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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6 the mode argument to scalar_int_mode.
7 (do_jump_by_parts_zero_rtx): Likewise.
8 (do_jump_by_parts_equality_rtx): Likewise.
9 (do_jump_by_parts_greater): Take a mode argument.
10 (do_jump_by_parts_equality): Likewise.
11 (do_jump_1): Update calls accordingly.
13 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14 Alan Hayward <alan.hayward@arm.com>
15 David Sherwood <david.sherwood@arm.com>
17 * is-a.h (safe_dyn_cast): New function.
18 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
19 (jump_table_for_label): Likewise.
20 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
21 instead of an rtx_insn *.
22 (shorten_branches): Use dyn_cast instead of LABEL_P and
23 JUMP_TABLE_DATA_P. Use jump_table_for_label and
24 rtx_jump_table_data::get_data_mode.
25 (final_scan_insn): Likewise.
27 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
28 Alan Hayward <alan.hayward@arm.com>
29 David Sherwood <david.sherwood@arm.com>
31 * combine.c (try_combine): Use is_a <scalar_int_mode> when
32 trying to combine a full-register integer set with a subreg
35 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
36 Alan Hayward <alan.hayward@arm.com>
37 David Sherwood <david.sherwood@arm.com>
39 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
40 that is always either address_mode or pointer_mode.
42 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
43 Alan Hayward <alan.hayward@arm.com>
44 David Sherwood <david.sherwood@arm.com>
46 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
47 when the two are known to be equal.
49 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
50 Alan Hayward <alan.hayward@arm.com>
51 David Sherwood <david.sherwood@arm.com>
53 * simplify-rtx.c (simplify_const_unary_operation): Use
54 is_a <scalar_int_mode> instead of checking for a nonzero
55 precision. Forcibly convert op_mode to a scalar_int_mode
56 in that case. More clearly differentiate the operand and
57 result modes and use the former when deciding what the value
58 of a count-bits operation should be. Use is_int_mode instead
59 of checking for a MODE_INT. Remove redundant check for whether
60 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
67 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
68 (widen_bswap): Likewise.
69 (expand_parity): Likewise.
70 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
71 (expand_ffs): Likewise.
72 (epand_unop): Check for scalar integer modes before calling the
75 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
76 Alan Hayward <alan.hayward@arm.com>
77 David Sherwood <david.sherwood@arm.com>
79 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
81 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
88 value_mode arguments. Use scalar_int_mode internally.
89 (store_bit_field_1): Rename the new integer mode from imode
90 to op0_mode and use it instead of GET_MODE (op0). Update calls
91 to store_split_bit_field, store_bit_field_using_insv and
92 store_fixed_bit_field.
93 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
94 Use scalar_int_mode internally. Use a bit count rather than a mode
95 when calculating the largest bit size for get_best_mode.
96 Update calls to store_split_bit_field and store_fixed_bit_field_1.
97 (store_fixed_bit_field_1): Add mode and value_mode arguments.
98 Remove assertion that OP0 has a scalar integer mode.
99 (store_split_bit_field): Add op0_mode and value_mode arguments.
100 Update calls to extract_fixed_bit_field.
101 (extract_bit_field_using_extv): Add an op0_mode argument.
102 Use scalar_int_mode internally.
103 (extract_bit_field_1): Rename the new integer mode from imode to
104 op0_mode and use it instead of GET_MODE (op0). Update calls to
105 extract_split_bit_field, extract_bit_field_using_extv and
106 extract_fixed_bit_field.
107 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
108 to extract_split_bit_field and extract_fixed_bit_field_1.
109 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
110 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
112 (extract_split_bit_field): Add an op0_mode argument. Update call
113 to extract_fixed_bit_field.
115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
116 Alan Hayward <alan.hayward@arm.com>
117 David Sherwood <david.sherwood@arm.com>
119 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
120 * explow.c (hard_function_value): Likewise.
121 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
122 convert_to_mode call outside the loop.
123 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
124 for the mode iterator. Require the mode specified by max_pieces
126 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
128 (copy_blkmode_to_reg): Likewise.
129 (set_storage_via_setmem): Likewise.
130 * optabs.c (prepare_cmp_insn): Likewise.
131 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
132 * stor-layout.c (finish_bitfield_representative): Likewise.
134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
135 Alan Hayward <alan.hayward@arm.com>
136 David Sherwood <david.sherwood@arm.com>
138 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
139 * expr.c (convert_move): Use them.
140 (convert_modes): Likewise.
141 (store_expr_with_bounds): Likewise.
143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
144 Alan Hayward <alan.hayward@arm.com>
145 David Sherwood <david.sherwood@arm.com>
147 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
148 parameter for the mode of "x". Remove the "known_x", "known_mode"
149 and "known_ret" arguments. Change the type of the mode argument
151 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
152 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
153 (reg_num_sign_bit_copies_for_combine): Likewise.
154 * rtlanal.c (nonzero_bits1): Likewise.
155 (num_sign_bit_copies1): Likewise.
156 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
157 (reg_num_sign_bit_copies_general): Likewise.
158 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
159 (reg_nonzero_bits_general): Likewise.
161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
162 Alan Hayward <alan.hayward@arm.com>
163 David Sherwood <david.sherwood@arm.com>
165 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
166 than in subroutines. Return 1 for non-integer modes.
167 (cached_num_sign_bit_copies): Change the type of the mode parameter
169 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
170 classes. Handle CONST_INT_P first and then check whether X also
171 has a scalar integer mode. Check the same thing for inner registers
172 of a SUBREG and for values that are being extended or truncated.
174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
175 Alan Hayward <alan.hayward@arm.com>
176 David Sherwood <david.sherwood@arm.com>
178 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
179 in subroutines. Return the mode mask for non-integer modes.
180 (cached_nonzero_bits): Change the type of the mode parameter
182 (nonzero_bits1): Likewise. Remove early exit for other mode
183 classes. Handle CONST_INT_P first and then check whether X
184 also has a scalar integer mode.
186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
187 Alan Hayward <alan.hayward@arm.com>
188 David Sherwood <david.sherwood@arm.com>
190 * combine.c (extract_left_shift): Add a mode argument and update
192 (make_compound_operation_int): Change the type of the mode parameter
193 to scalar_int_mode and update the call to extract_left_shift.
195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
196 Alan Hayward <alan.hayward@arm.com>
197 David Sherwood <david.sherwood@arm.com>
199 * combine.c (simplify_and_const_int): Change the type of the mode
200 parameter to scalar_int_mode.
201 (simplify_and_const_int_1): Likewise. Update recursive call.
203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
204 Alan Hayward <alan.hayward@arm.com>
205 David Sherwood <david.sherwood@arm.com>
207 * combine.c (simplify_compare_const): Check that the mode is a
208 scalar_int_mode (rather than VOIDmode) before testing its
210 (simplify_comparison): Move COMPARISON_P handling out of the
211 loop and restrict the latter part of the loop to scalar_int_modes.
212 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
213 and when considering SUBREG_REGs. Use is_int_mode instead of
214 checking GET_MODE_CLASS against MODE_INT.
216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
217 Alan Hayward <alan.hayward@arm.com>
218 David Sherwood <david.sherwood@arm.com>
220 * combine.c (try_widen_shift_mode): Move check for equal modes to...
221 (simplify_shift_const_1): ...here. Use scalar_int_mode for
222 shift_unit_mode and for modes involved in scalar shifts.
224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
225 Alan Hayward <alan.hayward@arm.com>
226 David Sherwood <david.sherwood@arm.com>
228 * combine.c (force_int_to_mode): New function, split out from...
229 (force_to_mode): ...here. Keep xmode up-to-date and use it
230 instead of GET_MODE (x).
232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
233 Alan Hayward <alan.hayward@arm.com>
234 David Sherwood <david.sherwood@arm.com>
236 * optabs-query.h (extraction_insn::struct_mode): Change type to
237 opt_scalar_int_mode and update comment.
238 (extraction_insn::field_mode): Change type to scalar_int_mode.
239 (extraction_insn::pos_mode): Likewise.
240 * combine.c (make_extraction): Update accordingly.
241 * optabs-query.c (get_traditional_extraction_insn): Likewise.
242 (get_optab_extraction_insn): Likewise.
243 * recog.c (simplify_while_replacing): Likewise.
244 * expmed.c (narrow_bit_field_mem): Change the type of the mode
245 parameter to opt_scalar_int_mode.
247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
248 Alan Hayward <alan.hayward@arm.com>
249 David Sherwood <david.sherwood@arm.com>
251 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
252 to a scalar_int_mode instead of a machine_mode.
253 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
254 (get_best_mode): Return a boolean and use a pointer argument to store
255 the selected mode. Replace the limit mode parameter with a bit limit.
256 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
257 for the values returned by bit_field_mode_iterator::next_mode.
258 (store_bit_field): Update call to get_best_mode.
259 (store_fixed_bit_field): Likewise.
260 (extract_fixed_bit_field): Likewise.
261 * expr.c (optimize_bitfield_assignment_op): Likewise.
262 * fold-const.c (optimize_bit_field_compare): Likewise.
263 (fold_truth_andor_1): Likewise.
264 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
265 Update for new type of m_mode.
266 (get_best_mode): As above.
268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
269 Alan Hayward <alan.hayward@arm.com>
270 David Sherwood <david.sherwood@arm.com>
272 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
273 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
274 (store_bit_field): Check is_a <scalar_int_mode> before calling
275 strict_volatile_bitfield_p.
276 (extract_bit_field): Likewise.
278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
279 Alan Hayward <alan.hayward@arm.com>
280 David Sherwood <david.sherwood@arm.com>
282 * target.def (cstore_mode): Return a scalar_int_mode.
283 * doc/tm.texi: Regenerate.
284 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
285 * targhooks.h (default_cstore_mode): Likewise.
286 * targhooks.c (default_cstore_mode): Likewise, using a forced
288 * expmed.c (emit_cstore): Expect the target of the cstore to be
291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
292 Alan Hayward <alan.hayward@arm.com>
293 David Sherwood <david.sherwood@arm.com>
295 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
296 * builtins.c (expand_builtin_signbit): Use it.
297 * cfgexpand.c (expand_debug_expr): Likewise.
298 * dojump.c (do_jump): Likewise.
299 (do_compare_and_jump): Likewise.
300 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
301 * expmed.c (make_tree): Likewise.
302 * expr.c (expand_expr_real_2): Likewise.
303 (expand_expr_real_1): Likewise.
304 (try_casesi): Likewise.
305 * fold-const-call.c (fold_const_call_ss): Likewise.
306 * fold-const.c (unextend): Likewise.
307 (extract_muldiv_1): Likewise.
308 (fold_single_bit_test): Likewise.
309 (native_encode_int): Likewise.
310 (native_encode_string): Likewise.
311 (native_interpret_int): Likewise.
312 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
313 * internal-fn.c (expand_addsub_overflow): Likewise.
314 (expand_neg_overflow): Likewise.
315 (expand_mul_overflow): Likewise.
316 (expand_arith_overflow): Likewise.
317 * match.pd: Likewise.
318 * stor-layout.c (layout_type): Likewise.
319 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
320 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
321 * tree-ssanames.c (get_range_info): Likewise.
322 * tree-switch-conversion.c (array_value_type) Likewise.
323 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
324 (vect_recog_divmod_pattern): Likewise.
325 (vect_recog_mixed_size_cond_pattern): Likewise.
326 * tree-vrp.c (extract_range_basic): Likewise.
327 (simplify_float_conversion_using_ranges): Likewise.
328 * tree.c (int_fits_type_p): Likewise.
329 * ubsan.c (instrument_bool_enum_load): Likewise.
330 * varasm.c (mergeable_string_section): Likewise.
331 (narrowing_initializer_constant_valid_p): Likewise.
332 (output_constant): Likewise.
334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
335 Alan Hayward <alan.hayward@arm.com>
336 David Sherwood <david.sherwood@arm.com>
338 * machmode.h (NARROWEST_INT_MODE): New macro.
339 * expr.c (alignment_for_piecewise_move): Use it instead of
340 GET_CLASS_NARROWEST_MODE (MODE_INT).
341 (push_block): Likewise.
342 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
344 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
347 Alan Hayward <alan.hayward@arm.com>
348 David Sherwood <david.sherwood@arm.com>
350 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
352 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
353 * optabs.c (expand_unop): Likewise.
355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
356 Alan Hayward <alan.hayward@arm.com>
357 David Sherwood <david.sherwood@arm.com>
359 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
360 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
361 (popcount_loc_descriptor): Likewise.
362 (bswap_loc_descriptor): Likewise.
363 (rotate_loc_descriptor): Likewise.
364 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
365 calling the functions above.
367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
368 Alan Hayward <alan.hayward@arm.com>
369 David Sherwood <david.sherwood@arm.com>
371 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
373 (try_combine): Likewise.
374 (simplify_if_then_else): Likewise.
375 * cse.c (cse_insn): Likewise.
376 * dwarf2out.c (mem_loc_descriptor): Likewise.
377 * emit-rtl.c (gen_lowpart_common): Likewise.
378 * simplify-rtx.c (simplify_truncation): Likewise.
379 (simplify_binary_operation_1): Likewise.
380 (simplify_const_relational_operation): Likewise.
381 (simplify_ternary_operation): Likewise.
382 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
385 Alan Hayward <alan.hayward@arm.com>
386 David Sherwood <david.sherwood@arm.com>
388 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
389 * reload.c (push_reload): Likewise.
390 (find_reloads): Likewise.
392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
393 Alan Hayward <alan.hayward@arm.com>
394 David Sherwood <david.sherwood@arm.com>
396 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
397 (make_compound_operation_int): Likewise.
398 (change_zero_ext): Likewise.
399 * expr.c (convert_move): Likewise.
400 (convert_modes): Likewise.
401 * fwprop.c (forward_propagate_subreg): Likewise.
402 * loop-iv.c (get_biv_step_1): Likewise.
403 * optabs.c (widen_operand): Likewise.
404 * postreload.c (move2add_valid_value_p): Likewise.
405 * recog.c (simplify_while_replacing): Likewise.
406 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
407 (simplify_binary_operation_1): Likewise. Remove redundant
410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
411 Alan Hayward <alan.hayward@arm.com>
412 David Sherwood <david.sherwood@arm.com>
414 * combine.c (combine_simplify_rtx): Add checks for
415 is_a <scalar_int_mode>.
416 (simplify_if_then_else): Likewise.
417 (make_field_assignment): Likewise.
418 (simplify_comparison): Likewise.
419 * ifcvt.c (noce_try_bitop): Likewise.
420 * loop-invariant.c (canonicalize_address_mult): Likewise.
421 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
424 Alan Hayward <alan.hayward@arm.com>
425 David Sherwood <david.sherwood@arm.com>
427 * simplify-rtx.c (simplify_binary_operation_1): Use
428 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
431 Alan Hayward <alan.hayward@arm.com>
432 David Sherwood <david.sherwood@arm.com>
434 * wide-int.h (int_traits<unsigned char>) New class.
435 (int_traits<unsigned short>) Likewise.
436 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
437 Use GET_MODE_UNIT_PRECISION and remove redundant test for
439 * combine.c (set_nonzero_bits_and_sign_copies): Use
440 is_a <scalar_int_mode>.
441 (find_split_point): Likewise.
442 (combine_simplify_rtx): Likewise.
443 (simplify_logical): Likewise.
444 (expand_compound_operation): Likewise.
445 (expand_field_assignment): Likewise.
446 (make_compound_operation): Likewise.
447 (extended_count): Likewise.
448 (change_zero_ext): Likewise.
449 (simplify_comparison): Likewise.
450 * dwarf2out.c (scompare_loc_descriptor): Likewise.
451 (ucompare_loc_descriptor): Likewise.
452 (minmax_loc_descriptor): Likewise.
453 (mem_loc_descriptor): Likewise.
454 (loc_descriptor): Likewise.
455 * expmed.c (init_expmed_one_mode): Likewise.
456 * lra-constraints.c (lra_constraint_offset): Likewise.
457 * optabs.c (prepare_libcall_arg): Likewise.
458 * postreload.c (move2add_note_store): Likewise.
459 * reload.c (operands_match_p): Likewise.
460 * rtl.h (load_extend_op): Likewise.
461 * rtlhooks.c (gen_lowpart_general): Likewise.
462 * simplify-rtx.c (simplify_truncation): Likewise.
463 (simplify_unary_operation_1): Likewise.
464 (simplify_binary_operation_1): Likewise.
465 (simplify_const_binary_operation): Likewise.
466 (simplify_const_relational_operation): Likewise.
467 (simplify_subreg): Likewise.
468 * stor-layout.c (bitwise_mode_for_mode): Likewise.
469 * var-tracking.c (adjust_mems): Likewise.
470 (prepare_call_arguments): Likewise.
472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
473 Alan Hayward <alan.hayward@arm.com>
474 David Sherwood <david.sherwood@arm.com>
476 * machmode.h (is_int_mode): New fuction.
477 * combine.c (find_split_point): Use it.
478 (combine_simplify_rtx): Likewise.
479 (simplify_if_then_else): Likewise.
480 (simplify_set): Likewise.
481 (simplify_shift_const_1): Likewise.
482 (simplify_comparison): Likewise.
483 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
484 * cse.c (notreg_cost): Likewise.
485 (cse_insn): Likewise.
486 * cselib.c (cselib_lookup_1): Likewise.
487 * dojump.c (do_jump_1): Likewise.
488 (do_compare_rtx_and_jump): Likewise.
489 * dse.c (get_call_args): Likewise.
490 * dwarf2out.c (rtl_for_decl_init): Likewise.
491 (native_encode_initializer): Likewise.
492 * expmed.c (emit_store_flag_1): Likewise.
493 (emit_store_flag): Likewise.
494 * expr.c (convert_modes): Likewise.
495 (store_field): Likewise.
496 (expand_expr_real_1): Likewise.
497 * fold-const.c (fold_read_from_constant_string): Likewise.
498 * gimple-ssa-sprintf.c (get_format_string): Likewise.
499 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
500 * optabs.c (expand_binop): Likewise.
501 (expand_unop): Likewise.
502 (expand_abs_nojump): Likewise.
503 (expand_one_cmpl_abs_nojump): Likewise.
504 * simplify-rtx.c (mode_signbit_p): Likewise.
505 (val_signbit_p): Likewise.
506 (val_signbit_known_set_p): Likewise.
507 (val_signbit_known_clear_p): Likewise.
508 (simplify_relational_operation_1): Likewise.
509 * tree.c (vector_type_mode): Likewise.
511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
512 Alan Hayward <alan.hayward@arm.com>
513 David Sherwood <david.sherwood@arm.com>
515 * machmode.h (smallest_mode_for_size): Fix formatting.
516 (smallest_int_mode_for_size): New function.
517 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
518 instead of smallest_mode_for_size.
519 * combine.c (make_extraction): Likewise.
520 * config/arc/arc.c (arc_expand_movmem): Likewise.
521 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
522 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
523 * config/s390/s390.c (s390_expand_insv): Likewise.
524 * config/sparc/sparc.c (assign_int_registers): Likewise.
525 * config/spu/spu.c (spu_function_value): Likewise.
526 (spu_function_arg): Likewise.
527 * coverage.c (get_gcov_type): Likewise.
528 (get_gcov_unsigned_t): Likewise.
529 * dse.c (find_shift_sequence): Likewise.
530 * expmed.c (store_bit_field_1): Likewise.
531 * expr.c (convert_move): Likewise.
532 (store_field): Likewise.
533 * internal-fn.c (expand_arith_overflow): Likewise.
534 * optabs-query.c (get_best_extraction_insn): Likewise.
535 * optabs.c (expand_twoval_binop_libfunc): Likewise.
536 * stor-layout.c (layout_type): Likewise.
537 (initialize_sizetypes): Likewise.
538 * targhooks.c (default_get_mask_mode): Likewise.
539 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
542 Alan Hayward <alan.hayward@arm.com>
543 David Sherwood <david.sherwood@arm.com>
545 * machmode.h (opt_mode::else_blk): New function.
546 (int_mode_for_mode): Declare.
547 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
548 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
550 * cfgexpand.c (expand_debug_expr): Likewise.
551 * combine.c (gen_lowpart_or_truncate): Likewise.
552 (gen_lowpart_for_combine): Likewise.
553 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
554 * config/avr/avr.c (avr_to_int_mode): Likewise.
555 (avr_out_plus_1): Likewise.
556 (avr_out_plus): Likewise.
557 (avr_out_round): Likewise.
558 * config/i386/i386.c (ix86_split_to_parts): Likewise.
559 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
560 (s390_expand_vcond): Likewise.
561 * config/spu/spu.c (spu_split_immediate): Likewise.
562 (spu_expand_mov): Likewise.
563 * dse.c (get_stored_val): Likewise.
564 * expmed.c (store_bit_field_1): Likewise.
565 (convert_extracted_bit_field): Use int_mode_for_mode instead of
567 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
568 (extract_low_bits): Likewise.
569 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
570 handling rather than repeating the check.
571 (emit_group_store): Likewise.
572 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
573 * optabs.c (expand_absneg_bit): Likewise.
574 (expand_copysign_absneg): Likewise.
575 (expand_copysign_bit): Likewise.
576 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
577 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
578 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
579 * var-tracking.c (prepare_call_arguments): Likewise.
580 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
581 int_mode_for_mode instead of mode_for_size.
582 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
585 Alan Hayward <alan.hayward@arm.com>
586 David Sherwood <david.sherwood@arm.com>
588 * machmode.h (int_mode_for_size): New function.
589 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
590 instead of mode_for_size.
591 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
593 * combine.c (expand_field_assignment): Use int_mode_for_size
594 instead of mode_for_size.
595 (make_extraction): Likewise.
596 (simplify_shift_const_1): Likewise.
597 (simplify_comparison): Likewise.
598 * dojump.c (do_jump): Likewise.
599 * dwarf2out.c (mem_loc_descriptor): Likewise.
600 * emit-rtl.c (init_derived_machine_modes): Likewise.
601 * expmed.c (flip_storage_order): Likewise.
602 (convert_extracted_bit_field): Likewise.
603 * expr.c (copy_blkmode_from_reg): Likewise.
604 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
605 * internal-fn.c (expand_mul_overflow): Likewise.
606 * lower-subreg.c (simple_move): Likewise.
607 * optabs-libfuncs.c (init_optabs): Likewise.
608 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
609 * tree.c (vector_type_mode): Likewise.
610 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
611 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
612 * tree-vect-generic.c (expand_vector_parallel): Likewise.
613 * tree-vect-stmts.c (vectorizable_load): Likewise.
614 (vectorizable_store): Likewise.
616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
617 Alan Hayward <alan.hayward@arm.com>
618 David Sherwood <david.sherwood@arm.com>
620 * coretypes.h (pod_mode): New type.
621 (scalar_int_mode_pod): New typedef.
622 * machmode.h (pod_mode): New class.
623 (int_n_data_t::m): Change type to scalar_int_mode_pod.
624 * genmodes.c (emit_mode_int_n): Update accordingly.
625 * lower-subreg.h (target_lower_subreg): Change type to
627 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
631 Alan Hayward <alan.hayward@arm.com>
632 David Sherwood <david.sherwood@arm.com>
634 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
635 machine_mode to scalar_int_mode.
636 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
637 (rs6000_option_override_internal): Remove cast to int.
638 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
639 machine_mode to scalar_int_mode.
640 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
641 (rs6000_option_override_internal): Remove cast to int.
642 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
643 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
645 * config/s390/s390.c (s390_expand_builtin): Likewise.
646 * coretypes.h (scalar_int_mode): New type.
647 (opt_scalar_int_mode): New typedef.
648 * machmode.h (scalar_int_mode): New class.
649 (scalar_int_mode::includes_p): New function.
650 (byte_mode): Change type to scalar_int_mode.
651 (word_mode): Likewise.
652 (ptr_mode): Likewise.
653 * emit-rtl.c (byte_mode): Likewise.
654 (word_mode): Likewise.
655 (ptr_mode): Likewise.
656 (init_derived_machine_modes): Update accordingly.
657 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
658 and MODE_PARTIAL_INT.
659 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
663 Alan Hayward <alan.hayward@arm.com>
664 David Sherwood <david.sherwood@arm.com>
666 * target.def (libgcc_floating_mode_supported_p): Take a
668 * doc/tm.texi: Regenerate.
669 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
671 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
672 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
676 Alan Hayward <alan.hayward@arm.com>
677 David Sherwood <david.sherwood@arm.com>
679 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
680 * doc/tm.texi: Regenerate.
681 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
682 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
683 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
684 * targhooks.h (default_floatn_mode): Likewise.
685 * targhooks.c (default_floatn_mode): Likewise.
686 * tree.c (build_common_tree_nodes): Update accordingly.
688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
689 Alan Hayward <alan.hayward@arm.com>
690 David Sherwood <david.sherwood@arm.com>
692 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
693 (mode_iterator::iterate_p): Likewise.
694 (mode_iterator::get_wider): Likewise.
695 * expr.c (init_expr_target): Use opt_scalar_float_mode.
697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
698 Alan Hayward <alan.hayward@arm.com>
699 David Sherwood <david.sherwood@arm.com>
701 * coretypes.h (opt_scalar_float_mode): New typedef.
702 * machmode.h (float_mode_for_size): New function.
703 * emit-rtl.c (double_mode): Delete.
704 (init_emit_once): Use float_mode_for_size.
705 * stor-layout.c (layout_type): Likewise.
706 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
709 Alan Hayward <alan.hayward@arm.com>
710 David Sherwood <david.sherwood@arm.com>
712 * output.h (assemble_real): Take a scalar_float_mode.
713 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
714 * config/arm/arm.md (consttable_4): Likewise.
715 (consttable_8): Likewise.
716 (consttable_16): Likewise.
717 * config/mips/mips.md (consttable_float): Likewise.
718 * config/s390/s390.c (s390_output_pool_entry): Likewise.
719 * varasm.c (assemble_real): Take a scalar_float_mode.
720 (output_constant_pool_2): Update accordingly.
721 (output_constant): Likewise.
723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
724 Alan Hayward <alan.hayward@arm.com>
725 David Sherwood <david.sherwood@arm.com>
727 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
728 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
729 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
730 (native_encode_real): Likewise.
731 (native_interpret_real): Likewise.
732 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
733 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
736 Alan Hayward <alan.hayward@arm.com>
737 David Sherwood <david.sherwood@arm.com>
739 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
740 <scalar_float_mode>. Simplify.
741 (gen_extend_conv_libfunc): Likewise.
743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
744 Alan Hayward <alan.hayward@arm.com>
745 David Sherwood <david.sherwood@arm.com>
747 * coretypes.h (scalar_float_mode): New type.
748 * machmode.h (mode_traits::from_int): Use machine_mode if
749 USE_ENUM_MODES is defined.
750 (is_a): New function.
752 (dyn_cast): Likewise.
753 (scalar_float_mode): New class.
754 (scalar_float_mode::includes_p): New function.
755 (is_float_mode): Likewise.
756 * gdbhooks.py (MachineModePrinter): New class.
757 (build_pretty_printer): Use it for scalar_float_mode.
758 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
759 (format_helper::format_helper): Turn into a template.
760 * genmodes.c (get_mode_class): New function.
761 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
762 or machine_mode if none.
763 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
764 as_a <scalar_float_mode>.
765 * dwarf2out.c (mem_loc_descriptor): Likewise.
766 (insert_float): Likewise.
767 (add_const_value_attribute): Likewise.
768 * simplify-rtx.c (simplify_immed_subreg): Likewise.
769 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
770 (expand_unop): Update accordingly.
771 (expand_abs_nojump): Likewise.
772 (expand_copysign_absneg): Take a scalar_float_mode.
773 (expand_copysign_bit): Likewise.
774 (expand_copysign): Update accordingly.
776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
777 Alan Hayward <alan.hayward@arm.com>
778 David Sherwood <david.sherwood@arm.com>
780 * coretypes.h (opt_mode): New class.
781 * machmode.h (opt_mode): Likewise.
782 (opt_mode::else_void): New function.
783 (opt_mode::require): Likewise.
784 (opt_mode::exists): Likewise.
785 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
786 (GET_MODE_2XWIDER_MODE): Likewise.
787 (mode_iterator::get_wider): Update accordingly.
788 (mode_iterator::get_2xwider): Likewise.
789 (mode_iterator::get_known_wider): Likewise, turning into a template.
790 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
791 forcing a wider mode to exist.
792 * config/cr16/cr16.h (LONG_REG_P): Likewise.
793 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
794 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
795 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
796 * lower-subreg.c (init_lower_subreg): Likewise.
797 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
798 on the final iteration.
799 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
800 a wider mode exists before asking for a move pattern.
801 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
802 forcing a wider mode to exist.
803 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
804 returning false if no such mode exists.
805 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
806 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
807 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
808 Avoid checking for a MODE_INT if we already know the mode is not a
810 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
811 forcing a wider mode to exist.
812 (expmed_mult_highpart_optab): Likewise.
813 (expmed_mult_highpart): Likewise.
814 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
816 * lto-streamer-in.c (lto_input_mode_table): Likewise.
817 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
818 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
819 * internal-fn.c (expand_mul_overflow): Update use of
820 GET_MODE_2XWIDER_MODE.
821 * omp-low.c (omp_clause_aligned_alignment): Likewise.
822 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
824 (convert_plusminus_to_widen): Likewise.
825 * tree-switch-conversion.c (array_value_type): Likewise.
826 * var-tracking.c (emit_note_insn_var_location): Likewise.
827 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
828 Return false inside rather than outside the loop if no wider mode
830 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
831 and GET_MODE_2XWIDER_MODE
832 (can_compare_p): Use else_void.
833 * gdbhooks.py (OptMachineModePrinter): New class.
834 (build_pretty_printer): Use it for opt_mode.
836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
837 Alan Hayward <alan.hayward@arm.com>
838 David Sherwood <david.sherwood@arm.com>
840 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
841 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * machmode.h (mode_traits): New structure.
848 (get_narrowest_mode): New function.
849 (mode_iterator::start): Likewise.
850 (mode_iterator::iterate_p): Likewise.
851 (mode_iterator::get_wider): Likewise.
852 (mode_iterator::get_known_wider): Likewise.
853 (mode_iterator::get_2xwider): Likewise.
854 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
855 (FOR_EACH_MODE): Likewise.
856 (FOR_EACH_MODE_FROM): Likewise.
857 (FOR_EACH_MODE_UNTIL): Likewise.
858 (FOR_EACH_WIDER_MODE): Likewise.
859 (FOR_EACH_2XWIDER_MODE): Likewise.
860 * builtins.c (expand_builtin_strlen): Use new mode iterators.
861 * combine.c (simplify_comparison): Likewise
862 * config/i386/i386.c (type_natural_mode): Likewise.
863 * cse.c (cse_insn): Likewise.
864 * dse.c (find_shift_sequence): Likewise.
865 * emit-rtl.c (init_derived_machine_modes): Likewise.
866 (init_emit_once): Likewise.
867 * explow.c (hard_function_value): Likewise.
868 * expmed.c (extract_fixed_bit_field_1): Likewise.
869 (extract_bit_field_1): Likewise.
870 (expand_divmod): Likewise.
871 (emit_store_flag_1): Likewise.
872 * expr.c (init_expr_target): Likewise.
873 (convert_move): Likewise.
874 (alignment_for_piecewise_move): Likewise.
875 (widest_int_mode_for_size): Likewise.
876 (emit_block_move_via_movmem): Likewise.
877 (copy_blkmode_to_reg): Likewise.
878 (set_storage_via_setmem): Likewise.
879 (compress_float_constant): Likewise.
880 * omp-low.c (omp_clause_aligned_alignment): Likewise.
881 * optabs-query.c (get_best_extraction_insn): Likewise.
882 * optabs.c (expand_binop): Likewise.
883 (expand_twoval_unop): Likewise.
884 (expand_twoval_binop): Likewise.
885 (widen_leading): Likewise.
886 (widen_bswap): Likewise.
887 (expand_parity): Likewise.
888 (expand_unop): Likewise.
889 (prepare_cmp_insn): Likewise.
890 (prepare_float_lib_cmp): Likewise.
891 (expand_float): Likewise.
892 (expand_fix): Likewise.
893 (expand_sfix_optab): Likewise.
894 * postreload.c (move2add_use_add2_insn): Likewise.
895 * reg-stack.c (reg_to_stack): Likewise.
896 * reginfo.c (choose_hard_reg_mode): Likewise.
897 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
898 * stor-layout.c (mode_for_size): Likewise.
899 (smallest_mode_for_size): Likewise.
900 (mode_for_vector): Likewise.
901 (finish_bitfield_representative): Likewise.
902 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
903 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
904 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
905 * var-tracking.c (prepare_call_arguments): Likewise.
907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
908 Alan Hayward <alan.hayward@arm.com>
909 David Sherwood <david.sherwood@arm.com>
911 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
912 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
913 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
914 * machmode.h (mode_size): Move earlier in file.
915 (mode_precision): Likewise.
916 (mode_inner): Likewise.
917 (mode_nunits): Likewise.
918 (mode_unit_size): Likewise.
919 (unit_unit_precision): Likewise.
920 (mode_wider): Likewise.
921 (mode_2xwider): Likewise.
922 (machine_mode): New class.
923 (mode_to_bytes): New function.
924 (mode_to_bits): Likewise.
925 (mode_to_precision): Likewise.
926 (mode_to_inner): Likewise.
927 (mode_to_unit_size): Likewise.
928 (mode_to_unit_precision): Likewise.
929 (mode_to_nunits): Likewise.
930 (GET_MODE_SIZE): Use mode_to_bytes.
931 (GET_MODE_BITSIZE): Use mode_to_bits.
932 (GET_MODE_PRECISION): Use mode_to_precision.
933 (GET_MODE_INNER): Use mode_to_inner.
934 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
935 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
936 (GET_MODE_NUNITS): Use mode_to_nunits.
937 * system.h (ALWAYS_INLINE): New macro.
938 * config/powerpcspe/powerpcspe-c.c
939 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
940 int for arg1_mode and arg2_mode.
942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
943 Alan Hayward <alan.hayward@arm.com>
944 David Sherwood <david.sherwood@arm.com>
946 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
947 Prefix mode names with E_ in case statements.
948 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
949 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
950 (aarch64_split_simd_move): Likewise.
951 (aarch64_gen_storewb_pair): Likewise.
952 (aarch64_gen_loadwb_pair): Likewise.
953 (aarch64_gen_store_pair): Likewise.
954 (aarch64_gen_load_pair): Likewise.
955 (aarch64_get_condition_code_1): Likewise.
956 (aarch64_constant_pool_reload_icode): Likewise.
957 (get_rsqrte_type): Likewise.
958 (get_rsqrts_type): Likewise.
959 (get_recpe_type): Likewise.
960 (get_recps_type): Likewise.
961 (aarch64_gimplify_va_arg_expr): Likewise.
962 (aarch64_simd_container_mode): Likewise.
963 (aarch64_emit_load_exclusive): Likewise.
964 (aarch64_emit_store_exclusive): Likewise.
965 (aarch64_expand_compare_and_swap): Likewise.
966 (aarch64_gen_atomic_cas): Likewise.
967 (aarch64_emit_bic): Likewise.
968 (aarch64_emit_atomic_swap): Likewise.
969 (aarch64_emit_atomic_load_op): Likewise.
970 (aarch64_evpc_trn): Likewise.
971 (aarch64_evpc_uzp): Likewise.
972 (aarch64_evpc_zip): Likewise.
973 (aarch64_evpc_ext): Likewise.
974 (aarch64_evpc_rev): Likewise.
975 (aarch64_evpc_dup): Likewise.
976 (aarch64_gen_ccmp_first): Likewise.
977 (aarch64_gen_ccmp_next): Likewise.
978 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
979 (alpha_emit_xfloating_libcall): Likewise.
980 (emit_insxl): Likewise.
981 (alpha_arg_type): Likewise.
982 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
983 (arc_preferred_simd_mode): Likewise.
984 (arc_secondary_reload): Likewise.
985 (get_arc_condition_code): Likewise.
986 (arc_print_operand): Likewise.
987 (arc_legitimate_constant_p): Likewise.
988 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
989 * config/arc/arc.md (casesi_load): Likewise.
990 (casesi_compact_jump): Likewise.
991 * config/arc/predicates.md (proper_comparison_operator): Likewise.
992 (cc_use_register): Likewise.
993 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
994 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
995 (arm_init_iwmmxt_builtins): Likewise.
996 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
997 (neon_expand_vector_init): Likewise.
998 (arm_attr_length_move_neon): Likewise.
999 (maybe_get_arm_condition_code): Likewise.
1000 (arm_emit_vector_const): Likewise.
1001 (arm_preferred_simd_mode): Likewise.
1002 (arm_output_iwmmxt_tinsr): Likewise.
1003 (thumb1_output_casesi): Likewise.
1004 (thumb2_output_casesi): Likewise.
1005 (arm_emit_load_exclusive): Likewise.
1006 (arm_emit_store_exclusive): Likewise.
1007 (arm_expand_compare_and_swap): Likewise.
1008 (arm_evpc_neon_vuzp): Likewise.
1009 (arm_evpc_neon_vzip): Likewise.
1010 (arm_evpc_neon_vrev): Likewise.
1011 (arm_evpc_neon_vtrn): Likewise.
1012 (arm_evpc_neon_vext): Likewise.
1013 (arm_validize_comparison): Likewise.
1014 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1015 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1016 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1017 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1018 (c6x_preferred_simd_mode): Likewise.
1019 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1020 (epiphany_rtx_costs): Likewise.
1021 * config/epiphany/predicates.md (proper_comparison_operator):
1023 * config/frv/frv.c (condexec_memory_operand): Likewise.
1024 (frv_emit_move): Likewise.
1025 (output_move_single): Likewise.
1026 (output_condmove_single): Likewise.
1027 (frv_hard_regno_mode_ok): Likewise.
1028 (frv_matching_accg_mode): Likewise.
1029 * config/h8300/h8300.c (split_adds_subs): Likewise.
1030 (h8300_rtx_costs): Likewise.
1031 (h8300_print_operand): Likewise.
1032 (compute_mov_length): Likewise.
1033 (output_logical_op): Likewise.
1034 (compute_logical_op_length): Likewise.
1035 (compute_logical_op_cc): Likewise.
1036 (h8300_shift_needs_scratch_p): Likewise.
1037 (output_a_shift): Likewise.
1038 (compute_a_shift_length): Likewise.
1039 (compute_a_shift_cc): Likewise.
1040 (expand_a_rotate): Likewise.
1041 (output_a_rotate): Likewise.
1042 * config/i386/i386.c (classify_argument): Likewise.
1043 (function_arg_advance_32): Likewise.
1044 (function_arg_32): Likewise.
1045 (function_arg_64): Likewise.
1046 (function_value_64): Likewise.
1047 (ix86_gimplify_va_arg): Likewise.
1048 (ix86_legitimate_constant_p): Likewise.
1049 (put_condition_code): Likewise.
1050 (split_double_mode): Likewise.
1051 (ix86_avx256_split_vector_move_misalign): Likewise.
1052 (ix86_expand_vector_logical_operator): Likewise.
1053 (ix86_split_idivmod): Likewise.
1054 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1055 (ix86_build_const_vector): Likewise.
1056 (ix86_build_signbit_mask): Likewise.
1057 (ix86_match_ccmode): Likewise.
1058 (ix86_cc_modes_compatible): Likewise.
1059 (ix86_expand_branch): Likewise.
1060 (ix86_expand_sse_cmp): Likewise.
1061 (ix86_expand_sse_movcc): Likewise.
1062 (ix86_expand_int_sse_cmp): Likewise.
1063 (ix86_expand_vec_perm_vpermi2): Likewise.
1064 (ix86_expand_vec_perm): Likewise.
1065 (ix86_expand_sse_unpack): Likewise.
1066 (ix86_expand_int_addcc): Likewise.
1067 (ix86_split_to_parts): Likewise.
1068 (ix86_vectorize_builtin_gather): Likewise.
1069 (ix86_vectorize_builtin_scatter): Likewise.
1070 (avx_vpermilp_parallel): Likewise.
1071 (inline_memory_move_cost): Likewise.
1072 (ix86_tieable_integer_mode_p): Likewise.
1073 (x86_maybe_negate_const_int): Likewise.
1074 (ix86_expand_vector_init_duplicate): Likewise.
1075 (ix86_expand_vector_init_one_nonzero): Likewise.
1076 (ix86_expand_vector_init_one_var): Likewise.
1077 (ix86_expand_vector_init_concat): Likewise.
1078 (ix86_expand_vector_init_interleave): Likewise.
1079 (ix86_expand_vector_init_general): Likewise.
1080 (ix86_expand_vector_set): Likewise.
1081 (ix86_expand_vector_extract): Likewise.
1082 (emit_reduc_half): Likewise.
1083 (ix86_emit_i387_round): Likewise.
1084 (ix86_mangle_type): Likewise.
1085 (ix86_expand_round_sse4): Likewise.
1086 (expand_vec_perm_blend): Likewise.
1087 (canonicalize_vector_int_perm): Likewise.
1088 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1089 (expand_vec_perm_1): Likewise.
1090 (expand_vec_perm_interleave3): Likewise.
1091 (expand_vec_perm_even_odd_pack): Likewise.
1092 (expand_vec_perm_even_odd_1): Likewise.
1093 (expand_vec_perm_broadcast_1): Likewise.
1094 (ix86_vectorize_vec_perm_const_ok): Likewise.
1095 (ix86_expand_vecop_qihi): Likewise.
1096 (ix86_expand_mul_widen_hilo): Likewise.
1097 (ix86_expand_sse2_abs): Likewise.
1098 (ix86_expand_pextr): Likewise.
1099 (ix86_expand_pinsr): Likewise.
1100 (ix86_preferred_simd_mode): Likewise.
1101 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1102 * config/i386/sse.md (*andnot<mode>3): Likewise.
1103 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1104 (*<code><mode>3): Likewise.
1105 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1106 (ia64_expand_atomic_op): Likewise.
1107 (ia64_arg_type): Likewise.
1108 (ia64_mode_to_int): Likewise.
1109 (ia64_scalar_mode_supported_p): Likewise.
1110 (ia64_vector_mode_supported_p): Likewise.
1111 (expand_vec_perm_broadcast): Likewise.
1112 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1113 (iq2000_function_arg_advance): Likewise.
1114 (iq2000_function_arg): Likewise.
1115 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1116 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1117 (m68k_libcall_value): Likewise.
1118 (m68k_function_value): Likewise.
1119 (sched_attr_op_type): Likewise.
1120 * config/mcore/mcore.c (mcore_output_move): Likewise.
1121 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1123 (microblaze_function_arg): Likewise.
1124 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1125 (mips_print_operand): Likewise.
1126 (mips_mode_ok_for_mov_fmt_p): Likewise.
1127 (mips_vector_mode_supported_p): Likewise.
1128 (mips_preferred_simd_mode): Likewise.
1129 (mips_expand_vpc_loongson_even_odd): Likewise.
1130 (mips_expand_vec_unpack): Likewise.
1131 (mips_expand_vi_broadcast): Likewise.
1132 (mips_expand_vector_init): Likewise.
1133 (mips_expand_vec_reduc): Likewise.
1134 (mips_expand_msa_cmp): Likewise.
1135 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1136 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1137 (cc_flags_for_mode): Likewise.
1138 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1139 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1140 (nds32_output_casesi_pc_relative): Likewise.
1141 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1142 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1143 (nvptx_gen_unpack): Likewise.
1144 (nvptx_gen_pack): Likewise.
1145 (nvptx_gen_shuffle): Likewise.
1146 (nvptx_gen_wcast): Likewise.
1147 (nvptx_preferred_simd_mode): Likewise.
1148 * config/pa/pa.c (pa_secondary_reload): Likewise.
1149 * config/pa/predicates.md (base14_operand): Likewise.
1150 * config/powerpcspe/powerpcspe-c.c
1151 (altivec_resolve_overloaded_builtin): Likewise.
1152 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1154 (rs6000_preferred_simd_mode): Likewise.
1155 (output_vec_const_move): Likewise.
1156 (rs6000_expand_vector_extract): Likewise.
1157 (rs6000_split_vec_extract_var): Likewise.
1158 (reg_offset_addressing_ok_p): Likewise.
1159 (rs6000_legitimate_offset_address_p): Likewise.
1160 (rs6000_legitimize_address): Likewise.
1161 (rs6000_emit_set_const): Likewise.
1162 (rs6000_const_vec): Likewise.
1163 (rs6000_emit_move): Likewise.
1164 (spe_build_register_parallel): Likewise.
1165 (rs6000_darwin64_record_arg_recurse): Likewise.
1166 (swap_selector_for_mode): Likewise.
1167 (spe_init_builtins): Likewise.
1168 (paired_init_builtins): Likewise.
1169 (altivec_init_builtins): Likewise.
1170 (do_load_for_compare): Likewise.
1171 (rs6000_generate_compare): Likewise.
1172 (rs6000_expand_float128_convert): Likewise.
1173 (emit_load_locked): Likewise.
1174 (emit_store_conditional): Likewise.
1175 (rs6000_output_function_epilogue): Likewise.
1176 (rs6000_handle_altivec_attribute): Likewise.
1177 (rs6000_function_value): Likewise.
1178 (emit_fusion_gpr_load): Likewise.
1179 (emit_fusion_p9_load): Likewise.
1180 (emit_fusion_p9_store): Likewise.
1181 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1182 (fusion_gpr_mem_load): Likewise.
1183 (fusion_addis_mem_combo_load): Likewise.
1184 (fusion_addis_mem_combo_store): Likewise.
1185 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1186 (fusion_gpr_mem_load): Likewise.
1187 (fusion_addis_mem_combo_load): Likewise.
1188 (fusion_addis_mem_combo_store): Likewise.
1189 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1191 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1192 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1193 (rs6000_preferred_simd_mode): Likewise.
1194 (output_vec_const_move): Likewise.
1195 (rs6000_expand_vector_extract): Likewise.
1196 (rs6000_split_vec_extract_var): Likewise.
1197 (reg_offset_addressing_ok_p): Likewise.
1198 (rs6000_legitimate_offset_address_p): Likewise.
1199 (rs6000_legitimize_address): Likewise.
1200 (rs6000_emit_set_const): Likewise.
1201 (rs6000_const_vec): Likewise.
1202 (rs6000_emit_move): Likewise.
1203 (rs6000_darwin64_record_arg_recurse): Likewise.
1204 (swap_selector_for_mode): Likewise.
1205 (paired_init_builtins): Likewise.
