (selftest::ix86_test_loading_unspec): New function.
22940 (selftest::ix86_run_selftests): Call the new functions.
22941 * emit-rtl.c (maybe_set_max_label_num): New function.
22942 * emit-rtl.h (maybe_set_max_label_num): New decl.
22943 * function.c (instantiate_decls): Guard call to
22944 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
22945 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
22947 * gensupport.c (gen_reader::gen_reader): Pass "false"
22948 for new "compact" param of rtx_reader.
22949 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
22950 rather than an empty string for NULL strings.
22951 * read-md.c: Potentially include config.h rather than bconfig.h.
22952 Wrap include of errors.h with #ifdef GENERATOR_FILE.
22953 (have_error): New global, copied from errors.c.
22954 (md_reader::read_name): Rename to...
22955 (md_reader::read_name_1): ...this, adding "out_loc" param,
22956 and converting "missing name or number" to returning false, rather
22958 (md_reader::read_name): Reimplement in terms of read_name_1.
22959 (md_reader::read_name_or_nil): New function.
22960 (md_reader::read_string): Handle "(nil)" by returning NULL.
22961 (md_reader::md_reader): Add new param "compact".
22962 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
22963 (md_reader::read_file): New method.
22964 * read-md.h (md_reader::md_reader): Add new param "compact".
22965 (md_reader::read_file): New method.
22966 (md_reader::is_compact): New accessor.
22967 (md_reader::read_name): Convert return type from void to file_location.
22968 (md_reader::read_name_or_nil): New decl.
22969 (md_reader::read_name_1): New decl.
22970 (md_reader::m_compact): New field.
22971 (noop_reader::noop_reader): Pass "false" for new "compact" param
22973 (rtx_reader::rtx_reader): Add new "compact" param.
22974 (rtx_reader::read_rtx_operand): Make virtual and convert return
22975 type from void to rtx.
22976 (rtx_reader::read_until): New decl.
22977 (rtx_reader::handle_any_trailing_information): New virtual function.
22978 (rtx_reader::postprocess): New virtual function.
22979 (rtx_reader::finalize_string): New virtual function.
22980 (rtx_reader::m_in_call_function_usage): New field.
22981 (rtx_reader::m_reuse_rtx_by_id): New field.
22982 * read-rtl-function.c: New file.
22983 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
22984 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
22985 (selftest::verify_three_block_rtl_cfg): New decl.
22986 * read-rtl-function.h: New file.
22987 * read-rtl.c: Potentially include config.h rather than bconfig.h.
22988 For host, include function.h, memmodel.h, and emit-rtl.h.
22989 (one_time_initialization): New function.
22990 (struct compact_insn_name): New struct.
22991 (compact_insn_names): New array.
22992 (find_code): Handle insn codes in compact dumps.
22993 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
22994 (bind_subst_iter_and_attr): Likewise.
22995 (add_condition_to_string): Likewise.
22996 (add_condition_to_rtx): Likewise.
22997 (apply_attribute_uses): Likewise.
22998 (add_current_iterators): Likewise.
22999 (apply_iterators): Likewise.
23000 (initialize_iterators): Guard usage of apply_subst_iterator with
23001 #ifdef GENERATOR_FILE.
23002 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23003 (md_reader::read_mapping): Likewise.
23004 (add_define_attr_for_define_subst): Likewise.
23005 (add_define_subst_attr): Likewise.
23006 (read_subst_mapping): Likewise.
23007 (check_code_iterator): Likewise.
23008 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23010 (one_time_initialization): New function.
23011 (rtx_reader::read_until): New method.
23012 (read_flags): New function.
23013 (parse_reg_note_name): New function.
23014 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23015 Handle reuse_rtx ids.
23016 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23017 Add parsing support for RTL dumps, mirroring the special-cases in
23018 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23019 values, and calling handle_any_trailing_information.
23020 (rtx_reader::read_rtx_operand): Convert return type from void
23021 to rtx, returning return_rtx. Handle case 'e'. Call
23022 finalize_string on XSTR and XTMPL fields.
23023 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23024 "(nil)" values were omitted. Call the postprocess vfunc on the
23026 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23027 class ctor. Initialize m_in_call_function_usage. Call
23028 one_time_initialization.
23029 * rtl-tests.c (selftest::test_uncond_jump): Call
23030 set_new_first_and_last_insn.