1206 (altivec_init_builtins): Likewise.
1207 (rs6000_expand_float128_convert): Likewise.
1208 (emit_load_locked): Likewise.
1209 (emit_store_conditional): Likewise.
1210 (rs6000_output_function_epilogue): Likewise.
1211 (rs6000_handle_altivec_attribute): Likewise.
1212 (rs6000_function_value): Likewise.
1213 (emit_fusion_gpr_load): Likewise.
1214 (emit_fusion_p9_load): Likewise.
1215 (emit_fusion_p9_store): Likewise.
1216 * config/rx/rx.c (rx_gen_move_template): Likewise.
1217 (flags_from_mode): Likewise.
1218 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1219 (s390_slb_comparison): Likewise.
1220 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1221 (s390_vector_mode_supported_p): Likewise.
1222 (s390_cc_modes_compatible): Likewise.
1223 (s390_match_ccmode_set): Likewise.
1224 (s390_canonicalize_comparison): Likewise.
1225 (s390_emit_compare_and_swap): Likewise.
1226 (s390_branch_condition_mask): Likewise.
1227 (s390_rtx_costs): Likewise.
1228 (s390_secondary_reload): Likewise.
1229 (__SECONDARY_RELOAD_CASE): Likewise.
1230 (s390_expand_cs): Likewise.
1231 (s390_preferred_simd_mode): Likewise.
1232 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1233 * config/sh/sh.c (sh_print_operand): Likewise.
1234 (dump_table): Likewise.
1235 (sh_secondary_reload): Likewise.
1236 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1237 * config/sh/sh.md (casesi_worker_1): Likewise.
1238 (casesi_worker_2): Likewise.
1239 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1240 (fcc_comparison_operator): Likewise.
1241 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1242 (emit_soft_tfmode_cvt): Likewise.
1243 (sparc_preferred_simd_mode): Likewise.
1244 (output_cbranch): Likewise.
1245 (sparc_print_operand): Likewise.
1246 (sparc_expand_vec_perm_bmask): Likewise.
1247 (vector_init_bshuffle): Likewise.
1248 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1249 (spu_vector_mode_supported_p): Likewise.
1250 (spu_expand_insv): Likewise.
1251 (spu_emit_branch_or_set): Likewise.
1252 (spu_handle_vector_attribute): Likewise.
1253 (spu_builtin_splats): Likewise.
1254 (spu_builtin_extract): Likewise.
1255 (spu_builtin_promote): Likewise.
1256 (spu_expand_sign_extend): Likewise.
1257 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1258 (tilegx_simd_int): Likewise.
1259 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1260 (tilepro_simd_int): Likewise.
1261 * config/v850/v850.c (const_double_split): Likewise.
1262 (v850_print_operand): Likewise.
1263 (ep_memory_offset): Likewise.
1264 * config/vax/vax.c (vax_rtx_costs): Likewise.
1265 (vax_output_int_move): Likewise.
1266 (vax_output_int_add): Likewise.
1267 (vax_output_int_subtract): Likewise.
1268 * config/visium/predicates.md (visium_branch_operator): Likewise.
1269 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1270 (visium_print_operand_address): Likewise.
1271 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1272 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1273 (xtensa_expand_conditional_branch): Likewise.
1274 (xtensa_copy_incoming_a7): Likewise.
1275 (xtensa_output_literal): Likewise.
1276 * dfp.c (decimal_real_maxval): Likewise.
1277 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1280 Alan Hayward <alan.hayward@arm.com>
1281 David Sherwood <david.sherwood@arm.com>
1283 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1284 (mode_nunits_inline): Likewise.
1285 (mode_inner_inline): Likewise.
1286 (mode_unit_size_inline): Likewise.
1287 (mode_unit_precision_inline): Likewise.
1288 (emit_insn_modes_h): Likewise. Also emit a #define of the
1290 (emit_mode_wider): Add an E_ prefix to mode names.
1291 (emit_mode_complex): Likewise.
1292 (emit_mode_inner): Likewise.
1293 (emit_mode_adjustments): Likewise.
1294 (emit_mode_int_n): Likewise.
1295 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1296 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1297 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1298 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1299 (CRC32_BUILTIN, ENTRY): Likewise.
1300 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1301 (aarch64_pop_regs): Likewise.
1302 (aarch64_process_components): Likewise.
1303 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1304 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1305 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1306 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1307 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1308 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1310 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1311 (pcmpistr): Likewise.
1312 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1313 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1314 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1316 * config/rl78/rl78.c (mduc_regs): Likewise.
1317 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1318 (htm_expand_builtin): Likewise.
1319 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1320 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1321 * config/xtensa/xtensa.c (print_operand): Likewise.
1322 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1323 (NUM_MODE_VECTOR_INT): Likewise.
1324 * genoutput.c (null_operand): Likewise.
1325 (output_operand_data): Likewise.
1326 * genrecog.c (print_parameter_value): Likewise.
1327 * lra.c (debug_operand_data): Likewise.
1329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1331 * dwarf2out.c (scompare_loc_descriptor_wide)
1332 (scompare_loc_descriptor_narrow): New functions, split out from...
1333 (scompare_loc_descriptor): ...here.
1334 * expmed.c (emit_store_flag_int): New function, split out from...
1335 (emit_store_flag): ...here.
1337 2017-08-30 Richard Biener <rguenther@suse.de>
1339 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1340 (dwarf2out_early_finish): Move setting of AT_pubnames from
1341 early debug output to early finish.
1343 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1345 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1346 and -mdata-region to the assembler.
1348 2017-08-30 Richard Biener <rguenther@suse.de>
1350 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1352 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1353 (dwarf2out_early_global_decl): Only generate a DIE for the
1354 abstract origin if it doesn't already exist or is a declaration DIE.
1355 (resolve_addr): Do not add the linkage name twice when
1356 generating a stub DIE for the DW_TAG_GNU_call_site target.
1358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1360 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1361 Use machine_mode rather than int for arg1_mode.
1363 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1366 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1367 that the second argument of the built-in functions to unpack
1368 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1369 switch statement instead a lot of if statements.
1370 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1371 Allow 64-bit values to be in Altivec registers as well as
1372 traditional floating point registers.
1373 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1375 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1377 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1378 MAX_REGS_PER_ADDRESS == 1.
1380 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1382 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1383 * config/i386/i386.c (ix86_option_override_internal): Simplify
1384 setting of opts->x_flag_entry.
1386 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1387 Jakub Jelinek <jakub@redhat.com>
1388 Richard Biener <rguenther@suse.de>
1390 PR tree-optimization/81503
1391 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1392 folded constant fits in the target type; reorder tests for clarity.
1394 2017-08-29 Martin Liska <mliska@suse.cz>
1396 * passes.def: Include pass_lower_switch.
1397 * stmt.c (dump_case_nodes): Remove and move to
1398 tree-switch-conversion.
1399 (case_values_threshold): Likewise.
1400 (expand_switch_as_decision_tree_p): Likewise.
1401 (emit_case_decision_tree): Likewise.
1402 (expand_case): Likewise.
1403 (balance_case_nodes): Likewise.
1404 (node_has_low_bound): Likewise.
1405 (node_has_high_bound): Likewise.
1406 (node_is_bounded): Likewise.
1407 (emit_case_nodes): Likewise.
1408 (struct simple_case_node): New struct.
1409 (add_case_node): Remove.
1410 (emit_case_dispatch_table): Use vector instead of case_list.
1411 (reset_out_edges_aux): Remove.
1412 (compute_cases_per_edge): Likewise.
1413 (expand_case): Build list of simple_case_node.
1414 (expand_sjlj_dispatch_table): Use it.
1415 * tree-switch-conversion.c (struct case_node): Moved from
1416 stmt.c and adjusted.
1417 (emit_case_nodes): Likewise.
1418 (node_has_low_bound): Likewise.
1419 (node_has_high_bound): Likewise.
1420 (node_is_bounded): Likewise.
1421 (case_values_threshold): Likewise.
1422 (reset_out_edges_aux): Likewise.
1423 (compute_cases_per_edge): Likewise.
1424 (add_case_node): Likewise.
1425 (dump_case_nodes): Likewise.
1426 (balance_case_nodes): Likewise.
1427 (expand_switch_as_decision_tree_p): Likewise.
1428 (emit_jump): Likewise.
1429 (emit_case_decision_tree): Likewise.
1430 (try_switch_expansion): Likewise.
1431 (do_jump_if_equal): Likewise.
1432 (emit_cmp_and_jump_insns): Likewise.
1433 (fix_phi_operands_for_edge): New function.
1434 (record_phi_operand_mapping): Likewise.
1435 (class pass_lower_switch): New pass.
1436 (pass_lower_switch::execute): New function.
1437 (make_pass_lower_switch): Likewise.
1438 (conditional_probability):
1439 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1440 * tree-pass.h: Add make_pass_lower_switch.
1442 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1445 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1448 2017-08-29 Richard Biener <rguenther@suse.de>
1450 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1451 we do not add a DW_AT_inline attribute twice.
1452 (gen_subprogram_die): Remove code setting DW_AT_inline on
1453 DECL_ABSTRACT_P nodes.
1455 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1457 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1458 calls to internal functions.
1459 (gimplify_modify_expr): Likewise.
1460 * tree-call-cdce.c (use_internal_fn): Likewise.
1461 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1462 (convert_to_divmod): Set the nothrow flag.
1463 * tree-if-conv.c (predicate_mem_writes): Likewise.
1464 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1465 (vectorizable_call): Likewise.
1466 (vectorizable_store): Likewise.
1467 (vectorizable_load): Likewise.
1468 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1469 (vect_recog_mask_conversion_pattern): Likewise.
1471 2017-08-29 Martin Liska <mliska@suse.cz>
1474 * gcc.c (driver_handle_option): Add new argument.
1475 * opts-common.c (handle_option): Pass
1476 target_option_override_hook.
1477 * opts-global.c (lang_handle_option): Add new option.
1478 (set_default_handlers): Add new argument.
1479 (decode_options): Likewise.
1480 * opts.c (target_handle_option): Likewise.
1481 (common_handle_option): Call target_option_override_hook.
1482 * opts.h (struct cl_option_handler_func): Add hook for
1483 target option override.
1484 (struct cl_option_handlers): Likewise.
1485 (set_default_handlers): Add new argument.
1486 (decode_options): Likewise.
1487 (common_handle_option): Likewise.
1488 (target_handle_option): Likewise.
1489 * toplev.c (toplev::main): Pass targetm.target_option.override
1492 2017-08-29 Richard Biener <rguenther@suse.de>
1493 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1495 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1496 life to the active subtree.
1498 2017-08-28 Jeff Law <law@redhat.com>
1500 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1501 derive_equivalences.
1502 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1504 (edge_info::derive_equivalences): New private member function
1506 * tree-ssa-dom.c (class edge_info): Changed from a struct
1507 to a class. Add ctor/dtor, methods and data members.
1508 (edge_info::edge_info): Renamed from allocate_edge_info.
1509 Initialize additional members.
1510 (edge_info::~edge_info): New.
1511 (free_dom_edge_info): Delete the edge info.
1512 (record_edge_info): Use new class & associated member functions.
1513 Tighten forms for testing for edge equivalences.
1514 (record_temporary_equivalences): Iterate over the simple
1515 equivalences rather than assuming there's only one per edge.
1516 (cprop_into_successor_phis): Iterate over the simple
1517 equivalences rather than assuming there's only one per edge.
1518 (optimize_stmt): Use operand_equal_p rather than pointer
1519 equality for mini-DSE code.
1521 2017-08-28 Nathan Sidwell <nathan@acm.org>
1523 * gcc.c (execute): Fold SIGPIPE handling into switch
1524 statement. Adjust internal error message.
1526 2017-08-28 Richard Biener <rguenther@suse.de>
1529 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1530 Do nothing for removed DIEs.
1532 2017-08-28 Richard Biener <rguenther@suse.de>
1534 PR tree-optimization/81977
1535 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1538 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1541 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1542 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1543 using targetm.gen_mem_thread_fence.
1545 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1548 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1549 predicate to register_operand. Reorder operands.
1550 (*btr<mode>): Ditto.
1551 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1552 (*btr<mode>_mask): Ditto.
1554 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1556 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1557 * config/rs6000/xmmintrin.h: New file.
1558 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1560 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1563 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1564 parameter and_insn and return it.
1565 (recombine_lvx_pattern): Insert a copy to ensure availability of
1566 the base register of the copied masking operation at the point of
1567 the instruction replacement.
1568 (recombine_stvx_pattern): Likewise.
1570 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1572 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1573 undocumented switches.
1574 (-mpower9-dform-vector): Likewise.
1575 (-mpower9-dform): Likewise.
1576 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1577 comments to delete references to -mpower9-dform* switches.
1578 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1579 Delete reference to -mpower9-dform* switches, test for
1580 -mpower9-vector instead.
1581 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1582 (OTHER_P9_VECTOR_MASKS): Likewise.
1583 (POWERPC_MASKS): Likewise.
1584 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1585 tests against -mpower9-dform* to -mpower9-vector. Delete code
1586 that checked for -mpower9-dform* consistancy with other options.
1587 Add test for -mpower9-misc to enable other power9 switches.
1588 (rs6000_init_hard_regno_mode_ok): Likewise.
1589 (rs6000_option_override_internal): Likewise.
1590 (rs6000_emit_prologue): Likewise.
1591 (rs6000_emit_epilogue): Likewise.
1592 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1593 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1594 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1596 (emit_fusion_p9_store): Likewise.
1597 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1598 resetting these macros if the assembler does not support ISA 3.0
1600 (TARGET_P9_DFORM_VECTOR): Likewise.
1601 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1602 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1604 2017-08-25 Alan Modra <amodra@gmail.com>
1606 PR rtl-optimization/81747
1607 * cse.c (cse_extended_basic_block): Don't attempt to record
1608 equivalences for degenerate conditional branches that branch
1609 to their fall-through.
1611 2017-08-24 Martin Sebor <msebor@redhat.com>
1614 * gimple-fold.c (size_must_be_zero_p): New function.
1615 (gimple_fold_builtin_memory_op): Call it.
1617 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1619 * config/rs6000/mm_malloc.h: New file.
1621 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1623 PR tree-optimization/81913
1624 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1625 analysis when either IVs in condition can wrap.
1627 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1629 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1630 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1632 2017-08-24 Richard Biener <rguenther@suse.de>
1635 * targhooks.c (default_target_can_inline_p): Properly
1636 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1637 is present and always compare.
1638 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1639 infer -mfpmath=sse from TARGET_SSE_P.
1640 (ix86_can_inline_p): Properly use target_option_default_node when
1641 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1643 2017-08-24 Richard Biener <rguenther@suse.de>
1646 * dwarf2out.c (output_die): Handle flag_generate_offload like
1648 (output_comp_unit): Likewise.
1649 (gen_array_type_die): Likewise.
1650 (dwarf2out_early_finish): Likewise.
1651 (note_variable_value_in_expr): Likewise.
1652 (dwarf2out_finish): Likewise. Adjust assert.
1653 * cgraphunit.c (symbol_table::compile): Move setting of
1654 flag_generate_offload earlier ...
1655 (symbol_table::finalize_compilation_unit): ... here, before
1656 early debug finalization.
1658 2017-08-24 Richard Biener <rguenther@suse.de>
1660 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1661 and ipa-fnsummary.h.
1662 (ix86_can_inline_p): When ix86_fpmath flags do not match
1663 check whether the callee uses FP math at all.
1665 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1668 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1669 instead of TYPE_PRECISION.
1671 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1672 Alan Hayward <alan.hayward@arm.com>
1673 David Sherwood <david.sherwood@arm.com>
1675 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1677 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1679 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1681 (ix86_frame::stack_realign_allocate): New field.
1682 (struct machine_frame_state): Modify comments.
1683 (machine_frame_state::sp_realigned_fp_end): New field.
1684 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1686 (sp_valid_at): Add assertion to assure no attempt to access invalid
1687 offset of a realigned stack.
1688 (fp_valid_at): Likewise.
1689 (choose_baseaddr): Modify comments.
1690 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1691 ix86_expand_prologue.
1692 (ix86_expand_prologue): Modify stack realignment and allocation.
1693 (ix86_expand_epilogue): Modify comments.
1694 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1695 avx2_runtime, avx512f, and avx512f_runtime.
1697 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1699 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1700 (mstackrealign): Do not init to -1.
1701 * config/i386/i386.c (ix86_option_override_internal):
1702 Check opts_set, not opts when setting default value of
1703 opts->x_ix86_force_align_arg_pointer.
1705 2017-08-23 Richard Biener <rguenther@suse.de>
1707 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1708 lang_hooks.decl_printable_name.
1709 * print-rtl-function.c (print_rtx_function): Likewise.
1710 * tree-pretty-print.c (dump_function_header): Likewise.
1712 2017-08-23 Richard Biener <rguenther@suse.de>
1715 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1716 -g0 at compile-time.
1718 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1721 * doc/sourcebuild.texi (Other hardware attributes):
1724 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1727 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1728 Fix single-use check.
1730 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1732 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1734 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1736 * doc/install.texi: Modify to add more details on running selected
1739 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1741 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1742 is combined with -mabi=ms.
1743 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1746 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1748 PR tree-optimization/81488
1749 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1750 and cached_basis fields.
1751 (MAX_SPREAD): New constant.
1752 (alloc_cand_and_find_basis): Initialize new fields.
1753 (clear_visited): New function.
1754 (create_phi_basis_1): Rename from create_phi_basis, set visited
1755 and cached_basis fields.
1756 (create_phi_basis): New wrapper function.
1757 (phi_add_costs_1): Rename from phi_add_costs, add spread
1758 parameter, set visited field, short-circuit when limits reached.
1759 (phi_add_costs): New wrapper function.
1760 (record_phi_increments_1): Rename from record_phi_increments, set
1762 (record_phi_increments): New wrapper function.
1763 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1764 (phi_incr_cost): New wrapper function.
1765 (all_phi_incrs_profitable_1): Rename from
1766 all_phi_incrs_profitable, set visited field.
1767 (all_phi_incrs_profitable): New wrapper function.
1769 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1770 Alan Hayward <alan.hayward@arm.com>
1771 David Sherwood <david.sherwood@arm.com>
1773 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1774 that takes the outer and inner modes.
1775 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1776 comparison as the canonical test for a paradoxical subreg.
1777 * combine.c (simplify_set): Use paradoxical_subreg_p.
1778 (make_extraction): Likewise.
1779 (force_to_mode): Likewise.
1780 (rtx_equal_for_field_assignment_p): Likewise.
1781 (gen_lowpart_for_combine): Likewise.
1782 (simplify_comparison): Likewise.
1783 * cse.c (equiv_constant): Likewise.
1784 * expmed.c (store_bit_field_1): Likewise.
1785 * final.c (alter_subreg): Likewise.
1786 * fwprop.c (propagate_rtx): Likewise.
1787 (forward_propagate_subreg): Likewise.
1788 * ira-conflicts.c (ira_build_conflicts): Likewise.
1789 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1790 * lra-constraints.c (curr_insn_transform): Likewise.
1791 (split_reg): Likewise.
1792 * lra-eliminations.c (move_plus_up): Likewise.
1793 (lra_eliminate_regs_1): Likewise.
1794 * recog.c (general_operand): Likewise.
1795 * ree.c (combine_reaching_defs): Likewise.
1796 * reload.c (push_reload): Likewise.
1797 (find_reloads): Likewise.
1798 * reload1.c (elimination_effects): Likewise.
1799 (compute_reload_subreg_offset): Likewise.
1800 (choose_reload_regs): Likewise.
1801 * rtlanal.c (subreg_lsb_1): Likewise.
1802 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1803 (simplify_subreg): Likewise.
1804 * var-tracking.c (track_loc_p): Likewise.
1805 * emit-rtl.c (byte_lowpart_offset): Likewise.
1806 (paradoxical_subreg_p): Delete out-of-line definition.
1808 2017-08-22 Jeff Law <law@redhat.com>
1810 PR tree-optimization/81741
1811 PR tree-optimization/71947
1812 * tree-ssa-dom.c: Include tree-inline.h.
1813 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1814 equivalences if one is more expensive to compute than the other.
1815 * tree-ssa-scopedtables.h (class const_or_copies): Make
1816 record_const_or_copy_raw method private.
1817 (class avail_exprs_stack): New method simplify_binary_operation.
1818 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1819 avail_exprs_stack::simplify_binary_operation as needed.
1820 (avail_exprs_stack::simplify_binary_operation): New function.
1822 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
1824 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
1825 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
1826 (DOT_SYMBOLS): Likewise.
1827 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
1828 (RELOCATABLE_NEEDS_FIXUP): Likewise.
1829 (RS6000_ABI_NAME): Likewise.
1830 (TARGET_CMODEL): Likewise.
1831 (TOC_SECTION_ASM_OP): Likewise.
1832 (SET_CMODEL): New macro.
1833 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
1835 2017-08-22 Richard Biener <rguenther@suse.de>
1837 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
1838 to allow for free-lang-data replacements similar to verify_type_variant.
1840 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1841 Alan Hayward <alan.hayward@arm.com>
1842 David Sherwood <david.sherwood@arm.com>
1844 * config/aarch64/aarch64.md (casesi): Use DImode rather than
1845 VOIDmode for the LABEL_REF.
1847 2017-08-22 Richard Biener <rguenther@suse.de>
1849 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
1851 2017-08-22 Richard Biener <rguenther@suse.de>
1853 * common.opt (feliminate-dwarf2-dups): Ignore.
1854 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
1855 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
1856 same_die_p_wrap, compute_section_prefix,
1857 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
1858 (comdat_symbol_id, comdat_symbol_number): Likewise.
1859 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
1861 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
1862 (output_die): Mark unreachable path unreachable.
1863 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
1864 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
1865 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
1866 (dwarf2out_early_finish): Likewise.
1868 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
1870 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
1872 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
1875 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
1876 not VAR_P. Filter attribute warnings with OPT_Wattributes.
1877 (avr_attribute_table) <io, io_low, address>: Initialize
1878 .decl_required with true.
1880 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
1882 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
1883 undocumented debugging options.
1884 (-mvsx-scalar-double): Likewise.
1885 (-mallow-df-permute): Likewise.
1886 (-mvectorize-builtins): Likewise.
1887 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1888 (rs6000_builtin_vectorized_function): Likewise.
1889 (rs6000_builtin_md_vectorized_function): Likewise.
1890 (rs6000_opt_vars): Likewise.
1892 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
1895 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
1896 (*btrq_imm): Rename from *btrq.
1897 (*btcq_imm): Rename from *btcq.
1898 (btsc): New code attribute.
1899 (*<btsc><mode>): New insn pattern.
1900 (*btr<mode>): Ditto.
1901 (*<btsc><mode>_mask): New insn_and_split pattern.
1902 (*btr<mode>_mask): Ditto.
1904 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1905 Alan Hayward <alan.hayward@arm.com>
1906 David Sherwood <david.sherwood@arm.com>
1908 * function.c (pad_below): Simplify padding calculation.
1910 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1911 Alan Hayward <alan.hayward@arm.com>
1912 David Sherwood <david.sherwood@arm.com>
1914 * target.def (function_prologue): Remove frame size argument.
1915 (function_epilogue): Likewise.
1916 * doc/tm.texi: Regenerate.
1917 * final.c (final_start_function): Update call to function_prologue.
1918 (final_end_function): Update call to function_epilogue.
1919 (default_function_pro_epilogue): Remove frame size argument.
1920 * output.h (default_function_pro_epilogue): Likewise.
1921 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
1922 (arm_output_function_prologue): Likewise.
1923 * config/frv/frv.c (frv_function_prologue): Likewise.
1924 (frv_function_epilogue): Likewise.
1925 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
1926 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
1927 (ia64_output_function_epilogue): Likewise.
1928 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
1929 (m32r_output_function_epilogue): Likewise.
1930 * config/microblaze/microblaze.c (microblaze_function_prologue)
1931 (microblaze_function_epilogue): Likewise.
1932 * config/mips/mips.c (mips_output_function_prologue): Likewise.
1933 (mips_output_function_epilogue): Likewise.
1934 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
1935 (mmix_target_asm_function_epilogue): Likewise.
1936 * config/msp430/msp430.c (msp430_start_function): Likewise.
1937 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
1938 (nds32_asm_function_epilogue): Likewise.
1939 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
1940 * config/pa/pa.c (pa_output_function_prologue): Likewise.
1941 (pa_output_function_epilogue): Likewise.
1942 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
1943 (rs6000_output_function_epilogue): Likewise.
1944 * config/rl78/rl78.c (rl78_start_function): Likewise.
1945 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
1946 (rs6000_output_function_epilogue): Likewise.
1947 * config/rx/rx.c (rx_output_function_prologue): Likewise.
1948 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
1949 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
1950 (sparc_asm_function_epilogue): Likewise.
1952 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1954 * tree.h (type_has_mode_precision_p): New function.
1955 * convert.c (convert_to_integer_1): Use it.
1956 * expr.c (expand_expr_real_2): Likewise.
1957 (expand_expr_real_1): Likewise.
1958 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
1959 * match.pd: Likewise.
1960 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
1961 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
1962 * tree-tailcall.c (process_assignment): Likewise.
1963 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1964 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
1965 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
1966 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1967 (vectorizable_assignment): Likewise.
1968 (vectorizable_shift): Likewise.
1969 (vectorizable_operation): Likewise.
1970 * tree-vrp.c (register_edge_assert_for_2): Likewise.
1972 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
1974 * match.pd: Add pow (C, x) simplification.
1976 2017-08-21 Richard Biener <rguenther@suse.de>
1978 PR tree-optimization/81900
1979 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
1980 for blocks with abnormal predecessors.
1981 (compute_antic): Do not set visited flag prematurely.
1983 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
1986 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
1988 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1990 * stor-layout.h (vector_type_mode): Move to...
1991 * tree.h (vector_type_mode): ...here.
1992 * stor-layout.c (vector_type_mode): Move to...
1993 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
1995 2017-08-21 Richard Biener <rguenther@suse.de>
1997 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
1998 register_external_die hooks.
1999 (debug_false_tree_charstarstar_uhwistar): Declare.
2000 (debug_nothing_tree_charstar_uhwi): Likewise.
2001 * debug.c (do_nothing_debug_hooks): Adjust.
2002 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2003 (debug_nothing_tree_charstar_uhwi): Likewise.
2004 * dbxout.c (dbx_debug_hooks): Adjust.
2005 (xcoff_debug_hooks): Likewise.
2006 * sdbout.c (sdb_debug_hooks): Likewise.
2007 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2008 * dwarf2out.c (macinfo_label_base): New global.
2009 (dwarf2out_register_external_die): New function for the
2010 register_external_die hook.
2011 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2012 (dwarf2_debug_hooks): Use them.
2013 (dwarf2_lineno_debug_hooks): Adjust.
2014 (struct die_struct): Add with_offset flag.
2015 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2016 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2017 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2018 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2019 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2020 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2021 defining section names for the early LTO debug variants.
2022 (reset_indirect_string): New helper.
2023 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2024 (print_dw_val): Add support for offsetted symbol references.
2025 (get_ultimate_context): Split out from is_cxx.
2026 (is_cxx): Use get_ultimate_context.
2027 (is_fortran): Add decl overload.
2028 (compute_comp_unit_symbol): Split out worker from
2029 compute_section_prefix.
2030 (compute_section_prefix): Call compute_comp_unit_symbol and
2031 set comdat_type_p here.
2032 (output_die): Skip DIE symbol output for the LTO added one.
2033 Handle DIE symbol references with offset.
2034 (output_comp_unit): Guard section name mangling properly.
2035 For LTO debug sections emit a symbol at the section beginning
2036 which we use to refer to its DIEs.
2037 (add_abstract_origin_attribute): For DIEs registered via
2038 dwarf2out_register_external_die directly refer to the early
2039 DIE rather than indirectly through the shadow one we created.
2040 Remove obsolete call to dwarf2out_abstract_function for
2041 non-function/block origins.
2042 (gen_array_type_die): When generating early LTO debug do
2043 not emit DW_AT_string_length.
2044 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2045 late when in LTO. As suggested place a gcc_unreachable for
2046 the DECL_ABSTRACT_P case.
2047 (gen_subprogram_die): Avoid another specification DIE
2048 for early built declarations/definitions for the late LTO case.
2049 (gen_variable_die): Add type references for late duplicated VLA dies
2051 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2052 we have the abstract instance already.
2053 (process_scope_var): Adjust decl DIE contexts in LTO which
2054 first puts them in limbo.
2055 (gen_decl_die): Do not generate type DIEs late apart from
2056 types for VLAs or for decls we do not yet have a DIE. Do not
2057 call dwarf2out_abstract_function late.
2058 (dwarf2out_early_global_decl): Make sure to create DIEs
2059 for abstract instances of a decl first.
2060 (dwarf2out_late_global_decl): Adjust comment.
2061 (output_macinfo_op): With multiple macro sections use
2062 macinfo_label_base to distinguish labels.
2063 (output_macinfo): Likewise. Update macinfo_label_base.
2064 Pass in the line info label.
2065 (note_variable_value_in_expr): When generating LTO resolve
2066 all variable values here by generating DIEs as needed.
2067 (init_sections_and_labels): Add early LTO debug flag parameter
2068 and generate different sections and names if set. Add generation
2069 counter for the labels so we can have multiple of them.
2070 (reset_dies): Helper to allow DIEs to be output multiple times.
2071 (dwarf2out_finish): When outputting DIEs to the fat part of an
2072 LTO object first reset DIEs.
2073 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2074 (modified_type_die): Check for decl_ultimate_origin being self
2076 (gen_type_die_with_usage): Likewise.
2077 (gen_typedef_die): Allow decl_ultimate_origin being self.
2078 (set_decl_abstract_flags): Remove.
2079 (set_block_abstract_flags): Likewise.
2080 (dwarf2out_abstract_function): Treat the early generated DIEs
2081 as the abstract copy and only add DW_AT_inline and
2082 DW_AT_artificial here and call set_decl_origin_self.
2083 If the DIE has an abstract origin don't do anything.
2084 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2085 if we have none yet (Go fails to build one, PR78628).
2086 (variably_modified_type_p): Prevent endless recursion for Ada
2087 cyclic pointer types.
2088 * lto-streamer-in.c: Include debug.h.
2089 (dref_queue): New global.
2090 (lto_read_tree_1): Stream in DIE references.
2091 (lto_input_tree): Register DIE references.
2092 (input_function): Stream DECL_DEBUG_ARGS.
2093 * lto-streamer-out.c: Include debug.h.
2094 (lto_write_tree_1): Output DIE references.
2095 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2096 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2097 (output_function): Stream DECL_DEBUG_ARGS.
2098 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2099 Stream DECL_ABSTRACT_ORIGIN.
2100 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2101 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2102 DECL_CONTEXT for file-scope decls.
2103 * lto-streamer.h (struct dref_entry): Declare.
2104 (dref_queue): Likewise.
2105 * cfgexpand.c (pass_expand::execute): Do not call the
2106 outlining_inline_function hook here.
2107 * lto-wrapper.c (debug_obj): New global.
2108 (tool_cleanup): Unlink it if required.
2109 (debug_objcopy): New function.
2110 (run_gcc): Handle early debug sections in the IL files by
2111 extracting them to separate files, partially linkin them and
2112 feeding the result back as result to the linker.
2113 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2114 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2115 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2116 sections into a separate segment.
2117 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2119 (darwin_asm_dwarf_section): Likewise.
2120 (darwin_asm_output_dwarf_offset): Likewise.
2121 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2123 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2124 Alan Hayward <alan.hayward@arm.com>
2125 David Sherwood <david.sherwood@arm.com>
2127 * read-md.h (md_reader::record_potential_iterator_use): Replace
2128 pointer argument with an rtx and an index.
2129 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2130 (apply_mode_iterator): Likewise.
2131 (apply_code_iterator): Likewise.
2132 (apply_int_iterator): Likewise.
2133 (apply_subst_iterator): Likewise.
2134 (record_iterator_use): Likewise.
2135 (record_attribute_use): Likewise.
2136 (md_reader::record_potential_iterator_use): Likewise. Update calls
2137 to record_iterator_use and apply_iterator.
2138 (iterator_use): Replace ptr with x and index.
2139 (attribute_use): Likewise.
2140 (apply_attribute_uses): Update calls to apply_iterator.
2141 (apply_iterators): Likewise. Update initialization of iterator_use.
2142 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2143 and record_potential_iterator_use.
2144 (rtx_reader::read_rtx_operand): Likewise.
2146 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2147 Alan Hayward <alan.hayward@arm.com>
2148 David Sherwood <david.sherwood@arm.com>
2150 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2153 2017-08-21 Richard Biener <rguenther@suse.de>
2156 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2157 at struct end conservatively when comparing common bases.
2159 2017-08-21 Richard Biener <rguenther@suse.de>
2161 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2162 (mem_ref_in_stmt): Remove.
2163 (determine_max_movement): Use ref index to get at the reference.
2164 (invariantness_dom_walker::before_dom_children): Deal with
2165 lim data already initialized.
2166 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2168 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2170 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2171 (TARGET_ISA_ROUND): Ditto.
2172 (TARGET_ROUND): Ditto.
2173 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2174 * config/i386/i386.md: Ditto.
2175 * config/i386/sse.md: Ditto.
2176 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2177 with OPTION_MASK_ISA_SSE4_1.
2179 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2182 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2183 __builtin_ia32_lzcnt_u16.
2185 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2188 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2189 (rs6000_set_current_function): Rewrite function to use it.
2191 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2194 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2195 and TYPE_WARN_IF_NOT_ALIGN.
2196 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2197 (handle_warn_if_not_align): New.
2198 (place_union_field): Call handle_warn_if_not_align.
2199 (place_field): Call handle_warn_if_not_align.
2200 Copy TYPE_WARN_IF_NOT_ALIGN.
2201 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2202 (layout_type): Likewise.
2203 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2205 (tree_decl_common): Add warn_if_not_align.
2206 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2207 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2208 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2209 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2210 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2211 * doc/extend.texi: Document warn_if_not_aligned attribute.
2212 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2214 2017-08-17 Martin Liska <mliska@suse.cz>
2217 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2218 (get_data_dependence): Use it as pointer type.
2219 (distribute_loop): Likewise.
2221 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2223 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2224 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2225 (p8_vmrgow_<mode>_direct): New define_insn.
2226 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2227 handle endianness for vmrgew and vmrgow permute patterns.
2229 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2231 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2232 * config/rs6000/rs6000-cpus.def: Remove comment.
2233 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2234 (POWERPC_MASKS): Likewise.
2235 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2236 use of TARGET_VSX_TIMODE.
2237 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2238 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2239 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2240 (rs6000_option_override_internal): Remove dead code.
2241 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2242 (rs6000_legitimize_reload_address): Likewise.
2243 (rs6000_legitimate_address_p): Likewise.
2244 (rs6000_opt_masks): Delete "vsx-timode".
2245 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2246 from function comment.
2247 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2248 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2249 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2251 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2252 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2253 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2256 (define_peephole2): Likewise.
2258 2017-08-17 Martin Sebor <msebor@redhat.com>
2261 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2262 past the end of an array.
2263 (test_pp_format): Add test cases.
2265 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2267 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2268 missing ECF_NOTHROW flags.
2270 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2273 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2274 operands residing in integer registers.
2275 (*vsx_le_perm_load_<mode>): Likewise.
2276 (*vsx_le_perm_store_<mode>): Likewise.
2277 (define_peephole2): Add peepholes to optimize the above.
2279 2017-08-17 Marek Polacek <polacek@redhat.com>
2282 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2283 to mimic what shorten_compare did. Change the return type to bool.
2284 (fold_cond_expr_with_comparison): Update call to
2285 operand_equal_for_comparison_p.
2286 (fold_ternary_loc): Likewise.
2288 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2290 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2292 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2294 2017-08-17 Richard Biener <rguenther@suse.de>
2296 * tree-ssa-structalias.c (solve_graph): When propagating
2297 to successors update the graphs succ edges and avoid duplicate work.
2299 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2302 * config/i386/i386.c (ix86_option_override_internal): Save target
2303 specific options after ix86_stack_protector_guard_reg was changed.
2305 2017-08-17 Richard Biener <rguenther@suse.de>
2307 PR tree-optimization/81827
2308 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2309 (new_var_info): Initialize it conservatively.
2310 (get_call_vi): Mark register vars.
2311 (new_scalar_tmp_constraint_exp): Likewise.
2312 (handle_rhs_call): Likewise.
2313 (handle_const_call): Likewise.
2314 (create_function_info_for): Likewise.
2315 (solve_constraints): Sort varinfos to separate register from
2316 non-register vars to pack points-to solution bitmaps during
2319 2017-08-17 Marek Polacek <polacek@redhat.com>
2321 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2323 2017-08-17 Richard Biener <rguenther@suse.de>
2325 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2326 to true when overflow is undefined and we saturated the result.
2328 2017-08-17 Alan Modra <amodra@gmail.com>
2331 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2332 Don't use store multiple if only one reg needs saving.
2333 (interesting_frame_related_regno): New function.
2334 (rs6000_frame_related): Don't emit frame info for regs that
2336 (rs6000_emit_epilogue): Likewise.
2338 2017-08-16 Nathan Sidwell <nathan@acm.org>
2340 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2341 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2342 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2343 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2344 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2345 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2346 (verify_type): Adjust for TYPE_BINFO move.
2347 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2348 process TYPE_BINFO directly.
2349 (hash_tree): Likewise.
2350 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2352 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2355 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2357 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2359 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2362 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2363 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2364 Add ix86_binary_operator_ok to insn constraint.
2365 (*iordi_1_bts): Ditto.
2366 (*xordi_1_btc): Ditto.
2367 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2368 Update corresponding peephole2 pattern.
2372 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2374 PR tree-optimization/81832
2375 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2376 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2378 2017-08-16 Marek Polacek <polacek@redhat.com>
2381 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2382 regarding size_zero_node.
2384 2017-08-16 Martin Liska <mliska@suse.cz>
2387 * config.gcc: Respect previously set extra_objs in case
2390 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2392 PR tree-optimization/81835
2393 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2394 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2395 not depend on the phi.
2397 2017-08-16 Alan Modra <amodra@gmail.com>
2399 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2402 2017-08-16 Alan Modra <amodra@gmail.com>
2404 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2405 (save_reg_p): ..into this. Update all callers.
2406 (first_reg_to_save): Simplify.
2408 2017-08-16 Alan Modra <amodra@gmail.com>
2410 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2413 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2417 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2418 constant count if that count is less than 32.
2420 2017-08-15 Nathan Sidwell <nathan@acm.org>
2422 * gcc.c (execute): Emit friendlier message if inferior is killed
2423 by an external cause.
2425 2017-08-15 Richard Biener <rguenther@suse.de>
2427 PR tree-optimization/81790
2428 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2429 CONSTRUCTORs from simplifying and VN.
2431 2017-08-14 Martin Sebor <msebor@redhat.com>
2433 * builtin-attrs.def: Add comments.
2435 2017-08-14 Martin Sebor <msebor@redhat.com>
2438 * doc/extend.texi (attribute nonstring): Document new attribute.
2440 2017-08-14 Martin Sebor <msebor@redhat.com>
2443 * tree-diagnostic.c (default_tree_printer): Handle %G.
2444 * gimple-pretty-print.h (percent_G_format): Declare new function.
2445 * gimple-pretty-print.c (percent_G_format): Define.
2446 * tree-pretty-print.c (percent_K_format): Add argument.
2448 2017-08-14 Martin Sebor <msebor@redhat.com>
2450 PR translation/79998
2451 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2452 Remove a stray space.
2454 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2457 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2458 (*iordi_1_bts): Ditto.
2459 (*xordi_1_btc): Ditto.
2461 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2464 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2465 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2467 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2468 quoted strings, and make more translator-friendly.
2469 (darwin_rs6000_override_options): Likewise.
2470 (rs6000_option_override_internal): Likewise.
2471 (rs6000_return_in_memory): Fix overlong line.
2472 (init_cmulative_args): Use quoted strings, and make more
2473 translator-friendly.
2474 (rs6000_pass_by_reference): Fix overlong line.
2475 (def_builtin): Use quoted strings.
2476 (altivec_expand_predicate_builtin): Use quoted strings, and make
2477 more translator-friendly.
2478 (htm_expand_builtin): Use quoted strings.
2479 (cpu_expand_builtin): Use quoted strings, and make more
2480 translator-friendly.
2481 (altivec_expand_builtin): Likewise.
2482 (paired_expand_predicate_builtin): Likewise.
2483 (rs6000_invalid_builtin): Likewise.
2484 (builtin_function_type): Use quoted strings.
2485 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2486 more translator-friendly.
2487 (rs6000_trampoline_init): Likewise.
2488 (rs6000_handle_altivec_attribute): Likewise.
2489 (rs6000_inner_target_options): Use quoted strings.
2490 (rs6000_disable_incompatible_switches): Likewise.
2491 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2492 strings, and make more translator-friendly.
2493 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2495 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2497 PR tree-optimization/81799
2498 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2499 cond_expr to simple gimple operand.
2501 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2504 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2506 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2510 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2511 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2512 TARGET_GASISR_PROLOGUES.
2513 * config/avr/avr.c (avr_option_override): Same.
2514 (avr_pass_pre_proep::execute): Same.
2516 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2519 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2520 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2522 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2524 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2525 define_insn_and_split. Split to a memory load from 0 in
2526 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2527 using PTR mode iterator.
2528 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2529 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2530 (*add_tp_<mode>): Redefine as define_insn_and_split.
2531 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2532 address space. Merge with *add_tp_x32 using PTR mode iterator.
2533 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2534 Split to an add with a memory load from 0 in
2535 DEFAULT_TLS_SEG_REG address space.
2537 2017-08-12 Andrew Pinski <apinski@cavium.com>
2539 * config/aarch64/aarch64-option-extensions.def (rdma):
2540 Fix feature string to what Linux prints out in /proc/cpuinfo.
2542 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2545 * dwarf2out.c (modified_type_die): For C typedef types that have
2546 an ultimate origin, process the ultimate origin instead of the
2548 (gen_typedef_die): Assert that input DECLs have no ultimate
2550 (gen_type_die_with_usage): For typedef variants that have an
2551 ultimate origin, just call gen_decl_die on the original DECL.
2552 (process_scope_var): Avoid creating DIEs for local typedefs and
2553 concrete static variables.
2555 2017-08-12 Alan Modra <amodra@gmail.com>
2559 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2560 match gnu-user.h startfile.
2561 (ENDFILE_LINUX_SPEC): Similarly.
2563 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2566 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2568 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2570 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2571 * config/aarch64/aarch64.md (mov<mode>): Change.
2572 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2573 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2574 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2576 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2578 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2579 for storage order barriers.
2581 2017-08-11 Martin Liska <mliska@suse.cz>
2584 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2585 variables of void type.
2587 2017-08-11 Martin Liska <mliska@suse.cz>
2589 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2590 TARGET_SUPPORTS_ALIASES.
2591 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2592 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2593 (optimize_weakref): Likewise.
2594 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2595 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2596 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2598 2017-08-11 Martin Liska <mliska@suse.cz>
2601 * config/i386/i386.c (make_resolver_func): Do complete
2602 refactoring of the function.
2604 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2607 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2608 * config/i386/i386.c (ix86_stack_protect_guard): Use
2609 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2610 * doc/invoke.texi (x86 Options): Document
2611 -mstack-protector-guard-symbol= option.
2613 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2615 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2616 * config/i386/i386.c (ix86_split_stack_guard): New function.
2617 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2618 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2619 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2620 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2621 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2622 (split_stack_space_check): Call ix86_split_stack_guard.
2624 2017-08-10 Martin Sebor <msebor@redhat.com>
2626 * print-tree.c (print_node): Print location using the established
2628 Replace spaces with colons.
2629 (debug_raw, debug): Ditto.
2631 2017-08-10 Martin Sebor <msebor@redhat.com>
2634 * pretty-print.c (pp_format): Correct the handling of %s precision.
2636 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2639 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2641 (ix86_finalize_stack_frame_flags): This. Also clear
2642 frame_pointer_needed if -fno-omit-frame-pointer is used without
2644 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2645 with ix86_finalize_stack_frame_flags.
2646 (ix86_expand_epilogue): Likewise.
2647 (ix86_expand_split_stack_prologue): Likewise.
2648 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2650 2017-08-10 Martin Liska <mliska@suse.cz>
2653 * c-attribs.c (handle_target_attribute):
2654 Report warning for an empty string argument of target attribute.
2656 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2659 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2661 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2662 or DECL_NONLOCAL labels.
2663 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2664 or DECL_NONLOCAL labels here.
2666 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2669 to indicate when early gimple folding has been disabled.
2670 (rs6000_gimple_fold_builtin): Add debug content.
2671 (rs6000_invalid_builtin): Fix whitespace.
2672 (rs6000_expand_builtin): Fix whitespace.
2673 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2675 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2678 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2679 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2681 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2683 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2684 (qdf24xx): Likewise.
2685 * config/aarch64/aarch64.md: Include falkor.md.
2686 * config/aarch64/falkor.md: New.
2688 2017-08-09 Marek Polacek <polacek@redhat.com>
2691 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2692 * diagnostic.c (emit_diagnostic): Add a comment.
2693 (emit_diagnostic_valist): New function.
2695 2017-08-09 Marek Polacek <polacek@redhat.com>
2698 * input.c (make_location): New overload.
2699 * input.h (make_location): Declare.
2701 2017-08-08 Alan Modra <amodra@gmail.com>
2702 H.J. Lu <hongjiu.lu@intel.com>
2705 * gcc.c (NO_PIE_SPEC): Delete.
2706 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2708 (LINK_PIE_SPEC): ..to here.
2709 (LINK_COMMAND_SPEC): Support -no-pie.
2710 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2711 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2713 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2714 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2715 (ENDFILE_CRTEND_SPEC): Similarly.
2717 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2720 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2721 (mstack-protector-guard-offset=): Ditto.
2722 * config/i386/i386.c (ix86_option_override): Handle
2723 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2725 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2726 ix86_stack_protect_guard_offset variables.
2727 (TARGET_STACK_PROTECT_GUARD): Always define.
2728 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2729 and -mstack-protector-guard-offset= options.
2731 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2733 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2734 boundary case for the last candidate.
2736 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2738 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2739 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2741 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2744 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2745 * config/aarch64/aarch64-builtins.c
2746 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2747 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2748 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2750 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2751 Andrew Pinski <pinskia@gmail.com>
2754 * internal-fn.def (XORSIGN): New.
2755 * optabs.def (xorsign_optab): New.
2756 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2757 (convert_expand_mult_copysign): New.
2758 (pass_optimize_widening_mul::execute): Call
2759 convert_expand_mult_copysign.
2761 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2763 PR tree-optimization/81354
2764 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2765 Insert on edges rather than explicitly creating landing pads.
2766 (analyze_candidates_and_replace): Commit edge inserts.
2768 2017-08-08 Richard Biener <rguenther@suse.de>
2771 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2772 (expand_simple_operations): Also look through ADDR_EXPRs with
2773 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2775 2017-08-08 Richard Biener <rguenther@suse.de>
2777 PR tree-optimization/81723
2778 * tree-vect-slp.c (struct bst_traits): New hash traits.
2779 (bst_fail): New global.
2780 (vect_build_slp_tree_2): New worker, split out from ...
2781 (vect_build_slp_tree): ... this now wrapping it with using
2782 bst_fail set to cache SLP tree build fails. Properly handle
2784 (vect_analyze_slp_instance): Allocate and free bst_fail.
2786 2017-08-08 Martin Liska <mliska@suse.cz>
2789 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2790 LABEL_DECLs that can be from a different function.
2792 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2794 PR tree-optimization/81744
2795 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2796 loop's number of iterations.
2798 2017-08-08 Martin Liska <mliska@suse.cz>
2800 * asan.c: Include header files.
2801 * attribs.c (build_decl_attribute_variant): New function moved
2803 (build_type_attribute_qual_variant): Likewise.
2804 (cmp_attrib_identifiers): Likewise.
2805 (simple_cst_list_equal): Likewise.
2806 (omp_declare_simd_clauses_equal): Likewise.
2807 (attribute_value_equal): Likewise.
2808 (comp_type_attributes): Likewise.
2809 (build_type_attribute_variant): Likewise.
2810 (lookup_ident_attribute): Likewise.
2811 (remove_attribute): Likewise.
2812 (merge_attributes): Likewise.
2813 (merge_type_attributes): Likewise.
2814 (merge_decl_attributes): Likewise.
2815 (merge_dllimport_decl_attributes): Likewise.
2816 (handle_dll_attribute): Likewise.
2817 (attribute_list_equal): Likewise.
2818 (attribute_list_contained): Likewise.
2819 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2820 (lookup_attribute_by_prefix): Likewise.
2821 * bb-reorder.c: Include header files.
2822 * builtins.c: Likewise.
2823 * calls.c: Likewise.
2824 * cfgexpand.c: Likewise.
2825 * cgraph.c: Likewise.
2826 * cgraphunit.c: Likewise.
2827 * convert.c: Likewise.
2828 * dwarf2out.c: Likewise.
2829 * final.c: Likewise.
2830 * fold-const.c: Likewise.
2831 * function.c: Likewise.
2832 * gimple-expr.c: Likewise.
2833 * gimple-fold.c: Likewise.
2834 * gimple-pretty-print.c: Likewise.
2835 * gimple.c: Likewise.
2836 * gimplify.c: Likewise.
2837 * hsa-common.c: Likewise.
2838 * hsa-gen.c: Likewise.
2839 * internal-fn.c: Likewise.
2840 * ipa-chkp.c: Likewise.
2841 * ipa-cp.c: Likewise.
2842 * ipa-devirt.c: Likewise.
2843 * ipa-fnsummary.c: Likewise.
2844 * ipa-inline.c: Likewise.
2845 * ipa-visibility.c: Likewise.
2847 * lto-cgraph.c: Likewise.
2848 * omp-expand.c: Likewise.
2849 * omp-general.c: Likewise.
2850 * omp-low.c: Likewise.
2851 * omp-offload.c: Likewise.
2852 * omp-simd-clone.c: Likewise.
2853 * opts-global.c: Likewise.
2854 * passes.c: Likewise.
2855 * predict.c: Likewise.
2856 * sancov.c: Likewise.
2857 * sanopt.c: Likewise.
2858 * symtab.c: Likewise.
2859 * toplev.c: Likewise.
2860 * trans-mem.c: Likewise.
2861 * tree-chkp.c: Likewise.
2862 * tree-eh.c: Likewise.
2863 * tree-into-ssa.c: Likewise.
2864 * tree-object-size.c: Likewise.
2865 * tree-parloops.c: Likewise.
2866 * tree-profile.c: Likewise.
2867 * tree-ssa-ccp.c: Likewise.
2868 * tree-ssa-live.c: Likewise.
2869 * tree-ssa-loop.c: Likewise.
2870 * tree-ssa-sccvn.c: Likewise.
2871 * tree-ssa-structalias.c: Likewise.
2872 * tree-ssa.c: Likewise.
2873 * tree-streamer-in.c: Likewise.
2874 * tree-vectorizer.c: Likewise.
2875 * tree-vrp.c: Likewise.
2877 * ubsan.c: Likewise.
2878 * varasm.c: Likewise.
2879 * varpool.c: Likewise.
2880 * tree.c: Remove functions moved to attribs.[ch].
2882 * config/aarch64/aarch64.c: Add attrs.h header file.
2883 * config/alpha/alpha.c: Likewise.
2884 * config/arc/arc.c: Likewise.
2885 * config/arm/arm.c: Likewise.
2886 * config/avr/avr.c: Likewise.
2887 * config/bfin/bfin.c: Likewise.
2888 * config/c6x/c6x.c: Likewise.
2889 * config/cr16/cr16.c: Likewise.
2890 * config/cris/cris.c: Likewise.
2891 * config/darwin.c: Likewise.
2892 * config/epiphany/epiphany.c: Likewise.
2893 * config/fr30/fr30.c: Likewise.
2894 * config/frv/frv.c: Likewise.
2895 * config/ft32/ft32.c: Likewise.
2896 * config/h8300/h8300.c: Likewise.
2897 * config/i386/winnt.c: Likewise.
2898 * config/ia64/ia64.c: Likewise.
2899 * config/iq2000/iq2000.c: Likewise.
2900 * config/lm32/lm32.c: Likewise.
2901 * config/m32c/m32c.c: Likewise.
2902 * config/m32r/m32r.c: Likewise.
2903 * config/m68k/m68k.c: Likewise.
2904 * config/mcore/mcore.c: Likewise.
2905 * config/microblaze/microblaze.c: Likewise.
2906 * config/mips/mips.c: Likewise.
2907 * config/mmix/mmix.c: Likewise.
2908 * config/mn10300/mn10300.c: Likewise.
2909 * config/moxie/moxie.c: Likewise.
2910 * config/msp430/msp430.c: Likewise.
2911 * config/nds32/nds32-isr.c: Likewise.
2912 * config/nds32/nds32.c: Likewise.
2913 * config/nios2/nios2.c: Likewise.
2914 * config/nvptx/nvptx.c: Likewise.
2915 * config/pa/pa.c: Likewise.
2916 * config/pdp11/pdp11.c: Likewise.
2917 * config/powerpcspe/powerpcspe.c: Likewise.
2918 * config/riscv/riscv.c: Likewise.
2919 * config/rl78/rl78.c: Likewise.
2920 * config/rx/rx.c: Likewise.
2921 * config/s390/s390.c: Likewise.
2922 * config/sh/sh.c: Likewise.
2923 * config/sol2.c: Likewise.
2924 * config/sparc/sparc.c: Likewise.
2925 * config/spu/spu.c: Likewise.
2926 * config/stormy16/stormy16.c: Likewise.
2927 * config/tilegx/tilegx.c: Likewise.
2928 * config/tilepro/tilepro.c: Likewise.
2929 * config/v850/v850.c: Likewise.
2930 * config/vax/vax.c: Likewise.
2931 * config/visium/visium.c: Likewise.
2932 * config/xtensa/xtensa.c: Likewise.
2934 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2937 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
2938 constraints since the -mupper-regs-* switches have been
2940 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
2941 into a vector from a double word element that was extracted from
2942 another vector, and eliminate extra XXPERMDI instructions.
2943 (vsx_concat_<mode>_2): Likewise.
2944 (vsx_concat_<mode>_3): Likewise.
2945 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
2946 concat to allow optimizing inserts from previous extracts.
2948 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
2950 * config/i386/i386.c (ix86_stack_protect_guard): Generate
2951 memory reference to a SSP offset in TLS address space.
2952 (ix86_print_operand) <case '@'>: Remove.
2953 (ix86_print_operand_punct_valid_p): Remove '@' code.
2954 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
2956 (stack_tls_protect_set_<mode>): Remove.
2957 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
2958 (stack_tls_protect_test_<mode>): Remove.
2959 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
2961 2017-08-07 Olivier Hainque <hainque@adacore.com>
2964 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
2966 2017-08-07 Douglas Rupp <rupp@adacore.com>
2968 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
2969 variable was referenced as multidir in command.
2971 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2974 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
2977 2017-08-07 Martin Liska <mliska@suse.cz>
2979 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
2980 * config/rl78/rl78.c: Add include of attribs.h.
2981 * config/sh/sh.c: Likewise.
2982 * config/v850/v850.c: Likewise.
2984 2017-08-07 Tom de Vries <tom@codesourcery.com>
2987 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
2989 2017-08-07 Martin Liska <mliska@suse.cz>
2991 * config/mips/mips.c: Include attribs.h.
2993 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
2996 * doc/invoke.texi: Document change in behvaior for -Ofast for
2999 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3001 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3003 (aarch64_pop_regs): Likewise.
3004 (aarch64_gen_load_pair): Likewise.
3005 (aarch64_save_callee_saves): Likewise.
3006 (aarch64_restore_callee_saves): Likewise.
3008 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3010 * config/i386/i386.c: Revert the last change.
3012 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3015 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3017 (ix86_finalize_stack_frame_flags): This. Also clear
3018 frame_pointer_needed if -fno-omit-frame-pointer is used without
3020 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3021 with ix86_finalize_stack_frame_flags.
3022 (ix86_expand_epilogue): Likewise.
3023 (ix86_expand_split_stack_prologue): Likewise.
3025 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3028 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3029 to P_AES for Westmere.
3031 2017-08-07 Jonathan Yong <10walls@gmail.com>
3033 * config/i386/mingw.opt (fset-stack-executable): Removed.
3034 * config/i386/cygming.opt (fset-stack-executable): Moved
3036 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3038 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3040 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3042 2017-08-07 Marek Polacek <polacek@redhat.com>
3045 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3047 2017-08-07 Martin Liska <mliska@suse.cz>
3049 * attribs.h (canonicalize_attr_name): New function.
3050 (cmp_attribs): Move from c-format.c and adjusted.
3051 (is_attribute_p): Moved from tree.h.
3052 * tree-inline.c: Add new includes.
3053 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3054 (private_is_attribute_p): Remove.
3055 (private_lookup_attribute): Likewise.
3056 (private_lookup_attribute_by_prefix): Simplify.
3057 (remove_attribute): Use is_attribute_p.
3058 * tree.h: Remove removed declarations.
3060 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3063 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3064 instead of computing it in the function. Formatting fix.
3065 (expand_case): Don't rely on default_edge being the first edge,
3066 clear it if removing it, pass default_edge to
3067 emit_case_dispatch_table.
3068 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3071 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3073 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3074 insn in the function, emit NOP after the insn.
3076 2017-08-06 Tom de Vries <tom@codesourcery.com>
3078 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3081 2017-08-06 Tom de Vries <tom@codesourcery.com>
3083 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3086 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3088 PR tree-optimization/57371
3089 * match.pd: New pattern.
3091 2017-08-04 Marek Polacek <polacek@redhat.com>
3094 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3095 perform the computation in offset_int.
3097 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3099 PR tree-optimization/81136
3100 * tree-vectorizer.h: Include tree-hash-traits.h.
3101 (vec_base_alignments): New typedef.
3102 (vec_info): Add a base_alignments field.
3103 (vect_record_base_alignments): Declare.
3104 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3106 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3107 (create_data_ref): Add an is_conditional_in_stmt argument.
3108 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3109 the is_conditional_in_stmt field.
3110 (data_ref_loc): Add an is_conditional_in_stmt field.
3111 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3112 (find_data_references_in_stmt): Update call to create_data_ref.
3113 (graphite_find_data_references_in_stmt): Likewise.
3114 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3115 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3116 (vect_record_base_alignment): New function.
3117 (vect_record_base_alignments): Likewise.
3118 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3119 for nested statements even if we fail to compute a misalignment.
3120 Use pooled base alignments for unconditional references.
3121 (vect_find_same_alignment_drs): Compare base addresses instead
3123 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3124 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3126 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3128 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3129 Add an explicit name for the enum. Use auto_vec for slp_instances
3131 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3133 (_bb_vec_info): Add a constructor and destructor.
3134 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3135 (destroy_loop_vec_info): Delete.
3136 (vect_destroy_datarefs): Likewise.
3137 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3138 (vec_info::vec_info): New function.
3139 (vec_info::~vec_info): Likewise.
3140 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3141 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3142 destroy_loop_vec_info.
3143 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3144 (_loop_vec_info::_loop_vec_info): ...this.
3145 (destroy_loop_vec_info): Replace with...
3146 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3147 the stmt_vec_infos. Leave handling of vec_info information to its
3148 destructor. Remove explicit vector releases.
3149 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3150 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3151 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3152 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3153 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3154 (destroy_bb_vec_info): Replace with...
3155 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3156 information to its destructor.
3157 (vect_slp_analyze_bb_1): Use new and delete instead of
3158 new_bb_vec_info and destroy_bb_vec_info.
3159 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3162 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3164 * tree-data-ref.h (subscript): Add access_fn field.
3165 (data_dependence_relation): Add could_be_independent_p.
3166 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3167 (same_access_functions): Move to tree-data-ref.c.
3168 * tree-data-ref.c (ref_contains_union_access_p): New function.
3169 (access_fn_component_p): Likewise.
3170 (access_fn_components_comparable_p): Likewise.
3171 (dr_analyze_indices): Add a reference to access_fn_component_p.
3172 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3174 (constant_access_functions): Likewise.
3175 (add_other_self_distances): Likewise.
3176 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3177 (initialize_data_dependence_relation): Use XCNEW and remove
3178 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3179 of access functions that have the same type. Allow the
3180 subsequence to end with different bases in some circumstances.
3181 Record the chosen access functions in SUB_ACCESS_FN.
3182 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3183 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3184 (subscript_dependence_tester_1): Likewise dra and drb.
3185 (build_classic_dist_vector): Update calls accordingly.
3186 (subscript_dependence_tester): Likewise.
3187 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3188 DDR_COULD_BE_INDEPENDENT_P.
3189 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3190 comp_alias_ddrs instead of may_alias_ddrs.
3191 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3193 (vect_analyze_data_ref_dependence): Use it if
3194 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3195 distance vectors if that fails.
3196 (dependence_distance_ge_vf): New function.
3197 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3198 LOOP_VINFO_MAY_ALIAS_DDRS.
3200 2017-08-04 Richard Biener <rguenther@suse.de>
3203 * fold-const.c (fold_binary_loc): Properly restrict
3204 minus_var0 && minus_var1 case when associating undefined overflow
3207 2017-08-04 Tom de Vries <tom@codesourcery.com>
3209 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3211 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3214 Don't start diagnostic messages with a capital letter.
3215 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3217 (rs6000_invalid_builtin): Likewise.
3218 (rs6000_trampoline_init): Likewise.
3220 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3223 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3224 after setting changeable df flags.
3226 2017-08-03 Richard Biener <rguenther@suse.de>
3228 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3229 up if the use is in USE - X.
3231 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3233 * toplev.c (dumpfile.h): New include.
3234 (internal_error_reentered): New static function. Use it...
3235 (internal_error_function): ...here to handle reentered internal_error.
3237 2017-08-03 Richard Biener <rguenther@suse.de>
3240 * fold-const.c (split_tree): Add minus_var and minus_con
3241 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3242 here but always use minus_*.
3243 (associate_trees): Assert we never associate with MINUS_EXPR
3244 and NULL first operand. Do not recurse for PLUS_EXPR operands
3245 when associating as MINUS_EXPR either.
3246 (fold_binary_loc): Track minus_var and minus_con.
3248 2017-08-03 Tom de Vries <tom@codesourcery.com>
3251 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3252 ACCEL_COMPILER, apply finish_options on
3253 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3255 2017-08-03 Tom de Vries <tom@codesourcery.com>
3258 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3259 function_entry_patch_area_size > 0.
3261 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3264 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3265 instead of 10??LU, perform unit multiplication in wide_int,
3266 don't change alloc_object_size_limit if the limit is larger
3269 PR tree-optimization/81655
3270 PR tree-optimization/81588
3271 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3272 the case when ranges[i].low and high are 1 for unsigned type with
3276 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3277 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3279 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3281 * tree-vrp.h: Add include guard.
3283 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3286 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3287 (ud2): New insn pattern.
3288 * config/i386/i386.c (ix86_expand_epilogue):
3289 For naked functions, generate ud2 instead of trap insn.
3291 2017-08-02 Marek Polacek <polacek@redhat.com>
3294 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3296 2017-08-02 Tom de Vries <tom@codesourcery.com>
3297 Cesar Philippidis <cesar@codesourcery.com>
3299 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3300 Add missing edge probabilities.
3302 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3304 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3307 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3310 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3311 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3312 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3313 prologue_seq sequences - if any.
3315 2017-08-02 Richard Biener <rguenther@suse.de>
3317 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3318 via vectors if supported, integer extracts via punning if supported
3319 or otherwise vector extracts.
3321 2017-08-02 Richard Biener <rguenther@suse.de>
3323 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3325 (bitmap_insert_into_set): ... this.
3327 2017-08-02 Richard Biener <rguenther@suse.de>
3329 PR tree-optimization/81633
3331 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3333 PR tree-optimization/71752
3334 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3336 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3338 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3339 (machine_function::call_ms2sysv_pad_out): Remove field.
3340 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3341 (ix86_compute_frame_layout): Likewise.
3343 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3346 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3347 attribute with interrupt attribute.
3349 2017-08-01 Andrew Pinski <apinski@cavium.com>
3351 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3352 BIT_INSERT_EXPR's operand 1
3353 to see if the types precision matches.
3355 2017-08-01 Martin Liska <mliska@suse.cz>
3358 * builtins.c (expand_builtin_memcpy_args): Remove.
3359 (expand_builtin_memcpy): Call newly added function
3360 expand_builtin_memory_copy_args.
3361 (expand_builtin_memcpy_with_bounds): Likewise.
3362 (expand_builtin_mempcpy): Remove last argument.
3363 (expand_builtin_mempcpy_with_bounds): Likewise.
3364 (expand_builtin_memory_copy_args): New function created from
3365 expand_builtin_mempcpy_args with small modifications.
3366 (expand_builtin_mempcpy_args): Remove.
3367 (expand_builtin_stpcpy): Remove unused argument.
3368 (expand_builtin): Likewise.
3369 (expand_builtin_with_bounds): Likewise.
3371 2017-08-01 Martin Liska <mliska@suse.cz>
3374 Make mempcpy more optimal (PR middle-end/70140).
3376 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3379 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3380 __builtin_vec_cmpne verify both arguments are compatible vectors
3381 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3382 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3383 move computation of aligned to after checking the argument types.
3387 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3388 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3391 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3392 Doug Rupp <rupp@adacore.com>
3393 Olivier Hainque <hainque@adacore.com>
3395 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3396 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3398 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3399 for TARGET_OS_CPP_BUILTIN.
3400 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3401 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3403 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3404 passing required abi options to the assembler for EABI configurations.
3405 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3406 of .text.hot and .text.unlikely sections for kernel modules when
3407 using ARM style exceptions.
3408 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3409 options. Add EXTRA_CC1_SPEC.
3410 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3412 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3414 (ARM_TARGET2_DWARF_FORMAT): Define.
3415 * config/arm/t-vxworks: Adjust multilib control to removal of the
3416 Diab command line options.
3418 2017-08-01 Martin Liska <mliska@suse.cz>
3420 PR gcov-profile/81561
3421 * gcov.c (unblock): Make unblocking safe as we need to preserve
3422 index correspondence of blocks and block_lists.
3424 2017-08-01 Richard Biener <rguenther@suse.de>
3426 PR tree-optimization/81181
3427 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3428 (compute_antic): ... end of iteration here.
3430 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3432 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3433 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3434 (ftree-slp-vectorize): Likewise.
3435 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3436 can no longer be set independent of flag_tree_loop_vectorize.
3437 * omp-general.c (emp_max_vf): Likewise.
3438 * opts.c (enable_fdo_optimizations): Remove references to
3439 flag_tree_vectorize, these are now implicit.
3440 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3441 and leave it for the options machinery.
3443 2017-08-01 Martin Liska <mliska@suse.cz>
3446 * builtins.c (expand_builtin_memcpy_args): Remove.
3447 (expand_builtin_memcpy): Call newly added function
3448 expand_builtin_memory_copy_args.
3449 (expand_builtin_memcpy_with_bounds): Likewise.
3450 (expand_builtin_mempcpy): Remove last argument.
3451 (expand_builtin_mempcpy_with_bounds): Likewise.
3452 (expand_builtin_memory_copy_args): New function created from
3453 expand_builtin_mempcpy_args with small modifications.
3454 (expand_builtin_mempcpy_args): Remove.
3455 (expand_builtin_stpcpy): Remove unused argument.
3456 (expand_builtin): Likewise.
3457 (expand_builtin_with_bounds): Likewise.
3459 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3462 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3463 print "ds:" only for immediates in generic address space.
3465 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3468 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3469 (ix86_function_ok_for_sibcall): Return false for naked functions.
3471 2017-08-01 Richard Biener <rguenther@suse.de>
3473 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3474 (compute_antic): Seed worklist with exit block predecessors.
3475 * cfganal.c (dfs_find_deadend): For a cycle return the source
3476 of the edge closing it.
3478 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3480 * config/aarch64/aarch64.c
3481 (aarch64_can_const_movi_rtx_p): Move 0 check.
3483 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3485 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3486 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3488 * match.pd: Ditto in address comparison pattern.
3490 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3492 PR tree-optimization/81627
3493 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3494 closed ssa form for store-store chain.
3496 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3498 PR tree-optimization/81620
3499 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3500 for store-store chain.
3502 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3504 PR tree-optimization/81588
3505 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3506 ranges[i].in_p, invert comparison code ccode. For >/>=,
3507 swap rhs1 and rhs2 and comparison code unconditionally,
3508 for </<= don't do that. Don't swap rhs1/rhs2 again if
3509 ranges[i].in_p, instead invert comparison code ccode if
3510 opcode or oe->rank is BIT_IOR_EXPR.
3513 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3514 a direct optab to conversion optab.
3515 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3516 with GET_MODE_INNER as last argument instead of optab_handler.
3517 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3518 vector extraction if possible and optab is available.
3519 * expr.c (store_constructor): Use convert_optab_handler instead
3520 of optab_handler. Use vector initialization from smaller
3521 vectors if possible and optab is available.
3522 * tree-vect-stmts.c (vectorizable_load): Likewise.
3523 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3525 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3526 of vec_init from half-sized vectors with the same element mode.
3527 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3528 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3529 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3530 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3531 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3532 after mode in gen_vec_extract* calls.
3533 (vec_extract<mode>): Renamed to ...
3534 (vec_extract<mode><ssescalarmodelower>): ... this.
3535 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3536 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3537 element mode after mode in gen_vec_init* calls.
3538 (VEC_INIT_HALF_MODE): New mode iterator.
3539 (vec_init<mode>): Renamed to ...
3540 (vec_init<mode><ssescalarmodelower>): ... this.
3541 (vec_init<mode><ssehalfvecmodelower>): New expander.
3542 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3543 (vec_extractv2sfsf): ... this.
3544 (vec_initv2sf): Renamed to ...
3545 (vec_initv2sfsf): ... this.
3546 (vec_extractv2si): Renamed to ...
3547 (vec_extractv2sisi): ... this.
3548 (vec_initv2si): Renamed to ...
3549 (vec_initv2sisi): ... this.
3550 (vec_extractv4hi): Renamed to ...
3551 (vec_extractv4hihi): ... this.
3552 (vec_initv4hi): Renamed to ...
3553 (vec_initv4hihi): ... this.
3554 (vec_extractv8qi): Renamed to ...
3555 (vec_extractv8qiqi): ... this.
3556 (vec_initv8qi): Renamed to ...
3557 (vec_initv8qiqi): ... this.
3558 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3559 (vec_init<mode>): Renamed to ...
3560 (vec_init<mode><VEC_base_l>): ... this.
3561 (vec_extract<mode>): Renamed to ...
3562 (vec_extract<mode><VEC_base_l>): ... this.
3563 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3564 (vec_initv2sfsf): ... this.
3565 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3566 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3567 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3568 element mode after mode in gen_vec_init* calls.
3569 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3570 (vec_init<mode><Vel>): ... this.
3571 (vec_extract<mode>): Renamed to ...
3572 (vec_extract<mode><Vel>): ... this.
3573 * config/aarch64/iterators.md (Vel): New mode attribute.
3574 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3575 Add element mode after mode in gen_vec_extract* calls.
3576 * config/s390/vector.md (non_vec_l): New mode attribute.
3577 (vec_extract<mode>): Renamed to ...
3578 (vec_extract<mode><non_vec_l>): ... this.
3579 (vec_init<mode>): Renamed to ...
3580 (vec_init<mode><non_vec_l>): ... this.
3581 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3582 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3584 * config/arm/iterators.md (V_elem_l): New mode attribute.
3585 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3586 (vec_extract<mode><V_elem_l>): ... this.
3587 (vec_extractv2di): Renamed to ...
3588 (vec_extractv2didi): ... this.
3589 (vec_init<mode>): Renamed to ...
3590 (vec_init<mode><V_elem_l>): ... this.
3591 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3592 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3593 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3594 Add element mode after gen_vec_extract* calls.
3595 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3596 (vec_init<mode><unitmode>): ... this.
3597 (vec_extract<mode>): Renamed to ...
3598 (vec_extract<mode><unitmode>): ... this.
3599 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3600 (vec_init<mode><unitmode>): ... this.
3601 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3602 (vec_initv2sfsf): ... this.
3603 (vec_extractv2sf): Renamed to ...
3604 (vec_extractv2sfsf): ... this.
3605 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3606 Add element mode after gen_vec_extract* calls.
3607 * config/mips/mips.md (unitmode): New mode iterator.
3608 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3609 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3610 * config/spu/spu.md (inner_l): New mode attribute.
3611 (vec_init<mode>): Renamed to ...
3612 (vec_init<mode><inner_l>): ... this.
3613 (vec_extract<mode>): Renamed to ...
3614 (vec_extract<mode><inner_l>): ... this.
3615 * config/sparc/sparc.md (veltmode): New mode iterator.
3616 (vec_init<VMALL:mode>): Renamed to ...
3617 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3618 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3619 (vec_initv2sisi): ... this.
3620 (vec_initv2sf): Renamed to ...
3621 (vec_initv2sfsf): ... this.
3622 (vec_extractv2sf): Renamed to ...
3623 (vec_extractv2sfsf): ... this.
3624 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3625 (vec_init<mode>): Renamed to ...
3626 (vec_init<mode><VEC_base_l>): ... this.
3627 (vec_extract<mode>): Renamed to ...
3628 (vec_extract<mode><VEC_base_l>): ... this.
3629 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3630 (vec_initv2sfsf): ... this.
3631 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3632 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3633 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3634 gen_vec_init* calls.
3636 2017-08-01 Richard Biener <rguenther@suse.de>
3638 PR tree-optimization/81297
3639 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3640 TREE_OVERFLOW from INTEGER_CSTs.
3642 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3644 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3646 2017-07-31 Carl Love <cel@us.ibm.com>
3648 * config/rs6000/rs6000-c: Add support for built-in functions
3649 vector signed char vec_xl_be (signed long long, signed char *);
3650 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3651 vector signed int vec_xl_be (signed long long, signed int *);
3652 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3653 vector signed long long vec_xl_be (signed long long, signed long long *);
3654 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3655 vector signed short vec_xl_be (signed long long, signed short *);
3656 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3657 vector double vec_xl_be (signed long long, double *);
3658 vector float vec_xl_be (signed long long, float *);
3659 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3660 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3661 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3663 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3664 (altivec_expand_builtin): Add switch statement to call
3665 altivec_expand_xl_be for each builtin.
3666 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3667 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3668 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3669 __builtin_vsx_le_be_v16qi.
3670 * doc/extend.texi: Update the built-in documentation file for the
3671 new built-in functions.
3673 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3676 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3678 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3680 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3682 * config.gcc: Add z14.
3683 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3684 CPU model numbers for z13s and z14.
3685 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3687 * config/s390/s390-opts.h (enum processor_type): Rename
3688 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3689 * config/s390/s390.c (processor_table): Add field for CPU name to
3690 be passed to Binutils.
3691 (s390_asm_output_machine_for_arch): Use the new field in
3692 processor_table for Binutils.
3693 (s390_expand_builtin): Replace arch12 with z14.
3694 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3695 (s390_get_sched_attrmask): Likewise.
3696 (s390_get_unit_mask): Likewise.
3697 * config/s390/s390.opt: Add z14 to processor_type enum.
3699 2017-07-31 Martin Jambor <mjambor@suse.cz>
3702 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3703 regardless of optimization level.
3705 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3706 Martin Liska <mliska@suse.cz>
3708 * predict.def: Remove old comment and adjust probability.
3709 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3712 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3715 * config/i386/i386.c (ix86_function_naked): New function.
3716 (ix86_can_use_return_insn_p): Return false for naked functions.
3717 (ix86_expand_prologue): Skip prologue for naked functions.
3718 (ix86_expand_epilogue): Skip epilogue for naked functions
3719 and emit trap instruction.
3720 (ix86_warn_func_return): New function.
3721 (ix86_attribute_table): Add "naked" attribute specification.
3722 (TARGET_WARN_FUNC_RETURN): Define.
3723 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3725 2017-07-31 Martin Liska <mliska@suse.cz>
3727 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3728 (dump_gimple_bb_header): Always dump BB info.
3729 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3731 2017-07-31 Martin Liska <mliska@suse.cz>
3734 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3735 also with current_function_decl non-null equality.
3737 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3740 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3741 change type to the element type, instead add eltype variable and
3742 use it where we are interested in the element type.
3744 PR tree-optimization/81603
3745 * ipa-polymorphic-call.c
3746 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3747 offset arithmetic in offset_int, bail out if the resulting bit offset
3748 doesn't fit into shwi.
3750 2017-07-31 Martin Liska <mliska@suse.cz>
3752 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3753 (gimplify_save_expr): Fix comment.
3755 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3758 * config/i386/i386.c (ix86_function_arg): Update arguments for
3760 (ix86_compute_frame_layout): Set the initial stack offset to
3761 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3762 INCOMING_FRAME_SP_OFFSET.
3763 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3764 stack before exception handler returns.
3765 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3766 the 'ERROR_CODE' for exception handler.
3768 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3770 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3771 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3772 (ASM_OUTPUT_REG_POP): Ditto.
3773 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3774 instead of asm_fprintf to output pure string.
3776 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3778 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3779 to imported_module_or_decl hook.
3780 (debug_nothing_tree_tree_tree_bool): Remove.
3781 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3782 * debug.c (do_nothing_debug_hooks): Use
3783 debug_nothing_tree_tree_tree_bool_bool instead of
3784 debug_nothing_tree_tree_tree_bool.
3785 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3786 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3787 * sdbout.c (sdb_debug_hooks): Likewise.
3788 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3789 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3791 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3792 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3793 attribute, don't add anything.
3795 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3797 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3798 (fold_build2_stat_loc): Likewise.
3799 (fold_build3_stat_loc): Likewise.
3800 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3801 (fold_build1_loc): Remove macro.
3802 (fold_build2_loc): Likewise.
3803 (fold_build3_loc): Likewise.
3805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3807 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3808 (gimple_build_debug_bind_source_stat): Likewise.
3809 * gimple.h (gimple_build_debug_bind): Remove macro.
3810 (gimple_build_debug_bind_source): Likewise.
3812 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3814 * bitmap.c (bitmap_alloc): Adjust.
3815 (bitmap_gc_alloc): Likewise.
3816 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3818 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3820 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3821 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3822 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
3823 (bitmap_gc_alloc_stat): Likewise.
3824 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
3826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3828 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
3829 * rtl.h (shallow_copy_rtx): Remove macro.
3831 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3833 * emit-rtl.c (gen_raw_REG): Adjust.
3834 * gengenrtl.c (gendef): Likewise.
3835 * rtl.c (rtx_alloc_stat): Remove _stat from name.
3836 * rtl.h (rtx_alloc): Remove macro.
3838 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3840 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
3841 (build_tree_list_stat): Likewise.
3842 * tree.h (build_tree_list): Remove macro.
3843 (build_tree_list_vec): Likewise.
3845 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3847 * tree.c (make_vector_stat): Remove _stat from name.
3848 (build_vector_stat): Likewise.
3849 * tree.h (make_vector_stat): Remove macro.
3850 (build_vector_stat): Likewise.
3852 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3854 * tree.h (build_var_debug_value): Remove prototype.
3856 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3858 * tree.c (tree_cons_stat): Remove _stat from name.
3859 * tree.h (tree_cons): Remove macro.
3861 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3863 * tree.c (build_vl_exp_stat): Remove _stat from name.
3864 * tree.h (build_vl_exp): Remove macro.
3866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3868 * tree.c (build_decl_stat): Remove _stat from name.
3869 * tree.h (build_decl): Remove macro.
3871 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3873 * gimple.c (gimple_build_with_ops_stat): Adjust.
3874 (gimple_alloc_stat): Remove _stat from name.
3875 * gimple.h (gimple_alloc): Remove macro.
3877 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3879 * tree.c (make_tree_vec_stat): Remove _stat from name.
3880 (grow_tree_vec_stat): Likewise.
3881 * tree.h (make_tree_vec_stat): Adjust prototype.
3882 (grow_tree_vec_stat): Likewise.
3883 (make_tree_vec): Remove macro.
3884 (grow_tree_vec): Likewise.
3886 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3888 * fold-const.c (fold_build1_stat_loc): Adjust.
3889 (fold_build2_stat_loc): Likewise.
3890 (fold_build3_stat_loc): Likewise.
3891 * tree.c (build0_stat): Remove _stat from name.
3892 (build1_stat): Likewise.
3893 (build2_stat): Likewise.
3894 (build3_stat): Likewise.
3895 (build4_stat): Likewise.
3896 (build5_stat): Likewise.
3897 * tree.h (build1_loc): Remove macro, and rename _stat function
3899 (build2_loc): Likewise.
3900 (build3_loc): Likewise.
3901 (build4_loc): Likewise.
3902 (build5_loc): Likewise.
3904 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3906 * tree.c (make_int_cst_stat): Remove _stat from name.
3907 * tree.h (make_int_cst_stat): Adjust prototype.
3908 (make_int_cst): Remove macro.
3910 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3912 * tree.c (make_tre_binfo_stat): Remove _stat from name.
3913 * tree.h (make_tree_binfo_stat): Adjust prototype.
3914 (make_tree_binfo): Remove.
3916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3918 * tree.c (copy_node_stat): Rename to copy_node.
3919 (build_distinct_type_copy): Adjust.
3920 * tree.h (copy_node_stat): Adjust prototype.
3921 (copy_node): Remove macro.
3923 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3925 * tree.c (make_node_stat): rename to make_node.
3926 (build_tree_list_stat): Adjust.
3927 (build0_stat): Likewise.
3928 (build2_stat): Likewise.
3929 (build3_stat): Likewise.
3930 (build4_stat): Likewise.
3931 (build5_stat): Likewise.
3932 (build_decl_stat): Likewise.
3933 * tree.h (make_node_stat): Adjust prototype.
3934 (make_node): remove macro.
3936 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
3938 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
3939 (PPC_FEATURE2_SCV): Likewise.
3940 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
3942 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3944 * config/aarch64/aarch64.c
3945 (aarch64_internal_mov_immediate): Add new special pattern.
3946 * config/aarch64/aarch64.md (*movdi_aarch64):
3947 Add reg/32bit const mov case.
3949 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3950 Richard Sandiford <richard.sandiford@linaro.org>
3952 * config/aarch64/aarch64.md (mov<mode>): Generalize.
3953 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3954 Add integer and movi cases.
3955 (movi-split-hf-df-sf split, fp16): New.
3956 (enabled): Added TARGET_FP_F16INST.
3957 * config/aarch64/iterators.md (GPF_HF): New.
3958 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
3960 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3962 * config/aarch64/aarch64.c
3963 (aarch64_simd_container_mode): Add prototype.
3964 (aarch64_expand_mov_immediate): Add HI support.
3965 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
3966 (aarch64_can_const_movi_rtx_p): New.
3967 (aarch64_preferred_reload_class):
3968 Remove restrictions of using FP registers for certain SIMD operations.
3969 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
3970 (aarch64_valid_floating_const): Add integer move validation.
3971 (aarch64_simd_imm_scalar_p): Remove.
3972 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
3973 (aarch64_legitimate_constant_p): Expand list of supported cases.
3974 * config/aarch64/aarch64-protos.h
3975 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
3976 (aarch64_reinterpret_float_as_int): New.
3977 (aarch64_simd_imm_scalar_p): Remove.
3978 * config/aarch64/constraints.md (Uvi): New.
3979 (Dd): Split into Ds and new Dd.
3980 * config/aarch64/aarch64.md (*movsi_aarch64):
3982 (*movdi_aarch64): Add SIMD mov case.
3984 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3986 * tree-predcom.c: (struct chain): Handle store-store chain in which
3987 stores for elimination only store loop invariant values.
3988 (execute_pred_commoning_chain): Ditto.
3989 (prepare_initializers_chain_store_elim): Ditto.
3990 (prepare_finalizers): Ditto.
3991 (is_inv_store_elimination_chain): New function.
3992 (initialize_root_vars_store_elim_1): New function.
3994 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3996 * tree-predcom.c: Revise general description of the pass.
3997 (enum chain_type): New enum type for store elimination.
3998 (struct chain): New field supporting store elimination.
3999 (struct component): Ditto.
4000 (dump_chain): Dump store-stores chain.
4001 (release_chain): Release resources.
4002 (split_data_refs_to_components): Compute and create component
4003 contains only stores for elimination.
4004 (get_chain_last_ref_at): New function.
4005 (make_invariant_chain): Initialization.
4006 (make_rooted_chain): Specify chain type in parameter and record it.
4007 (add_looparound_copies): Skip for store-stores chain.
4008 (determine_roots_comp): Compute type of chain and pass it to
4010 (initialize_root_vars_store_elim_2): New function.
4011 (finalize_eliminated_stores): New function.
4012 (remove_stmt): Handle store for elimination.
4013 (execute_pred_commoning_chain): Execute predictive commoning on
4015 (determine_unroll_factor): Skip unroll for store-stores chain.
4016 (prepare_initializers_chain_store_elim): New function.
4017 (prepare_initializers_chain): Hanlde store-store chain.
4018 (prepare_finalizers_chain, prepare_finalizers): New function.
4019 (tree_predictive_commoning_loop): Return integer value indicating
4020 if loop is unrolled or lcssa form is corrupted.
4021 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4023 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4025 * tree-predcom.c (initialize_root): Delete.
4026 (execute_pred_commoning_chain): Initialize root vars and replace
4027 reference of non-combined chain directly, rather than call above
4030 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4032 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4033 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4035 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4037 * tree-predcom.c (struct chain): New field init_seq.
4038 (release_chain): Release init_seq.
4039 (prepare_initializers_chain): Record intialization stmts in above
4041 (insert_init_seqs): New function.
4042 (tree_predictive_commoning_loop): Call insert_init_seqs.
4044 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4046 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4048 2017-07-28 Richard Biener <rguenther@suse.de>
4050 * match.pd: Remove superfluous :c.
4051 * genmatch.c (simplify::id): Add member.
4052 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4054 (current_id): New global.
4055 (dt_node::parent): Move from ...
4056 (dt_operand::parent): ... here. Add for_id member.
4057 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4058 (decision_tree::find_node): Relax order requirement when
4059 merging DT_TRUE nodes to ones inbetween the current simplify
4060 and the one we try to merge with. Add diagnostic whenever
4061 we need to enforce pattern order by not merging.
4062 (decision_tree::insert): Set current_id.
4063 (decision_tree::print_node): Dump parent node and for_id.
4064 (parser::last_id): Add member.
4065 (parser::push_simplify): Assign unique id.
4066 (parser::parser): Initialize last_id.
4068 2017-07-28 Martin Liska <mliska@suse.cz>
4071 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4072 gimple_build_debug_bind.
4074 2017-07-28 Richard Biener <rguenther@suse.de>
4076 PR tree-optimization/81502
4077 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4079 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4081 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4082 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4083 for BIT_FIELD_REF args.
4084 * fold-const.c (make_bit_field_ref): Likewise.
4085 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4087 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4090 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4091 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4092 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4093 Or it into SANITIZER_UNDEFINED.
4094 * ubsan.c: Include gimple-fold.h and varasm.h.
4095 (ubsan_expand_ptr_ifn): New function.
4096 (instrument_pointer_overflow): New function.
4097 (maybe_instrument_pointer_overflow): New function.
4098 (instrument_object_size): Formatting fix.
4099 (pass_ubsan::execute): Call instrument_pointer_overflow
4100 and maybe_instrument_pointer_overflow.
4101 * internal-fn.c (expand_UBSAN_PTR): New function.
4102 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4103 * sanitizer.def (__ubsan_handle_pointer_overflow,
4104 __ubsan_handle_pointer_overflow_abort): New builtins.
4105 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4106 * internal-fn.def (UBSAN_PTR): New internal function.
4107 * opts.c (sanitizer_opts): Add pointer-overflow.
4108 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4109 * fold-const.c (build_range_check): Compute pointer range check in
4110 integral type if pointer arithmetics would be needed. Formatting
4113 2017-07-28 Martin Liska <mliska@suse.cz>
4116 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4117 parameters that are of a variable-length.
4119 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4121 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4123 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4124 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4125 (CRT_CALL_STATIC_FUNCTION): Likewise.
4126 (ASM_DEFAULT_SPEC): New define.
4127 (ASM_SPEC32): Likewise.
4128 (ASM_SPEC64): Likewise.
4129 (ASM_SPEC_COMMON): Likewise.
4130 (ASM_SPEC): Likewise.
4131 (INVALID_64BIT): Likewise.
4132 (LINK_OS_DEFAULT_SPEC): Likewise.
4133 (LINK_OS_SPEC32): Likewise.
4134 (LINK_OS_SPEC64): Likewise.
4135 (POWERPC_LINUX): Likewise.
4136 (PTRDIFF_TYPE): Likewise.
4137 (RESTORE_FP_PREFIX): Likewise.
4138 (RESTORE_FP_SUFFIX): Likewise.
4139 (SAVE_FP_PREFIX): Likewise.
4140 (SAVE_FP_SUFFIX): Likewise.
4141 (SIZE_TYPE): Likewise.
4142 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4143 (TARGET_64BIT): Likewise.
4144 (TARGET_64BIT): Likewise.
4145 (TARGET_AIX): Likewise.
4146 (WCHAR_TYPE_SIZE): Likewise.
4147 (WCHAR_TYPE): Undefine.
4148 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4149 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4150 (CPP_OS_RTEMS_SPEC): Delete.
4151 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4152 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4154 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4156 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4158 PR tree-optimization/81578
4159 * tree-parloops.c (build_new_reduction): Bail out if
4160 reduction_code isn't one of the standard OpenMP reductions.
4161 Move the details printing after that decision.
4163 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4165 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4166 related to reload_in_progress.
4167 (splat_input_operand): Likewise.
4168 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4170 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4172 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4173 (TARGET_INSTANTIATE_DECLS): Likewise.
4174 (legitimate_indexed_address_p): Delete reload_in_progress code.
4175 (rs6000_debug_legitimate_address_p): Likewise.
4176 (rs6000_eliminate_indexed_memrefs): Likewise.
4177 (rs6000_emit_le_vsx_store): Likewise.
4178 (rs6000_emit_move_si_sf_subreg): Likewise.
4179 (rs6000_emit_move): Likewise.
4180 (register_to_reg_type): Likewise.
4181 (rs6000_pre_atomic_barrier): Likewise.
4182 (rs6000_machopic_legitimize_pic_address): Likewise.
4183 (rs6000_allocate_stack_temp): Likewise.
4184 (rs6000_address_for_fpconvert): Likewise.
4185 (rs6000_address_for_altivec): Likewise.
4186 (rs6000_secondary_memory_needed_rtx): Delete function.
4187 (rs6000_check_sdmode): Likewise.
4188 (rs6000_alloc_sdmode_stack_slot): Likewise.
4189 (rs6000_instantiate_decls): Likewise.
4190 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4191 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4192 Delete reload_in_progress.
4193 (*vec_reload_and_plus_<mptrsize>): Likewise.
4194 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4195 (vsx_div_v2di): Likewise.
4196 (vsx_udiv_v2di): Likewise.
4198 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4200 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4201 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4202 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4203 (rs6000_debug_reg_global): Delete print of LRA status.
4204 (rs6000_option_override_internal): Delete dead LRA related code.
4205 (rs6000_lra_p): Delete function.
4206 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4208 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4210 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4211 * config/riscv/rtems.h: New file.
4213 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4214 Sudakshina Das <sudi.das@arm.com>
4216 * config/aarch64/aarch64.md
4217 (define_split for and<mode>3nr_compare): Move
4218 non aarch64_logical_operand to a register.
4219 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4220 register immediate operand to a register.
4221 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4223 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4226 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4228 2017-07-27 Richard Biener <rguenther@suse.de>
4230 PR tree-optimization/81573
4231 PR tree-optimization/81494
4232 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4233 multi defuse cycle case.
4235 2017-07-27 Richard Biener <rguenther@suse.de>
4237 PR tree-optimization/81571
4238 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4241 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4243 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4244 earlier and only if MASK_FPU is set. Adjust formatting.
4246 2017-07-27 Martin Liska <mliska@suse.cz>
4248 * opt-functions.awk: Add validation of value of Init.
4249 * optc-gen.awk: Pass new argument.
4251 2017-07-27 Martin Liska <mliska@suse.cz>
4253 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4254 Fix wrong condition.
4256 2017-07-27 Martin Liska <mliska@suse.cz>
4258 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4259 BBs and edges seen by autoFDO.
4261 2017-07-27 Richard Biener <rguenther@suse.de>
4263 PR tree-optimization/81502
4264 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4265 with incompatible but same sized type.
4266 (execute_update_addresses_taken): Likewise.