23031 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23032 * selftest-rtl.c: New file.
23033 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23034 (selftest::get_insn_by_uid): New decl.
23035 * selftest-run-tests.c (selftest::run_tests): Call
23036 read_rtl_function_c_tests.
23037 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23038 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23041 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23043 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23044 operands in a special way. Assert that pos+len <= mode precision.
23046 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23048 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23049 3 argument Alias with unlimited for the negative form.
23050 (fno-vect-cost-model): Removed.
23052 2017-01-05 Martin Liska <mliska@suse.cz>
23054 * hsa-gen.c (gen_hsa_divmod): New function.
23055 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23057 2017-01-05 Martin Liska <mliska@suse.cz>
23060 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23063 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23065 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23066 small constant length operands.
23068 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23070 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23071 between loop iterations.
23073 2017-01-05 Martin Liska <mliska@suse.cz>
23076 * gimplify.c (gimplify_decl_expr): Compare to
23077 asan_poisoned_variables instread of checking flags.
23078 (gimplify_target_expr): Likewise.
23079 (gimplify_expr): Likewise.
23080 (gimplify_function_tree): Conditionally initialize
23081 asan_poisoned_variables.
23083 2017-01-04 Jeff Law <law@redhat.com>
23085 PR tree-optimizatin/78812
23086 * rtl.h (contains_mem_rtx_p): Prototype.
23087 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23088 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23089 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23090 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23091 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23093 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23095 * input.c (assert_char_at_range): Default-initialize actual_range.
23097 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23099 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23100 to match the caller.
23102 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23104 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23105 insns after final jump in test to emit dummy move.
23107 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23109 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23110 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23112 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23114 * multiple_target.c (create_dispatcher_calls): Init e_next.
23115 * tree-ssa-loop-split.c (split_loop): Init border.
23116 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23119 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23124 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23125 (altivec_register_operand): Do not return true if the operand
23126 contains a SUBREG mixing SImode and SFmode.
23127 (vsx_register_operand): Likewise.
23128 (vsx_reg_sfsubreg_ok): New predicate.
23129 (vfloat_operand): Do not return true if the operand contains a
23130 SUBREG mixing SImode and SFmode.
23131 (vint_operand): Likewise.
23132 (vlogical_operand): Likewise.
23133 (gpc_reg_operand): Likewise.
23134 (int_reg_operand): Likewise.
23135 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23136 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23137 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23139 (rs6000_emit_move_si_sf_subreg): New helper function.
23140 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23141 fixup SUBREGs involving SImode and SFmode.
23142 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23143 numbers for the new peephole2 optimization.
23144 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23145 the GLIBC math library that do AND/IOR/XOR operations on single
23146 precision floating point.
23147 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23148 target macros to say whether we need to avoid SUBREGs mixing
23150 (TARGET_ALLOW_SF_SUBREG): Likewise.
23151 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23152 (UNSPEC_SI_FROM_SF): Likewise.
23153 (iorxor): Change spacing.
23154 (and_ior_xor): New iterator for AND, IOR, and XOR.
23155 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23156 (movdi_from_sf_zero_ext): Likewise.
23157 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23158 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23159 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23160 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23161 (fms<mode>4): Likewise.
23162 (fnma<mode>4): Likewise.
23163 (fnms<mode>4): Likewise.
23164 (nfma<mode>4): Likewise.
23165 (nfms<mode>4): Likewise.
23167 2017-01-04 Marek Polacek <polacek@redhat.com>
23170 * doc/invoke.texi: Document -Wpointer-compare.
23172 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23174 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23177 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23178 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23181 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23184 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23185 documentation of the powerpc_popcntb_ok attribute.
23186 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23187 code to issue warning messages if a requested CPU configuration is
23188 not supported by the binary (assembler and loader) toolchain.
23189 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23190 made to define a built-in function that has been disabled.
23191 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23192 made to define a built-in function that has been disabled.
23193 (altivec_init_builtins): Add comment explaining why definition
23194 of the DST built-in functions is not preceded by an assertion
23195 check. Add assertions to prevent ICE if attempts are made to
23196 define an altivec predicate or an abs* built-in function that has
23198 (htm_init_builtins): Add comment explaining why definition of the
23199 htm built-in functions is not preceded by an assertion check.