4268 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4270 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4271 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4273 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4276 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4277 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4278 Change s_operand to memory_operand.
4280 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4282 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4283 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4284 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4285 Emit instructions rather than returning an expression. Handle TFmode
4286 and KFmode by casting to TImode.
4287 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4288 (rs6000_emit_le_vsx_store): Likewise.
4289 * config/rs6000/vsx.md (VSX_TI): New iterator.
4290 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4291 (*vsx_le_undo_permute_<mode>): Likewise.
4292 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4293 emit the split sequence.
4294 (*vsx_le_perm_store_<mode>): Likewise.
4296 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4298 PR tree-optimization/81555
4299 PR tree-optimization/81556
4300 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4301 if true, force CHANGED for the recursive invocation.
4302 (reassociate_bb): Remember original length of ops array, pass
4303 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4305 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4306 attributes for noipa attribute. For naked attribute use
4307 lookup_attribute first before lookup_attribute_spec.
4308 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4310 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4311 for functions with noipa attribute.
4312 (cgraph_externally_visible_p): Return true for functions with noipa
4314 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4315 for functions with noipa attribute.
4316 * doc/extend.texi: Document noipa function attribute.
4317 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4318 also for functions with noipa attribute.
4319 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4321 2017-07-26 Andrew Pinski <apinski@cavium.com>
4323 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4324 vec_unalign_load_cost and vec_unalign_store_cost.
4326 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4328 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4329 -mvsx-small-integer option.
4330 (ISA_3_0_MASKS_IEEE): Likewise.
4331 (OTHER_VSX_VECTOR_MASKS): Likewise.
4332 (POWERPC_MASKS): Likewise.
4333 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4334 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4335 code, only testing for DImode being allowed in non-VSX floating
4337 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4338 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4340 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4341 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4342 TARGET_P8_VECTOR test.
4343 (rs6000_secondary_reload_simple_move): Likewise.
4344 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4345 since TARGET_P9_VECTOR was already tested.
4346 (rs6000_opt_masks): Remove -mvsx-small-integer.
4347 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4348 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4350 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4351 test for TARGET_VEXTRACTUB was used, and that uses
4353 (p9 extract splitter): Likewise.
4354 (vsx_extract_<mode>_di_p9): Likewise.
4355 (vsx_extract_<mode>_store_p9): Likewise.
4356 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4357 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4358 the elimination of TARGET_VSX_SMALL_INTEGER.
4359 (vsx_extract_<mode>_p8): Likewise.
4360 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4361 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4362 (vsx_set_<mode>_p9): Likewise.
4363 (vsx_set_v4sf_p9): Likewise.
4364 (vsx_set_v4sf_p9_zero): Likewise.
4365 (vsx_insert_extract_v4sf_p9): Likewise.
4366 (vsx_insert_extract_v4sf_p9_2): Likewise.
4367 * config/rs6000/rs6000.md (sign extend splitter): Change
4368 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4369 (floatsi<mode>2_lfiwax_mem): Likewise.
4370 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4371 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4372 since a test for TARGET_P9_VECTOR was used.
4373 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4374 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4375 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4376 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4377 TARGET_P8_VECTOR test.
4378 (fix_trunc<mode>si2_stfiwx): Likewise.
4379 (fix_trunc<mode>si2_internal): Likewise.
4380 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4381 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4383 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4384 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4385 TARGET_P8_VECTOR test.
4386 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4387 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4388 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4390 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4391 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4392 since a test for TARGET_P9_VECTOR was used.
4393 (splitter for loading small constants): Likewise.
4395 2017-07-26 Andrew Pinski <apinski@cavium.com>
4397 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4400 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4403 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4404 (fp_valid_at): Likewise.
4406 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4408 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4409 (qdf24xx_addrcost_table): Likewise.
4410 (cortexa57_tunings): Update to use generic_branch_cost.
4411 (cortexa72_tunings): Likewise.
4412 (cortexa73_tunings): Likewise.
4413 (qdf24xx_tunings): Likewise.
4415 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4417 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4418 (thunderx2t99_branch_cost): Likewise.
4419 (cortexa35_tunings): Update to use generic_branch_cost.
4420 (cortexa53_tunings): Likewise.
4421 (cortexa57_tunings): Likewise.
4422 (cortexa72_tunings): Likewise.
4423 (cortexa73_tunings): Likewise.
4424 (thunderx2t99_tunings): Likewise.
4426 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4428 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4429 (sparc_option_override): Honour MASK_FSMULD.
4430 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4431 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4432 * config/sparc/sparc.opt (mfsmuld): New option.
4433 * doc/invoke.texi (mfsmuld): Document option.
4435 2017-07-26 Marek Polacek <polacek@redhat.com>
4438 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4440 2017-07-26 Richard Biener <rguenther@suse.de>
4442 * gimple-match-head.c (do_valueize): Return OP if valueize
4444 (get_def): New helper to get at the def stmt of a SSA name
4446 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4447 do_valueize to get at the def stmt.
4448 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4450 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4453 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4455 2017-07-26 Martin Liska <mliska@suse.cz>
4458 * function.c (expand_function_start): Make expansion of
4459 nonlocal_goto_save_area after parm_birth_insn.
4461 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4463 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4464 from all CPU target flags enable members.
4466 2017-07-26 Richard Biener <rguenther@suse.de>
4468 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4469 (decision_tree::gen): Make 'type' const.
4470 (write_predicate): Likewise.
4472 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4474 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4475 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4476 (rs6000_option_override_internal): Likewise.
4477 (rs6000_expand_vector_set): Likewise.
4478 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4479 (TARGET_UPPER_REGS_SF): Likewise.
4480 (TARGET_UPPER_REGS_DI): Likewise.
4481 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4482 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4483 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4484 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4485 (Splitters for DI constants in Altivec registers): Likewise.
4486 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4487 (vsx_set_v4sf_p9): Likewise.
4488 (vsx_set_v4sf_p9_zero): Likewise.
4489 (vsx_insert_extract_v4sf_p9): Likewise.
4490 (vsx_insert_extract_v4sf_p9_2): Likewise.
4492 2017-07-25 Carl Love <cel@us.ibm.com>
4494 * doc/extend.texi: Update the built-in documentation file for the
4495 existing built-in functions
4496 vector signed char vec_cnttz (vector signed char);
4497 vector unsigned char vec_cnttz (vector unsigned char);
4498 vector signed short vec_cnttz (vector signed short);
4499 vector unsigned short vec_cnttz (vector unsigned short);
4500 vector signed int vec_cnttz (vector signed int);
4501 vector unsigned int vec_cnttz (vector unsigned int);
4502 vector signed long long vec_cnttz (vector signed long long);
4503 vector unsigned long long vec_cnttz (vector unsigned long long);
4505 2017-07-25 Andrew Pinski <apinski@cavium.com>
4507 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4508 accesses where the use is for the first operand of a BIT_INSERT.
4510 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4513 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4514 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4516 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4518 * config/i386/gstabs.h: Delete.
4519 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4521 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4523 * config/i386/i386.c (ix86_decompose_address): Do not check for
4524 register RTX when looking at index_reg or base_reg.
4525 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4527 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4529 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4530 to update EH info here.
4532 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4534 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4536 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4538 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4540 2017-07-25 Torsten Duwe <duwe@suse.de>
4542 * common.opt: Introduce -fpatchable-function-entry
4543 command line option, and its variables function_entry_patch_area_size
4544 and function_entry_patch_area_start.
4545 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4546 including a two-value parser.
4547 * target.def (print_patchable_function_entry): New target hook.
4548 * targhooks.h (default_print_patchable_function_entry): New function.
4549 * targhooks.c (default_print_patchable_function_entry): Likewise.
4550 * toplev.c (process_options): Switch off IPA-RA if
4551 patchable function entries are being generated.
4552 * varasm.c (assemble_start_function): Look at the
4553 patchable-function-entry command line switch and current
4554 function attributes and maybe generate NOP instructions by
4555 calling the print_patchable_function_entry hook.
4556 * doc/extend.texi: Document patchable_function_entry attribute.
4557 * doc/invoke.texi: Document -fpatchable_function_entry
4558 command line option.
4559 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4561 * doc/tm.texi: Re-generate.
4563 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4566 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4567 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4569 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4571 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4573 2017-07-25 Richard Biener <rguenther@suse.de>
4575 PR tree-optimization/81455
4576 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4577 not walk in cycles when looking for guards.
4579 2017-07-25 Richard Biener <rguenther@suse.de>
4581 PR tree-optimization/81529
4582 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4583 when optimizing backedge uses.
4585 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4587 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4589 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4590 to dl_section_ref. On AIX, append an expression to subtract
4591 the size of the section length to dl_section_ref.
4593 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4595 * configure.ac: If any of the config.* scripts fail, exit 1.
4596 * configure: Regenerate.
4598 2017-07-25 Richard Biener <rguenther@suse.de>
4601 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4602 of immediate uses, be more verbose on errors.
4604 2017-07-25 Richard Biener <rguenther@suse.de>
4606 PR tree-optimization/81510
4607 * tree-vect-loop.c (vect_is_simple_reduction): When the
4608 reduction stmt is not inside the loop bail out.
4610 2017-07-25 Richard Biener <rguenther@suse.de>
4612 PR tree-optimization/81303
4613 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4614 profitability check against LOOP_VINFO_NITERSM1.
4616 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4618 * domwalk.c (cmp_bb_postorder): Simplify.
4619 (sort_bbs_postorder): New function. Use it...
4620 (dom_walker::walk): ...here to optimize common cases.
4622 2017-07-25 Martin Liska <mliska@suse.cz>
4625 * ipa-visibility.c (function_and_variable_visibility): Make the
4626 redirection just on target that supports aliasing.
4627 Fix GNU coding style.
4629 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4632 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4634 * config/arm/rtems.h: Likewise.
4635 * config/bfin/rtems.h: Likewise.
4636 * config/i386/rtemself.h: Likewise.
4637 * config/lm32/rtems.h: Likewise.
4638 * config/m32c/rtems.h: Likewise.
4639 * config/m68k/rtemself.h: Likewise.
4640 * config/microblaze/rtems.h: Likewise.
4641 * config/mips/rtems.h: Likewise.
4642 * config/moxie/rtems.h: Likewise.
4643 * config/nios2/rtems.h: Likewise.
4644 * config/powerpcspe/rtems.h: Likewise.
4645 * config/rs6000/rtems.h: Likewise.
4646 * config/rtems.h: Likewise.
4647 * config/sh/rtems.h: Likewise.
4648 * config/sh/rtemself.h: Likewise.
4649 * config/sparc/rtemself.h: Likewise.
4651 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4654 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4655 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4656 * tree-ssa-structalias.c (alias_get_name): Same.
4658 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4661 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4662 instructions if no du chain is found.
4664 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4666 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4668 2017-07-25 Richard Biener <rguenther@suse.de>
4671 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4674 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4676 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4678 (ISA_2_7_MASKS_SERVER): Likewise.
4679 (ISA_3_0_MASKS_IEEE): Likewise.
4680 (OTHER_P8_VECTOR_MASKS): Likewise.
4681 (OTHER_VSX_VECTOR_MASKS): Likewise.
4682 (POWERPC_MASKS): Likewise.
4683 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4684 duplicate list of options.
4685 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4686 explicit -mupper-regs options.
4687 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4688 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4689 alias for -mupper-regs-df.
4690 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4691 (rs6000_init_hard_regno_mode_ok): Likewise.
4692 (rs6000_option_override_internal): Likewise.
4693 (rs6000_opt_masks): Likewise.
4694 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4695 options in terms of whether -mvsx or -mpower8-vector was used.
4696 (TARGET_UPPER_REGS_DI): Likewise.
4697 (TARGET_UPPER_REGS_SF): Likewise.
4698 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4699 -mupper-regs-* options.
4701 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4703 * passes.c (emergency_dump_function): Print some empty lines and a
4704 header before the RTL dump.
4706 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4708 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4710 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4713 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4714 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4716 2017-07-24 Carl Love <cel@us.ibm.com>
4718 * config/rs6000/rs6000-c.c: Add support for built-in functions
4719 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4720 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4721 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4722 vec_extract_fp_from_shortl): Add defines for the two builtins.
4723 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4724 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4726 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4727 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4728 * doc/extend.texi: Update the built-in documentation file for the
4729 new built-in function.
4731 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4734 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4736 * doc/generic.texi: Likewise.
4737 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4738 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4740 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4742 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4743 (aarch64_mls_elt_merge<mode>): Likewise.
4745 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4747 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4748 having __cxa_atexit.
4750 2017-07-23 Michael Collison <michael.collison@arm.com>
4752 * config/arm/arm.c (arm_option_override): Deprecate
4753 use of -mstructure-size-boundary.
4754 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4755 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4757 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4760 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4761 Reduce cost estimate for direct moves.
4763 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4766 * config/i386/i386.c (ix86_option_override_internal): Disable
4767 BMI, BMI2 and TBM instructions for -m16.
4769 2017-07-21 Carl Love <cel@us.ibm.com>
4771 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4772 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4773 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4774 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4775 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4776 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4777 VMULOSW): New enum "unspec" values.
4778 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4779 altivec_vmulosw): New patterns.
4780 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4781 VMULOSW): Add definitions.
4783 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4785 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4786 (qdf24xx): Likewise.
4787 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4788 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4789 (AARCH64_FL_V8_1): Renumber.
4790 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4791 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4792 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4793 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4794 rdma to feature modifiers list.
4796 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4799 * ipa-visibility (function_and_variable_visibility): Convert
4800 recursive PLT call to direct call if appropriate.
4802 2017-07-21 Andrew Pinski <apinski@cavium.com>
4804 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4805 operand 1 to see if the types precision matches.
4806 * fold-const.c (operand_equal_p): Likewise.
4808 2017-07-21 Richard Biener <rguenther@suse.de>
4810 PR tree-optimization/81303
4811 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4812 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4813 (vect_peeling_hash_get_lowest_cost): Adjust.
4814 (vect_enhance_data_refs_alignment): Likewise. Use
4815 vect_get_peeling_costs_all_drs to compute the penalty for no
4816 peeling to match up costs.
4818 2017-07-21 Richard Biener <rguenther@suse.de>
4820 PR tree-optimization/81500
4821 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4822 we didn't identify a reduction path.
4824 2017-07-21 Tom de Vries <tom@codesourcery.com>
4825 Cesar Philippidis <cesar@codesourcery.com>
4827 PR gcov-profile/81442
4828 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
4831 2017-07-21 Tom de Vries <tom@codesourcery.com>
4834 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
4836 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
4837 nvptx_override_options_after_change.
4839 2017-07-21 Ulrich Drepper <drepper@redhat.com>
4841 * dwarf2out.c (output_file_names): Avoid double testing for
4844 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
4846 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
4848 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4850 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
4852 (try_crossjump_to_edge): Do not punt on partitioned functions.
4854 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4856 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4857 Put all BBs reachable only via paths crossing cold region to cold
4859 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
4861 2016-07-21 Richard Biener <rguenther@suse.de>
4863 PR tree-optimization/81303
4864 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
4865 into account prologue and epilogue iterations when raising
4866 min_profitable_iters to sth at least covering one vector iteration.
4868 2017-07-21 Tamar Christina <tamar.christina@arm.com>
4870 * config/arm/arm.c (arm_test_cpu_arch_dat):
4873 2017-07-20 Nathan Sidwell <nathan@acm.org>
4875 Remove TYPE_METHODS.
4876 * tree.h (TYPE_METHODS): Delete.
4877 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
4878 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
4879 (dbxout_type_methods): Scan TYPE_FIELDS.
4880 (dbxout_type): Don't check TYPE_METHODS here.
4881 * function.c (use_register_for_decl): Always ignore register for
4882 class types when not optimizing.
4883 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
4884 * tree.c (free_lang_data_in_type): Stitch out member functions and
4885 templates from TYPE_FIELDS.
4886 (build_distinct_type_copy, verify_type_variant,
4887 verify_type): Member fns are on TYPE_FIELDS.
4888 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
4889 * tree-pretty-print.c (dump_generic_node): Likewise.
4891 2017-07-20 Jakub Jelinek <jakub@redhat.com>
4894 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
4895 V2TImode and V4TImode.
4896 (ix86_expand_vector_extract): Likewise.
4897 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
4898 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
4899 (ssescalarmode): Handle V4TImode and V2TImode.
4900 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
4901 (*vec_extractv2ti, *vec_extractv4ti): New insns.
4902 (VEXTRACTI128_MODE): New mode iterator.
4903 (splitter for *vec_extractv?ti first element): New.
4904 (VEC_INIT_MODE): New mode iterator.
4905 (vec_init<mode>): Consolidate 3 expanders into one using
4906 VEC_INIT_MODE mode iterator.
4908 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4910 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
4911 non_spilled_static_chain_regno_p.
4913 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4915 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
4917 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
4919 * bb-reorder.c (connect_traces): Allow copying of blocks within
4922 2017-07-20 Richard Biener <rguenther@suse.de>
4924 * gimple.h (gimple_phi_result): Add gphi * overload.
4925 (gimple_phi_result_ptr): Likewise.
4926 (gimple_phi_arg): Likewise. Adjust index assert to only
4927 allow actual argument accesses rather than all slots available
4929 (gimple_phi_arg_def): Add gphi * overload.
4930 * tree-phinodes.c (make_phi_node): Initialize only actual
4932 (resize_phi_node): Clear memory not covered by old node,
4933 do not initialize excess argument slots.
4934 (reserve_phi_args_for_new_edge): Initialize new argument slot
4937 2017-07-20 Bin Cheng <bin.cheng@arm.com>
4939 PR tree-optimization/81388
4941 2016-07-21 Bin Cheng <bin.cheng@arm.com>
4943 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
4944 by removing computation of may_be_zero.
4946 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4947 Tom de Vries <tom@codesourcery.com>
4950 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
4951 when gimple level profile disagrees with what RTL expander did.
4953 2017-07-20 Richard Biener <rguenther@suse.de>
4955 PR tree-optimization/61171
4956 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
4957 (vect_analyze_stmt): Add slp instance parameter.
4958 (vectorizable_reduction): Likewise.
4959 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
4960 (vect_is_simple_reduction): Deal with chains not detected
4961 as SLP reduction chain, specifically not properly associated
4962 chains containing a mix of plus/minus.
4963 (get_reduction_op): Remove.
4964 (get_initial_defs_for_reduction): Simplify, pass in whether
4965 this is a reduction chain, pass in the SLP node for the PHIs.
4966 (vect_create_epilog_for_reduction): Get the SLP instance as
4968 (vectorizable_reduction): Get the SLP instance as arg.
4969 During analysis remember the SLP node with the PHIs in the
4970 instance. Simplify getting at the vectorized reduction PHIs.
4971 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
4972 through SLP instance.
4973 (vect_slp_analyze_operations): Likewise.
4974 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
4975 (vect_transform_stmt): Likewise.
4977 2017-07-20 Tom de Vries <tom@codesourcery.com>
4979 PR tree-optimization/81489
4980 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
4981 read of phi arg location to before loop that modifies phi.
4983 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4985 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
4988 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4991 * except.c (execute): Fix ordering issue.
4993 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4995 PR rtl-optimization/81423
4996 * combine.c (make_compound_operation_int): Don't try to optimize
4997 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
4999 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5001 PR rtl-optimization/81423
5002 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5003 with a constant that is -1 in the truncated to mode.
5005 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5007 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5008 (determine_unlikely_bbs): ... here.
5009 * predict.h (propagate_unlikely_bbs_forward): Declare.
5010 * cfgexpand.c (pass_expand::execute): Use it.
5011 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5013 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5014 propagate_unlikely_bbs_forward.
5016 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5019 * except.c (maybe_add_nop_after_section_switch): New function.
5022 2017-07-19 Tom de Vries <tom@codesourcery.com>
5024 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5026 2017-07-19 Tom de Vries <tom@codesourcery.com>
5028 * gimple.h (gimple_phi_arg): Make assert more strict.
5030 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5032 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5033 * config/rs6000/mmintrin.h: New file.
5034 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5036 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5038 PR tree-optimization/81346
5039 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5041 2017-07-19 Tom de Vries <tom@codesourcery.com>
5043 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5045 2017-07-19 Tom de Vries <tom@codesourcery.com>
5047 * config/nvptx/nvptx-modes.def: Add V2DImode.
5048 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5049 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5050 (nvptx_output_mov_insn): Handle lack of mov.b128.
5051 (nvptx_print_operand): Handle 'H' and 'L' codes.
5052 (nvptx_vector_mode_supported): Allow V2DImode.
5053 (nvptx_preferred_simd_mode): New function.
5054 (nvptx_data_alignment): New function.
5055 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5056 nvptx_preferred_simd_mode.
5057 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5059 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5061 2017-07-19 Tom de Vries <tom@codesourcery.com>
5063 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5064 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5065 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5066 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5067 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5068 (mov<VECIM>_insn): New define_insn.
5069 (define_expand "mov<VECIM>): New define_expand.
5071 2017-07-19 Tom de Vries <tom@codesourcery.com>
5073 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5075 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5077 PR tree-optimization/81346
5078 * fold-const.h (fold_div_compare, range_check_type): Declare.
5079 * fold-const.c (range_check_type): New function.
5080 (build_range_check): Use range_check_type.
5081 (fold_div_compare): No longer static, rewritten into
5082 a match.pd helper function.
5083 (fold_comparison): Don't call fold_div_compare here.
5084 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5087 2017-07-19 Nathan Sidwell <nathan@acm.org>
5089 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5090 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5091 * tree.c (find_decls_types_r, verify_type): Use
5092 TYPE_{MIN,MAX}_VALUE_RAW.
5093 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5094 (hash_tree): Likewise.
5095 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5097 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5100 2017-07-18 Tom de Vries <tom@codesourcery.com>
5103 * omp-expand.c (expand_omp_for_static_chunk): Handle
5104 equal-argument loop exit phi.
5106 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5109 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5110 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5111 operand 2 predicate.
5112 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5113 operand 2 predicate.
5114 (ror,rol -> rorx splitters): Use const_int_operand as
5115 operand 2 predicate.
5117 2017-06-18 Richard Biener <rguenther@suse.de>
5119 PR tree-optimization/81410
5120 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5121 the gap in the ! slp_perm SLP case after each group.
5123 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5126 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5129 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5132 * predict.c (set_even_probabilities): Cleanup; do not affect
5133 probabilities that are already known.
5134 (combine_predictions_for_bb): Call even when count is set.
5136 2017-07-18 Nathan Sidwell <nathan@acm.org>
5138 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5141 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5144 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5145 optimization for loop niter analysis.
5147 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5150 * config/avr/avr.c (avr_optimize_casesi): Don't use
5151 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5153 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5155 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5156 body_cost_vec from _vect_peel_extended_info.
5157 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5158 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5161 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5163 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5165 2017-07-18 Richard Biener <rguenther@suse.de>
5167 PR tree-optimization/80620
5168 PR tree-optimization/81403
5169 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5170 info when re-using a VN table entry.
5172 2017-07-18 Richard Biener <rguenther@suse.de>
5174 PR tree-optimization/81418
5175 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5176 vectype_in. Verify that with lane-reducing reduction operations
5177 we have a single def-use cycle.
5179 2017-07-17 Carl Love <cel@us.ibm.com>
5181 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5183 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5184 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5185 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5186 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5187 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5188 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5189 VMULOSW): New enum "unspec" values.
5190 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5191 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5192 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5193 altivec_vmulosw): New patterns.
5194 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5195 VMULOSW): Add definitions.
5196 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5198 * config/alpha/alpha.c: Include predict.h.
5200 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5202 * tree-vrp.c (compare_assert_loc): Fix comparison function
5203 to return predictable results.
5205 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5207 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5210 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5211 * doc/invoke.texi (mexpand-adddi): Update text.
5213 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5215 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5216 that also clobbers the CC register. The old expand code is moved
5218 (*arc_clzsi2): ... here.
5219 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5220 the CC register. The old expand code is moved to ...
5221 (arc_ctzsi2): ... here.
5223 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5225 * config/arc/arc.opt (mindexed-loads): Use initial value
5226 TARGET_INDEXED_LOADS_DEFAULT.
5227 (mauto-modify-reg): Use initial value
5228 TARGET_AUTO_MODIFY_REG_DEFAULT.
5229 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5230 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5231 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5232 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5234 2017-07-17 Martin Liska <mliska@suse.cz>
5237 * opts.c (finish_options): Do not allow -fgnu-tm
5238 w/ -fsanitize={kernel-,}address. Say sorry.
5240 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5243 * tree-loop-distribution.c (classify_partition): Only assert on
5244 numer of iterations.
5245 (merge_dep_scc_partitions): Delete prameter. Update function call.
5246 (distribute_loop): Remove code handling loop with unknown niters.
5247 (pass_loop_distribution::execute): Skip loop with unknown niters.
5249 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5252 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5253 function sort_partitions_by_post_order.
5255 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5257 PR tree-optimization/81374
5258 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5259 the max index of basic blocks, rather than number of basic blocks.
5261 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5263 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5265 (arc_legitimate_pic_operand_p): Likewise.
5266 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5268 (arc_needs_pcl_p): Likewise.
5269 (arc_legitimate_pc_offset_p): Likewise.
5270 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5271 function is also used in constrains.md.
5272 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5273 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5274 PLUS. Only return true/false in known cases, otherwise assert.
5275 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5276 is already called in arc_legitimate_constant_p.
5277 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5279 (LEGITIMATE_PIC_OPERAND_P): Use
5280 arc_raw_symbolic_reference_mentioned_p function.
5281 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5286 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5287 Andrew Burgess <andrew.burgess@embecosm.com>
5289 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5290 (arc_return_address_register): New function.
5291 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5292 (arc_handle_fndecl_attribute): Add naked attribute.
5293 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5294 (TARGET_WARN_FUNC_RETURN): Likewise.
5295 (arc_allocate_stack_slots_for_args): New function.
5296 (arc_warn_func_return): Likewise.
5297 (machine_function): Change type fn_type.
5298 (arc_compute_function_type): Consider new naked function type,
5299 change function return type.
5300 (arc_must_save_register): Adapt to handle new
5301 arc_compute_function_type's return type.
5302 (arc_expand_prologue): Likewise.
5303 (arc_expand_epilogue): Likewise.
5304 (arc_return_address_regs): Delete.
5305 (arc_return_address_register): New function.
5306 (arc_epilogue_uses): Use above function.
5307 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5308 (arc_function_type): Change encoding, add naked type.
5309 (ARC_INTERRUPT_P): Change to handle the new encoding.
5310 (ARC_FAST_INTERRUPT_P): Likewise.
5311 (ARC_NORMAL_P): Define.
5312 (ARC_NAKED_P): Likewise.
5313 (arc_compute_function_type): Delete prototype.
5314 * config/arc/arc.md (in_ret_delay_slot): Use
5315 arc_return_address_register function.
5316 (simple_return): Likewise.
5317 (p_return_i): Likewise.
5319 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5321 PR tree-optimization/81428
5322 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5323 can't be built for those types.
5325 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5327 Remove stuff dead since r239246.
5329 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5330 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5331 (avr_inform_devices): Remove dead stuff.
5333 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5335 * config/arm/arm_neon.h: Fix softp typo.
5337 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5339 PR tree-optimization/81365
5340 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5341 aggregate moves onto bb predecessor edges, make sure there are no
5342 loads that could alias the lhs in between the start of bb and the
5345 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5348 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5349 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5350 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5352 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5354 PR tree-optimization/81396
5355 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5356 (init_symbolic_number): Initialize it to 1.
5357 (perform_symbolic_merge): Add n_ops from both operands into the new
5359 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5360 without base_addr as useless if they need more than one operation.
5361 (bswap_replace): Handle !bswap case for NULL base_addr.
5363 2017-07-17 Tom de Vries <tom@codesourcery.com>
5366 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5369 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5371 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5372 conditional builtin define __FIX_LEON3FT_B2BST.
5374 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5376 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5377 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5380 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5382 PR rtl-optimization/81424
5383 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5384 to remove potential trapping from operands if -fnon-call-exceptions.
5386 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5388 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5389 profile_proability for scalling.
5390 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5392 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5394 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5396 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5398 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5399 fixpoint arithmetics.
5401 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5403 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5404 fixpoint arithmetics.
5406 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5408 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5409 fixpoint arithmetics.
5411 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5413 * profile-count.h (profile_probability::from_reg_br_prob_note,
5414 profile_probability::to_reg_br_prob_note): New functions.
5415 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5416 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5417 * predict.c (probability_reliable_p): Update.
5418 (edge_probability_reliable_p): Update.
5419 (br_prob_note_reliable_p): Update.
5420 (invert_br_probabilities): Update.
5421 (add_reg_br_prob_note): New function.
5422 (combine_predictions_for_insn): Update.
5423 * asan.c (asan_clear_shadow): Update.
5424 * cfgbuild.c (compute_outgoing_frequencies): Update.
5425 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5426 (update_br_prob_note): Update.
5427 (rtl_verify_edges): Update.
5428 (purge_dead_edges): Update.
5429 (fixup_reorder_chain): Update.
5430 * emit-rtl.c (try_split): Update.
5431 * ifcvt.c (cond_exec_process_insns): Update.
5432 (cond_exec_process_if_block): Update.
5433 (dead_or_predicable): Update.
5434 * internal-fn.c (expand_addsub_overflow): Update.
5435 (expand_neg_overflow): Update.
5436 (expand_mul_overflow): Update.
5437 * loop-doloop.c (doloop_modify): Update.
5438 * loop-unroll.c (compare_and_jump_seq): Update.
5439 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5440 * predict.h: Update.
5441 * reorg.c (mostly_true_jump): Update.
5443 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5444 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5445 * config/arc/arc.c: (emit_unlikely_jump): Update.
5446 * config/arm/arm.c: (emit_unlikely_jump): Update.
5447 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5448 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5449 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5450 (ix86_print_operand): Update.
5451 (ix86_split_fp_branch): Update.
5452 (predict_jump): Update.
5453 * config/ia64/ia64.c (ia64_print_operand): Update.
5454 * config/mmix/mmix.c (mmix_print_operand): Update.
5455 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5456 (rs6000_expand_split_stack_prologue): Update.
5457 * config/rs6000/rs6000.c: Update.
5458 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5459 (s390_expand_vec_movstr): Update.
5460 (s390_expand_cs_tdsi): Update.
5461 (s390_expand_split_stack_prologue): Update.
5462 * config/sh/sh.c (sh_print_operand): Update.
5463 (expand_cbranchsi4): Update.
5464 (expand_cbranchdi4): Update.
5465 * config/sparc/sparc.c (output_v9branch): Update.
5466 * config/spu/spu.c (get_branch_target): Update.
5467 (ea_load_store_inline): Update.
5468 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5469 * config/tilepro/tilepro.c: Update.
5471 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5473 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5474 (gimplify_save_expr): Likewise.
5476 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5478 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5480 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5482 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5484 * timevar.def (TV_IPA_FNSUMMARY): Define.
5486 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5488 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5489 to back store errata sensitive sequence from being generated.
5490 (sqrtdf2_fix): Likewise.
5492 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5494 * tree-ssa-threadupdate.c (compute_path_counts,
5495 update_joiner_offpath_counts): Use profile_probability.
5497 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5500 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5502 * config/arm/arm-c.c (arm_cpu_builtins): Define
5503 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5505 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5507 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5508 array entries to represent __ieee128 versions of the
5509 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5510 scalar_extract_sig, and scalar_insert_exp built-in functions.
5511 (altivec_resolve_overloaded_builtin): Add special case handling
5512 for the __builtin_scalar_insert_exp function, as represented by
5513 the P9V_BUILTIN_VEC_VSIEDP constant.
5514 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5515 exponent support for __ieee128 argument.
5516 (VSESQP): Add scalar extract signature support for __ieee128
5518 (VSTDCNQP): Add scalar test negative support for __ieee128
5520 (VSIEQP): Add scalar insert exponent support for __int128 argument
5521 with __ieee128 result.
5522 (VSIEQPF): Add scalar insert exponent support for __ieee128
5523 argument with __ieee128 result.
5524 (VSTDCQP): Add scalar test data class support for __ieee128
5526 (VSTDCNQP): Add overload support for scalar test negative with
5528 (VSTDCQP): Add overload support for scalar test data class
5530 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5532 (UNSPEC_VSX_SIEXPQP): New constant.
5533 (xsxexpqp): New insn for VSX scalar extract exponent quad
5535 (xsxsigqp): New insn for VSX scalar extract significand quad
5537 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5538 precision with floating point argument.
5539 (xststdcqp): New expand for VSX scalar test data class quad
5541 (xststdcnegqp): New expand for VSX scalar test negative quad
5543 (xststdcqp): New insn to match expansions for VSX scalar test data
5544 class quad precision and VSX scalar test negative quad precision.
5545 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5546 special case operand checking to enforce that second operand of
5547 VSX scalar test data class with quad precision argument is a 7-bit
5549 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5550 prototypes and descriptions of __ieee128 versions of
5551 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5552 scalar_test_data_class, and scalar_test_neg built-in functions.
5554 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5556 PR tree-optimization/81162
5557 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5558 replace a negate with an add.
5560 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5562 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5564 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5566 * config/arm/arm-c.c (arm_cpu_builtins): Define
5567 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5569 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5571 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5572 (armv8-r): Set ARM Cortex-R52 as default CPU.
5573 * config/arm/arm-tables.opt: Regenerate.
5574 * config/arm/arm-tune.md: Regenerate.
5575 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5577 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5578 extension for -mcpu=cortex-r52.
5580 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5582 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5583 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5584 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5585 (fp-armv8): Define it as FP_ARMv8 only.
5586 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5587 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5589 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5590 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5591 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5592 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5593 than TARGET_FPU_ARMV8.
5594 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5595 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5596 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5598 * config/arm/neon.md (neon_vrint): Likewise.
5599 (neon_vcvt): Likewise.
5600 (neon_<fmaxmin_op><mode>): Likewise.
5601 (<fmaxmin><mode>3): Likewise.
5602 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5603 * config/arm/predicates.md (arm_cond_move_operator): Check against
5604 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5606 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5608 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5611 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5613 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5614 loop in comment with memset.
5616 2017-07-14 Martin Liska <mliska@suse.cz>
5618 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5619 * dwarf2out.c (is_java): Remove the function.
5620 (output_pubname): Remove usage of the function.
5621 (lower_bound_default): Remove usage of DW_LANG_Java.
5622 (gen_compile_unit_die): Likewise.
5623 * gcc.c: Remove compiler defaults for .java and .zip files.
5624 * gimple-expr.c (remove_suffix): Change as there's no longer
5625 extension than 4-letter one.
5626 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5627 (gimplify_save_expr): Likewise.
5628 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5629 as it's possible even for other languages than Java.
5630 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5631 * lto-opts.c (lto_write_options): Remove reference to Java.
5632 * opts.c (strip_off_ending): Update file extension handling.
5633 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5634 * tree-eh.c (lower_resx): Likewise.
5635 * tree.c (free_lang_data_in_type): Remove dead code.
5636 (find_decls_types_r): Likewise.
5637 (build_common_builtin_nodes): Remove Java from a comment.
5638 (verify_type): Remove dead code.
5639 * varasm.c (assemble_external): Remove Java from a comment.
5641 2017-07-14 Martin Liska <mliska@suse.cz>
5643 * opts.c (finish_options): Add quotes.
5644 (common_handle_option): Likewise.
5646 2017-07-14 Martin Liska <mliska@suse.cz>
5648 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5649 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5651 * dwarf2out.c (lower_bound_default): Do not handle
5653 (gen_compile_unit_die): Likewise.
5654 * gcc.c: Remove default extension binding for GNU Pascal.
5655 * stmt.c: Remove Pascal language from a comment.
5656 * xcoffout.c: Likewise.
5658 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5661 * diagnostic-show-locus.c (fixit_cmp): New function.
5662 (layout::layout): Sort m_fixit_hints.
5663 (column_range::column_range): Assert that the values are valid.
5664 (struct char_span): New struct.
5665 (correction::overwrite): New method.
5666 (struct source_line): New struct.
5667 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5668 calls in terms of classes source_line and char_span, and
5669 correction::overwrite.
5670 (selftest::test_overlapped_fixit_printing_2): New function.
5671 (selftest::diagnostic_show_locus_c_tests): Call it.
5673 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5675 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5676 early if there is no lhs.
5678 2017-07-13 Martin Liska <mliska@suse.cz>
5680 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5681 (gen_reference_type_die): Likewise.
5682 * stor-layout.c: Remove Pascal-related comment.
5684 2017-07-13 Martin Liska <mliska@suse.cz>
5686 * opts.c (finish_options): Add quotes to error messages.
5687 (parse_sanitizer_options): Likewise.
5689 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5691 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5693 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5695 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5697 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5699 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5701 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5703 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5706 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5707 provides the hardware capability bits, define the macro
5708 __BUILTIN_CPU_SUPPORTS__.
5709 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5710 if GLIBC does not provide the hardware capability bits. Add a
5711 gcc_unreachable call if the built-in cpu function is neither
5712 __builtin_cpu_is nor __builtin_cpu_supports.
5713 (rs6000_get_function_versions_dispatcher): Change the warning
5714 that an old GLIBC is used which does not export the capability
5715 bits to be an error.
5716 * doc/extend.texi (target_clones attribute): Document the
5717 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5718 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5719 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5720 the macros defined by GCC if the newer GLIBC is available.
5722 2017-07-12 Jeff Law <law@redhat.com>
5724 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5725 remaining includes slightly.
5726 * config/riscv/riscv-builtins.c: Include profile-count.h.
5728 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5731 * config/avr/avr.c (avr_set_current_function): In diagnostic
5732 messages: Quote keywords and (parts of) identifiers.
5733 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5736 2017-07-12 Carl Love <cel@us.ibm.com>
5738 * config/rs6000/rs6000-c.c: Add support for built-in functions
5739 vector bool char vec_revb (vector bool char);
5740 vector bool short vec_revb (vector short char);
5741 vector bool int vec_revb (vector bool int);
5742 vector bool long long vec_revb (vector bool long long);
5743 * doc/extend.texi: Update the built-in documentation file for the
5744 new built-in functions.
5746 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5748 * config/s390/s390.md: Remove movcc splitter.
5750 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5752 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5753 load/store on condition.
5755 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5758 * config/avr/avr.c (avr_encode_section_info)
5759 [progmem && !TREE_READONLY]: Error if progmem object needs
5762 2017-07-11 Michael Collison <michael.collison@arm.com>
5764 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5767 2017-07-11 Carl Love <cel@us.ibm.com>
5769 * config/rs6000/rs6000-c.c: Add support for builtins
5770 vector unsigned int vec_parity_lsbb (vector signed int);
5771 vector unsigned int vec_parity_lsbb (vector unsigned int);
5772 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5773 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5774 vector unsigned long long vec_parity_lsbb (vector signed long long);
5775 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5776 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5777 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5778 * doc/extend.texi: Update the built-in documentation file for the
5779 new built-in functions.
5781 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5783 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5784 (layout::m_primary_loc): New field.
5785 (layout::layout): Initialize new field. Move location filtering
5786 logic from here to...
5787 (layout::maybe_add_location_range): ...this new method. Add
5788 support for filtering to just the lines already specified by other
5790 (layout::will_show_line_p): New method.
5791 (gcc_rich_location::add_location_if_nearby): New method.
5792 (selftest::test_add_location_if_nearby): New test function.
5793 (selftest::diagnostic_show_locus_c_tests): Call it.
5794 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5797 2017-07-11 Tom de Vries <tom@codesourcery.com>
5799 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5800 (bb_first_real_insn): New function.
5801 (nvptx_single): Add extra initialization of broadcasted condition
5804 2017-07-11 Nathan Sidwell <nathan@acm.org>
5806 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5808 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5810 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5811 Binutils doc as TEXI will mess them up.
5812 * doc/invoke.texi (AVR Options): Same here.
5814 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5816 * config/sparc/sparc.opt (mfix-ut700): New option.
5817 (mfix-gr712rc): Likewise.
5818 (sparc_fix_b2bst): New variable.
5819 * doc/invoke.texi (SPARC options): Document them.
5820 (ARM options): Fix warnings.
5821 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5822 instructions to prevent sequences that can trigger the store-store
5823 errata for certain LEON3FT processors.
5824 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
5825 (sparc_option_override): Set sparc_fix_b2bst appropriately.
5826 * config/sparc/sparc.md (fix_b2bst): New attribute.
5827 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
5829 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
5832 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
5834 (*rsqrtsf2_sse): Ditto.
5836 (div<mode>3): Macroize insn from divdf3 and divsf3
5837 using MODEF mode iterator.
5839 2017-07-10 Martin Sebor <msebor@redhat.com>
5841 PR tree-optimization/80397
5842 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
5843 instead of testing for equality to INTEGER_TYPE.
5845 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
5847 * config.gcc: Remove uclibc from arc target spec.
5849 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
5851 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
5853 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5856 * lto-wrapper.c (remove_option): New function.
5857 (merge_and_complain): Merge PIC/PIE options more realistically.
5859 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5861 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
5865 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
5866 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
5867 * config.in: Regenerate.
5868 * configure: Regenerate.
5869 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
5870 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
5871 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
5872 (TARGET_GASISR_PROLOGUES): ...target mask.
5873 * common/config/avr/avr-common.c
5874 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
5875 Set -mgas-isr-prologues.
5876 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
5877 INSERT_PASS_BEFORE for it.
5878 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
5879 * config/avr/avr.c (avr_option_override)
5880 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
5881 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
5882 (avr_attribute_table) <no_gccisr>: Add new function attribute.
5883 (avr_set_current_function) <is_no_gccisr>: Init machine field.
5884 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
5886 (make_avr_pass_pre_proep): New function.
5887 (emit_push_sfr) <treg>: Add argument to function and use it
5889 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
5890 and set machine->gasisr.yes.
5891 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
5892 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
5893 __gcc_isr.n_pushed to .L__stack_usage.
5894 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
5895 (avr_asm_final_postscan_insn): ...this new static function.
5896 * config/avr/avr.h (machine_function)
5897 <is_no_gccisr, use_L__stack_usage>: New fields.
5898 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
5899 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
5900 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
5901 (gasisr, *gasisr): New expander and insn.
5902 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5903 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
5904 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
5906 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
5908 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
5911 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5913 Move jump-tables out of .text again.
5916 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
5917 (ASM_OUTPUT_ADDR_VEC): New function.
5918 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
5919 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
5920 INSN_ADDRESSes as asm comment.
5921 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
5922 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
5923 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
5924 * config/avr/avr.md (*tablejump): Adjust comment.
5925 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
5926 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
5928 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
5929 (avr_output_addr_vec): New proto.
5930 (avr_log_t) <insn_addresses>: New field.
5932 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
5935 * config/i386/i386.c (ix86_function_arg_advance): Set
5936 outgoing_args_on_stack to true if there are outgoing arguments
5938 (ix86_function_arg): Likewise.
5939 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
5940 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
5941 * config/i386/i386.h (machine_function): Add
5942 outgoing_args_on_stack.
5944 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
5946 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5948 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
5950 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
5952 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
5953 (REAL_H): Remove $(MACHMODE_H).
5954 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
5956 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
5957 $(MACHMODE_H) and double-int.h.
5958 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
5960 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
5963 2017-07-07 Andrew Pinski <apinski@cavium.com>
5965 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
5966 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
5968 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5970 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
5971 Add warning if GCC was not configured to link against a GLIBC that
5972 exports the hardware capability bits.
5973 (make_resolver_func): Make resolver function private and not a
5974 COMDAT function. Create the name with clone_function_name instead
5975 of make_unique_name.
5978 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
5979 correct operand in doing the split.
5981 2017-07-07 Carl Love <cel@us.ibm.com>
5983 * config/rs6000/rs6000-c: Add support for built-in function
5984 vector unsigned short vec_pack_to_short_fp32 (vector float,
5986 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
5987 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
5988 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
5989 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
5990 (convert_4f32_8i16): Add define_expand.
5991 * doc/extend.texi: Update the built-in documentation file for the
5992 new built-in function.
5994 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5996 * config/sparc/m8.md: New file.
5997 * config/sparc/sparc.md: Include m8.md.
5999 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6001 * config/sparc/sparc.opt: New option -mvis4b.
6002 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6003 (sparc_option_override): Handle VIS4B.
6004 (enum sparc_builtins): Define
6005 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6006 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6007 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6008 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6009 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6010 (check_constant_argument): New function.
6011 (sparc_vis_init_builtins): Define builtins
6012 __builtin_vis_dictunpack{8,16,32},
6013 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6014 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6015 __builtin_vis_fpcmpde{8,16,32}shl and
6016 __builtin_vis_fpcmpur{8,16,32}shl.
6017 (sparc_expand_builtin): Check that the constant operands to
6018 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6019 constant and in range.
6020 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6022 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6023 (SPARC_IMM5_P): Likewise.
6024 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6025 (enabled): Handle vis4b.
6026 (UNSPEC_DICTUNPACK): New unspec.
6027 (UNSPEC_FPCMPSHL): Likewise.
6028 (UNSPEC_FPUCMPSHL): Likewise.
6029 (UNSPEC_FPCMPDESHL): Likewise.
6030 (UNSPEC_FPCMPURSHL): Likewise.
6031 (cpu_feature): New CPU feature `vis4b'.
6032 (dictunpack{8,16,32}): New insns.
6033 (FPCSMODE): New mode iterator.
6034 (fpcscond): New code iterator.
6035 (fpcsucond): Likewise.
6036 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6037 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6038 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6039 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6040 * config/sparc/constraints.md: Define constraints `q' for unsigned
6041 2-bit integer constants and `t' for unsigned 5-bit integer
6043 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6045 (imm5_operand_dictunpack16): Likewise.
6046 (imm5_operand_dictunpack32): Likewise.
6047 (imm2_operand): Likewise.
6048 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6049 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6050 ditunpack* and fpcmp*shl builtins.
6052 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6054 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6055 * config.in: Add HAVE_AS_SPARC6 define.
6056 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6058 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6060 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6061 (CPP_CPU_SPEC): Handle m8.
6062 (ASM_CPU_SPEC): Likewise.
6063 * config/sparc/sparc-opts.h (enum processor_type): Add
6065 * config/sparc/sparc.c (m8_costs): New struct.
6066 (sparc_option_override): Handle TARGET_CPU_m8.
6067 (sparc32_initialize_trampoline): Likewise.
6068 (sparc64_initialize_trampoline): Likewise.
6069 (sparc_issue_rate): Likewise.
6070 (sparc_register_move_cost): Likewise.
6071 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6072 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6073 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6074 (CPP_CPU_SPEC): Handle M8.
6075 (ASM_CPU_SPEC): Likewise.
6076 (AS_M8_FLAG): Define.
6077 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6078 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6079 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6081 * configure: Regenerate.
6082 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6085 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6087 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6089 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6090 ("*movdi_insn_sp32"): Do not set v3pipe.
6091 ("*movsi_insn"): Likewise.
6092 ("*movdi_insn_sp64"): Likewise.
6093 ("*movsf_insn"): Likewise.
6094 ("*movdf_insn_sp32"): Likewise.
6095 ("*movdf_insn_sp64"): Likewise.
6096 ("*zero_extendsidi2_insn_sp64"): Likewise.
6097 ("*sign_extendsidi2_insn"): Likewise.
6098 ("*mov<VM32:mode>_insn"): Likewise.
6099 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6100 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6101 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6102 ("<vlop:code><VL:mode>3"): Likewise.
6103 ("*not_<vlop:code><VL:mode>3"): Likewise.
6104 ("*nand<VL:mode>_vis"): Likewise.
6105 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6106 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6107 ("one_cmpl<VL:mode>2"): Likewise.
6108 ("faligndata<VM64:mode>_vis"): Likewise.
6109 ("alignaddrsi_vis"): Likewise.
6110 ("alignaddrdi_vis"): Likweise.
6111 ("alignaddrlsi_vis"): Likewise.
6112 ("alignaddrldi_vis"): Likewise.
6113 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6114 ("bmaskdi_vis"): Likewise.
6115 ("bmasksi_vis"): Likewise.
6116 ("bshuffle<VM64:mode>_vis"): Likewise.
6117 ("cmask8<P:mode>_vis"): Likewise.
6118 ("cmask16<P:mode>_vis"): Likewise.
6119 ("cmask32<P:mode>_vis"): Likewise.
6120 ("pdistn<P:mode>_vis"): Likewise.
6121 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6123 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6125 * config/sparc/sparc.md ("subtype"): New insn attribute.
6126 ("*wrgsr_sp64"): Set insn subtype.
6127 ("*rdgsr_sp64"): Likewise.
6128 ("alignaddrsi_vis"): Likewise.
6129 ("alignaddrdi_vis"): Likewise.
6130 ("alignaddrlsi_vis"): Likewise.
6131 ("alignaddrldi_vis"): Likewise.
6132 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6133 ("fexpand_vis"): Likewise.
6134 ("fpmerge_vis"): Likewise.
6135 ("faligndata<VM64:mode>_vis"): Likewise.
6136 ("bshuffle<VM64:mode>_vis"): Likewise.
6137 ("cmask8<P:mode>_vis"): Likewise.
6138 ("cmask16<P:mode>_vis"): Likewise.
6139 ("cmask32<P:mode>_vis"): Likewise.
6140 ("fchksm16_vis"): Likewise.
6141 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6142 ("fmean16_vis"): Likewise.
6143 ("fp<plusminus_insn>64_vis"): Likewise.
6144 ("<plusminus_insn>v8qi3"): Likewise.
6145 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6146 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6147 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6148 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6149 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6150 ("*movqi_insn"): Likewise.
6151 ("*movhi_insn"): Likewise.
6152 ("*movsi_insn"): Likewise.
6153 ("movsi_pic_gotdata_op"): Likewise.
6154 ("*movdi_insn_sp32"): Likewise.
6155 ("*movdi_insn_sp64"): Likewise.
6156 ("movdi_pic_gotdata_op"): Likewise.
6157 ("*movsf_insn"): Likewise.
6158 ("*movdf_insn_sp32"): Likewise.
6159 ("*movdf_insn_sp64"): Likewise.
6160 ("*zero_extendhisi2_insn"): Likewise.
6161 ("*zero_extendqihi2_insn"): Likewise.
6162 ("*zero_extendqisi2_insn"): Likewise.
6163 ("*zero_extendqidi2_insn"): Likewise.
6164 ("*zero_extendhidi2_insn"): Likewise.
6165 ("*zero_extendsidi2_insn_sp64"): Likewise.
6166 ("ldfsr"): Likewise.
6167 ("prefetch_64"): Likewise.
6168 ("prefetch_32"): Likewise.
6169 ("tie_ld32"): Likewise.
6170 ("tie_ld64"): Likewise.
6171 ("*tldo_ldub_sp32"): Likewise.
6172 ("*tldo_ldub1_sp32"): Likewise.
6173 ("*tldo_ldub2_sp32"): Likewise.
6174 ("*tldo_ldub_sp64"): Likewise.
6175 ("*tldo_ldub1_sp64"): Likewise.
6176 ("*tldo_ldub2_sp64"): Likewise.
6177 ("*tldo_ldub3_sp64"): Likewise.
6178 ("*tldo_lduh_sp32"): Likewise.
6179 ("*tldo_lduh1_sp32"): Likewise.
6180 ("*tldo_lduh_sp64"): Likewise.
6181 ("*tldo_lduh1_sp64"): Likewise.
6182 ("*tldo_lduh2_sp64"): Likewise.
6183 ("*tldo_lduw_sp32"): Likewise.
6184 ("*tldo_lduw_sp64"): Likewise.
6185 ("*tldo_lduw1_sp64"): Likewise.
6186 ("*tldo_ldx_sp64"): Likewise.
6187 ("*mov<VM32:mode>_insn"): Likewise.
6188 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6189 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6191 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6193 * config/sparc/sparc.md ("type"): New insn type viscmp.
6194 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6196 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6197 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6198 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6199 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6201 ("n7_vis_logical_11cycle"): Likewise.
6202 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6203 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6204 * config/sparc/niagara.md ("niag_vis"): Likewise.
6205 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6206 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6208 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6210 * config/sparc/sparc.md: New instruction type `bmask'.
6211 (bmaskdi_vis): Use the `bmask' type.
6212 (bmasksi_vis): Likewise.
6213 * config/sparc/ultra3.md (us3_array): Likewise.
6214 * config/sparc/niagara7.md (n7_array): Likewise.
6215 * config/sparc/niagara4.md (n4_array): Likewise.
6216 * config/sparc/niagara2.md (niag2_vis): Likewise.
6217 (niag3_vis): Likewise.
6218 * config/sparc/niagara.md (niag_vis): Likewise.
6220 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6222 * ipa-comdats.c: Remove optimize check from gate.
6223 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6224 for functions not optimized.
6225 (ipa_fn_summary_read): Skip optimize check.
6226 (ipa_fn_summary_write): Likewise.
6227 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6229 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6231 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6232 (check_callers): Give up on caller which is not optimized.
6233 (inline_small_functions): Likewise.
6234 (ipa_inline): Do not give up when not optimizing.
6235 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6236 away unoptimizes cdtors.
6237 (whole_program_function_and_variable_visibility): Do
6238 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6239 * ipa.c (process_references): Do not check optimize.
6240 (symbol_table::remove_unreachable_nodes): Update optimize check.
6241 (set_writeonly_bit): Update optimize check.
6242 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6243 (pass_ipa_single_use::gate): Remove.
6245 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6247 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6248 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6249 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6250 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6251 permute_load, permute_store, adjust_extract, adjust_splat,
6252 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6253 replace_swap_with_copy, dump_swap_insn_table,
6254 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6255 recombine_lvx_pattern, recombine_stvx_pattern,
6256 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6257 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6258 to file rs6000-p8swap.c.
6259 * config/rs6000/rs6000-p8swap.c: New file.
6260 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6261 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6262 and rs6000*-*-* targets.
6264 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6266 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6268 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6270 * lto-wrapper.c (merge_and_complain): Do not merge
6271 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6272 fsigned_zeros, ftrapping_math, fwrapv.
6273 (append_compiler_options): Do not track these options.
6274 (append_linker_options): Likewie
6276 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6278 * cgraphunit.c (cgraph_node::finalize_function): When
6279 !flag_toplevel_reorde set no_reorder flag.
6280 (varpool_node::finalize_decl): Likewise.
6281 (symbol_table::compile): Drop no toplevel reorder path.
6283 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6285 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6286 edges; zero probability is not better than uninitialized.
6288 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6290 * asan.h (asan_sanitize_allocas_p): Declare.
6291 * asan.c (asan_sanitize_allocas_p): New function.
6292 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6293 (handle_builtin_alloca): Likewise.
6294 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6295 if !asan_sanitize_allocas_p.
6296 * params.def (asan-instrument-allocas): Add new option.
6297 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6298 * opts.c (common_handle_option): Disable allocas sanitization for
6301 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6303 * asan.c: Include gimple-fold.h.
6304 (get_last_alloca_addr): New function.
6305 (handle_builtin_stackrestore): Likewise.
6306 (handle_builtin_alloca): Likewise.
6307 (asan_emit_allocas_unpoison): Likewise.
6308 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6309 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6310 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6311 (instrument_builtin_call): Pass gimple iterator to
6312 get_mem_refs_of_builtin_call.
6313 (last_alloca_addr): New global.
6314 * asan.h (asan_emit_allocas_unpoison): Declare.
6315 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6316 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6317 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6318 if function calls alloca.
6319 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6320 * gimple-fold.h (replace_call_with_value): Declare.
6321 * internal-fn.c: Include asan.h.
6322 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6323 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6325 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6327 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6328 (C_SELFTEST_FLAGS): New.
6329 (CPP_SELFTEST_FLAGS): New.
6330 (SELFTEST_DEPS): New, from deps of s-selftest.
6331 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6332 (CPP_SELFTEST_DEPS): New.
6333 (selftest): Add dependency on s-selftest-c++.
6334 (s-selftest): Rename to...
6335 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6336 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6337 than SELFTEST_FLAGS.
6338 (selftest-gdb): Rename to...
6339 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6341 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6342 (selftest-valgrind): Rename to...
6343 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6345 (selftest-valgrind): Reintroduce as an alias for
6346 selftest-c-valgrind.
6347 (s-selftest-c++): New.
6348 (selftest-c++-gdb): New.
6349 (selftest-c++-valgrind): New.
6351 2017-07-06 Olivier Hainque <hainque@adacore.com>
6353 * gcc.c (process_command): When deciding if undefined variables
6354 should be ignored when processing specs, accept "gcc -v" as well.
6356 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6358 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6359 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6361 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6363 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6364 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6365 * config/arm/arm-tables.opt: Regenerate.
6366 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6368 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6370 2017-07-06 Carl Love <cel@us.ibm.com>
6372 * ChangeLog: Clean up from mid air collision
6374 2017-07-06 Carl Love <cel@us.ibm.com>
6376 * config/rs6000/rs6000-c.c: Add support for built-in functions
6377 vector signed int vec_subc (vector signed int, vector signed int);
6378 vector signed __int128 vec_subc (vector signed __int128,
6379 vector signed __int128);
6380 vector unsigned __int128 vec_subc (vector unsigned __int128,
6381 vector unsigned __int128);
6382 vector signed int vec_sube (vector signed int, vector signed int,
6384 vector unsigned int vec_sube (vector unsigned int,
6385 vector unsigned int,
6386 vector unsigned int);
6387 vector signed __int128 vec_sube (vector signed __int128,
6388 vector signed __int128,
6389 vector signed__int128);
6390 vector unsigned __int128 vec_sube (vector unsigned __int128,
6391 vector unsigned __int128,
6392 vector unsigned __int128);
6393 vector signed int vec_subec (vector signed int, vector signed int,
6395 vector unsigned int vec_subec (vector unsigned int,
6396 vector unsigned int,
6397 vector unsigned int);
6398 vector signed __int128 vec_subec (vector signed __int128,
6399 vector signed __int128,
6400 vector signed__int128);
6401 vector unsigned __int128 vec_subec (vector unsigned __int128,
6402 vector unsigned __int128,
6403 vector unsigned __int128);
6404 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6405 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6406 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6407 BU_ALTIVEC_OVERLOAD_X definitions.
6408 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6409 * doc/extend.texi: Update the built-in documentation file for the new
6412 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6415 * diagnostic-show-locus.c (layout::layout): Use start and finish
6416 spelling location for the start and finish of each range.
6417 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6418 Add unused aspect param.
6419 * input.c (expand_location_1): Add "aspect" param, and use it
6420 to access the correct part of the location.
6421 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6423 (expand_location_to_spelling_point): Likewise.
6424 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6425 param, and pass it to expand_location_1.
6427 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6429 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6430 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6431 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6432 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6433 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6434 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6435 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6436 _mm_maskz_getmant_ss): New intrinsics.
6437 (__builtin_ia32_getexpss128_mask): Changed to ...
6438 __builtin_ia32_getexpss128_round ... this.
6439 (__builtin_ia32_getexpsd128_mask): Changed to ...
6440 __builtin_ia32_getexpsd128_round ... this.
6441 * config/i386/i386-builtin-types.def
6442 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6443 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6444 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6445 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6446 __builtin_ia32_getmantss_mask_round): New builtins.
6447 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6448 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6449 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6450 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6451 * config/i386/sse.md
6452 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6453 avx512f_sgetexp<mode><mask_scalar_name>
6454 <round_saeonly_scalar_name> ... this.
6455 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6456 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6457 vgetexp<ssescalarmodesuffix>
6458 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6459 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6460 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6461 avx512f_vgetmant<mode><mask_scalar_name>
6462 <round_saeonly_scalar_name> ... this.
6463 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6464 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6465 vgetmant<ssescalarmodesuffix>
6466 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6467 %0<mask_scalar_operand4>, %1, %2
6468 <round_saeonly_scalar_mask_op4>, %3} ... this.
6469 * config/i386/subst.md (mask_scalar_operand4,
6470 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6471 round_saeonly_scalar_nimm_predicate): New subst attributes.
6473 2017-07-06 Julia Koval <julia.koval@intel.com>
6475 * config/i386/i386.c (ix86_erase_embedded_rounding):
6476 Remove code for old rounding pattern.
6478 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6480 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6482 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6484 * doc/sourcebuild.texi (Test Directives, Variants of
6485 dg-require-support): Add documentation for dg-require-stack-check.
6487 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6489 * config/i386/subst.md (mask_scalar, round_scalar,
6490 round_saeonly_scalar): New meta-templates.
6491 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6492 round_scalar_mask_operand3, round_scalar_mask_op3,
6493 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6494 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6495 round_saeonly_scalar_constraint,
6496 round_saeonly_scalar_prefix): New subst attribute.
6497 * config/i386/sse.md
6498 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6499 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6500 <round_scalar_name> ... this.
6501 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6502 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6503 <round_scalar_name> ... this.
6504 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6505 <sse>_vm<code><mode>3<mask_scalar_name>
6506 <round_saeonly_scalar_name> ... this.
6507 (v<plusminus_mnemonic><ssescalarmodesuffix>
6508 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6509 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6510 v<plusminus_mnemonic><ssescalarmodesuffix>
6511 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6512 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6513 (v<multdiv_mnemonic><ssescalarmodesuffix>
6514 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6515 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6516 v<multdiv_mnemonic><ssescalarmodesuffix>
6517 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6518 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6519 (v<maxmin_float><ssescalarmodesuffix>
6520 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6521 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6522 v<maxmin_float><ssescalarmodesuffix>
6523 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6524 %0<mask_scalar_operand3>, %1, %<iptr>2
6525 <round_saeonly_scalar_mask_op3>} ... this.
6527 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6529 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6530 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6532 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6533 Alan Hayward <alan.hayward@arm.com>
6534 David Sherwood <david.sherwood@arm.com>
6536 * combine.c (simplify_if_then_else): Remove "enum" before
6538 * compare-elim.c (can_eliminate_compare): Likewise.
6539 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6541 (aarch64_lookup_simd_builtin_type): Likewise.
6542 (aarch64_simd_builtin_type): Likewise.
6543 (aarch64_init_simd_builtin_types): Likewise.
6544 (aarch64_simd_expand_args): Likewise.
6545 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6547 (aarch64_reverse_mask): Likewise.
6548 (aarch64_simd_emit_reg_reg_move): Likewise.
6549 (aarch64_gen_adjusted_ldpstp): Likewise.
6550 (aarch64_ccmp_mode_to_code): Likewise.
6551 (aarch64_operands_ok_for_ldpstp): Likewise.
6552 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6553 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6555 (aarch64_min_divisions_for_recip_mul): Likewise.
6556 (aarch64_reassociation_width): Likewise.
6557 (aarch64_get_condition_code_1): Likewise.
6558 (aarch64_simd_emit_reg_reg_move): Likewise.
6559 (aarch64_simd_attr_length_rglist): Likewise.
6560 (aarch64_reverse_mask): Likewise.
6561 (aarch64_operands_ok_for_ldpstp): Likewise.
6562 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6563 (aarch64_gen_adjusted_ldpstp): Likewise.
6564 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6566 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6567 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6568 (arm_lookup_simd_builtin_type): Likewise.
6569 (arm_simd_builtin_type): Likewise.
6570 (arm_init_simd_builtin_types): Likewise.
6571 (arm_expand_builtin_args): Likewise.
6572 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6573 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6574 (ft32_setup_incoming_varargs): Likewise.
6575 (ft32_function_arg): Likewise.
6576 (ft32_function_arg_advance): Likewise.
6577 (ft32_pass_by_reference): Likewise.
6578 (ft32_arg_partial_bytes): Likewise.
6579 (ft32_valid_pointer_mode): Likewise.
6580 (ft32_addr_space_pointer_mode): Likewise.
6581 (ft32_addr_space_legitimate_address_p): Likewise.
6582 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6584 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6585 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6586 (iamcu_alignment): Likewise.
6587 (canonicalize_vector_int_perm): Likewise.
6588 (ix86_noce_conversion_profitable_p): Likewise.
6589 (ix86_mpx_bound_mode): Likewise.
6590 (ix86_operands_ok_for_move_multiple): Likewise.
6591 * config/microblaze/microblaze-protos.h
6592 (microblaze_expand_conditional_branch_reg): Likewise.
6593 * config/microblaze/microblaze.c
6594 (microblaze_expand_conditional_branch_reg): Likewise.
6595 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6597 (rs6000_reassociation_width): Likewise.
6598 (rs6000_invalid_binary_op): Likewise.
6599 (fusion_p9_p): Likewise.
6600 (emit_fusion_p9_load): Likewise.
6601 (emit_fusion_p9_store): Likewise.
6602 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6604 (riscv_hard_regno_mode_ok_p): Likewise.
6605 (riscv_address_insns): Likewise.
6606 (riscv_split_symbol): Likewise.
6607 (riscv_legitimize_move): Likewise.
6608 (riscv_function_value): Likewise.
6609 (riscv_hard_regno_nregs): Likewise.
6610 (riscv_expand_builtin): Likewise.
6611 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6612 (riscv_build_integer): Likewise.
6613 (riscv_split_integer): Likewise.
6614 (riscv_legitimate_constant_p): Likewise.
6615 (riscv_cannot_force_const_mem): Likewise.
6616 (riscv_regno_mode_ok_for_base_p): Likewise.
6617 (riscv_valid_base_register_p): Likewise.
6618 (riscv_valid_offset_p): Likewise.
6619 (riscv_valid_lo_sum_p): Likewise.
6620 (riscv_classify_address): Likewise.
6621 (riscv_legitimate_address_p): Likewise.
6622 (riscv_address_insns): Likewise.
6623 (riscv_load_store_insns): Likewise.
6624 (riscv_force_binary): Likewise.
6625 (riscv_split_symbol): Likewise.
6626 (riscv_force_address): Likewise.
6627 (riscv_legitimize_address): Likewise.
6628 (riscv_move_integer): Likewise.
6629 (riscv_legitimize_const_move): Likewise.
6630 (riscv_legitimize_move): Likewise.
6631 (riscv_address_cost): Likewise.
6632 (riscv_subword): Likewise.
6633 (riscv_output_move): Likewise.
6634 (riscv_canonicalize_int_order_test): Likewise.
6635 (riscv_emit_int_order_test): Likewise.
6636 (riscv_function_arg_boundary): Likewise.
6637 (riscv_pass_mode_in_fpr_p): Likewise.
6638 (riscv_pass_fpr_single): Likewise.
6639 (riscv_pass_fpr_pair): Likewise.
6640 (riscv_get_arg_info): Likewise.
6641 (riscv_function_arg): Likewise.
6642 (riscv_function_arg_advance): Likewise.
6643 (riscv_arg_partial_bytes): Likewise.
6644 (riscv_function_value): Likewise.
6645 (riscv_pass_by_reference): Likewise.
6646 (riscv_setup_incoming_varargs): Likewise.
6647 (riscv_print_operand): Likewise.
6648 (riscv_elf_select_rtx_section): Likewise.
6649 (riscv_save_restore_reg): Likewise.
6650 (riscv_for_each_saved_reg): Likewise.
6651 (riscv_register_move_cost): Likewise.
6652 (riscv_hard_regno_mode_ok_p): Likewise.
6653 (riscv_hard_regno_nregs): Likewise.
6654 (riscv_class_max_nregs): Likewise.
6655 (riscv_memory_move_cost): Likewise.
6656 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6657 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6658 (rl78_addr_space_address_mode): Likewise.
6659 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6661 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6662 (rs6000_reassociation_width): Likewise.
6663 (rs6000_invalid_binary_op): Likewise.
6664 (fusion_p9_p): Likewise.
6665 (emit_fusion_p9_load): Likewise.
6666 (emit_fusion_p9_store): Likewise.
6667 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6668 (ok_for_simple_move_operands): Likewise.
6669 (ok_for_simple_move_strict_operands): Likewise.
6670 (ok_for_simple_arith_logic_operands): Likewise.
6671 (visium_legitimize_reload_address): Likewise.
6672 (visium_select_cc_mode): Likewise.
6673 (output_cbranch): Likewise.
6674 (visium_split_double_move): Likewise.
6675 (visium_expand_copysign): Likewise.
6676 (visium_expand_int_cstore): Likewise.
6677 (visium_expand_fp_cstore): Likewise.
6678 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6679 (visium_function_arg): Likewise.
6680 (visium_function_arg_advance): Likewise.
6681 (visium_libcall_value): Likewise.
6682 (visium_setup_incoming_varargs): Likewise.
6683 (visium_legitimate_constant_p): Likewise.
6684 (visium_legitimate_address_p): Likewise.
6685 (visium_legitimize_address): Likewise.
6686 (visium_secondary_reload): Likewise.
6687 (visium_register_move_cost): Likewise.
6688 (visium_memory_move_cost): Likewise.
6689 (prepare_move_operands): Likewise.
6690 (ok_for_simple_move_operands): Likewise.
6691 (ok_for_simple_move_strict_operands): Likewise.
6692 (ok_for_simple_arith_logic_operands): Likewise.
6693 (visium_function_value_1): Likewise.
6694 (rtx_ok_for_offset_p): Likewise.
6695 (visium_legitimize_reload_address): Likewise.
6696 (visium_split_double_move): Likewise.
6697 (visium_expand_copysign): Likewise.
6698 (visium_expand_int_cstore): Likewise.
6699 (visium_expand_fp_cstore): Likewise.
6700 (visium_split_cstore): Likewise.
6701 (visium_select_cc_mode): Likewise.
6702 (visium_split_cbranch): Likewise.
6703 (output_cbranch): Likewise.
6704 (visium_print_operand_address): Likewise.
6705 * expmed.c (flip_storage_order): Likewise.
6706 * expmed.h (emit_cstore): Likewise.
6707 (flip_storage_order): Likewise.
6708 * genrecog.c (validate_pattern): Likewise.
6709 * hsa-gen.c (gen_hsa_addr): Likewise.
6710 * internal-fn.c (expand_arith_overflow): Likewise.
6711 * ira-color.c (allocno_copy_cost_saving): Likewise.
6712 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6713 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6714 (process_invariant_for_inheritance): Likewise.
6715 * lra-eliminations.c (move_plus_up): Likewise.
6716 * omp-low.c (lower_oacc_reductions): Likewise.
6717 * simplify-rtx.c (simplify_subreg): Likewise.
6718 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6719 (TARGET_CHKP_BOUND_MODE): Likewise..
6720 * targhooks.c (default_chkp_bound_mode): Likewise.
6721 (default_setup_incoming_vararg_bounds): Likewise.
6722 * targhooks.h (default_chkp_bound_mode): Likewise.
6723 (default_setup_incoming_vararg_bounds): Likewise.
6724 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6725 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6726 (have_whole_vector_shift): Likewise.
6727 * tree-vect-stmts.c (vectorizable_load): Likewise.
6728 * doc/tm.texi: Regenerate.
6730 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6732 Graceful degrade if Binutils PR21472 is not available.
6735 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6736 .rodata in flash test fails.
6737 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6738 * confgure: Regenerate.
6739 * config.in: Regenerate.
6740 * config/avr/avr.c (avr_asm_named_section)
6741 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6742 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6743 (avr_asm_init_sections): Same.
6745 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6747 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6748 (fma<VH:mode>4_intrinsic): Likewise.
6749 (*fmsub<VCVTF:mode>4): Likewise.
6750 (*fmsub<VH:mode>4_intrinsic): Likewise.
6752 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6755 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6756 Don't depend on "optimize > 0".
6757 (out_movhi_r_mr, out_movqi_mr_r): Same.
6758 (out_movhi_mr_r, out_movqi_r_mr): Same.
6759 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6760 io_address_operand on "optimize > 0".
6762 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6764 * tree-loop-distribution.c: Add general explanantion on the pass.
6765 (generate_loops_for_partition): Mark distributed loop.
6766 (pg_add_dependence_edges): New parameter. Handle alias data
6767 dependence specially and record it in the parameter if asked.
6768 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6769 (init_partition_graph_vertices, add_partition_graph_edge): New.
6770 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6771 (free_partition_graph_vdata, build_partition_graph): New.
6772 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6773 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6774 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6775 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6776 (version_for_distribution_p, finalize_partitions): New.
6777 (distribute_loop): Handle alias data dependence specially. Factor
6778 out loop fusion code as functions and call these functions.
6780 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6782 * tree-loop-distribution.c (classify_partition): New parameter and
6783 better handle reduction statement.
6784 (rdg_build_partitions): Revise comment.
6785 (distribute_loop): Compute statements in all partitions and pass it
6786 to classify_partition.
6788 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6790 * tree-loop-distribution.c (enum partition_type): New.
6791 (struct partition): New field type.
6792 (partition_merge_into): Add parameter. Update partition type.
6793 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6794 (build_rdg_partition_for_vertex): Compute partition type.
6795 (rdg_build_partitions): Dump partition type.
6796 (distribute_loop): Update calls to partition_merge_into.
6798 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6800 * tree-loop-distribution.c (struct ddr_hasher): New.
6801 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6803 (classify_partition): Call get_data_dependence.
6804 (pg_add_dependence_edges): Ditto.
6805 (distribute_loop): Release data dependence hash table.
6807 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6809 * tree-loop-distribution.c (ref_base_address): Delete.
6810 (similar_memory_accesses): Rename ...
6811 (share_memory_accesses): ... to this. Check if partitions access
6812 the same memory reference.
6813 (distribute_loop): Call share_memory_accesses.
6815 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6817 * tree-loop-distribution.c (struct partition): New field recording
6819 (partition_alloc, partition_free): Init and release data refs.
6820 (partition_merge_into): Merge data refs.
6821 (build_rdg_partition_for_vertex): Collect data refs for partition.
6822 (pg_add_dependence_edges): Change parameters from vector to bitmap.
6824 (distribute_loop): Remve data refs from vertice data of partition
6827 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6829 * tree-loop-distribution.c (params.h): Include header file.
6830 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
6831 (datarefs_vec): New global var.
6832 (create_rdg_vertices): Use datarefs_vec directly.
6833 (free_rdg): Don't free data references.
6834 (build_rdg): Update use. Don't free data references.
6835 (distribute_loop): Compute global variable for data references.
6836 Bail out if there are too many data references.
6838 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6840 * tree-loop-distribution.c (loop_nest): New global var.
6841 (build_rdg): Use loop directly, rather than loop nest.
6842 (pg_add_dependence_edges): Remove loop nest parameter. Use global
6844 (distribute_loop): Compute global variable loop nest. Update use.
6846 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6848 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
6849 (partition_merge_into): New parameter. Dump reason for fusion.
6850 (distribute_loop): Update use of partition_merge_into.
6852 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6854 * tree-loop-distribution.c (bb_top_order_index): New.
6855 (bb_top_order_index_size, bb_top_order_cmp): New.
6856 (stmts_from_loop): Use topological order.
6857 (pass_loop_distribution::execute): Compute and release topological
6858 order for basic blocks.
6860 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6862 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
6865 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6867 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
6868 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
6869 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
6870 * internal-fn.def (LOOP_DIST_ALIAS): New.
6871 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
6872 (fold_loop_internal_call): ... this.
6873 (vect_loop_dist_alias_call): New function.
6874 (set_uid_loop_bbs): Call fold_loop_internal_call.
6875 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
6878 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6881 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
6882 Require dead FLAGS_REG at the beginning of a peephole.
6884 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6887 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
6888 arguments in the call to __builtin_ia32_sbb_u32.
6889 (_subborrow_u64): Swap _X and _Y arguments in the call to
6890 __builtin_ia32_sbb_u64.
6892 2017-07-04 Jakub Jelinek <jakub@redhat.com>
6895 * tree-vrp.c (compare_assert_loc): Turn into a function template
6896 with stable template parameter. Only test if a->e is NULL,
6897 !a->e == !b->e has been verified already. Use e == NULL or
6898 e != NULL instead of e or ! e tests. If stable is true, don't use
6899 iterative_hash_expr, on the other side allow a or b or both NULL
6900 and sort the NULLs last.
6901 (process_assert_insertions): Sort using compare_assert_loc<false>
6902 instead of compare_assert_loc, later sort using
6903 compare_assert_loc<true> before calling process_assert_insertions_for
6904 in a loop. Use break instead of continue once seen NULL pointer.
6906 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6908 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6909 Cortex-R7 and Cortex-R8 processors.
6911 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6913 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
6914 uninitialized while src is not.
6916 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
6918 * common/config/arm/arm-common.c: Adjust include path for
6920 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
6921 (arm-cpu.h): Create in build directory. Adjust dependency rules.
6922 (arm-cpu-data.h): Likewise.
6923 (arm-cpu-cdata.h): Likewise.
6924 * config/arm/arm-cpu.h: Delete.
6925 * config/arm/arm-cpu-cdata.h: Delete.
6926 * config/arm/arm-cpu-data.h: Delete.
6928 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
6930 * config/arm/arm-cpus.in (cortex-a55): New.
6931 (cortex-a75): Likewise.
6932 (cortex-a75.cortex-a55): Likewise.
6933 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
6935 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
6936 * config/arm/arm-cpu-cdata.h: Regenerate.
6937 * config/arm/arm-cpu-data.h: Regenerate.
6938 * config/arm/arm-cpu.h: Regenerate.
6939 * config/arm/arm-tables.opt: Regenerate.
6940 * config/arm/arm-tune.md: Regenerate.
6942 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6944 * haifa-sched.c (sched_create_recovery_edges): Update profile.
6946 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6948 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
6951 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
6953 PR tree-optimization/81292
6954 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
6955 full_string_p, also call adjust_related_strinfos if the adjustment
6956 is simple, otherwise invalidate related strinfos.
6958 2017-07-04 Martin Liska <mliska@suse.cz>
6961 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
6962 newly created variable as DECL_IGNORED_P.
6964 2017-07-04 Martin Liska <mliska@suse.cz>
6967 * ipa-inline.c (inline_small_functions):
6968 Use xstrdup_for_dump.
6970 2017-07-04 Tom de Vries <tom@codesourcery.com>
6972 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
6974 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6977 * config/darwin.c (darwin_function_switched_text_sections):
6980 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
6982 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
6984 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6986 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
6988 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6990 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
6991 min_profitable_iters, and th as inclusive lower bounds.
6992 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
6993 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
6994 for min_profitable_iters and min_profitable_estimate.
6995 (vect_transform_loop): Treat th as an inclusive lower bound.
6996 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6998 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7001 * config/darwin.c (darwin_function_switched_text_sections):
7002 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7003 in two pieces, and suppress the use of buf.
7005 2017-07-03 Nathan Sidwell <nathan@acm.org>
7007 * hash-table.h (hash_table_mod1): Fix indentation.
7009 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7012 * predict.c (force_edge_cold): Be more careful about propagation
7014 * profile-count.h (profile_probability::guessed,
7015 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7017 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7019 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7021 * doc/invoke.texi (rcpc architecture extension): Document it.
7023 2017-07-03 Richard Biener <rguenther@suse.de>
7025 PR tree-optimization/60510
7026 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7027 the scalar reduction PHI and use it.
7028 (vectorizable_reduction): Properly guard the single_defuse_cycle
7029 path for non-SLP reduction chains where we cannot use it.
7030 Rework reduc_def/index and vector type deduction. Rework
7031 vector operand gathering during reduction op code-gen.
7032 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7033 chains dissolve the chain and leave it to non-SLP reduction
7036 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7038 * tree-data-ref.h (dr_alignment): Declare.
7039 * tree-data-ref.c (dr_alignment): New function.
7040 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7041 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7043 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7045 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7047 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7048 and base_misalignment fields.
7049 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7050 * tree-data-ref.c: Include builtins.h.
7051 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7052 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7053 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7054 * tree-vect-data-refs.c: Include tree-cfg.h.
7055 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7056 fields instead of calculating an alignment here.
7057 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7058 innermost_loop_behavior fields.
7060 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7062 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7064 (DR_STEP_ALIGNMENT): New macro.
7065 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7066 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7067 (create_data_ref): Print it.
7068 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7069 to tell whether the step preserves vector (mis)alignment.
7070 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7071 Move the check for an integer step and generalise to all INTEGER_CST.
7072 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7073 Print the outer step alignment.
7075 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7077 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7078 with offset_alignment.
7079 (DR_ALIGNED_TO): Delete.
7080 (DR_OFFSET_ALIGNMENT): New macro.
7081 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7082 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7083 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7084 (create_data_ref): Likewise.
7085 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7086 (vect_analyze_data_refs): Likewise.
7087 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7088 creating dummy innermost behavior.
7090 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7092 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7093 with a "innermost_loop_behavior *" and refeence tree.
7094 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7095 (create_data_ref): Update call accordingly.
7096 * tree-predcom.c (find_looparound_phi): Likewise.
7098 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7100 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7101 fields with dr_wrt_vec_loop.
7102 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7103 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7104 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7105 (vect_dr_behavior): New function.
7106 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7107 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7108 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7109 track whether the step preserves the misalignment.
7110 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7111 Use vect_dr_behavior.
7112 (vect_setup_realignment): Update call accordingly.
7113 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7114 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7115 call to vect_create_addr_base_for_vector_ref.
7116 (vect_create_cond_for_align_checks): Likewise.
7117 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7118 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7119 (vect_recog_mask_conversion_pattern): Likewise.
7120 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7121 (new_stmt_vec_info): Remove redundant zeroing.
7123 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7125 * common/config/arm/arm-common.c (arm_be8_option): New function.
7126 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7127 (ISA_ARMv6): Add isa_bit_be8.
7128 * config/arm/arm.h (arm_be8_option): Add prototype.
7129 (BE8_SPEC_FUNCTION): New define.
7130 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7131 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7132 (mlittle-endian): Similarly.
7133 (mbe8, mbe32): New options.
7134 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7135 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7137 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7139 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7141 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7143 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7144 (cleanup_tree_cfg_bb): Use it.
7145 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7147 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7149 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7152 * loop-doloop.c (add_test): Update profile.
7154 2017-07-03 Martin Liska <mliska@suse.cz>
7157 * sanopt.c (rewrite_usage_of_param): New function.
7158 (sanitize_rewrite_addressable_params): Likewise.
7159 (pass_sanopt::execute): Call rewrite_usage_of_param.
7161 2017-07-03 Richard Biener <rguenther@suse.de>
7163 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7164 back to using VIEW_CONVERT_EXPR.
7166 2017-07-03 Martin Liska <mliska@suse.cz>
7169 * doc/extend.texi: Document when a resolver function is
7170 generated for target_clones.
7172 2017-07-03 Martin Liska <mliska@suse.cz>
7174 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7175 and shadow memory of auto variables which are subject of
7176 use-after-scope sanitization.
7177 (asan_expand_mark_ifn): Add do set only when is_poison.
7179 2016-07-03 Richard Biener <rguenther@suse.de>
7181 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7183 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7184 (vectorizable_reduction): Perform reduction PHI creation when
7185 visiting a reduction PHI and adjust and simplify code generation
7186 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7187 (vect_transform_loop): Visit reduction PHIs.
7188 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7189 defs into the SLP tree.
7190 (vect_build_slp_tree): Reduction defs terminate the recursion.
7191 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7193 (vect_get_vec_defs_for_stmt_copy): Export.
7194 (vect_get_vec_defs): Likewise.
7195 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7197 (vect_get_vec_defs_for_stmt_copy): Declare.
7198 (vect_get_vec_defs): Likewise.
7200 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7202 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7203 parameter with a "loop" parameter and use it instead of the
7204 loop containing DR_STMT. Don't check simple_iv when doing
7205 BB analysis. Describe the two analysis modes in the comment.
7207 2017-07-03 Tom de Vries <tom@codesourcery.com>
7209 PR tree-optimization/69468
7210 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7211 (find_same_succ_bb): Handle ignore_edge_flags.
7213 2017-07-03 Tom de Vries <tom@codesourcery.com>
7215 PR tree-optimization/81192
7216 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7218 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7220 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7222 2017-07-03 Tom de Vries <tom@codesourcery.com>
7224 PR tree-optimization/81192
7225 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7226 BB_SAME_SUCC (bb) == NULL.
7228 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7230 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7233 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7235 * dumpfile.c: Include profile-count.h
7236 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7238 (insert_cond_bb): Update profile.
7239 * tree-cfg.h (insert_cond_bb): Update prototype.
7240 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7241 * tree-dump.c: Do not include tree-cfg.
7243 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7245 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7247 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7249 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7252 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7254 * tree-complex.c (expand_complex_div_wide): update profile.
7256 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7257 Alan Hayward <alan.hayward@arm.com>
7258 David Sherwood <david.sherwood@arm.com>
7260 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7261 (CORETYPES_H): New define.
7262 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7263 (insn-modes-inline.h, s-modes-inline-h): New rules.
7264 (generated_files): Add insn-modes-inline.h.
7265 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7266 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7267 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7268 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7269 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7270 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7271 (build/gencodes.o, build/genconditions.o): Likewise.
7272 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7273 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7274 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7275 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7276 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7277 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7278 * coretypes.h: Include everything up to real.h for generators.
7279 Include insn-modes.h first. Include wide-int-print.h after
7280 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7281 * machmode.h: Don't include insn-modes.h here.
7282 * function-tests.c: Remove includes of signop.h, machmode.h,
7283 double-int.h and wide-int.h.
7285 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7287 * optc-save-gen.awk: Likewise.
7288 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7289 * godump.c: Remove include of wide-int-print.h.
7290 * pretty-print.h: Likewise.
7291 * wide-int-print.cc: Likewise.
7292 * wide-int.cc: Likewise.
7293 * hash-map-tests.c: Remove include of signop.h.
7294 * hash-set-tests.c: Likewise.
7295 * rtl-tests.c: Likewise.
7296 * mkconfig.sh: Remove include of machmode.h.
7297 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7299 (emit_insn_modes_inline_h): ...this new function. Emit the code
7300 into an insn-modes-inline.h header file, adding appropriate
7301 include guards and end comments.
7302 (emit_insn_modes_c_header): Remove include of machmode.h.
7303 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7305 (main): Handle -i flag and call emit_insn_modes_inline_h when
7308 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7310 * tree-ssa-strlen.c (strinfo): Rename the length field to
7311 nonzero_chars. Add a full_string_p field.
7312 (compare_nonzero_chars, zero_length_string_p): New functions.
7313 (get_addr_stridx): Add an offset_out parameter.
7314 Use compare_nonzero_chars.
7315 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7316 (new_strinfo): Update after above changes to strinfo.
7317 (set_endptr_and_length): Set full_string_p.
7318 (get_string_length): Update after above changes to strinfo.
7319 (unshare_strinfo): Update call to new_strinfo.
7320 (maybe_invalidate): Likewise.
7321 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7322 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7323 as a uhwi instead of an shwi. Update after above changes to
7324 strinfo and new_strinfo.
7325 (zero_length_string): Assert that chainsi contains full strings.
7326 Use zero_length_string_p. Update call to new_strinfo.
7327 (adjust_related_strinfos): Update after above changes to strinfo.
7328 Copy full_string_p from origsi.
7329 (adjust_last_stmt): Use zero_length_string_p.
7330 (handle_builtin_strlen): Update after above changes to strinfo and
7331 new_strinfo. Install the lhs as the string length if the previous
7332 entry didn't describe a full string.
7333 (handle_builtin_strchr): Update after above changes to strinfo
7335 (handle_builtin_strcpy): Likewise.
7336 (handle_builtin_strcat): Likewise.
7337 (handle_builtin_malloc): Likewise.
7338 (handle_pointer_plus): Likewise.
7339 (handle_builtin_memcpy): Likewise. Track nonzero characters
7340 that aren't necessarily followed by a nul terminator.
7341 (handle_char_store): Likewise.
7343 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7345 PR tree-optimization/80769
7346 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7347 for malloc and calloc. Document the new invariant that all related
7348 strinfos have delayed lengths or none do.
7349 (verify_related_strinfos): Move earlier in file.
7350 (set_endptr_and_length): New function, split out from...
7351 (get_string_length): ...here. Also set the lengths of related
7353 (zero_length_string): Assert that chainsi has known (rather than
7355 (adjust_related_strinfos): Likewise.
7357 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7359 PR tree-optimization/81136
7360 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7361 assert that two references with the same misalignment have the same
7362 compile-time misalignment if those compile-time misalignments
7365 2017-07-01 Andi Kleen <ak@linux.intel.com>
7367 * print-tree.c (print_node): Print all attributes.
7369 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7371 * cfg.c (scale_bbs_frequencies): New function.
7372 * cfg.h (scale_bbs_frequencies): Declare it.
7373 * cfgloopanal.c (single_likely_exit): Cleanup.
7374 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7376 (scale_loop_profile): Likewise.
7377 (loop_version): Likewise.
7378 (create_empty_loop_on_edge): Update.
7379 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7380 scale_loop_frequencies, scale_loop_profile, loopify,
7381 loop_version): Update prototypes.
7382 * modulo-sched.c (sms_schedule): Update.
7383 * predict.c (unlikely_executed_edge_p): Also check probability.
7384 (probably_never_executed_edge_p): Fix typo.
7385 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7386 * tree-parloops.c (gen_parallel_loop): Update.
7387 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7388 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7389 * tree-ssa-loop-split.c (split_loop): Update.
7390 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7391 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7392 (vect_loop_versioning): Update.
7393 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7395 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7397 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7399 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7401 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7402 to keep profile consistent.
7404 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7406 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7407 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7408 * profile-count.h (max_safe_multiplier): Make unsigned.
7409 (profile_count::guessed_zero): New.
7411 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7413 * bb-reorder.c (fix_up_crossing_landing_pad,
7414 fix_crossing_conditional_branches): Use make_single_succ_edge
7415 to keep profile consistent.
7417 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7419 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7422 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7425 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7426 the right scopes, make sure cond_jump isn't preserved between multiple
7427 iterations. Search for fallthru edge whenever there are 3+ edges and
7428 use find_fallthru_edge for it.
7430 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7432 Patch by Alexander Monakov <amonakov@ispras.ru>
7433 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7434 probabilities consistently.
7436 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7438 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7439 emit_cmp_and_jump_insns.
7441 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7444 * tree-inline.c (expand_call_inline): Combine profile statuses.
7446 2017-06-30 Andrew Pinski <apinski@cavium.com>
7448 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7449 fold_stmt returned true.
7451 2017-06-30 Nathan Sidwell <nathan@acm.org>
7453 * ggc.h (empty_string): Delete.
7454 * cfgexpand.c (expand_asm_stmt): Use plain "".
7455 * optabs.c (expand_asm_memory_barrier): Likewise.
7456 * stringpool.c (empty_string): Delete.
7457 (digit_vector, digit_string): Delete.
7458 (ggc_alloc_string): Use plain "", don't optimize single digit
7459 strings. Use ggc_alloc_atomic.
7461 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7463 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7464 comparison set and one other set, use the cost of the non-comparison
7467 2017-06-30 Nathan Sidwell <nathan@acm.org>
7469 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7472 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7474 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7475 loops. Remove now unneeded calls to gimple_switch_set_label() that
7476 just set removed labels to NULL_TREE.
7478 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7480 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7481 (set_range_info): ...here. Only call set_range_info_raw if domain
7483 (set_nonzero_bits): Call set_range_info_raw.
7484 * tree-ssanames.h (set_range_info_raw): New.
7486 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7489 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7490 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7491 of nonimmediate_operand and <store_mask_constraint> instead of m
7492 for the input operand. For V8FI iterator, always split if input
7493 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7494 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7495 <store_mask_predicate> instead of register_operand and
7496 <store_mask_constraint> instead of v for the input operand. Make
7497 sure both operands aren't MEMs for if not <mask_applied>.
7499 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7501 * lto-wrapper.c (copy_file) Close both file descriptors before
7504 2017-06-30 Martin Liska <mliska@suse.cz>
7507 * multiple_target.c (create_dispatcher_calls): Make ifunc
7508 also for function that don't have calls or are not referenced.
7510 2017-06-30 Richard Biener <rguenther@suse.de>
7512 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7513 analyze the first scalar stmt. Move vector type computation
7514 for the BB case here from ...
7515 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7516 live operation processing in the SLP case properly.
7518 2017-06-30 Richard Biener <rguenther@suse.de>
7520 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7522 2017-06-30 Martin Liska <mliska@suse.cz>
7525 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7526 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7528 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7530 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7533 2017-06-30 Richard Biener <rguenther@suse.de>
7535 PR tree-optimization/81249
7536 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7537 condition reduction result to original scalar type.
7539 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7541 * profile-count.h (enum profile_quality): Fix typos and whitespace
7544 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7546 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7547 type for branch probabilities.
7549 2017-06-29 Julian Brown <julian@codesourcery.com>
7550 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7552 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7553 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7554 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7555 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7557 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7559 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7560 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7561 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7562 CC usage from generic code to here.
7563 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7564 CC usage into the target macros.
7566 2017-06-29 Maya Rashish <coypu@sdf.org>
7568 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7571 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7573 * arm/arm-builtins.c: Include profile-count.h
7574 * except.c (sjlj_emit_function_enter): Use
7575 profile_probability::unlikely.