23201 2017-01-04 Jeff Law <law@redhat.com>
23203 PR tree-optimizatin/67955
23204 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23205 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23206 the points-to solution does not include pt_null. Use DECL_PT_UID
23209 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23211 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23212 Use gen_int_mode instead of gen_lopwart for const_int operands.
23214 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23216 PR tree-optimization/71563
23217 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23218 out of range value - has low bits known to be zero.
23220 2017-01-04 Alan Modra <amodra@gmail.com>
23222 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23223 * configure: Regenerate.
23224 * config.in: Regenerate.
23226 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23229 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23230 a substring of the message, but strcmp with the whole message. Ifdef
23231 ENABLE_NLS, translate the message first using dgettext.
23233 2017-01-03 Jeff Law <law@redhat.com>
23235 PR tree-optimizatin/78856
23236 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23237 (mark_threaded_blocks): Remove code to truncate thread paths that
23238 cross multiple loop headers. Instead invalidate the cached loop
23239 iteration information and handle case of a thread path walking
23240 into an irreducible region.
23242 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23245 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23246 assertions. Add support for doing the signbit if the IEEE 128-bit
23247 floating point value is in a GPR.
23248 * config/rs6000/rs6000.md (Fsignbit): Delete.
23249 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23250 Update the length attribute if the value is in a GPR.
23251 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23252 the sign or zero extension instruction, since the value is always 0/1.
23253 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23256 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23257 extracting SImode to a GPR register so that we can generate a
23258 store, limit the vector to be in a traditional Altivec register
23259 for the vextuwrx instruction.
23261 2017-01-03 Ian Lance Taylor <iant@google.com>
23263 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23265 2017-01-03 Martin Sebor <msebor@redhat.com>
23267 PR tree-optimization/78696
23268 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23269 precision. Use MPFR for %f for greater fidelity. Correct handling
23271 (pass_sprintf_length::compute_format_length): Set width and precision
23272 specified by asrerisk to void_node for vararg functions.
23273 (try_substitute_return_value): Adjust dump output.
23275 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23277 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23279 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23281 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23282 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23283 -mlra/-mno-lra was passed to the compiler.
23285 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23287 PR rtl-optimization/65618
23288 * emit-rtl.c (try_split): Move initialization of "before" and
23289 "after" to just before the call to emit_insn_after_setloc.
23291 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23293 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23295 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23297 * dwarf2out.c (gen_enumeration_type_die): When
23298 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23300 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23302 PR tree-optimization/78965
23303 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23304 Change first argument from const call_info & to call_info &. For %n
23305 set info.nowrite to false.
23307 PR middle-end/78901
23308 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23309 possibly throwing calls.
23311 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23312 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23313 and fns handling, rather than in a separate case SSA_NAME.
23315 2017-01-02 Jeff Law <law@redhat.com>
23317 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23318 fixes for first_period and second_period variables.
23320 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23323 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23324 (*insvqi_1): New insn pattern.
23325 (*insvqi_1_mem_rex64): Ditto.
23326 (*insvqi_2): Ditto.
23327 (*insvqi_3): Rename from *insvqi.
23329 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23331 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23333 * doc/cfg.texi (Edges): Remove reference to Java.
23334 (Maintaining the CFG): Ditto.
23336 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23338 PR middle-end/77674
23339 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23340 transparent aliases.
23342 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23344 PR middle-end/77484
23345 * predict.def (PRED_CALL): Update hitrate.
23346 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23347 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23348 into direct/indirect/polymorphic variants.
23350 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23352 Update copyright years.
23354 * gcc.c (process_command): Update copyright notice dates.
23355 * gcov-dump.c (print_version): Ditto.
23356 * gcov.c (print_version): Ditto.
23357 * gcov-tool.c (print_version): Ditto.
23358 * gengtype.c (create_file): Ditto.
23359 * doc/cpp.texi: Bump @copying's copyright year.
23360 * doc/cppinternals.texi: Ditto.
23361 * doc/gcc.texi: Ditto.
23362 * doc/gccint.texi: Ditto.
23363 * doc/gcov.texi: Ditto.
23364 * doc/install.texi: Ditto.
23365 * doc/invoke.texi: Ditto.
23367 Copyright (C) 2017 Free Software Foundation, Inc.
23369 Copying and distribution of this file, with or without modification,
23370 are permitted in any medium without royalty provided the copyright
23371 notice and this notice are preserved.