7577 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7579 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7580 and tocrel_offset be pointer args rather than implicitly using
7582 (legitimate_constant_pool_address_p, rs6000_emit_move,
7583 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7584 tocrel_offset and use in toc_relative_expr_p call.
7585 (print_operand, print_operand_address): Use static tocrel_base_oac
7586 and tocrel_offset_oac.
7587 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7590 2017-06-29 Maya Rashish <coypu@sdf.org>
7592 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7594 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7596 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7597 objects, take into account only the alignment of 'op0' and 'mode1' if
7600 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7602 * ccmp.c (ccmp_tree_comparison_p): New function.
7603 (ccmp_candidate_p): Update to use above function.
7604 (get_compare_parts): New function.
7605 (expand_ccmp_next): Update to use new functions.
7606 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7608 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7609 take mode as argument.
7610 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7611 * expr.c (expand_expr_real_1): Pass mode as argument.
7613 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7615 * combine.c (combine_instructions): Print insns to dump_file, together
7618 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7620 * asan.c (asan_emit_stack_protection): Update.
7621 (create_cond_insert_point): Update.
7622 * auto-profile.c (afdo_propagate_circuit): Update.
7623 * basic-block.h (struct edge_def): Turn probability to
7624 profile_probability.
7625 (EDGE_FREQUENCY): Update.
7626 * bb-reorder.c (find_traces_1_round): Update.
7627 (better_edge_p): Update.
7628 (sanitize_hot_paths): Update.
7629 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7630 (make_single_succ_edge): Update.
7631 (check_bb_profile): Update.
7632 (dump_edge_info): Update.
7633 (update_bb_profile_for_threading): Update.
7634 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7635 probabilitycount to 0.
7636 * cfgbuild.c (compute_outgoing_frequencies): Update.
7637 * cfgcleanup.c (try_forward_edges): Update.
7638 (outgoing_edges_match): Update.
7639 (try_crossjump_to_edge): Update.
7640 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7641 (expand_gimple_tailcall): Update.
7642 (construct_init_block): Use make_single_succ_edge.
7643 (construct_exit_block): Use make_single_succ_edge.
7644 * cfghooks.c (verify_flow_info): Update.
7645 (redirect_edge_succ_nodup): Update.
7646 (split_edge): Update.
7647 (account_profile_record): Update.
7648 * cfgloopanal.c (single_likely_exit): Update.
7649 * cfgloopmanip.c (scale_loop_profile): Update.
7650 (set_zero_probability): Remove.
7651 (duplicate_loop_to_header_edge): Update.
7652 * cfgloopmanip.h (loop_version): Update prototype.
7653 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7654 (force_nonfallthru_and_redirect): Update.
7655 (update_br_prob_note): Update.
7656 (rtl_verify_edges): Update.
7657 (purge_dead_edges): Update.
7658 (rtl_lv_add_condition_to_bb): Update.
7659 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7660 * cgraphunit.c (init_lowered_empty_function): Update.
7661 (cgraph_node::expand_thunk): Update.
7662 * cilk-common.c: Include profile-count.h
7663 * dojump.c (inv): Remove.
7664 (jumpifnot): Update.
7665 (jumpifnot_1): Update.
7666 (do_jump_1): Update.
7668 (do_jump_by_parts_greater_rtx): Update.
7669 (do_compare_rtx_and_jump): Update.
7670 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7671 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7672 * dwarf2cfi.c: Include profile-count.h
7673 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7674 (sjlj_emit_dispatch_table): Likewise.
7675 * explow.c: Include profile-count.h
7676 * expmed.c (emit_store_flag_force): Update.
7677 (do_cmp_and_jump): Update.
7678 * expr.c (compare_by_pieces_d::generate): Update.
7679 (compare_by_pieces_d::finish_mode): Update.
7680 (emit_block_move_via_loop): Update.
7681 (store_expr_with_bounds): Update.
7682 (store_constructor): Update.
7683 (expand_expr_real_2): Update.
7684 (expand_expr_real_1): Update.
7685 * expr.h (try_casesi, try_tablejump): Update prototypes.
7686 * gimple-pretty-print.c (dump_probability): Update.
7687 (dump_profile): New.
7688 (dump_gimple_label): Update.
7689 (dump_gimple_bb_header): Update.
7690 * graph.c (draw_cfg_node_succ_edges): Update.
7691 * hsa-gen.c (convert_switch_statements): Update.
7692 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7693 (find_if_case_1): Update.
7694 (find_if_case_2): Update.
7695 * internal-fn.c (expand_arith_overflow_result_store): Update.
7696 (expand_addsub_overflow): Update.
7697 (expand_neg_overflow): Update.
7698 (expand_mul_overflow): Update.
7699 (expand_vector_ubsan_overflow): Update.
7700 * ipa-cp.c (good_cloning_opportunity_p): Update.
7701 * ipa-split.c (split_function): Use make_single_succ_edge.
7702 * ipa-utils.c (ipa_merge_profiles): Update.
7703 * loop-doloop.c (add_test): Update.
7704 (doloop_modify): Update.
7705 * loop-unroll.c (compare_and_jump_seq): Update.
7706 (unroll_loop_runtime_iterations): Update.
7707 * lra-constraints.c (lra_inheritance): Update.
7708 * lto-streamer-in.c (input_cfg): Update.
7709 * lto-streamer-out.c (output_cfg): Update.
7710 * mcf.c (adjust_cfg_counts): Update.
7711 * modulo-sched.c (sms_schedule): Update.
7712 * omp-expand.c (expand_omp_for_init_counts): Update.
7713 (extract_omp_for_update_vars): Update.
7714 (expand_omp_ordered_sink): Update.
7715 (expand_omp_for_ordered_loops): Update.
7716 (expand_omp_for_generic): Update.
7717 (expand_omp_for_static_nochunk): Update.
7718 (expand_omp_for_static_chunk): Update.
7719 (expand_cilk_for): Update.
7720 (expand_omp_simd): Update.
7721 (expand_omp_taskloop_for_outer): Update.
7722 (expand_omp_taskloop_for_inner): Update.
7723 * omp-simd-clone.c (simd_clone_adjust): Update.
7724 * optabs.c (expand_doubleword_shift): Update.
7725 (expand_abs): Update.
7726 (emit_cmp_and_jump_insn_1): Update.
7727 (expand_compare_and_swap_loop): Update.
7728 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7729 * predict.c (predictable_edge_p): Update.
7730 (edge_probability_reliable_p): Update.
7731 (set_even_probabilities): Update.
7732 (combine_predictions_for_insn): Update.
7733 (combine_predictions_for_bb): Update.
7734 (propagate_freq): Update.
7735 (estimate_bb_frequencies): Update.
7736 (force_edge_cold): Update.
7737 * profile-count.c (profile_count::dump): Add missing space into dump.
7738 (profile_count::debug): Add newline.
7739 (profile_count::differs_from_p): Explicitly convert to unsigned.
7740 (profile_count::stream_in): Update.
7741 (profile_probability::dump): New member function.
7742 (profile_probability::debug): New member function.
7743 (profile_probability::differs_from_p): New member function.
7744 (profile_probability::differs_lot_from_p): New member function.
7745 (profile_probability::stream_in): New member function.
7746 (profile_probability::stream_out): New member function.
7747 * profile-count.h (profile_count_quality): Rename to ...
7748 (profile_quality): ... this one.
7749 (profile_probability): New.
7750 (profile_count): Update.
7751 * profile.c (compute_branch_probabilities): Update.
7752 * recog.c (peep2_attempt): Update.
7753 * sched-ebb.c (schedule_ebbs): Update.
7754 * sched-rgn.c (find_single_block_region): Update.
7755 (compute_dom_prob_ps): Update.
7756 (schedule_region): Update.
7757 * sel-sched-ir.c (compute_succs_info): Update.
7758 * stmt.c (struct case_node): Update.
7759 (do_jump_if_equal): Update.
7760 (get_outgoing_edge_probs): Update.
7761 (conditional_probability): Update.
7762 (emit_case_dispatch_table): Update.
7763 (expand_case): Update.
7764 (expand_sjlj_dispatch_table): Update.
7765 (emit_case_nodes): Update.
7766 * targhooks.c: Update.
7767 * tracer.c (better_p): Update.
7768 (find_best_successor): Update.
7769 * trans-mem.c (expand_transaction): Update.
7770 * tree-call-cdce.c: Update.
7771 * tree-cfg.c (gimple_split_edge): Upate.
7772 (move_sese_region_to_fn): Upate.
7773 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7774 * tree-eh.c (lower_resx): Upate.
7775 (cleanup_empty_eh_move_lp): Upate.
7776 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7777 * tree-inline.c (copy_edges_for_bb): Update.
7778 (copy_cfg_body): Update.
7779 * tree-parloops.c (gen_parallel_loop): Update.
7780 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7781 (gimple_gen_time_profiler): Update.
7782 * tree-ssa-dce.c (remove_dead_stmt): Update.
7783 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7784 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7785 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7786 (unloop_loops): Update.
7787 (try_peel_loop): Update.
7788 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7789 * tree-ssa-loop-split.c (connect_loops): Update.
7790 (split_loop): Update.
7791 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7792 (hoist_guard): Update.
7793 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7794 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7795 (value_replacement): Update.
7796 * tree-ssa-reassoc.c (branch_fixup): Update.
7797 * tree-ssa-tail-merge.c (replace_block_by): Update.
7798 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7799 (create_edge_and_update_destination_phis): Update.
7800 (compute_path_counts): Update.
7801 (recompute_probabilities): Update.
7802 (update_joiner_offpath_counts): Update.
7803 (freqs_to_counts_path): Update.
7804 (duplicate_thread_path): Update.
7805 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7806 (struct switch_conv_info): Update.
7807 (gen_inbound_check): Update.
7808 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7809 (vect_do_peeling): Update.
7810 (vect_loop_versioning): Update.
7811 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7812 (optimize_mask_stores): Update.
7813 * ubsan.c (ubsan_expand_null_ifn): Update.
7814 * value-prof.c (gimple_divmod_fixed_value): Update.
7815 (gimple_divmod_fixed_value_transform): Update.
7816 (gimple_mod_pow2): Update.
7817 (gimple_mod_pow2_value_transform): Update.
7818 (gimple_mod_subtract): Update.
7819 (gimple_mod_subtract_transform): Update.
7820 (gimple_ic): Update.
7821 (gimple_stringop_fixed_value): Update.
7822 (gimple_stringops_transform): Update.
7823 * value-prof.h: Update.
7825 2017-06-29 Carl Love <cel@us.ibm.com>
7827 * config/rs6000/rs6000-c.c: Add support for built-in functions
7828 vector signed int vec_signed (vector float);
7829 vector signed long long vec_signed (vector double);
7830 vector signed int vec_signed2 (vector double, vector double);
7831 vector signed int vec_signede (vector double);
7832 vector signed int vec_signedo (vector double);
7833 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
7834 instruction generator.
7835 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7836 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
7837 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
7839 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
7840 vunsignede_v2df): Add define_expands.
7841 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
7842 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
7843 VEC_UNSIGNEDO): Add definitions.
7844 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7845 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
7846 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
7847 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
7848 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
7849 * config/rs6000/altivec.h (vec_signed, vec_signed2,
7850 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
7851 vec_unsignede, vec_unsignedo): Add builtin defines.
7852 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
7854 * doc/extend.texi: Update the built-in documentation file for the
7855 new built-in functions.
7857 2017-06-29 Richard Biener <rguenther@suse.de>
7859 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
7860 reduction chains to LOOP_VINFO_REDUCTIONS.
7861 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
7862 SLP reductions after processing reduction chains.
7864 2017-06-29 Nathan Sidwell <nathan@acm.org>
7866 * builtins.c (fold_builtin_FUNCTION): Use
7867 lang_hooks.decl_printable_name.
7869 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
7872 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
7873 with only one label.
7874 * stmt.c (expand_case): Assert NCASES is greater than one.
7876 2017-06-29 Richard Biener <rguenther@suse.de>
7878 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
7880 (group_case_labels): Likewise.
7881 (find_taken_edge): Push sanity checking on val to workers...
7882 (find_taken_edge_cond_expr): ... here
7883 (find_taken_edge_switch_expr): ... and here, handle cases
7884 with just a default label.
7885 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
7886 (group_case_labels): Likewise.
7887 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
7888 group_case_labels does anything cleanup the CFG again.
7890 2017-06-29 Bin Cheng <bin.cheng@arm.com>
7892 PR tree-optimization/81196
7893 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
7894 exit condition comparing two IVs.
7896 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
7898 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
7899 profile to the dummy entry at the end of the list of architectures.
7900 * config/arm/arm-cpu-cdata.h: Regenerated.
7902 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7903 Michael Collison <michael.collison@arm.com>
7906 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
7908 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
7909 (*aarch64_reg_<mode>3_minus_mask): New pattern.
7910 (*aarch64_<optab>_reg_di3_mask2): New pattern.
7911 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
7912 of shift when the shift amount is masked with constant equal to
7913 the size of the mode.
7914 * config/aarch64/predicates.md (subreg_lowpart_operator): New
7917 2017-06-29 Martin Liska <mliska@suse.cz>
7919 * config/i386/i386.opt: Change range from [1,5] to [0,5].
7921 2017-06-29 Yury Gribov <tetra2005@gmail.com>
7924 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
7927 (edge_growth_cache_entry::edge_growth_cache_entry): New
7929 (reset_edge_growth_cache): Update to use constructor.
7931 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7933 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
7934 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7935 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7937 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
7939 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
7940 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
7942 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
7944 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
7945 (*-linux-uclibc*): Add t-uclibc tmake_file.
7946 * config/t-musl: New.
7947 * config/t-uclibc: New.
7949 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
7951 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
7953 (gen_comm_data): Emit architectural setting of arch_prof.
7954 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
7956 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
7957 (armv8-m.base, armv8-m.main): Likewise.
7958 * arm-protos.h (arm_build_target): Add profile field.
7959 (arch_option): Likewise.
7960 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
7962 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
7963 arm_active_target.profile.
7965 2017-06-28 Richard Biener <rguenther@suse.de>
7968 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
7969 TYPE_OVERFLOW_WRAPS.
7970 * match.pd (negate_expr_p): Likewise.
7971 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
7972 fold_build2, not fold_binary.
7974 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7976 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
7977 Convert memory address to Pmode.
7978 (aarch64_print_operand): Assert MEM operands are always Pmode.
7980 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7983 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
7984 Remove redundant if.
7985 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
7986 * config/arm/aarch-common-protos.h
7987 (aarch_forward_to_shift_is_not_shifted_re): Remove.
7988 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
7990 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
7993 * multiple_target.c (create_dispatcher_calls): Set the default
7994 clone to be static, not public.
7996 2017-06-28 Richard Biener <rguenther@suse.de>
7998 * tree-vect-loop.c (vectorizable_reduction): Move special
7999 cond reduction IV var creation ...
8000 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8001 parameter. Use STMT_VINFO_VECTYPE.
8002 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8005 2017-06-28 Martin Liska <mliska@suse.cz>
8008 * ipa-visibility.c (non_local_p): Handle visibility.
8010 2017-06-28 Martin Liska <mliska@suse.cz>
8013 * common.opt: Add IntegerRange to various options.
8014 * opt-functions.awk (integer_range_info): New function.
8015 * optc-gen.awk: Add integer_range_info to cl_options struct.
8016 * opts-common.c (decode_cmdline_option): Handle
8017 CL_ERR_INT_RANGE_ARG.
8018 (cmdline_handle_error): Likewise.
8019 * opts.c (print_filtered_help): Show valid interval in
8020 when --help is provided.
8021 * opts.h (struct cl_option): Add range_min and range_max fields.
8022 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8024 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8026 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8027 (x * C EQ/NE y * C): New transformation.
8029 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8031 * genmultilib (combination_space): Accept '+' in option names.
8033 2017-06-28 Martin Liska <mliska@suse.cz>
8036 * asan.c (instrument_derefs): Bail out inner references
8037 that are hard register variables.
8039 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8042 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8043 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8045 2017-06-28 Richard Biener <rguenther@suse.de>
8047 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8048 (vect_get_slp_defs): Adjust.
8049 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8051 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8053 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8054 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8055 (vectorizable_reduction): Adjust.
8056 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8058 (vect_get_slp_defs): Likewise.
8059 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8060 (vectorizable_bswap): Adjust.
8061 (vectorizable_call): Likewise.
8062 (vectorizable_conversion): Likewise.
8063 (vectorizable_assignment): Likewise.
8064 (vectorizable_shift): Likewise.
8065 (vectorizable_operation): Likewise.
8066 (vectorizable_store): Likewise.
8067 (vectorizable_condition): Likewise.
8068 (vectorizable_comparison): Likewise.
8070 2017-06-28 Michael Collison <michael.collison@arm.com>
8073 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8075 (arm_gen_movmemqi): Removed unused variable 'i'.
8076 Convert 'for' loop into 'while' loop.
8077 (arm_expand_prologue): Remove last unnecessary set of insn.
8078 (thumb_pop): Remove unused variable 'pushed_words'.
8079 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8081 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8083 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8084 * config/s390/s390-protos.h: Add prototype of
8085 s390_rel_address_ok_p.
8086 * config/s390/s390.c (s390_got_symbol): New function.
8087 (s390_rel_address_ok_p): New function.
8088 (legitimize_pic_address): Use s390_rel_address_ok_p.
8089 (s390_load_got): Use s390_got_symbol.
8090 (s390_option_override): Issue error if
8091 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8092 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8094 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8096 2017-06-27 Andrew Pinski <apinski@cavium.com>
8098 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8099 (X * copysign (1.0, X)): New pattern.
8100 (X * copysign (1.0, -X)): New pattern.
8101 (copysign (-1.0, CST)): New pattern.
8103 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8105 * genmultilib (combination_space): Remove variable.
8106 Validate reuse rules against regular expression for any sequence
8107 of multilib options in any order.
8109 2017-06-27 Michael Collison <michael.collison@arm.com>
8111 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8112 call aarch64_split_simd_combine.
8113 * (aarch64_combine_internal<mode>): Delete pattern.
8114 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8115 Allow register and subreg operands.
8117 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8119 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8120 specific need, just fallback on defaults.
8121 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8123 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8124 Olivier Hainque <hainque@adacore.com>
8126 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8128 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8129 targets. Pick a default if no particular attempt applied.
8130 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8133 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8135 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8136 (x86_64-wrs-vxworks7): Likewise.
8138 2017-06-27 Marek Polacek <polacek@redhat.com>
8141 * ubsan.c (instrument_null): Check get_base_address's result for null.
8143 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8145 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8147 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8149 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8150 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8151 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8152 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8154 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8155 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8156 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8157 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8158 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8159 BUILT_IN_FEUPDATEENV): New builtins.
8160 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8161 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8162 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8163 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8165 (builtin_structptr_types): Adjust size.
8166 * tree.c (builtin_structptr_types): Add four entries.
8168 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8169 Olivier Hainque <hainque@adacore.com>
8171 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8172 (TLS_SYM): New local macro, forcing reference to __tls__ on
8173 link command lines for VxWorks 7 RTPs, triggering initialization
8175 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8176 OS features TLS support, true for RTPs on VxWorks 7.
8177 * config/vxworks.c (vxworks_override_options): Setup emutls
8180 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8182 * predict.c (test_prediction_value_range): Use -1U instead of -1
8183 to avoid narrowing conversion warning.
8184 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8185 to avoid narrowing conversion warning.
8186 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8188 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8190 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8192 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8193 64bit configurations.
8194 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8195 (SIZE_TYPE): Likewise.
8196 * config/vxworks.c (vxworks_emutls_var_fields): Use
8197 long_unsigned_type_node instead of unsigned_type_node as the offset
8198 field type, which is "pointer" mode in emutls.c.
8200 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8203 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8206 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8207 gimple_vuse copying separately from gimple_vdef copying.
8209 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8211 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8214 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8215 Olivier Hainque <hainque@adacore.com>
8217 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8218 for all vxworks7 targets.
8219 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8220 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8221 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8222 variations for VX6/VX7 and 32/64bits later on in ...
8223 (VXWORKS_LIB_SPEC): Leverage new macros.
8224 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8225 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8227 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8229 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8230 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8232 2017-06-26 Carl Love <cel@us.ibm.com>
8234 * config/rs6000/rs6000-c.c: Add support for built-in functions
8235 vector bool char vec_reve (vector bool char);
8236 vector signed char vec_reve (vector signed char);
8237 vector unsigned char vec_reve (vector unsigned char);
8238 vector bool int vec_reve (vector bool int);
8239 vector signed int vec_reve (vector signed int);
8240 vector unsigned int vec_reve (vector unsigned int);
8241 vector bool long long vec_reve (vector bool long long);
8242 vector signed long long vec_reve (vector signed long long);
8243 vector unsigned long long vec_reve (vector unsigned long long);
8244 vector bool short vec_reve (vector bool short);
8245 vector signed short vec_reve (vector signed short);
8246 vector double vec_reve (vector double);
8247 vector float vec_reve (vector float);
8248 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8249 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8250 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8251 (altivec_vreve): New pattern.
8252 * config/rs6000/altivec.h (vec_reve): New define.
8253 * doc/extend.texi (vec_rev): Update the built-in documentation file
8254 for the new built-in functions.
8256 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8258 PR tree-optimization/71815
8259 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8261 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8263 (slsr_process_phi): Likewise.
8264 (replace_uncond_cands_and_profitable_phis): Don't replace a
8265 multiply candidate with a stride of 1 (copy or cast).
8266 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8268 (lowest_cost_path): Likewise.
8269 (total_savings): Likewise.
8271 2017-06-26 Richard Biener <rguenther@suse.de>
8274 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8275 Use def_builtin_pure for all gather builtins.
8277 2017-06-26 Richard Biener <rguenther@suse.de>
8279 PR tree-optimization/81203
8280 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8283 2017-06-26 Marek Polacek <polacek@redhat.com>
8286 * doc/invoke.texi: Document -Wmultistatement-macros.
8288 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8290 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8291 arm_neon_ok_no_float_abi effective target.
8293 2017-06-26 Richard Biener <rguenther@suse.de>
8295 PR tree-optimization/80928
8296 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8297 (copy_bbs): Set BB_DUPLICATED flag early.
8298 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8300 (execute_on_shrinking_pred): Likewise.
8301 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8302 BB_DUPLICATED blocks.
8303 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8304 iterate over all PHIs considering removal of *gsi.
8306 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8308 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8311 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8313 * config/rs6000/rs6000-string.c: (expand_block_clear,
8314 do_load_for_compare, select_block_compare_mode,
8315 compute_current_alignment, expand_block_compare,
8316 expand_strncmp_align_check, expand_strn_compare,
8317 expand_block_move, rs6000_output_load_multiple)
8318 Move functions related to string/block move/compare
8320 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8321 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8322 for this function which is now used in two files.
8323 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8324 * config.gcc: Add rs6000-string.o to extra_objs for
8325 targets powerpc*-*-* and rs6000*-*-*.
8327 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8330 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8331 32-bit, since indexed is not valid for DImode.
8332 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8333 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8334 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8335 (define_peephole2 for Altivec d-form store): Likewise.
8338 * multiple_target.c (create_dispatcher_calls): Only create the
8339 dispatcher call if the function is the default clone of a
8342 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8345 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8346 a call, force the call to not be a tail call.
8348 2017-06-23 Jeff Law <law@redhat.com>
8350 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8353 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8355 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8356 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8357 handling for early expansion of vector shifts (sl,sr,sra,rl).
8358 (builtin_function_type): Add vector shift right instructions
8359 to the unsigned argument list.
8361 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8363 rtl-optimizatoin/79286
8364 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8365 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8366 trap. PIC register plus a const unspec without offset can never trap.
8368 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8370 * tree.h (builtin_structptr_type): New type.
8371 (builtin_structptr_types): Declare new array.
8372 * tree.c (builtin_structptr_types): New array.
8373 (free_lang_data, build_common_tree_nodes): Use it.
8375 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8378 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8381 2017-06-22 Matt Turner <mattst88@gmail.com>
8383 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8384 Lake models to skylake case. Assume skylake for unknown
8385 models with clflushopt.
8387 2017-06-22 Jeff Law <law@redhat.com>
8389 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8390 frame sizes that do not satisfy aarch64_uimm12_shift.
8392 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8394 * profile-count.h (apply_probability,
8395 apply_scale, probability_in): Fix checks for zero.
8397 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8399 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8400 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8402 2016-06-22 Richard Biener <rguenther@suse.de>
8404 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8405 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8406 REDUC_MAX_EXPR support.
8407 (vectorizable_reduction): Likewise.
8408 (vect_create_epilog_for_reduction): Likewise.
8410 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8412 * match.pd (A / (1 << B) -> A >> B): New.
8413 * generic-match-head.c: Include optabs-tree.h.
8414 * gimple-match-head.c: Likewise.
8415 * optabs-tree.h (target_supports_op_p): New.
8416 * optabs-tree.c (target_supports_op_p): New.
8418 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8420 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8421 $gcc_cv_ld --help output.
8422 (gcc_cv_ld_demangle): Likewise.
8423 (gcc_cv_ld_eh_frame_hdr): Likewise.
8424 (gcc_cv_ld_pie): Likewise.
8425 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8426 (gcc_cv_ld_buildid): Likewise.
8427 (gcc_cv_ld_sysroot): Likewise.
8428 (ld_bndplt_support): Likewise.
8429 (ld_pushpopstate_support): Likewise.
8430 * configure: Regenerate.
8431 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8433 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8436 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8437 operands indexes to avoid gap between operands and match_dups.
8439 2017-06-21 Andrew Pinski <apinski@cavium.com>
8441 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8442 Increment Arith_shift and Arith_shift_reg by 1.
8443 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8445 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8446 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8447 (aarch64_strip_extend): Add new argument and test for it.
8448 (aarch64_cheap_mult_shift_p): New function.
8449 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8450 add a cost if it is true.
8451 Update calls to aarch64_strip_extend.
8452 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8454 2017-06-21 Andrew Pinski <apinski@cavium.com>
8456 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8458 (thunderxt88): Likewise.
8459 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8460 (thunderx_prefetch_tune): New variable.
8461 (thunderx2t99_prefetch_tune): Update for the correct values.
8462 (thunderxt88_tunings): New variable.
8463 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8464 generic_prefetch_tune.
8465 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8467 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8469 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8470 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8471 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8472 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8473 (aarch64_atomic_cas<mode>, GPI): Likewise.
8475 2017-06-21 Martin Liska <mliska@suse.cz>
8477 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8478 statements on cold and hot labels.
8479 * predict.c (tree_estimate_probability_bb): Remove the
8480 prediction from this place.
8482 2017-06-21 Martin Liska <mliska@suse.cz>
8484 PR tree-optimization/79489
8485 * gimplify.c (maybe_add_early_return_predict_stmt): New
8487 (gimplify_return_expr): Call the function.
8488 * predict.c (tree_estimate_probability_bb): Remove handling
8490 * predict.def: Update comment about early return predictor.
8491 * gimple-predict.h (is_gimple_predict): New function.
8492 * predict.def: Change default value of early return to 66.
8493 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8495 * passes.def: Put pass_strip_predict_hints to the beginning of
8498 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8500 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8501 FUNCTION_DECL declarations.
8502 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8504 (dwaf2out_decl): Likewise.
8505 * godump.c (go_early_global_decl): Skip call to the real debug hook
8506 for FUNCTION_DECL declarations.
8507 * passes.c (rest_of_decl_compilation): Skip call to the
8508 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8509 -fdump-go-spec is passed.
8511 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8513 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8514 Reorder for compactness.
8515 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8516 (def_builtin_pure, def_builtin_pure2): New functions.
8517 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8519 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8521 * match.pd (nop_convert): New predicate.
8522 ((A +- CST1) +- CST2): Allow some NOP conversions.
8524 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8527 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8528 with ctors/dtors if GOVD_SHARED is set.
8530 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8532 * config/aarch64/aarch64.md (movti_aarch64):
8533 Emit mov rather than orr.
8534 (movtf_aarch64): Likewise.
8535 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8536 Emit mov rather than orr.
8538 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8540 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8541 Swap alternatives, make integer dup more expensive.
8543 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8545 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8546 Return true for non-tls symbols.
8548 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8550 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8551 (cortex-a75): Likewise.
8552 (cortex-a75.cortex-a55): Likewise.
8553 * config/aarch64/aarch64-tune.md: Regenerate.
8554 * doc/invoke.texi (-mtune): Document new values for -mtune.
8556 2017-06-21 Tom de Vries <tom@codesourcery.com>
8558 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8560 (Effective-Target Keywords, Other attributes): Suggest using
8561 dg-add-options stack_size feature to get stack limit in stack_size
8562 effective target documentation.
8564 2017-06-21 Julian Brown <julian@codesourcery.com>
8565 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8567 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8568 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8569 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8571 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8572 attribute type list for neon_multiply.
8573 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8574 attribute type list for neon_multiply.
8575 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8576 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8577 attribute type list for neon_multiply.
8578 * config/arm/types.md (crypto_pmull): Add.
8579 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8580 attribute type list.
8582 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8584 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8587 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8589 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8590 to make sure not to dereference a NULL cost_classes_ptr pointer.
8592 2017-06-20 Carl Love <cel@us.ibm.com>
8594 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8595 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8596 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8597 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8598 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8599 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8600 VMULOSW): New enum "unspec" values.
8601 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8602 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8603 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8604 altivec_vmulosw): New patterns.
8605 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8606 VMULOSW): Add definitions.
8608 2017-06-20 Julia Koval <julia.koval@intel.com>
8610 * config/i386/i386.c: Fix rounding expand for new pattern.
8611 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8613 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8615 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8616 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8618 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8620 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8623 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8625 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8626 architecture, then by implementer ID.
8627 * config/aarch64/aarch64-tune.md: Regenerate.
8629 2017-06-20 Richard Biener <rguenther@suse.de>
8632 * fold-const.c (split_tree): Fold to type before negating.
8634 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8636 * diagnostic-show-locus.c
8637 (selftest::test_fixit_deletion_affecting_newline): New function.
8638 (selftest::diagnostic_show_locus_c_tests): Call it.
8640 2017-06-20 Andreas Schwab <schwab@suse.de>
8643 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8646 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8648 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8649 __ARM_FEATURE_COPROC according to support.
8651 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8653 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8654 Rewritten to avoid overflow for > 32-bit pointers.
8657 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8658 by removing enum keyword.
8659 (ubsan_type_descriptor): Likewise. Formatting fix.
8662 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8663 splitter): Require TARGET_SSE2 in the condition.
8665 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8668 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8669 for doing vector set of SFmode on ISA 3.0.
8670 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8671 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8673 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8674 SFmode value into a V4SF variable that was extracted from another
8675 V4SF variable without converting the element to double precision
8676 and back to single precision vector format.
8677 (vsx_insert_extract_v4sf_p9_2): Likewise.
8679 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8681 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8682 in UWHI to avoid undefined overflow.
8685 * ubsan.h (enum ubsan_encode_value_phase): New.
8686 (ubsan_encode_value): Change second argument to
8687 enum ubsan_encode_value_phase with default value of
8688 UBSAN_ENCODE_VALUE_GENERIC.
8689 * ubsan.c (ubsan_encode_value): Change second argument to
8690 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8691 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8692 create_tmp_var_raw instead of create_tmp_var and use a
8694 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8695 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8696 ubsan_encode_value callers.
8699 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8700 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8701 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8703 2017-06-19 Richard Biener <rguenther@suse.de>
8706 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8707 estimates if we changed anything.
8709 2017-06-19 Richard Biener <rguenther@suse.de>
8711 PR tree-optimization/80887
8712 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8713 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8714 simplified lookups, then reset mprts_hook.
8715 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8717 (try_to_simplify): Likewise.
8719 2017-06-19 Martin Liska <mliska@suse.cz>
8722 * gimplify.c (gimplify_switch_expr):
8723 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8725 2017-06-19 Martin Liska <mliska@suse.cz>
8727 * doc/install.texi: Document that PGO runs in 4 stages.
8729 2017-06-19 Martin Liska <mliska@suse.cz>
8732 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8733 to dispatcher function name.
8734 * multiple_target.c (replace_function_decl): New function.
8735 (create_dispatcher_calls): Redirect both edges and references.
8737 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8739 * profile-count.c (profile_count::dump): Dump quality.
8740 (profile_count::differs_from_p): Update for unsigned val.
8741 * profile-count.h (profile_count_quality): New enum.
8742 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8744 2017-06-19 Richard Biener <rguenther@suse.de>
8746 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8747 struct function as arg.
8748 (estimate_numbers_of_iterations): Export overload with loop arg.
8749 (free_numbers_of_iterations_estimates_loop): Use an overload of
8750 free_numbers_of_iterations_estimates instead.
8751 * tree-cfg.c (remove_bb): Adjust.
8752 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8753 * tree-parloops.c (gen_parallel_loop): Likewise.
8754 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8756 (tree_unroll_loops_completely): Likewise.
8757 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8758 Use an overload instead and export.
8759 (estimated_loop_iterations): Adjust.
8760 (max_loop_iterations): Likewise.
8761 (likely_max_loop_iterations): Likewise.
8762 (estimate_numbers_of_iterations): Take struct function as arg
8764 (loop_exits_before_overflow): Adjust.
8765 (free_numbers_of_iterations_estimates_loop): Use an overload.
8766 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8767 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8769 2017-06-19 Richard Biener <rguenther@suse.de>
8772 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8773 RANGE_EXPR conservatively.
8775 2017-06-16 Carl Love <cel@us.ibm.com>
8777 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8778 definitions for vec_float, vec_float2, vec_floato,
8779 vec_floate built-ins.
8780 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8781 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8783 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8784 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8785 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8786 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8787 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8788 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8789 vec_floato): Add builtin defines.
8790 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8791 Update the built-in documentation file for the new built-in
8794 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8796 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8797 (mthumb): Mark as the negative of -marm.
8799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8801 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8803 (ARM Options, -mtune): Document that this accepts the same
8804 extension options as -mcpu.
8805 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8809 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8810 permitted extensions.
8812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8814 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8815 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8816 (armv8-m.main): Add option +nodsp.
8817 * config/arm/arm-cpu-cdata.h: Regenerated.
8819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8821 * config/arm/t-fuchsia: New file.
8822 * config.gcc (arm*-*-fuchsia*): Use it.
8824 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8826 * config/arm/t-symbian: Rewrite for new option infrastructure.
8828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8830 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
8831 (MULTILIB_REQUIRED): Likewise.
8833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8835 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
8836 (MULTILIB_RESUE): Likewise.
8837 (MULTILIB_MATCHES): Likewise.
8838 (MULTLIB_REQUIRED): Likewise.
8840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8842 * config/arm/t-rtems: Rewrite for new option framework.
8844 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8846 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
8847 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
8848 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
8849 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
8850 * config/arm/t-multilib: ... here.
8851 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
8852 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
8853 armv7-a and armv8*-a when A-profile libraries have not been built.
8854 * config/arm/t-rmprofile: Rewrite.
8856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8858 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
8859 with a backslash. Remove the backslash after substituting unescaped
8861 * doc/fragments.texi (MULTILIB_REUSE): Document it.
8863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8865 * config.gcc: (arm*-*-*): When building a-profile libraries, force
8866 the driver to pass through the default setting of -mfloat-abi.
8867 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
8869 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
8870 (all_feat_combs): New rule.
8871 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
8873 * config/arm/t-aprofile: Rewrite.
8875 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8877 * config/arm/arm.h (FPUTYPE_AUTO): Define.
8878 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
8879 fpu is not specified by the user/command-line.
8880 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
8881 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
8882 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
8883 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
8884 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
8885 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
8887 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8889 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
8890 * config/arm/t-arm-elf: Rewritten.
8892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8894 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
8895 have some floating-point instructions.
8896 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
8897 (TARGET_MAYBE_HARD_FLOAT): New macro.
8898 * config/arm/arm-builtins.c (arm_init_builtins): Use
8899 TARGET_MAYBE_HARD_FLOAT.
8900 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
8902 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8904 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
8905 (configargs.h): Include it.
8906 (arm_print_hint_for_fpu_option): New function.
8907 (arm_parse_fpu_option): New function.
8908 (candidate_extension): New class.
8909 (arm_canon_for_multilib): New function.
8910 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
8911 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
8912 (ARCH_CANONICAL_SPECS): New macro.
8913 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
8915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8917 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
8918 are set after handling multilib fragments. Set target_cpu_default2
8921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8923 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
8925 (arm*-*-*): Set target_cpu_default2 to a quoted string.
8926 * config/arm/parsecpu.awk (check_cpu): Validate any extension
8928 (check_arch): Likewise.
8929 * config/arm/arm.c (arm_configure_build_target): Handle
8930 TARGET_CPU_DEFAULT being a string constant. Scan any feature
8931 options in the default.
8933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8935 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
8936 when an option is an alias of another.
8937 * config/arm/parsecpu.awk (optalias): New parser token.
8938 (gen_comm_data): Mark non-alias options as such. Emit entries
8939 for extension aliases.
8940 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
8941 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
8942 (armv6kz, armv6zk, armv6t2): Likewise.
8943 (armv7): Make vfpv3-d16 an alias.
8944 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
8946 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
8947 Sort in canonical order.
8948 (armv8-a): Sort in canonical order.
8949 (armv8.1-a, armv8.2-a): Likewise.
8950 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
8952 (cortex-a9): Sort in canonical order.
8953 * config/arm/arm.c (selftests.h): Include it.
8954 (arm_test_cpu_arch_data): New function.
8955 (arm_run_self_tests): New function.
8956 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
8957 (targetm): Move declaration to the end of the file.
8958 * arm-cpu-cdata.h: Regenerated.
8960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8962 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
8963 call to target_mode_check describing the type of option passed.
8964 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
8965 (arm_target_thumb_only): Use arm_parse_arch_option_name or
8966 arm_parse_cpu_option_name to match parameters against list of
8968 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
8969 arm_arch_core_flags data structure.
8970 * config/arm/arm-cpu_cdata.h: Regenerated.
8972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8974 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
8976 (arm_print_hint_for_cpu_option): Likewise.
8977 (arm_print_hint_for_arch_option): Likewise.
8978 (arm_parse_cpu_option_name): Likewise.
8979 (arm_parse_arch_option_name): Likewise.
8980 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
8981 of entries in the all_fpus list.
8982 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
8983 (arm_parse_cpu_option_name): Declare.
8984 (arm_parse_arch_option_name): Declare.
8985 (arm_parse_option_features): Declare.
8986 (arm_intialize_isa): Declare.
8987 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
8989 (gen_comm_data): ... here. Make definitions non-static.
8990 * config/arm/arm-cpu-data.h: Regenerated.
8991 * config/arm/arm-cpu-cdata.h: Regenerated.
8993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8995 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
8996 (cpu_arch_extension): New structure.
8997 (cpu_arch_option, arch_option, cpu_option): New structures.
8998 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9000 (gen_data): Generate new format data tables.
9001 * config/arm/arm.c (cpu_tune): New structure.
9002 (cpu_option, processors): Delete.
9003 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9004 (arm_print_hint_for_cpu_option): ... this and ...
9005 (arm_print_hint_for_arch_option): ... this.
9006 (arm_parse_arch_cpu_name): Delete. Replace with ...
9007 (arm_parse_cpu_option_name): ... this and ...
9008 (arm_parse_arch_option_name): ... this.
9009 (arm_unrecognized_feature): Change type of target parameter to
9011 (arm_parse_arch_cpu_features): Delete. Replace with ...
9012 (arm_parse_option_features): ... this.
9013 (arm_configure_build_target): Rework to use new configuration data
9015 (arm_print_tune_info): Rework for new configuration data tables.
9016 * config/arm/arm-cpu-data.h: Regenerated.
9017 * config/arm/arm-cpu.h: Regenerated.
9019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9021 * Makefile.in (OBJS): Move sbitmap.o from here ...
9022 (OBJS-libcommon): ... to here.
9024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9026 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9027 (ISA_ALL_CRYPTO): New macro.
9028 (ISA_ALL_SIMD): New macro
9029 (ISA_ALL_FP): New macro.
9030 * config/arm/arm.c (fpu_bitlist): Update initializer.
9031 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9033 (arm9e): Add fpu. Add option for nofp
9034 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9035 (arm926ej-s, arm1026ej-s): Likewise.
9036 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9037 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9038 neon-fp16, neon-vfpv4, nofp and nosimd.
9039 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9040 (cortex-a8): Add fpu. Add option for nofp.
9041 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9042 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9043 (cortex-r4f): Add fpu.
9044 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9045 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9047 (cortex-r8): Likewise.
9048 (cortex-m4): Add fpu. Add option for nofp.
9049 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9050 (cortex-a17.cortex-a7): Likewise.
9051 (cortex-a32): Add fpu. Add options for crypto and nofp.
9052 (cortex-a35, cortex-a53): Likewise.
9053 (cortex-a57): Add fpu. Add option for crypto.
9054 (cortex-a72, cortex-a73): Likewise.
9055 (exynos-m1): Likewise.
9056 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9057 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9058 (cortex-m33): Add fpu. Add option for nofp.
9059 * config/arm/arm-cpu-cdata.h: Regenerated
9060 * config/arm/arm-cpu-data.h: Regenerated.
9062 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9064 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9065 (armv5te, armv5tej): Likewise.
9066 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9067 (armv7): Add options fp and vfpv3-d16.
9068 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9069 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9071 (armv7ve): Likewise.
9072 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9073 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9074 (armv8-a): Add nocrypto option.
9075 (armv8.1-a, armv8.2-a): Likewise.
9076 (armv8-m.main): add options fp, fp.dp and nofp.
9078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9080 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9082 (armv8-a+crc): Delete.
9083 (armv8.1-a): Add options simd, crypto and nofp.
9084 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9085 (armv8.2-a+fp16): Delete.
9086 (armv8-m.main): Add option dsp.
9087 (armv8-m.main+dsp): Delete.
9088 (cortex-a8): Add fpu. Add nofp option.
9089 (cortex-a9): Add fpu. Add nofp and nosimd options.
9090 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9091 link to main cpu and architecture data structures.
9092 (gen_comm_data): Only put isa attributes from the main architecture
9094 (option): New statement for architecture and CPU entries.
9095 * arm.c (struct cpu_option): New structure.
9096 (struct processors): Add entry for options.
9097 (arm_unrecognized_feature): New function.
9098 (arm_parse_arch_cpu_name): Ignore any characters after the first
9100 (arm_parse_arch_cpu_feature): New function.
9101 (arm_configure_build_target): Separate out any CPU and architecture
9102 features and parse separately. Don't error out if -mfpu=auto is
9103 used with only an architecture string.
9104 (arm_print_asm_arch_directives): New function.
9105 (arm_file_start): Call it.
9106 * config/arm/arm-cpu-cdata.h: Regenerated.
9107 * config/arm/arm-cpu-data.h: Likewise.
9108 * config/arm/arm-tables.opt: Likewise.
9110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9112 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9113 assembler when it is not -mfpu=auto.
9115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9117 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9118 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9119 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9120 (ASM_CPU_SPEC): Rewrite.
9121 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9122 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9123 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9124 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9125 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9126 copied string is NUL-terminated. Also strip any characters prefixed
9128 (arm_rewrite_selected_arch): New function.
9129 (arm_rewrite_march): New function.
9131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9133 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9134 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9135 (march, mcpu, mtune): Convert to string-based options.
9136 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9137 (arm_parse_arch_cpu_name): New function.
9138 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9139 identify selected architecture or CPU.
9140 (arm_option_save): New function.
9141 (TARGET_OPTION_SAVE): Redefine.
9142 (arm_option_restore): Restore string options.
9143 (arm_option_print): Print string options.
9145 2017-06-16 Martin Sebor <msebor@redhat.com>
9147 PR tree-optimization/80933
9148 PR tree-optimization/80934
9149 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9150 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9151 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9152 (gimple_fold_builtin): Call them.
9154 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9156 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9157 as unlikely; update profile.
9159 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9161 * predict.c (force_edge_cold): Handle declaring edges impossible
9164 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9166 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9168 (try_unroll_loop_completely): Fix reporting.
9170 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9172 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9174 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9177 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9178 if given a non-constant argument for an intrinsic which requires a
9181 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9183 * profile.c (compare_freqs): New function.
9184 (branch_prob): Sort edge list.
9185 (find_spanning_tree): Assume that the list is priority sorted.
9187 2017-06-16 Richard Biener <rguenther@suse.de>
9189 PR tree-optimization/81090
9190 * passes.def (pass_record_bounds): Remove.
9191 * tree-pass.h (make_pass_record_bounds): Likewise.
9192 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9193 make_pass_record_bounds): Likewise.
9194 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9195 not free niter estimates at the beginning but at the end.
9196 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9198 2017-06-16 Richard Biener <rguenther@suse.de>
9200 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9201 initializer to workaround ICE in host GCC 4.8.
9203 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9205 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9207 (clone_inlined_nodes): Update.
9209 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9211 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9212 prefetch settings, and enable prefetching by default at -O3.
9214 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9216 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9217 Set flag_prefetch_loop_arrays according to tuning data.
9219 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9221 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9223 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9224 [Unrelated to main purpose of the patch] Place the pointer field last
9225 to enable type checking errors when tune structure are wrongly merged.
9226 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9227 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9228 (thunderx2t99_prefetch_tune): New tune constants.
9229 (tune_params *_tunings): Update all tunings (no functional change).
9230 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9231 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9232 from tunings structures.
9234 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9237 * ubsan.c (instrument_null): Add T argument, use it instead
9238 of computing it based on IS_LHS.
9239 (instrument_object_size): Likewise.
9240 (pass_ubsan::execute): Adjust instrument_null and
9241 instrument_object_size callers to pass gimple_get_lhs or
9242 gimple_assign_rhs1 result to it. Use instrument_null instead of
9243 calling get_base_address and instrument_mem_ref. Handle
9244 aggregate call arguments for object-size sanitization.
9246 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9248 PR tree-optimization/81089
9249 * tree-vrp.c (is_masked_range_test): Validate operands of
9252 2017-06-15 Martin Sebor <msebor@redhat.com>
9255 * dumpfile.c (dump_register): Avoid calling memset to initialize
9256 a class with a default ctor.
9257 * gcc.c (struct compiler): Remove const qualification.
9258 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9259 * hash-table.h: Ditto.
9260 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9262 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9263 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9265 * params.h (struct param_info): Make struct members non-const.
9266 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9267 with default initialization.
9268 * vec.h (vec_copy_construct, vec_default_construct): New helper
9270 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9271 with vec_copy_construct.
9272 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9273 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9274 * doc/invoke.texi (-Wclass-memaccess): Document.
9276 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9278 * emit-rtl.h (is_leaf): Update comment about local
9281 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9284 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9285 for a variable to have a section before checking if the section has a
9287 Set section to.persistent if persistent attribute is set.
9288 Warn if .persistent attribute is used on an automatic variable.
9290 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9292 PR rtl-optimization/80474
9293 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9295 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9297 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9300 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9302 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9303 (casesi): Emit bounds checking as RTL.
9304 (casesi_internal_mips16_<mode>): Remove bounds checking.
9306 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9308 * config/xtensa/xtensa.c (xtensa_option_override): Append
9309 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9310 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9311 xtensa_doloop_hooks): Define unconditionally.
9312 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9314 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9315 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9316 for it in xtensa_option_override.
9317 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9318 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9320 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9322 * doc/cppopts.texi: Document '-' special value to -MF.
9324 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9326 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9327 (cortex_a53_fconst): Likewise.
9328 (cortex_a53_fpmul): Likewise.
9329 (cortex_a53_f_load_64): Likewise.
9330 (cortex_a53_f_load_many): Likewise.
9331 (cortex_a53_advsimd_alu): Likewise.
9332 (cortex_a53_advsimd_alu_q): Likewise.
9333 (cortex_a53_advsimd_mul): Likewise.
9334 (cortex_a53_advsimd_mul_q): Likewise.
9335 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9336 Add missing fmul, r2f_cvt and fconst cases.
9338 2017-06-14 Richard Biener <rguenther@suse.de>
9341 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9343 (fold_binary_loc): When associating do not treat pre-existing
9344 TREE_OVERFLOW on literal constants as a reason to allow
9345 TREE_OVERFLOW on associated literal constants.
9347 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9349 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9350 (MASK_FEATURES): New macro.
9351 * config/sparc/sparc.c (sparc_option_override): Remove the special
9352 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9354 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9356 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9357 a division of 0 if non-call exceptions are enabled.
9359 2017-06-14 Andrew Pinski <apinski@cavium.com>
9360 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9363 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9364 Improve vector initialization code gen for only variable case.
9366 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9368 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9370 2017-06-14 Richard Biener <rguenther@suse.de>
9372 PR tree-optimization/81083
9373 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9376 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9378 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9379 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9380 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9381 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9382 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9383 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9385 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9387 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9388 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9390 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9392 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9394 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9396 * config/rs6000/t-rtems: Don't handle SPE.
9398 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9400 * config/rs6000/t-linux: Don't handle SPE.
9402 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9404 * config/rs6000/eabispe.h: Delete file.
9406 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9408 * config/rs6000/t-spe: Delete file.
9410 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9412 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9413 (rs6000_legitimate_offset_address_p): Return false for anything in
9414 V2SImode or V2SFmode.
9416 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9418 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9419 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9420 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9422 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9423 (rs6000_legitimate_offset_address_p): Ditto.
9424 (rs6000_emit_move): Ditto.
9425 (rs6000_init_builtins): Remove V4HI_type_node.
9427 2017-06-13 Martin Liska <mliska@suse.cz>
9430 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9431 (gate_asan): Likewise.
9432 * asan.h (asan_no_sanitize_address_p): Remove the function.
9433 (sanitize_flags_p): New function.
9434 * builtins.def: Fix coding style.
9435 * common.opt: Use renamed enum value.
9436 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9437 * doc/extend.texi: Document no_sanitize attribute.
9438 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9439 to SANITIZE_UNDEFINED_NONDEFAULT.
9440 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9441 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9442 Use sanitize_flags_p.
9443 * gimplify.c (gimplify_function_tree): Likewise.
9444 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9445 * opts.c (parse_no_sanitize_attribute): New function.
9446 (common_handle_option): Use renamed enum value.
9447 * opts.h (parse_no_sanitize_attribute): Declare.
9448 * tree.c (sanitize_flags_p): New function.
9449 * tree.h: Declared here.
9450 * tsan.c: Use sanitize_flags_p.
9451 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9452 (instrument_mem_ref): Likewise.
9453 (instrument_bool_enum_load): Likewise.
9454 (do_ubsan_in_current_function): Remove the function.
9455 (pass_ubsan::execute): Use sanitize_flags_p.
9456 * ubsan.h: Remove do_ubsan_in_current_function
9457 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9458 (dump_function_to_file): Use it here.
9460 2017-06-13 Martin Jambor <mjambor@suse.cz>
9462 PR tree-optimization/80803
9463 PR tree-optimization/81063
9464 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9465 (propagate_subaccesses_across_link): Enqueue subtree whenever
9466 necessary instead of relying on the caller.
9468 2017-06-13 Martin Jambor <mjambor@suse.cz>
9470 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9471 that have a first_link.
9472 (sort_and_splice_var_accesses): Do not check first_link before
9474 (subtree_mark_written_and_enqueue): Likewise.
9475 (propagate_all_subaccesses): Likewise and do not stop at first
9476 parent with a first_link.
9478 2017-06-13 Martin Jambor <mjambor@suse.cz>
9480 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9483 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9485 * match.pd: New pattern.
9487 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9489 * tree-vrp.c (is_masked_range_test): New function.
9490 (register_edge_assert_for): Determine ranges for
9493 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9495 PR tree-optimization/67328
9496 * fold-const.c (maskable_range_p): New function.
9497 (build_range_check): Generate bittests if possible.
9499 2017-06-13 Martin Liska <mliska@suse.cz>
9501 * gimple-pretty-print.c (dump_probability): Add new argument.
9502 (dump_edge_probability): Dump both probability and count.
9503 (dump_gimple_label): Likewise.
9504 (dump_gimple_bb_header): Likewise.
9506 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9509 * config/avr/avr-devices.c: Fix indentation.
9510 * config/avr/gen-avr-mmcu-specs.c: Dito.
9512 2017-06-13 Richard Biener <rguenther@suse.de>
9514 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9515 instead get vector type from stmt_info.
9516 (vectorizable_reduction): Adjust. Remove dead code.
9518 2017-06-13 Richard Biener <rguenther@suse.de>
9521 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9522 case of C * (x * C2 + C3).
9523 (fold_addr_of_array_ref_difference): Properly fold index difference.
9525 2017-06-12 David S. Miller <davem@davemloft.net>
9528 * config/sparc/sparc.md (return expander): Emit frame blockage if
9529 function uses alloca.
9531 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9533 * combine.c (make_field_assignment): Check len rather than the mode
9534 precision when calling force_to_mode.
9536 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9538 Support multilibs and devices that see flash in RAM address range.
9541 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9542 (avr_mcu_t) <flash_pm_offset>: New field.
9543 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9544 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9545 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9546 (AVR_TINY_PM_OFFSET): Remove macro.
9547 * config/avr/avr.opt (-mshort-calls): New option.
9548 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9549 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9550 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9551 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9552 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9553 instead of avr_arch->have_jmp_call.
9554 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9555 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9556 avr_arch->flash_pm_offset to define.
9557 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9558 new field flash_pm_offset. Add entry for avrxmega3.
9559 (avr_texinfo): Add entry for avrxmega3.
9560 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9561 attiny212, attiny214,
9562 attiny412, attiny414, attiny416, attiny417,
9563 attiny814, attiny816, attiny817,
9564 attiny1614, attiny1616, attiny1617,
9565 attiny3214, attiny3216, attiny3217.
9566 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9567 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9568 (avr_print_operand_address) [AVR_TINY]: Same.
9569 (avr_asm_init_sections) <readonly_data_section>: Only patch
9570 callback if avr_arch->flash_pm_offset = 0.
9571 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9572 for rodata if avr_arch->flash_pm_offset != 0.
9573 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9574 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9575 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9576 (m_options): Append opt_rcall.
9577 (m_dirnames): Append dir_rcall.
9578 * config/avr/t-multilib: Regenerate.
9580 * configure.ac [target=avr]: Check whether avrxmega3 default
9581 linker description file works as needed.
9582 * configure: Regenerate.
9583 * doc/avr-mmcu.texi: Regenerate.
9584 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9585 <__AVR_ARCH__>: Document avrxmega3 and 103.
9586 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9587 <__AVR_SHORT_CALLS__>: Document it.
9588 <__AVR_PM_BASE_ADDRESS__>: Document it.
9589 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9590 (AVR Variable Attributes) <progmem>: Document this is
9591 not needed for avrxmega3.
9592 (AVR Named Address Spaces) <__flash>: Dito.
9594 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9596 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9598 2017-06-12 Doug Rupp <rupp@adacore.com>
9600 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9601 Append vxworks-stdint.h to the tm_file list.
9602 * config/vxworks-stdint.h: New file.
9604 2017-06-12 Martin Liska <mliska@suse.cz>
9606 PR tree-optimization/81041
9607 * tree-profile.c (gimple_gen_ic_func_profiler):
9608 Create an extra BB in profile-generate
9609 (gimple_gen_time_profiler): Likewise.
9611 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9613 PR tree-optimization/81003
9614 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9615 (update_range_test): Use it instead of force_gimple_operand_gsi.
9617 2017-06-12 Richard Biener <rguenther@suse.de>
9619 PR tree-optimization/81053
9620 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9621 with backedge value not defined in loop. Simplify def stmt
9624 2017-06-11 Tom de Vries <tom@codesourcery.com>
9627 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9629 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9630 nvptx_cannot_force_const_mem.
9632 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9634 * opts.c (finish_options): Move test for flag_split_stack after
9635 it has been initialized.
9637 2017-06-11 Jason Merrill <jason@redhat.com>
9639 * tree.h (id_equal): New.
9640 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9641 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9642 instead of strcmp of IDENTIFIER_POINTER.
9644 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9646 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9647 (mark_all_inlined_calls_cdtor): Fix formating.
9648 (inline_transform): Rescale profile before inlining.
9650 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9652 * cgraph.h (cgraph_edge::clone): Update prototype.
9653 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9654 (cgraph_node::create_clone): Update.
9655 (cgraph_node::create_version_clone): Update.
9656 * tree-inline.c (copy_bb): Update.
9657 (expand_call_inline): Update.
9659 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9661 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9662 factored out from ...
9663 (rs6000_emit_prologue): ... here.
9665 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9667 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9668 factored out from ...
9669 (rs6000_emit_prologue): ... here.
9671 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9673 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9675 (handle_missing_profiles): Fix computation of tp_first_run.
9676 (counts_to_freqs): Do not touch freqs when count is 0.
9678 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9680 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9683 2017-06-10 Tom de Vries <tom@codesourcery.com>
9685 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9686 attributes): Document signal effective target.
9688 2017-06-10 Tom de Vries <tom@codesourcery.com>
9690 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9691 Document effective target stack_size.
9693 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9695 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9696 to the edit_context if they can be auto-applied.
9698 2017-06-9 Ian Lance Taylor <iant@golang.org>
9700 * opts.c (finish_options): If -fsplit-stack, disable implicit
9701 -forder-blocks-and-partition.
9702 * doc/invoke.texi (Optimize Options): Document that when using
9703 -fsplit-stack -forder-blocks-and-partition is not implicitly
9706 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9708 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9709 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9710 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9711 * builtins.def (abort, trap, unreachable): Declare cold.
9712 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9713 * tree-core.h (ECF_COLD): New.
9714 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9715 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9717 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9719 * predict.c (unlikely_executed_stmt_p): Cleanup.
9721 2017-06-09 Richard Biener <rguenther@suse.de>
9723 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9724 model if the ref is always written to.
9726 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9728 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9730 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9732 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9735 2017-06-09 Tom de Vries <tom@codesourcery.com>
9738 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9739 "target cannot support label values" when encountering LABEL_REF.
9741 2017-06-09 Martin Liska <mliska@suse.cz>
9743 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9744 (gimple_gen_ic_func_profiler): Emit direct comparison
9745 of __gcov_indirect_call_callee with NULL.
9746 (gimple_gen_time_profiler): Change probability from
9747 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9749 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9751 * profile.c (edge_gcov_counts): Turn to pointer.
9752 (compute_branch_probabilities, compute_branch_probabilities): Update.
9753 (branch_prob): Do not clear edge_gcov_count.
9754 * profile.h (edge_gcov_counts): Turn to pointer.
9755 (edge_gcov_count): Update.
9757 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9759 * gimple.h (gimple_check_failed): Mark cold.
9761 2017-06-09 Richard Biener <rguenther@suse.de>
9763 PR tree-optimization/66623
9764 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9765 refactor check_reduction into two parts, properly computing
9766 whether we have to check reduction validity for outer loop
9769 2017-06-09 Richard Biener <rguenther@suse.de>
9771 PR tree-optimization/79483
9772 * graphite-scop-detection.c (order): New global.
9773 (get_order): Compute bb to order mapping that satisfies code
9774 generation constraints.
9775 (cmp_pbbs): New helper.
9776 (build_scops): Start domwalk at entry block, sort generated
9779 2017-06-09 Richard Biener <rguenther@suse.de>
9782 * ipa-polymorphic-call.c
9783 (ipa_polymorphic_call_context::restrict_to_inner_class):
9784 Skip FIELD_DECLs with error_mark_node type.
9785 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9788 2017-06-09 Martin Liska <mliska@suse.cz>
9790 * predict.c (struct branch_predictor): New struct.
9791 (test_prediction_value_range): New test.
9792 (predict_c_tests): New function.
9793 * selftest-run-tests.c (selftest::run_tests): Run the function.
9794 * selftest.h: Declare new tests.
9796 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9799 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9800 gen_add3_insn did not fail.
9801 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9802 r0, construct that number in a temporary reg and add that reg to r0.
9803 If asked to put the result in r0 as well, fail.
9805 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9807 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9808 for early expansion of vec_eqv.
9810 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9813 * ubsan.c (instrument_null): Avoid pointless code temporary.
9814 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9816 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9820 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
9823 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
9826 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9828 * predict.c (unlikely_executed_edge_p): Move ahead.
9829 (probably_never_executed_edge_p): Use it.
9831 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9834 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
9835 gimple_call_builtin_p call.
9837 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9839 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
9840 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
9841 rtl_check_failed_type2, rtl_check_failed_code1,
9842 rtl_check_failed_code2, rtl_check_failed_code_mode,
9843 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
9844 rtvec_check_failed_bounds, rtl_check_failed_flag,
9845 _fatal_insn_not_found, _fatal_insn): Likewise.
9846 * tree.h (tree_contains_struct_check_failed,
9847 tree_check_failed, tree_not_check_failed,
9848 tree_class_check_failed, tree_range_check_failed,
9849 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
9850 tree_vec_elt_check_failed, phi_node_elt_check_failed,
9851 tree_operand_check_failed, omp_clause_check_failed,
9852 omp_clause_operand_check_failed, omp_clause_range_check_failed):
9855 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9857 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
9858 flag_branch_probabilities.
9859 * ipa-inline.c (edge_badness): Likewise.
9860 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9861 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
9862 * predict.c (maybe_hot_frequency_p): Likewise.
9863 (probably_never_executed): Likewise.
9864 * sched-ebb.c (schedule_ebbs): Likewise.
9865 * sched-rgn.c (find_single_block_region): Likewise.
9866 * tracer.c (tail_duplicate): Likewise.
9868 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9870 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
9871 longer requires x_flag_profile_use.
9873 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9875 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
9876 instead of flag_reorder_blocks_and_partition.
9877 * dbxout.c (dbxout_function_end): Likewise.
9878 * dwarf2out.c (gen_subprogram_die): Likewise.
9879 * haifa-sched.c (sched_create_recovery_edges): Likewise.
9880 * hw-doloop.c (reorg_loops): Likewise.
9881 * varasm.c (assemble_start_function,
9882 assemble_end_function): Likewise.
9883 (decide_function_section): Do not check for
9884 flag_reorder_blocks_and_partition.
9886 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9888 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9890 (chkp_get_hard_register_fake_addr_expr): Ditto.
9891 (chkp_build_addr_expr): Add check for hard reg case.
9892 (chkp_parse_array_and_component_ref): Ditto.
9893 (chkp_find_bounds_1): Ditto.
9894 (chkp_process_stmt): Don't generate bounds store for
9897 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9899 * predict.c (maybe_hot_bb_p): Do not check profile status.
9900 (maybe_hot_edge_p): Likewise.
9901 (probably_never_executed): Check for zero counts even if profile
9903 (unlikely_executed_edge_p): New function.
9904 (unlikely_executed_stmt_p): New function.
9905 (unlikely_executed_bb_p): New function.
9906 (set_even_probabilities): Use unlikely predicates.
9907 (combine_predictions_for_bb): Likewise.
9908 (predict_paths_for_bb): Likewise.
9909 (predict_paths_leading_to_edge): Likewise.
9910 (determine_unlikely_bbs): New function.
9911 (estimate_bb_frequencies): Use it.
9912 (compute_function_frequency): Use zero counts even if profile is
9914 * profile-count.h: Fix typo.
9916 2017-08-08 Julia Koval <julia.koval@intel.com>
9918 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
9919 _mm512_mask_cvtsepi16_storeu_epi8,
9920 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
9921 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
9922 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
9923 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
9924 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
9925 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
9926 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
9927 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
9928 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
9929 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
9930 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
9931 __builtin_ia32_pmovuswb256mem_mask,
9932 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
9933 __builtin_ia32_pmovwb512mem_mask): New builtins.
9935 2017-08-08 Julia Koval <julia.koval@intel.com>
9937 PR target/73350,80862
9938 * config/i386/subst.md (round): Fix round pattern.
9939 * config/i386/i386.c (ix86_erase_embedded_rounding):
9940 Fix erasing rounding for the fixed pattern.
9942 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9944 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
9946 2017-06-08 Martin Liska <mliska@suse.cz>
9948 PR gcov-profile/80911
9949 * gcov.c (block_info::block_info): New constructor.
9951 2017-06-07 Carl Love <cel@us.ibm.com>
9953 * config/rs6000/rs6000-c: The return type of the following
9954 built-in functions was implemented as int not long long. Fix sign
9955 of return value for the unsigned version of vec_mulo and vec_mule.
9956 vector unsigned long long vec_bperm (vector unsigned long long,
9957 vector unsigned char)
9958 vector signed long long vec_mule (vector signed int,
9960 vector unsigned long long vec_mule (vector unsigned int,
9961 vector unsigned int)
9962 vector signed long long vec_mulo (vector signed int,
9964 vector unsigned long long vec_mulo (vector unsigned int,
9965 vector unsigned int)
9966 * doc/extend.texi: Fix the documentation for the built-in
9969 2017-06-07 Carl Love <cel@us.ibm.com>
9972 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
9975 2017-06-07 Carl Love <cel@us.ibm.com>
9977 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
9978 support, Generate doublehv for signed int/float for BE case only.
9980 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
9982 * doc/invoke.texi (mcx16): Rewrite.
9984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9986 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
9987 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
9988 *mov<mode>_softfloat, and an anonymous splitter): Use
9989 nonimmediate_operand instead of rs6000_nonimmediate_operand.
9991 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9993 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
9995 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9996 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
9997 (rs6000_debug_reg_global): Adjust.
9998 (rs6000_init_hard_regno_mode_ok): Adjust.
9999 (rs6000_dbx_register_number): Adjust.
10000 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10001 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10002 Remove SPE_ACC and SPEFSCR.
10003 (REG_ALLOC_ORDER): Ditto.
10004 (FRAME_POINTER_REGNUM): Change to 111.
10005 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10006 (REG_CLASS_NAMES): Ditto.
10007 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10008 (REGISTER_NAMES): Ditto.
10009 (ADDITIONAL_REG_NAMES): Ditto.
10010 (rs6000_reg_names): Ditto.
10011 * config/rs6000/rs6000.md: Renumber some register number
10014 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10016 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10018 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10019 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10021 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10022 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10023 Delete the SPE high registers.
10024 (REG_ALLOC_ORDER): Ditto.
10025 (enum reg_class): Remove SPE_HIGH_REGS.
10026 (REG_CLASS_NAMES): Ditto.
10027 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10028 (REGISTER_NAMES): Ditto.
10029 (rs6000_reg_names): Ditto.
10030 * doc/tm.texi.in: Remove SPE as example.
10031 * doc/tm.texi: Regenerate.
10033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10035 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10036 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10037 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10038 * config/rs6000/rs6000.md (type): Remove "brinc".
10040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10042 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10043 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10044 * config/rs6000/linuxspe.h: Delete file.
10045 * config/rs6000/rs6000.md: Don't include spe.md.
10046 * config/rs6000/spe.h: Delete file.
10047 * config/rs6000/spe.md: Delete file.
10048 * config/rs6000/t-rs6000: Remove spe.md.
10050 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10052 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10053 (reg_or_none500mem_operand): Delete.
10054 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10055 instead of reg_or_none500mem_operand.
10057 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10059 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10060 handling of SPE flags.
10061 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10065 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10067 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10068 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10069 paired_divv2sf3): Similar.
10070 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10071 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10072 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10074 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10075 Rename the paired_* instruction patterns.
10076 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10078 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10079 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10080 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10081 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10082 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10083 PAIRED_VECTOR_MODE.
10084 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10085 (spe_func_has_64bit_regs_p): Delete.
10086 (spe_expand_predicate_builtin): Delete.
10087 (spe_expand_evsel_builtin): Delete.
10088 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10089 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10090 (invalid_e500_subreg): Delete.
10091 (rs6000_legitimize_address): Always force_reg op2 as well, for
10092 paired single memory accesses.
10093 (rs6000_member_type_forces_blk): Delete.
10094 (rs6000_spe_function_arg): Delete.
10095 (rs6000_expand_unop_builtin): Delete SPE handling.
10096 (rs6000_expand_binop_builtin): Ditto.
10097 (spe_expand_stv_builtin): Delete.
10098 (bdesc_2arg_spe): Delete.
10099 (spe_expand_builtin): Delete.
10100 (spe_expand_predicate_builtin): Delete.
10101 (spe_expand_evsel_builtin): Delete.
10102 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10103 (spe_init_builtins): Delete.
10104 (spe_func_has_64bit_regs_p): Delete.
10105 (savres_routine_name): Delete "info" parameter. Adjust callers.
10106 (rs6000_emit_stack_reset): Ditto.
10107 (rs6000_dwarf_register_span): Delete.
10108 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10109 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10110 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10112 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10114 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10115 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10116 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10117 mulv2sf3, divv2sf3): Delete expanders.
10119 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10121 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10123 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10125 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10126 * config/rs6000/rs6000.c: Ditto.
10128 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10130 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10131 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10132 comparison_operator.
10134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10136 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10137 * config/rs6000/rs6000.opt: Ditto.
10138 * config/rs6000/t-rtems: Ditto.
10140 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10142 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10143 TARGET_E500_SINGLE by 0, simplify.
10144 * config/rs6000/rs6000.c: Ditto.
10145 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10146 (spe_build_register_parallel): Delete.
10147 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10148 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10149 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10150 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10151 (E500_CONVERT): Delete.
10152 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10154 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10156 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10157 * config/rs6000/dfp.md: Ditto.
10158 (negdd2, *negdd2_fpr): Merge.
10159 (absdd2, *absdd2_fpr): Merge.
10160 (negtd2, *negtd2_fpr): Merge.
10161 (abstd2, *abstd2_fpr): Merge.
10162 * config/rs6000/e500.h: Delete file.
10163 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10164 TARGET_FPRS by 1 and simplify.
10165 * config/rs6000/rs6000-c.c: Ditto.
10166 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10167 TARGET_DF_SPE by 0.
10168 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10170 * config/rs6000/rs6000.md: Ditto.
10171 (floatdidf2, *floatdidf2_fpr): Merge.
10172 (move_from_CR_gt_bit): Delete.
10173 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10174 (E500_CR_IOR_COMPARE): Delete.
10175 (All patterns that require !TARGET_FPRS): Delete.
10176 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10178 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10180 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10182 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10184 * graphds.c (add_edge): Intitialize edge's attached data.
10185 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10186 pointer parameter. Call pointed function on each edge during
10187 graph traversing. Skip traversing the edge when the function
10189 (graphds_dfs, graphds_scc): Ditto.
10190 (for_each_edge): New parameter. Pass the new parameter to callback
10192 * graphds.h (skip_edge_callback): New function pointer type.
10193 (graphds_dfs, graphds_scc): New function pointer parameter.
10194 (graphds_edge_callback, for_each_edge): New parameter.
10196 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10198 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10199 out code checking if runtime alias check is possible to below ...
10200 Call the new function.
10201 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10202 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10204 2017-06-07 Marek Polacek <polacek@redhat.com>
10207 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10208 TYPE_OVERFLOW_WRAPS check.
10210 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10212 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10213 if versioning is required.
10214 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10215 peeling with the check for versioning.
10217 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10219 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10220 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10221 Set true to new parameter if new ssa variable is defined.
10222 (vect_gen_vector_loop_niters): Refactor. Set range information
10223 for the new vector loop bound variable.
10224 (vect_do_peeling): Ditto.
10226 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10228 * tree-affine.c (ssa.h): Include header file.
10229 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10230 has wrapping overflow behavior.
10232 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10234 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10236 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10238 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10239 (tree_to_aff_combination): ... here.
10241 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10243 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10244 reg_pressure model function.
10245 (ivopts_global_cost_for_size): Delete.
10246 (determine_set_costs, iv_ca_recount_cost): Call new model function
10247 ivopts_estimate_reg_pressure.
10249 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10251 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10252 expensive than udiv. Remove floating point cases from mod.
10254 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10256 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10257 Increase idiv cost.
10259 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10261 * config/aarch64/aarch64.md
10262 (copysignsf3): Fix mask generation.
10264 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10266 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10268 (class dump_manager): Add register_dumps method.
10269 * dumpfile.c: Include langhooks.h.
10270 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10271 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10272 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10273 (dump_manager::dump_register): Start with 512 entries instead of 32.
10274 (dump_manager::register_dumps): New method.
10275 * toplev.c (general_init): Instead of invoking register_dumps
10276 langhook, invoke register_dumps method on the dump manager.
10277 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10280 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10282 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10283 Say that # requires an associated define_split to exist, and that
10284 the define_split must be suitable for use after register allocation.
10286 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10288 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10289 (compute_outgoing_frequencies): Also initialize zero counts.
10290 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10291 around loops; preserve more of profile when nothing changes.
10293 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10295 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10297 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10298 * config/arm/arm-cpu-cdata.h: Regenerate.
10299 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10300 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10301 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10302 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10303 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10305 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10307 * config/arm/t-rmprofile: Likewise.
10308 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10310 2017-06-06 David S. Miller <davem@davemloft.net>
10313 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10314 blockage if function uses alloca.
10316 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10318 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10319 New "uid" fields to hold pretty-print IDs of group and ref.
10320 Memory references are now identified as <group_id>:<ref_id>
10321 instead of using [random] addresses.
10322 (dump_mem_details): Simplify, no functional change.
10323 (dump_mem_ref): Simplify and make output more concise.
10324 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10325 (find_or_create_group): Initialize group uid.
10326 (record_ref): Initialize ref uid. Improve debug output.
10327 (prune_group_by_reuse, should_issue_prefetch_p,)
10328 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10329 (mark_nontemporal_store, determine_loop_nest_reuse):
10330 Improve debug output.
10332 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10334 * dbgcnt.def (prefetch): New debug counter.
10335 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10336 (schedule_prefetches): Stop issueing prefetches if debug counter
10339 2017-06-06 Tom de Vries <tom@codesourcery.com>
10341 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10342 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10344 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10346 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10347 Use aarch64_reg_or_zero predicate for operand 4.
10348 (aarch64_compare_and_swap<mode> define_insn_and_split):
10349 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10350 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10352 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10354 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10355 (arm_compute_save_core_reg_mask): This.
10356 (thumb1_compute_save_reg_mask): Rename into ...
10357 (thumb1_compute_save_core_reg_mask): This.
10358 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10359 (arm_compute_frame_layout): Likewise.
10361 2017-06-06 Richard Biener <rguenther@suse.de>
10363 PR tree-optimization/80974
10364 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10365 keep or clear leaders SSA info.
10367 2017-06-06 Tom de Vries <tom@codesourcery.com>
10369 * config/nvptx/nvptx.c (split_mode_p): New function.
10370 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10372 2017-06-06 Tom de Vries <tom@codesourcery.com>
10374 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10376 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10379 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10382 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10384 * shrink-wrap.c (handle_simple_exit): Update profile.
10385 (try_shrink_wrapping): Upate profile.
10387 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10389 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10390 (tree_guess_outgoing_edge_probabilities): New.
10391 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10392 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10394 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10396 * ipa-split.c (split_function): Initialize return bb profile.
10398 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10400 * profile.c (compute_branch_probabilities): Also initialize
10401 EXIT_BLOCK profile.
10403 2017-06-06 Richard Biener <rguenther@suse.de>
10405 PR tree-optimization/80928
10406 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10407 (vect_analyze_loop_operations): Properly guard analysis for
10409 (vect_transform_loop): Likewise.
10410 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10411 (vect_model_induction_cost): Do not cost for pure SLP.
10412 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10413 of induction in inner loop vectorization.
10414 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10415 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10416 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10418 (vect_analyze_slp_cost_1): Cost induction.
10419 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10420 (vect_get_slp_vect_defs): Likewise.
10421 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10422 (vect_transform_stmt): Handle SLP reductions.
10423 * tree-vectorizer.h (vectorizable_induction): Adjust.
10425 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10427 * config/rs6000/rs6000.c (make_resolver_func): Update
10428 init_lowered_empty_function call.
10430 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10432 * doc/invoke.texi: Document the -fprofile-abs-path option.
10433 * common.opt (fprofile-abs-path): New option.
10434 * gcov-io.h (gcov_write_filename): Declare.
10435 * gcov-io.c (gcov_write_filename): New function.
10436 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10437 * profile.c (output_location): Likewise.
10439 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10441 * shring-wrap.c: Revert accidental commit.
10443 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10445 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10447 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10449 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10451 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10452 profile in callgraph edge.
10453 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10454 (apply_scale): Likewise.
10455 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10456 Also scale profile when inlining function with zero profile.
10457 (initialize_cfun): Update exit block profile even when it is zero.
10458 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10459 when profile is read.
10461 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10463 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10464 (CLONE_*): New constants to define the processors we can generate
10465 code for with the target_clone attribute.
10466 (rs6000_clone_map): New array to identify which clone processors
10467 the current program is running on.
10468 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10469 target_clone attribute.
10470 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10471 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10472 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10473 (cpu_expand_builtin): Add support for target_clone attribute.
10474 (rs6000_valid_attribute_p): Allow "default" attribute.
10475 (get_decl_name): New debug function to simplify printing the
10476 current function name in debugging statements.
10477 (rs6000_clone_priority): New functions to support the target_clone
10478 attribute, and be able to generate code to switch between ISA 2.05
10479 through ISA 3.0 (power6 through power9).
10480 (rs6000_compare_version_priority): Likewise.
10481 (rs6000_get_function_versions_dispatcher): Likewise.
10482 (make_resolver_func): Likewise.
10483 (add_condition_to_bb): Likewise.
10484 (dispatch_function_versions): Likewise.
10485 (rs6000_generate_version_dispatcher_body): Likewise.
10486 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10487 (fusion_gpr_load_p): Fix a spacing issue.
10488 * doc/extend.texi (Common Function Attributes): Document that the
10489 PowerPC supports the target_clone attribute.
10491 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10493 * config/arm/arm.h: explain F symbol found in description of ARM
10494 register allocation in its legend.
10496 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10498 * config/mips/frame-header-opt.c: Include profile-count.h.
10499 * config/riscv/riscv.c: Include profile-count.h
10501 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10503 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10505 (sm_set_flag_if_changed): Add bbs field.
10506 (execute_sm_if_changed_flag_set): Pass BBS.
10507 (execute_sm): Update.
10509 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10511 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10514 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10516 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10517 (peephole2): New peephole2 to emit the above.
10518 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10520 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10522 * config/aarch64/aarch64.c (define_peephole2 above
10523 *sub_<shift>_<mode>): New peephole.
10525 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10527 * config/i386/i386.c (make_resolver_func): Update.
10528 * Makefile.in: Add profile-count.h and profile-count.o
10529 * auto-profile.c (afdo_indirect_call): Update to new API.
10530 (afdo_set_bb_count): Update.
10531 (afdo_propagate_edge): Update.
10532 (afdo_propagate_circuit): Update.
10533 (afdo_calculate_branch_prob): Update.
10534 (afdo_annotate_cfg): Update.
10535 * basic-block.h: Include profile-count.h
10536 (struct edge_def): Turn count to profile_count.
10537 (struct basic_block_def): Likewie.
10538 (REG_BR_PROB_BASE): Move to profile-count.h
10539 (RDIV): Move to profile-count.h
10540 * bb-reorder.c (max_entry_count): Turn to profile_count.
10541 (find_traces): Update.
10542 (rotate_loop):Update.
10543 (connect_traces):Update.
10544 (sanitize_hot_paths):Update.
10545 * bt-load.c (migrate_btr_defs): Update.
10546 * cfg.c (RDIV): Remove.
10547 (init_flow): Use alloc_block.
10548 (alloc_block): Uninitialize count.
10549 (unchecked_make_edge): Uninitialize count.
10550 (check_bb_profile): Update.
10551 (dump_edge_info): Update.
10552 (dump_bb_info): Update.
10553 (update_bb_profile_for_threading): Update.
10554 (scale_bbs_frequencies_int): Update.
10555 (scale_bbs_frequencies_gcov_type): Update.
10556 (scale_bbs_frequencies_profile_count): New.
10557 * cfg.h (update_bb_profile_for_threading): Update.
10558 (scale_bbs_frequencies_profile_count): Declare.
10559 * cfgbuild.c (compute_outgoing_frequencies): Update.
10560 (find_many_sub_basic_blocks): Update.
10561 * cfgcleanup.c (try_forward_edges): Update.
10562 (try_crossjump_to_edge): Update.
10563 * cfgexpand.c (expand_gimple_tailcall): Update.
10564 (construct_exit_block): Update.
10565 * cfghooks.c (verify_flow_info): Update.
10566 (dump_bb_for_graph): Update.
10567 (split_edge): Update.
10568 (make_forwarder_block): Update.
10569 (duplicate_block): Update.
10570 (account_profile_record): Update.
10571 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10572 (get_estimated_loop_iterations): Update.
10573 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10574 (single_likely_exit): Update.
10575 * cfgloopmanip.c (scale_loop_profile): Update.
10577 (set_zero_probability): Update.
10578 (lv_adjust_loop_entry_edge): Update.
10579 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10580 (purge_dead_edges): Update.
10581 (rtl_account_profile_record): Update.
10582 * cgraph.c (cgraph_node::create): Uninitialize count.
10583 (symbol_table::create_edge): Uninitialize count.
10584 (cgraph_update_edges_for_call_stmt_node): Update.
10585 (cgraph_edge::dump_edge_flags): Update.
10586 (cgraph_node::dump): Update.
10587 (cgraph_edge::maybe_hot_p): Update.
10588 * cgraph.h: Include profile-count.h
10589 (create_clone), create_edge, create_indirect_edge): Update.
10590 (cgraph_node): Turn count to profile_count.
10591 (cgraph_edge0: Likewise.
10592 (make_speculative, clone): Update.
10593 (create_edge): Update.
10594 (init_lowered_empty_function): Update.
10595 * cgraphclones.c (cgraph_edge::clone): Update.
10596 (duplicate_thunk_for_node): Update.
10597 (cgraph_node::create_clone): Update.
10598 * cgraphunit.c (cgraph_node::analyze): Update.
10599 (cgraph_node::expand_thunk): Update.
10600 * final.c (dump_basic_block_info): Update.
10601 * gimple-streamer-in.c (input_bb): Update.
10602 * gimple-streamer-out.c (output_bb): Update.
10603 * graphite.c (print_global_statistics): Update.
10604 (print_graphite_scop_statistics): Update.
10605 * hsa-brig.c: Include basic-block.h.
10606 * hsa-dump.c: Include basic-block.h.
10607 * hsa-gen.c (T sum_slice): Update.
10608 (convert_switch_statements):Update.
10609 * hsa-regalloc.c: Include basic-block.h.
10610 * ipa-chkp.c (chkp_produce_thunks): Update.
10611 * ipa-cp.c (struct caller_statistics): Update.
10612 (init_caller_stats): Update.
10613 (gather_caller_stats): Update.
10614 (ipcp_cloning_candidate_p): Update.
10615 (good_cloning_opportunity_p): Update.
10616 (get_info_about_necessary_edges): Update.
10617 (dump_profile_updates): Update.
10618 (update_profiling_info): Update.
10619 (update_specialized_profile): Update.
10620 (perhaps_add_new_callers): Update.
10621 (decide_about_value): Update.
10622 (ipa_cp_c_finalize): Update.
10623 * ipa-devirt.c (struct odr_type_warn_count): Update.
10624 (struct decl_warn_count): Update.
10625 (struct final_warning_record): Update.
10626 (possible_polymorphic_call_targets): Update.
10627 (ipa_devirt): Update.
10628 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10629 * ipa-icf.c (sem_function::merge): Update.
10630 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10631 * ipa-inline.c (compute_uninlined_call_time): Update.
10632 (compute_inlined_call_time): Update.
10633 (want_inline_small_function_p): Update.
10634 (want_inline_self_recursive_call_p): Update.
10635 (edge_badness): Update.
10636 (lookup_recursive_calls): Update.
10637 (recursive_inlining): Update.
10638 (inline_small_functions): Update.
10639 (dump_overall_stats): Update.
10640 (dump_inline_stats): Update.
10641 * ipa-profile.c (ipa_profile_generate_summary): Update.
10642 (ipa_propagate_frequency): Update.
10643 (ipa_profile): Update.
10644 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10645 * ipa-utils.c (ipa_merge_profiles): Update.
10646 * loop-doloop.c (doloop_modify): Update.
10647 * loop-unroll.c (report_unroll): Update.
10648 (unroll_loop_runtime_iterations): Update.
10649 * lto-cgraph.c (lto_output_edge): Update.
10650 (lto_output_node): Update.
10651 (input_node): Update.
10652 (input_edge): Update.
10653 (merge_profile_summaries): Update.
10654 * lto-streamer-in.c (input_cfg): Update.
10655 * lto-streamer-out.c (output_cfg): Update.
10656 * mcf.c (create_fixup_graph): Update.
10657 (adjust_cfg_counts): Update.
10658 (sum_edge_counts): Update.
10659 * modulo-sched.c (sms_schedule): Update.
10660 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10661 * predict.c (maybe_hot_count_p): Update.
10662 (probably_never_executed): Update.
10663 (dump_prediction): Update.
10664 (combine_predictions_for_bb): Update.
10665 (propagate_freq): Update.
10666 (handle_missing_profiles): Update.
10667 (counts_to_freqs): Update.
10668 (rebuild_frequencies): Update.
10669 (force_edge_cold): Update.
10670 * predict.h: Include profile-count.h
10671 (maybe_hot_count_p, counts_to_freqs): UPdate.
10672 * print-rtl-function.c: Do not include cfg.h
10673 * print-rtl.c: Include basic-block.h
10674 * profile-count.c: New file.
10675 * profile-count.h: New file.
10676 * profile.c (is_edge_inconsistent): Update.
10677 (correct_negative_edge_counts): Update.
10678 (is_inconsistent): Update.
10679 (set_bb_counts): Update.
10680 (read_profile_edge_counts): Update.
10681 (compute_frequency_overlap): Update.
10682 (compute_branch_probabilities): Update; Initialize and deinitialize
10684 (branch_prob): Update.
10685 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10686 (edge_gcov_count): New.
10687 (bb_gcov_count): New.
10688 * shrink-wrap.c (try_shrink_wrapping): Update.
10689 * tracer.c (better_p): Update.
10690 * trans-mem.c (expand_transaction): Update.
10691 (ipa_tm_insert_irr_call): Update.
10692 (ipa_tm_insert_gettmclone_call): Update.
10693 * tree-call-cdce.c: Update.
10694 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10695 (gimple_duplicate_sese_tail): Update.
10696 (gimple_account_profile_record): Update.
10697 (execute_fixup_cfg): Update.
10698 * tree-inline.c (copy_bb): Update.
10699 (copy_edges_for_bb): Update.
10700 (initialize_cfun): Update.
10701 (freqs_to_counts): Update.
10702 (copy_cfg_body): Update.
10703 (expand_call_inline): Update.
10704 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10705 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10706 (try_unroll_loop_completely): Update.
10707 (try_peel_loop): Update.
10708 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10709 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10710 * tree-ssa-loop-split.c (connect_loops): Update.
10711 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10712 * tree-ssa-reassoc.c (branch_fixup): Update.
10713 * tree-ssa-tail-merge.c (replace_block_by): Update.
10714 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10715 (compute_path_counts): Update.
10716 (update_profile): Update.
10717 (recompute_probabilities): Update.
10718 (update_joiner_offpath_counts): Update.
10719 (estimated_freqs_path): Update.
10720 (freqs_to_counts_path): Update.
10721 (clear_counts_path): Update.
10722 (ssa_fix_duplicate_block_edges): Update.
10723 (duplicate_thread_path): Update.
10724 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10725 (struct switch_conv_info): Update.
10726 * tree-tailcall.c (decrease_profile): Update.
10727 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10728 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10729 * value-prof.c (check_counter): Update.
10730 (gimple_divmod_fixed_value): Update.
10731 (gimple_mod_pow2): Update.
10732 (gimple_mod_subtract): Update.
10733 (gimple_ic_transform): Update.
10734 (gimple_stringop_fixed_value): Update.
10735 * value-prof.h (gimple_ic): Update.
10737 2017-06-02 Carl Love <cel@us.ibm.com>
10739 * config/rs6000/rs6000-c: Add support for built-in functions
10740 vector double vec_doublee (vector signed int);
10741 vector double vec_doublee (vector unsigned int);
10742 vector double vec_doublee (vector float);
10743 vector double vec_doubleh (vector signed int);
10744 vector double vec_doubleh (vector unsigned int);
10745 vector double vec_doubleh (vector float);
10746 vector double vec_doublel (vector signed int);
10747 vector double vec_doublel (vector unsigned int);
10748 vector double vec_doublel (vector float);
10749 vector double vec_doubleo (vector signed int);
10750 vector double vec_doubleo (vector unsigned int);
10751 vector double vec_doubleo (vector float);.
10752 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10753 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10755 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10756 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10757 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10759 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10760 vec_doublel, vec_doubleh.
10761 * doc/extend.texi: Update the built-in documentation file for the
10762 new built-in functions.
10764 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10767 * ipa-inline-analysis.c (free_growth_caches): Set
10768 edge_removal_hook_holder to NULL after removing it.
10770 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10772 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10773 comparision with zero.
10775 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10776 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10777 for early expansion of vec_min and vec_max builtins.
10778 (builtin_function_type): Add min/max unsigned variants to those
10779 identified as having unsigned arguments.
10781 2017-06-02 Olivier Hainque <hainque@adacore.com>
10783 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10785 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10787 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10788 Use VALL_F16 iterator rather than VALL.
10790 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10792 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10793 Emit CBNZ inside loop when doing a strong exchange and comparing
10794 against zero. Generate the CC flags after the loop.
10796 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10798 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10799 (dl_section_ref): New.
10800 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10801 On AIX, append an expression to subtract the size of the
10802 section length to dl_section_ref.
10804 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10807 for early expansion of vector absolute builtins.
10809 2017-06-02 Richard Biener <rguenther@suse.de>
10811 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10812 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10814 2017-06-02 Richard Biener <rguenther@suse.de>
10816 PR tree-optimization/80948
10817 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10818 stmt order as well.
10820 2017-06-02 Richard Biener <rguenther@suse.de>
10822 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
10824 * tree-vect-stmts.c (process_use): Do not mark backedge defs
10825 for inductions as relevant.
10827 2017-06-02 Richard Biener <rguenther@suse.de>
10829 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
10830 (vectorizable_induction): ... this. Remove dead code.
10832 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
10834 * builtins. (expand_builtin_alloca): Remove second parameter and
10835 infer its value from the first parameter instead.
10836 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
10838 2017-06-02 Jakub Jelinek <jakub@redhat.com>
10840 PR rtl-optimization/80903
10841 * loop-doloop.c (add_test): Unshare sequence.
10843 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
10845 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
10847 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
10849 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
10851 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
10852 xlogue_layout::get_instance, logue_layout::xlogue_layout,
10853 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
10854 (xlogue_layout::get_stub_rtx): Make static.
10855 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
10856 (xlogue_layout::compute_stub_managed_regs): Rename to...
10857 (xlogue_layout::count_stub_managed_regs): ...this.
10858 (xlogue_layout::is_stub_managed_reg): New function.
10859 (xlogue_layout::m_stub_names): Rename to...
10860 (xlogue_layout::s_stub_names): ...this, make static.
10861 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
10862 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
10863 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
10864 xlogue_layout::s_stub_names): Instantiate statics.
10865 (stub_managed_regs): Remove.
10866 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
10867 (disable_call_ms2sysv_xlogues): Rename to...
10868 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
10869 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
10871 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
10872 change after reload_completed.
10873 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
10875 (ix86_expand_prologue): Likewise.
10876 (ix86_expand_epilogue): Likewise.
10877 (ix86_expand_split_stack_prologue): Likewise.
10878 (ix86_compute_frame_layout): Remove frame parameter ...
10879 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
10880 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
10882 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
10883 (ix86_frame): Move from here ...
10884 * config/i386/i386.h (ix86_frame): ... to here.
10885 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
10886 complete ix86_frame data structure instead. Remove some_ld_name.
10888 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10890 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
10891 symbols that hold a DECL_VALUE_EXPR.
10893 2017-06-01 Martin Jambor <mjambor@suse.cz>
10895 PR tree-optimization/80898
10896 * tree-sra.c (process_subtree_disqualification): Removed.
10897 (disqualify_candidate): Do not acll
10898 process_subtree_disqualification.
10899 (subtree_mark_written_and_enqueue): New function.
10900 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
10901 RHS has been disqualified and re-queue LHS if necessary. Apart
10902 from that, ignore disqualified RHS.
10904 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10906 * config/s390/s390.c (s390_emit_epilogue): Disable early return
10907 address fetch for z10 or later.
10909 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10911 * config/arc/arc.md (tst_movb): Add guard when splitting.
10913 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10915 * config/arc/arc.c (arc_can_eliminate): Test against
10916 arc_frame_pointer_needed.
10918 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10920 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
10921 to prevent store reordering.
10922 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
10923 (type): Add block type.
10924 (stack_tie): Define special instruction to be used in
10927 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10929 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
10930 constraint. It is not valid for the pattern.
10931 (noncommutative_binary_comparison): Likewise.
10933 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10935 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
10938 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10940 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
10941 be used by the reg-alloc.
10943 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10945 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
10946 reg-alloc when having mul64 or mul32x16 instructions.
10947 (mulsidi3): Likewise.
10948 (umulsidi3): Likewise.
10949 (mulsi32x16): New pattern.
10950 (mulsi64): Likewise.
10951 (mulsidi64): Likewise.
10952 (umulsidi64): Likewise.
10953 (MUL32x16_REG): Define.
10954 (mul64_600): Use MUL32x16_REG.
10955 (mac64_600): Likewise.
10956 (umul64_600): Likewise.
10957 (umac64_600): Likewise.
10959 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10961 * config/arc/arc.md (mulsi3_700): Make it commutative.
10963 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
10965 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
10967 (*sign_extendsidi2_insn): Likewise for movstosw.
10969 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10971 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
10972 the type of the input discriminant value. Convert the
10973 discriminant value of signedness vary.
10975 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
10977 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
10978 Add to -Wall section.
10980 2017-06-01 Richard Biener <rguenther@suse.de>
10982 PR middle-end/66313
10983 * fold-const.c (fold_plusminus_mult_expr): If the factored
10984 factor may be zero use a wrapping type for the inner operation.
10985 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
10986 and handle moved defs.
10987 (process_assignment): Properly guard the unary op case. Return a
10988 tri-state indicating that moving the stmt before the call may allow
10989 to continue. Pass through to_move.
10990 (find_tail_calls): Handle moving unrelated defs before
10993 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10996 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
10997 splitter result in the canonical way.
10999 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11001 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11002 also for 32bit target. Update insn attributes.
11003 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11005 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11007 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11008 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11009 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11010 (_mm_maskz_min_ss): New intrinsics.
11012 2017-05-31 Martin Liska <mliska@suse.cz>
11014 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11015 Change comment style to one we normally use.
11016 (vectorizable_reduction): Likewise.
11017 (vectorizable_induction): Likewise.
11018 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11019 (vectorizable_call): Likewise.
11020 (vectorizable_simd_clone_call): Likewise.
11021 (vectorizable_conversion): Likewise.
11022 (vectorizable_assignment): Likewise.
11023 (vectorizable_shift): Likewise.
11024 (vectorizable_operation): Likewise.
11025 (vectorizable_store): Likewise.
11026 (vectorizable_load): Likewise.
11027 * tree-vectorizer.h: Likewise.
11029 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11031 * passes.c (emergency_dump_function): New.
11032 * tree-pass.h (emergency_dump_function): Declare.
11033 * plugin.c (plugins_internal_error_function): Remove.
11034 * plugin.h (plugins_internal_error_function): Remove declaration.
11035 * toplev.c (internal_error_function): New static function. Use it...
11036 (general_init): ...here.
11038 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11040 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11041 (arc_rtx_costs): Add costs for new patterns.
11042 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11043 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11045 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11047 * tree-ssa-strlen.c (get_next_strinfo): New function.
11048 (get_stridx_plus_constant): Use it.
11049 (zero_length_string): Likewise.
11050 (adjust_related_strinfos): Likewise.
11051 (adjust_last_stmt): Likewise.
11053 2017-05-31 Richard Biener <rguenther@suse.de>
11056 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11057 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11059 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11061 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11062 loop_vinfo argument and use of dependence distance vectors.
11063 Check instead whether the two references differ only in their
11064 initial value and assume that they have the same alignment if the
11065 difference is a multiple of the vector alignment.
11066 (vect_analyze_data_refs_alignment): Update call accordingly.
11068 2017-05-31 Martin Liska <mliska@suse.cz>
11071 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11073 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11075 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11076 (create_intersect_range_checks): Move from ...
11077 * tree-data-ref.c (create_intersect_range_checks_index)
11078 (create_intersect_range_checks): ... to here.
11079 (create_runtime_alias_checks): New function factored from ...
11080 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11081 here. Call above function.
11082 * tree-data-ref.h (create_runtime_alias_checks): New function.
11084 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11086 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11087 segment length for dr_b and compute it in wide_int.
11089 2017-05-31 Richard Biener <rguenther@suse.de>
11091 PR tree-optimization/80906
11092 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11093 and pass through iv_map.
11094 (copy_bb_and_scalar_dependences): Adjust.
11095 (translate_pending_phi_nodes): Likewise.
11096 (copy_loop_close_phi_args): Handle code-generating IVs instead
11099 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11101 * diagnostic-color.c (color_dict): Add "type-diff".
11102 (parse_gcc_colors): Update comment.
11103 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11104 -fdiagnostics-show-template-tree and -fno-elide-type.
11105 (GCC_COLORS): Add type-diff to example.
11107 (-fdiagnostics-show-template-tree): New.
11108 (-fno-elide-type): New.
11109 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11110 the pp_format_decoder callback. Call any m_format_postprocessor's
11112 (pretty_printer::pretty_printer): Initialize
11113 m_format_postprocessor.
11114 (pretty_printer::~pretty_printer): Delete any
11115 m_format_postprocessor.
11116 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11117 (class format_postprocessor): New class.
11118 (struct pretty_printer::format_decoder): Document the new parameters.
11119 (struct pretty_printer::m_format_postprocessor): New field.
11120 * tree-diagnostic.c (default_tree_printer): Update for new
11121 bool and const char ** params.
11122 * tree-diagnostic.h (default_tree_printer): Likewise.
11124 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11126 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11127 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11129 rs6000_gen_cell_microcode code.
11130 (rs6000_final_prescan_insn): Delete.
11131 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11132 "warn-cell-microcode" entries.
11133 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11134 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11135 throughout. Change cc_reg_not_micro_cr0_operand to
11136 cc_reg_not_cr0_operand throughout.
11137 (*extendhi<mode>2_noload): Delete.
11138 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11139 (mwarn-cell-microcode): Delete.
11140 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11141 -mgen-cell-microcode and -mwarn-cell-microcode.
11143 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11146 * config/i386/constraints.md (Yd): New constraint.
11148 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11149 and (?Yd, r) alternatives. Update insn attributes.
11150 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11151 and (?*Yd, r) alternatives. Update insn attributes.
11152 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11154 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11156 * gimplify.c (gimplify_modify_expr): Don't create a
11157 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11160 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11162 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11164 2017-05-30 Richard Biener <rguenther@suse.de>
11166 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11167 and reduc_def fields.
11168 (STMT_VINFO_REDUC_TYPE): New define.
11169 (STMT_VINFO_REDUC_DEF): Likewise.
11170 (vect_force_simple_reduction): Adjust prototype.
11171 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11172 (vect_is_simple_reduction): Remove check_reduction argument.
11173 (vect_force_simple_reduction): Adjust and set
11174 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11175 (vectorizable_reduction): Do not re-do reduction analysis
11176 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11177 * tree-parloops.c (gather_scalar_reductions): Adjust.
11179 2017-05-30 Richard Biener <rguenther@suse.de>
11181 PR middle-end/80901
11182 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11185 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11187 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11188 Introduce unknown_misalignment parameter and remove vf.
11189 (vect_peeling_hash_get_lowest_cost):
11190 Pass unknown_misalignment parameter.
11191 (vect_enhance_data_refs_alignment):
11192 Fix unsupportable data ref treatment.
11194 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11196 * tree-vect-data-refs.c (vect_get_data_access_cost):
11197 Workaround for SLP handling.
11198 (vect_enhance_data_refs_alignment):
11199 Compute costs for doing no peeling at all, compare to the best
11200 peeling costs so far and avoid peeling if cheaper.
11202 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11204 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11205 Return peeling info and set costs to zero for unlimited cost
11207 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11208 with unknown misalignment. Compute and costs for unknown
11209 misalignment, compare them to the costs for known misalignment
11210 and choose the cheapest for peeling.
11212 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11214 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11215 (vect_get_peeling_costs_all_drs): Create function.
11216 (vect_peeling_hash_get_lowest_cost):
11217 Use vect_get_peeling_costs_all_drs.
11218 (vect_peeling_supportable): Create function.
11219 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11221 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11223 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11224 DR_HAS_NEGATIVE_STEP.
11225 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11226 (vect_enhance_data_refs_alignment): Use.
11227 (vect_duplicate_ssa_name_ptr_info): Use.
11228 * tree-vectorizer.h (dr_misalignment): Use.
11229 (known_alignment_for_access_p): Use.
11231 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11234 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11236 (has_section_name): New function.
11238 2017-05-30 Martin Liska <mliska@suse.cz>
11241 * auto-profile.c (get_function_decl_from_block): Fix
11244 2017-05-30 Richard Biener <rguenther@suse.de>
11246 PR middle-end/80876
11247 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11249 2017-05-30 Martin Liska <mliska@suse.cz>
11251 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11252 * dumpfile.h (struct dump_file_info): Remove ctors.
11254 2017-05-30 Martin Liska <mliska@suse.cz>
11256 * predict.def: Fix GNU coding style.
11258 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11260 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11261 Mark 'to' argument with ATTRIBUTE_UNUSED.
11263 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11265 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11266 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11267 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11270 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11272 * doc/install.texi (Options specification): Restore entry of
11273 --enable-sjlj-exceptions.
11275 2017-05-27 Michael Eager <eager@eagercon.com>
11278 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11280 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11282 * config/microblaze/microblaze.h
11283 (FIXED_REGISTERS): Update in macro.
11284 (CALL_USED_REGISTERS): Update in macro.
11286 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11288 * doc/install.texi: Add links to macOS binary distributions.
11290 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11294 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11296 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11298 2017-05-26 Martin Liska <mliska@suse.cz>
11300 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11302 2017-05-26 Martin Liska <mliska@suse.cz>
11304 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11305 always leading ';; '.
11306 (dump_bb_info): Likewise.
11307 (brief_dump_cfg): Likewise.
11308 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11309 * dumpfile.c: Remove usage of TDF_VERBOSE.
11310 * dumpfile.h (enum dump_kind): Likewise.
11311 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11312 * print-tree.c (debug_verbose): Remove.
11313 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11314 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11315 * tree-diagnostic.c (default_tree_printer): Replace
11316 TDF_DIAGNOSTIC with TDF_SLIM.
11318 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11320 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11321 in parameter loop, rather than loop_vinfo.
11322 (create_intersect_range_checks): Ditto.
11323 (vect_create_cond_for_alias_checks): Update call to above functions.
11325 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11327 PR tree-optimization/80815
11328 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11329 for merging runtime alias checks. Handle negative DR_STEPs.
11331 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11333 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11335 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11336 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11337 out code pruning runtime alias checks.
11338 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11339 factored out from above.
11340 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11342 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11344 (prune_runtime_alias_test_list): New decalaration.
11346 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11348 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11349 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11350 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11351 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11352 (operator==, comp_dr_with_seg_len_pair): Ditto.
11353 (vect_prune_runtime_alias_test_list): Ditto.
11355 2017-05-26 Martin Liska <mliska@suse.cz>
11358 * params.def: Bound partial-inlining-entry-probability param.
11360 2017-05-26 Marek Polacek <polacek@redhat.com>
11363 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11366 2017-05-26 Richard Biener <rguenther@suse.de>
11368 PR tree-optimization/80842
11369 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11372 2017-05-26 Richard Biener <rguenther@suse.de>
11374 PR tree-optimization/80844
11375 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11377 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11379 * doc/md.texi (Machine Constraints): Update x86 family
11380 machine constraints section to match 'config/i386/constraints.md'.
11382 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11384 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11386 2017-05-25 Nathan Sidwell <nathan@acm.org>
11388 * doc/invoke.texi (--enable-languages): Update documentation.
11390 2017-05-25 Martin Liska <mliska@suse.cz>
11392 * dumpfile.c: Add TDF_FOLDING.
11393 * dumpfile.h (enum dump_kind): Likewise.
11394 * genmatch.c (dt_simplify::gen_1): Use it.
11396 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11398 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11400 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11402 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11403 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11405 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11407 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11408 * match.pd (X == C): Rewrite it here.
11409 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11410 with_certain_nonzero_bits2): New predicates.
11411 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11413 2017-05-24 Nathan Sidwell <nathan@acm.org>
11415 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11418 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11421 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11423 * config/powerpcspe: New port. Files are copied from the rs6000
11424 port, with "rs6000" in filenames replaced by "powerpcspe".
11426 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11428 PR rtl-optimization/80754
11429 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11431 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11433 * config/sparc/sparc.md (length): Return the correct value for -mflat
11434 sibcalls to match output_sibcall.
11436 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11440 * config/rs6000/rs6000.c (struct machine_function): Add new field
11442 (rs6000_get_separate_components): Init that field, use it.
11443 (rs6000_components_for_bb): Use the field.
11445 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11447 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11449 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11451 PR middle-end/80823
11452 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11454 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11457 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11458 address against address_operand predicate.
11459 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11461 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11463 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11464 parameters passed indirectly.
11466 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11468 * config/i386/i386.md (*movdi_internal): Remove SSE4
11469 alternative 18 (?r, *v). Update insn attributes.
11470 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11471 Update insn attributes.
11472 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11473 Update insn attributes.
11474 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11475 alternative 1 (r, v). Remove isa attribute.
11476 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11477 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11478 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11480 2017-05-23 Tom de Vries <tom@codesourcery.com>
11482 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11485 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11487 * cgraphunit.c (symbol_table::process_new_functions): Update.
11488 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11489 (inline_generate_summary): Rename to ...
11490 (ipa_fn_summary_generate): ... this one.
11491 (inline_read_summary): Rename to ...
11492 (ipa_fn_summary_read): ... this one.
11493 (inline_write_summary): Rename to ...
11494 (ipa_fn_summary_write): ... this one.
11495 (inline_free_summary): Rename to ...
11496 (ipa_free_fn_summary): ... this one.
11497 (pass_data_local_fn_summary, pass_local_fn_summary,
11498 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11499 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11500 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11501 make_pass_ipa_fn_summary): New.
11502 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11503 inline_write_summary, inline_free_summary): Remove.
11504 (ipa_free_fn_summary) : New.
11505 * ipa-inline.c (ipa_inline): Update.
11506 (pass_ipa_inline): Do not generate summaries.
11507 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11509 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11510 and add pass_ipa_fn_summary.
11511 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11513 (make_pass_inline_parameters): Remove.
11515 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11517 * omp-low.c (struct omp_context): Remove "default_kind" member.
11520 * omp-offload.c (execute_oacc_device_lower): Remove the
11521 parallelism dimensions function attributes for unparallelized
11522 OpenACC kernels constructs.
11524 2017-05-23 Martin Liska <mliska@suse.cz>
11526 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11528 (cgraph_edge::make_speculative): Likewise.
11529 (cgraph_edge::resolve_speculation): Likewise.
11530 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11531 (cgraph_node::dump): Likewise.
11532 * cgraph.h: Likewise.
11533 * cgraphunit.c (analyze_functions): Likewise.
11534 (symbol_table::compile): Likewise.
11535 * ipa-cp.c (print_all_lattices): Likewise.
11536 (determine_versionability): Likewise.
11537 (initialize_node_lattices): Likewise.
11538 (ipcp_verify_propagated_values): Likewise.
11539 (estimate_local_effects): Likewise.
11540 (update_profiling_info): Likewise.
11541 (create_specialized_node): Likewise.
11542 (perhaps_add_new_callers): Likewise.
11543 (decide_about_value): Likewise.
11544 (decide_whether_version_node): Likewise.
11545 (identify_dead_nodes): Likewise.
11546 (ipcp_store_bits_results): Likewise.
11547 * ipa-devirt.c (dump_targets): Likewise.
11548 (ipa_devirt): Likewise.
11549 * ipa-icf.c (sem_item::dump): Likewise.
11550 (sem_function::equals): Likewise.
11551 (sem_variable::equals): Likewise.
11552 (sem_item_optimizer::read_section): Likewise.
11553 (sem_item_optimizer::execute): Likewise.
11554 (congruence_class::dump): Likewise.
11555 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11556 (dump_inline_summary): Likewise.
11557 (estimate_node_size_and_time): Likewise.
11558 (inline_analyze_function): Likewise.
11559 * ipa-inline-transform.c (inline_call): Likewise.
11560 * ipa-inline.c (report_inline_failed_reason): Likewise.
11561 (want_early_inline_function_p): Likewise.
11562 (edge_badness): Likewise.
11563 (update_edge_key): Likewise.
11564 (inline_small_functions): Likewise.
11565 * ipa-profile.c (ipa_profile): Likewise.
11566 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11567 (ipa_make_edge_direct_to_target): Likewise.
11568 (remove_described_reference): Likewise.
11569 (ipa_impossible_devirt_target): Likewise.
11570 (propagate_controlled_uses): Likewise.
11571 (ipa_print_node_params): Likewise.
11572 (ipcp_transform_function): Likewise.
11573 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11574 (propagate_pure_const): Likewise.
11575 * ipa-reference.c (generate_summary): Likewise.
11576 (read_write_all_from_decl): Likewise.
11577 (propagate): Likewise.
11578 (ipa_reference_read_optimization_summary): Likewise.
11579 * ipa-utils.c (ipa_merge_profiles): Likewise.
11580 * ipa.c (walk_polymorphic_call_targets): Likewise.
11581 (symbol_table::remove_unreachable_nodes): Likewise.
11582 (ipa_single_use): Likewise.
11583 * passes.c (execute_todo): Likewise.
11584 * predict.c (drop_profile): Likewise.
11585 * symtab.c (symtab_node::get_dump_name): New function.
11586 (symtab_node::dump_name): Likewise.
11587 (symtab_node::dump_asm_name): Likewise.
11588 (symtab_node::dump_references): Likewise.
11589 (symtab_node::dump_referring): Likewise.
11590 (symtab_node::dump_base): Likewise.
11591 (symtab_node::debug_symtab): Likewise.
11592 * tree-sra.c (convert_callers_for_node): Likewise.
11593 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11594 * value-prof.c (init_node_map): Likewise.
11596 2017-05-23 Martin Liska <mliska@suse.cz>
11598 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11599 and symtab_node::debug_symtab to symbol_table::debug.
11600 * cgraphunit.c (analyze_functions): Use the renamed function.
11601 (symbol_table::compile): Likewise.
11602 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11603 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11604 * passes.c (execute_todo): Likewise.
11605 * symtab.c (symbol_table::dump): New function.
11606 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11608 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11610 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11611 that nonconst implies exec.
11613 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11615 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11616 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11617 (inline_edge_summary_vec): Turn into ...
11618 (ipa_call_summaries): ... this one.
11619 (redirect_to_unreachable, edge_set_predicate,
11620 evaluate_properties_for_edge, inline_summary_alloc,
11621 reset_ipa_call_summary, reset_inline_summary,
11622 inline_summary_t::duplicate): Update.
11623 (inline_edge_duplication_hook): Turn to ...
11624 (ipa_call_summary_t::duplicate): ... this one.
11625 (inline_edge_removal_hook): Turn to ...
11626 (ipa_call_summary_t::remove): ... this one.
11627 (dump_inline_edge_summary): Turn to ...
11628 (dump_ipa_call_summary): ... this one.
11629 (estimate_function_body_sizes): Update.
11630 (inline_update_callee_summaries): Update.
11631 (remap_edge_change_prob): Update.
11632 (remap_edge_summaries): Update.
11633 (inline_merge_summary): Update.
11634 (do_estimate_edge_time): Update.
11635 (inline_generate_summary): Update.
11636 (inline_read_section): Update.
11637 (inline_read_summary): Update.
11638 (inline_free_summary): Update.
11639 * ipa-inline.c (can_inline_edge_p): Update.
11640 (compute_inlined_call_time): Update.
11641 (want_inline_small_function_p): Update.
11642 (edge_badness): Update.
11643 (early_inliner): Update.
11644 * ipa-inline.h (inline_edge_summary): Turn to ...
11645 (ipa_call_summary): ... this one.
11646 (ipa_call_summary_t): New class.
11647 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11648 (ipa_call_summaries): New.
11649 (inline_edge_summary): Remove.
11650 (estimate_edge_growth): Update.
11651 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11652 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11653 * ipa-split.c (execute_split_functions): Update.
11654 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11656 2017-05-23 Tom de Vries <tom@codesourcery.com>
11658 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11659 attributes): Document rdrand effective target.
11661 2017-05-23 Tom de Vries <tom@codesourcery.com>
11663 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11664 attributes): Sort alphabetically.
11666 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11668 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11670 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11673 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11674 V2DF/V2DI splat into two separate patterns, one that handles
11675 registers, and the other that only handles memory. Drop support
11676 for splatting from a GPR on ISA 2.07 and then splitting the
11677 splat into direct move and splat.
11678 (vsx_splat_<mode>_reg): Likewise.
11679 (vsx_splat_<mode>_mem): Likewise.
11681 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11683 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11685 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11687 PR middle-end/80809
11688 * omp-low.c (finish_taskreg_remap): New function.
11689 (finish_taskreg_scan): If unit size of ctx->record_type
11690 is non-constant, unshare the size expression and replace
11691 decls in it with possible outer var refs.
11693 PR middle-end/80809
11694 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11695 GOVD_SHARED rather than GOVD_PRIVATE with it.
11696 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11697 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11699 PR middle-end/80853
11700 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11701 as last argument to build_outer_var_ref for pointer bases of array
11702 section reductions.
11704 2017-05-19 Martin Sebor <msebor@redhat.com>
11706 * print-tree.c (print_node): Print DECL_READ_P flag.
11708 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11710 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11711 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11712 * cgraph.c: Likewise.
11713 * cgraphunit.c: Likewise.
11714 * gengtype.c: Likewise.
11715 * ipa-cp.c: Likewise.
11716 * ipa-devirt.c: Likewise.
11717 * ipa-icf.c: Likewise.
11718 * ipa-predicate.c: Likewise.
11719 * ipa-profile.c: Likewise.
11720 * ipa-prop.c: Likewise.
11721 * ipa-split.c: Likewise.
11723 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11724 edge_predicate_pool, dump_inline_hints,
11725 inline_summary::account_size_time, redirect_to_unreachable,
11726 edge_set_predicate, set_hint_predicate,
11727 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11728 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11729 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11730 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11731 ipa_call_summary_t::remove, initialize_growth_caches,
11732 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11733 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11734 mark_modified, unmodified_parm_1, unmodified_parm,
11735 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11736 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11737 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11738 will_be_nonconstant_predicate, record_modified_bb_info,
11739 get_minimal_bb, record_modified, param_change_prob,
11740 phi_result_unknown_predicate, predicate_for_phi_result,
11741 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11742 estimate_function_body_sizes, compute_inline_parameters,
11743 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11744 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11745 inline_update_callee_summaries, remap_edge_change_prob,
11746 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11747 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11748 inline_analyze_function, inline_summary_t::insert,
11749 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11750 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11751 inline_free_summary): Move to ipa-fnsummary.h
11752 (predicate_t): Remove.
11753 * ipa-fnsummary.c: New file.
11754 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11755 (enum inline_hints_vals, inline_hints, agg_position_info,
11756 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11757 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11758 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11759 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11760 inline_read_summary, inline_write_summary, inline_free_summary,
11761 inline_analyze_function, initialize_inline_failed,
11762 inline_merge_summary, inline_update_overall_summary,
11763 compute_inline_parameters): Move to ipa-fnsummary.h
11764 * ipa-fnsummary.h: New file.
11765 * ipa-inline-transform.h: Include ipa-inline.h.
11766 * ipa-inline.c: LIkewise.
11768 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11770 * ipa-inline.c (edge_badness): Use inlined_time instead of
11771 inline_summaries->get.
11773 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11775 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11777 2017-05-22 Nathan Sidwell <nathan@acm.org>
11779 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11780 (fdump-lang): Document 'raw' option.
11781 * dumpfile.h (TDI_tu): Delete.
11782 * dumpfile.c (dump_files): Remove translation-unit.
11783 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11785 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11787 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11788 command option from $(AWK) call.
11789 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11790 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11791 [FORMAT]: Remove handling of variable.
11792 * config/avr/t-multilib: Regenerate.
11794 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11796 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11798 (dump_inline_summary): Do not print self_time.
11799 (estimate_function_body_sizes): Do not set self_time.
11800 (compute_inline_parameters): Likewise.
11801 (inline_read_section, inline_write_summary): Do not stream self_time.
11802 * ipa-inline.h (inline_summary): Drop self_time.
11804 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11806 * ipa-inline-analysis.c (account_size_time): Rename to ...
11807 (inline_summary::account_size_time): ... this one.
11808 (reset_ipa_call_summary): Turn to ...
11809 (ipa_call_summary::reset): ... this one.
11810 (reset_inline_summary): Turn to ...
11811 (inline_summary::reset): ... this one.
11812 (inline_summary_t::remove): Update.
11813 (inline_summary_t::duplicate): Update.
11814 (ipa_call_summary_t::remove): Update.
11815 (dump_inline_summary): Update.
11816 (estimate_function_body_sizes): Update.
11817 (compute_inline_parameters): Update.
11818 (estimate_node_size_and_time): Update.
11819 (inline_merge_summary): Update.
11820 (inline_update_overall_summary): Update.
11821 (inline_read_section): Update.
11822 (inline_write_summary): Update.
11823 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
11824 add account_size_time and reset member functions.
11825 (ipa_call_summary): Add reset function.
11826 * ipa-predicate.h (predicate::operator &): Constify.
11828 2017-05-22 Richard Biener <rguenther@suse.de>
11830 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
11832 2017-05-19 Jason Merrill <jason@redhat.com>
11834 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
11836 2017-05-19 Marek Polacek <polacek@redhat.com>
11839 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
11840 TYPE_OVERFLOW_WRAPS checks.
11842 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
11844 * tree-core.h (enum omp_clause_default_kind): Add
11845 "OMP_CLAUSE_DEFAULT_PRESENT".
11846 * tree-pretty-print.c (dump_omp_clause): Handle it.
11847 * gimplify.c (enum gimplify_omp_var_data): Add
11848 "GOVD_MAP_FORCE_PRESENT".
11849 (gimplify_adjust_omp_clauses_1): Map it to
11850 "GOMP_MAP_FORCE_PRESENT".
11851 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
11853 * gimplify.c (oacc_default_clause): Clarify.
11855 2017-05-19 Nathan Sidwell <nathan@acm.org>
11857 LANG_HOOK_REGISTER_DUMPS
11858 * toplev.c (general_init): Call register dump lang hook.
11859 * doc/invoke.texi: Document -fdump-lang option family.
11860 * dumpfile.c (dump_files): Remove class dump here.
11861 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
11862 * dumpfile.h (tree_dump_index): Remove TDI_class.
11863 * langhooks-def.h (lhd_register_dumps): Declare.
11864 (LANG_HOOKS_REGISTER_DUMPS): Define.
11865 (LANG_HOOKS_INITIALIZER): Add it.
11866 * langhooks.c (lhd_register_dumps): Define.
11867 * langhooks.h (struct lang_hooks): Add register_dumps.
11869 2017-05-19 Nathan Sidwell <nathan@acm.org>
11871 * context.h (context::set_passes): New.
11872 * context.c (context::context): Do not create pass manager.
11873 * toplev.c (general_init): Create pass manager here.
11875 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
11877 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
11878 use this splitter if two add or or instructions would also work for
11879 the constant we want to generate.
11881 2017-05-19 Richard Biener <rguenther@suse.de>
11884 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
11885 predicate evaluation.
11887 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11889 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
11891 * ipa-inline.c (want_inline_small_function_p): Do not cast to
11894 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11896 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11897 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11898 (inline_edge_summary_vec): Turn into ...
11899 (ipa_call_summaries): ... this one.
11900 (redirect_to_unreachable, edge_set_predicate,
11901 evaluate_properties_for_edge, inline_summary_alloc,
11902 reset_ipa_call_summary, reset_inline_summary,
11903 inline_summary_t::duplicate): Update.
11904 (inline_edge_duplication_hook): Turn to ...
11905 (ipa_call_summary_t::duplicate): ... this one.
11906 (inline_edge_removal_hook): Turn to ...
11907 (ipa_call_summary_t::remove): ... this one.
11908 (dump_inline_edge_summary): Turn to ...
11909 (dump_ipa_call_summary): ... this one.
11910 (estimate_function_body_sizes): Update.
11911 (inline_update_callee_summaries): Update.
11912 (remap_edge_change_prob): Update.
11913 (remap_edge_summaries): Update.
11914 (inline_merge_summary): Update.
11915 (do_estimate_edge_time): Update.
11916 (inline_generate_summary): Update.
11917 (inline_read_section): Update.
11918 (inline_read_summary): Update.
11919 (inline_free_summary): Update.
11920 * ipa-inline.c (can_inline_edge_p): Update.
11921 (compute_inlined_call_time): Update.
11922 (want_inline_small_function_p): Update.
11923 (edge_badness): Update.
11924 (early_inliner): Update.
11925 * ipa-inline.h (inline_edge_summary): Turn to ...
11926 (ipa_call_summary): ... this one.
11927 (ipa_call_summary_t): New class.
11928 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11929 (ipa_call_summaries): New.
11930 (inline_edge_summary): Remove.
11931 (estimate_edge_growth): Update.
11932 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11933 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11934 * ipa-split.c (execute_split_functions): Update.
11935 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11937 2017-05-19 Richard Biener <rguenther@suse.de>
11939 PR middle-end/80764
11940 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
11942 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
11944 * config/rs6000/rs6000.c (struct machine_function): Add field
11945 fpr_is_wrapped_separately.
11946 (rs6000_get_separate_components): Use 64 components. Handle the
11947 new FPR components.
11948 (rs6000_components_for_bb): Handle the FPR components.
11949 (rs6000_emit_prologue_components): Handle the FPR components.
11950 (rs6000_emit_epilogue_components): Handle the FPR components.
11951 (rs6000_set_handled_components): Handle the FPR components.
11952 (rs6000_emit_prologue): Don't output prologue code for those FPRs
11953 that are already separately shrink-wrapped.
11954 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
11955 that are already separately shrink-wrapped.
11957 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11960 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
11963 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
11964 (define_peephole2 for Altivec d-form load): Add peepholes to catch
11965 cases where the register allocator uses a move and an offsettable
11966 memory operation to/from a FPR register on ISA 2.06/2.07.
11967 (define_peephole2 for Altivec d-form store): Likewise.
11969 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
11972 * config/i386/mmx.md (*mov<mode>_internal): Enable
11973 alternatives 11, 12, 13 and 14 also for 32bit targets.
11974 Remove alternatives 15, 16, 17 and 18.
11975 * config/i386/sse.md (vec_concatv2di): Change
11976 alternative (!x, *y) to (x, ?!*Yn).
11978 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
11980 * dumpfile.h (enum dump_kind): Remove stray comma.
11982 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11984 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
11985 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
11986 predicate::num_conditions
11987 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
11988 (CHANGED): turn into predicate::changed.
11989 (agg_position_info): Move to ipa-predicate.h
11990 (add_condition, predicate::add_clause, predicate::operator &=,
11991 predicate::or_with, predicate::evaluate, predicate::probability,
11992 dump_condition, dump_clause, predicate::dump,
11993 predicate::remap_after_duplication, predicate::remap_after_inlining,
11994 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
11995 (evaluate_conditions_for_known_args): Update.
11996 (set_cond_stmt_execution_predicate): Update.
11997 * ipa-inline.h: Include ipa-predicate.h
11998 (condition, inline_param_summary, conditions, agg_position_info,
11999 predicate): Move to ipa-predicate.h
12000 * ipa-predicate.c: New file.
12001 * ipa-predicate.h: New file.
12003 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12005 * final.c (leaf_function_p): Check we are not in a sequence.
12007 2017-05-18 Martin Liska <mliska@suse.cz>
12009 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12010 * dumpfile.c (dump_register): Use new enum dump_kind.
12011 (get_dump_file_name): Likewise.
12012 (dump_enable_all): Likewise.
12013 (dump_switch_p_1): Likewise.
12014 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12015 * dumpfile.h (enum dump_kind): New enum type.
12016 (struct dump_file_info): Create constructor and
12017 format fields and comments.
12018 * passes.c (pass_manager::register_one_dump_file):
12020 * statistics.c (statistics_early_init): Likewise.
12021 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12022 TDF_TREE with TDF_SLIM.
12023 (gather_memory_references_ref): Likewise.
12025 2017-05-18 Martin Liska <mliska@suse.cz>
12027 * vec.h (struct vnull): Use it.
12029 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12031 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12032 (true_predicate, false_predicate, true_predicate_p,
12033 false_predicate_p): Remove.
12034 (single_cond_predicate, not_inlined_predicate): Turn to member function
12036 (add_condition): Update.
12037 (add_clause): Turn to...
12038 (predicate::add_clause): ... this one; update; allow passing NULL
12040 (and_predicates): Turn to ...
12041 (predicate::operator &=): ... this one.
12042 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12043 (or_predicates): Turn to ...
12044 (predicate::or_with): ... this one.
12045 (evaluate_predicate): Turn to ...
12046 (predicate::evaluate): ... this one.
12047 (predicate_probability): Turn to ...
12048 (predicate::probability): ... this one.
12049 (dump_condition): Update.
12050 (dump_predicate): Turn to ...
12051 (predicate::dump): ... this one.
12052 (account_size_time): Update.
12053 (edge_set_predicate): Update.
12054 (set_hint_predicate): UPdate.
12055 (evaluate_conditions_for_known_args): Update.
12056 (evaluate_properties_for_edge): Update.
12057 (remap_predicate_after_duplication): Turn to...
12058 (predicate::remap_after_duplication): ... this one.
12059 (remap_hint_predicate_after_duplication): Update.
12060 (inline_summary_t::duplicate): UPdate.
12061 (dump_inline_edge_summary): Update.
12062 (dump_inline_summary): Update.
12063 (set_cond_stmt_execution_predicate): Update.
12064 (set_switch_stmt_execution_predicate): Update.
12065 (compute_bb_predicates): Update.
12066 (will_be_nonconstant_expr_predicate): Update.
12067 (will_be_nonconstant_predicate): Update.
12068 (phi_result_unknown_predicate): Update.
12069 (predicate_for_phi_result): Update.
12070 (array_index_predicate): Update.
12071 (estimate_function_body_sizes): Update.
12072 (estimate_node_size_and_time): Update.
12073 (estimate_ipcp_clone_size_and_time): Update.
12074 (remap_predicate): Rename to ...
12075 (predicate::remap_after_inlining): ... this one.
12076 (remap_hint_predicate): Update.
12077 (inline_merge_summary): Update.
12078 (inline_update_overall_summary): Update.
12079 (estimate_size_after_inlining): Update.
12080 (read_predicate): Rename to ...
12081 (predicate::stream_in): ... this one.
12082 (read_inline_edge_summary): Update.
12083 (write_predicate): Rename to ...
12084 (predicate::stream_out): ... this one.
12085 (write_inline_edge_summary): Update.
12086 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12087 (clause_t): Turn to uint32_t
12088 (predicate): Turn to class; implement constructor and operators
12090 (size_time_entry): Update.
12091 (inline_summary): Update.
12092 (inline_edge_summary): Update.
12094 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12096 * fold-const.c (fold_binary_loc): Move transformation...
12097 * match.pd (C - X CMP X): ... here.
12099 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12101 * config/sparc/sparc.c (sparc_option_override): Set function
12102 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12103 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12105 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12107 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12109 2017-05-18 Marek Polacek <polacek@redhat.com>
12112 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12113 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12114 gimple_assign_load_p.
12116 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12118 PR middle-end/80692
12119 * real.c (do_compare): Give decimal_do_compare preference over
12120 comparing just the signs.
12122 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12124 * doc/md.texi (Canonicalization of Instructions): Describe the
12125 canonical form of instructions that inherently set a condition
12128 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12130 PR middle-end/80775
12131 * tree-cfg.c: Move deletion of unreachable case statements to after
12132 the merging of consecutive case labels.
12134 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12136 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12137 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12138 restoring of callee-saved registers.
12140 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12142 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12143 * config/visium/visium.c (single_set_and_flags): Likewise.
12144 * config/visium/visium.md (Substitutions): Likewise.
12146 2017-05-17 Martin Liska <mliska@suse.cz>
12148 * cfg.c: Introduce dump_flags_t type and
12149 use it instead of int type.
12151 * cfghooks.c: Likewise.
12152 * cfghooks.h (struct cfg_hooks): Likewise.
12153 * cfgrtl.c: Likewise.
12154 * cfgrtl.h: Likewise.
12155 * cgraph.c (cgraph_node::get_body): Likewise.
12156 * coretypes.h: Likewise.
12157 * domwalk.c: Likewise.
12158 * domwalk.h: Likewise.
12159 * dumpfile.c (struct dump_option_value_info): Likewise.
12160 (dump_enable_all): Likewise.
12161 (dump_switch_p_1): Likewise.
12162 (opt_info_switch_p): Likewise.
12163 * dumpfile.h (enum tree_dump_index): Likewise.
12164 (struct dump_file_info): Likewise.
12165 * genemit.c: Likewise.
12166 * generic-match-head.c: Likewise.
12167 * gengtype.c (open_base_files): Likewise.
12168 * gimple-pretty-print.c: Likewise.
12169 * gimple-pretty-print.h: Likewise.
12170 * graph.c (print_graph_cfg): Likewise.
12171 * graphite-scop-detection.c (dot_all_sese): Likewise.
12172 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12173 * loop-unroll.c (report_unroll): Likewise.
12174 * passes.c (pass_manager::register_one_dump_file): Likewise.
12175 * print-tree.c: Likewise.
12176 * statistics.c: Likewise.
12177 * tree-cfg.c: Likewise.
12178 * tree-cfg.h: Likewise.
12179 * tree-dfa.c: Likewise.
12180 * tree-dfa.h: Likewise.
12181 * tree-dump.c (dump_function): Likewise.
12182 * tree-dump.h (struct dump_info): Likewise.
12183 * tree-pretty-print.c: Likewise.
12184 * tree-pretty-print.h: Likewise.
12185 * tree-ssa-live.c: Likewise.
12186 * tree-ssa-live.h: Likewise.
12187 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12188 * tree-vect-loop.c: Likewise.
12189 * tree-vect-slp.c: Likewise.
12191 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12192 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12194 PR tree-optimization/80457
12195 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12196 of all arguments to a statement as scalar_to_vec operations.
12197 (vectorizable_call): Adjust call to vect_model_simple_cost for
12199 (vectorizable_conversion): Likewise.
12200 (vectorizable_assignment): Likewise.
12201 (vectorizable_shift): Likewise.
12202 (vectorizable_operation): Likewise.
12203 (vectorizable_comparison): Likewise.
12204 (vect_is_simple_cond): Record the def types for operands.
12205 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12206 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12207 for statement argument count.
12209 2017-05-16 Carl Love <cel@us.ibm.com>
12211 * config/rs6000/rs6000-c: Add support for built-in functions
12212 vector unsigned long long vec_bperm (vector unsigned long long,
12213 vector unsigned char)
12214 vector signed long long vec_mule (vector signed int,
12216 vector unsigned long long vec_mule (vector unsigned int,
12217 vector unsigned int)
12218 vector signed long long vec_mulo (vector signed int,
12220 vector unsigned long long vec_mulo (vector unsigned int,
12221 vector unsigned int)
12222 vector signed char vec_sldw (vector signed char,
12223 vector signed char,
12225 vector unsigned char vec_sldw (vector unsigned char,
12226 vector unsigned char,
12228 vector signed short vec_sldw (vector signed short,
12229 vector signed short,
12231 vector unsigned short vec_sldw (vector unsigned short,
12232 vector unsigned short,
12234 vector signed int vec_sldw (vector signed int,
12237 vector unsigned int vec_sldw (vector unsigned int,
12238 vector unsigned int,
12240 vector signed long long vec_sldw (vector signed long long,
12241 vector signed long long,
12243 vector unsigned long long vec_sldw (vector unsigned long long,
12244 vector unsigned long long,
12246 * config/rs6000/rs6000-c: Add support for built-in functions
12247 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12248 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12249 * doc/extend.texi: Update the built-in documentation for the
12250 new built-in functions.
12252 2017-05-16 Marek Polacek <polacek@redhat.com>
12256 * tree.c (save_expr): Don't fold the expression.
12258 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12260 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12261 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12262 and (?*y,m). Update insn attributes.
12264 2017-05-16 Martin Liska <mliska@suse.cz>
12266 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12267 flags argument of print_gimple_stmt, print_gimple_expr,
12268 print_generic_stmt and print_generic_expr.
12269 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12270 * coretypes.h: Likewise.
12271 * except.c (dump_eh_tree): Likewise.
12272 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12273 * gimple-pretty-print.h: Likewise.
12274 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12275 (backprop::push_to_worklist): Likewise.
12276 (backprop::pop_from_worklist): Likewise.
12277 (backprop::process_use): Likewise.
12278 (backprop::intersect_uses): Likewise.
12279 (note_replacement): Likewise.
12280 * gimple-ssa-store-merging.c
12281 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12282 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12283 (pass_store_merging::execute): Likewise.
12284 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12285 (ssa_base_cand_dump_callback): Likewise.
12286 (dump_incr_vec): Likewise.
12287 (replace_refs): Likewise.
12288 (replace_mult_candidate): Likewise.
12289 (create_add_on_incoming_edge): Likewise.
12290 (create_phi_basis): Likewise.
12291 (insert_initializers): Likewise.
12292 (all_phi_incrs_profitable): Likewise.
12293 (introduce_cast_before_cand): Likewise.
12294 (replace_one_candidate): Likewise.
12295 * gimplify.c (gimplify_expr): Likewise.
12296 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12297 (set_rename): Likewise.
12298 (rename_uses): Likewise.
12299 (copy_loop_phi_nodes): Likewise.
12300 (add_close_phis_to_merge_points): Likewise.
12301 (copy_loop_close_phi_args): Likewise.
12302 (copy_cond_phi_args): Likewise.
12303 (graphite_copy_stmts_from_block): Likewise.
12304 (translate_pending_phi_nodes): Likewise.
12305 * graphite-poly.c (print_pdr): Likewise.
12306 (dump_gbb_cases): Likewise.
12307 (dump_gbb_conditions): Likewise.
12308 (print_scop_params): Likewise.
12309 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12310 (build_cross_bb_scalars_use): Likewise.
12311 (gather_bbs::before_dom_children): Likewise.
12312 * hsa-dump.c (dump_hsa_immed): Likewise.
12313 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12314 (get_replacement_map): Likewise.
12315 * ipa-inline-analysis.c (dump_condition): Likewise.
12316 (estimate_function_body_sizes): Likewise.
12317 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12318 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12319 * ipa-prop.c (ipa_dump_param): Likewise.
12320 (ipa_print_node_jump_functions_for_edge): Likewise.
12321 (ipa_modify_call_arguments): Likewise.
12322 (ipa_modify_expr): Likewise.
12323 (ipa_dump_param_adjustments): Likewise.
12324 (ipa_dump_agg_replacement_values): Likewise.
12325 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12326 * ipa-pure-const.c (check_stmt): Likewise.
12327 (pass_nothrow::execute): Likewise.
12328 * ipa-split.c (execute_split_functions): Likewise.
12329 * omp-offload.c (dump_oacc_loop_part): Likewise.
12330 (dump_oacc_loop): Likewise.
12331 * trans-mem.c (tm_log_emit): Likewise.
12332 (tm_memopt_accumulate_memops): Likewise.
12333 (dump_tm_memopt_set): Likewise.
12334 (dump_tm_memopt_transform): Likewise.
12335 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12336 (print_loop): Likewise.
12337 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12338 (chkp_gather_checks_info): Likewise.
12339 (chkp_get_check_result): Likewise.
12340 (chkp_remove_check_if_pass): Likewise.
12341 (chkp_use_outer_bounds_if_possible): Likewise.
12342 (chkp_reduce_bounds_lifetime): Likewise.
12343 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12344 (chkp_mark_completed_bounds): Likewise.
12345 (chkp_register_incomplete_bounds): Likewise.
12346 (chkp_mark_invalid_bounds): Likewise.
12347 (chkp_maybe_copy_and_register_bounds): Likewise.
12348 (chkp_build_returned_bound): Likewise.
12349 (chkp_get_bound_for_parm): Likewise.
12350 (chkp_build_bndldx): Likewise.
12351 (chkp_get_bounds_by_definition): Likewise.
12352 (chkp_generate_extern_var_bounds): Likewise.
12353 (chkp_get_bounds_for_decl_addr): Likewise.
12354 * tree-chrec.c (chrec_apply): Likewise.
12355 * tree-data-ref.c (dump_data_reference): Likewise.
12356 (dump_subscript): Likewise.
12357 (dump_data_dependence_relation): Likewise.
12358 (analyze_overlapping_iterations): Likewise.
12359 * tree-inline.c (expand_call_inline): Likewise.
12360 (tree_function_versioning): Likewise.
12361 * tree-into-ssa.c (dump_defs_stack): Likewise.
12362 (dump_currdefs): Likewise.
12363 (dump_names_replaced_by): Likewise.
12364 (dump_update_ssa): Likewise.
12365 (update_ssa): Likewise.
12366 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12367 * tree-parloops.c (build_new_reduction): Likewise.
12368 (try_create_reduction_list): Likewise.
12369 (ref_conflicts_with_region): Likewise.
12370 (oacc_entry_exit_ok_1): Likewise.
12371 (oacc_entry_exit_single_gang): Likewise.
12372 * tree-pretty-print.h: Likewise.
12373 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12374 (get_scalar_evolution): Likewise.
12375 (add_to_evolution): Likewise.
12376 (get_loop_exit_condition): Likewise.
12377 (analyze_evolution_in_loop): Likewise.
12378 (analyze_initial_condition): Likewise.
12379 (analyze_scalar_evolution): Likewise.
12380 (instantiate_scev): Likewise.
12381 (number_of_latch_executions): Likewise.
12382 (gather_chrec_stats): Likewise.
12383 (final_value_replacement_loop): Likewise.
12384 (scev_const_prop): Likewise.
12385 * tree-sra.c (dump_access): Likewise.
12386 (disqualify_candidate): Likewise.
12387 (create_access): Likewise.
12388 (reject): Likewise.
12389 (maybe_add_sra_candidate): Likewise.
12390 (create_access_replacement): Likewise.
12391 (analyze_access_subtree): Likewise.
12392 (analyze_all_variable_accesses): Likewise.
12393 (sra_modify_assign): Likewise.
12394 (initialize_constant_pool_replacements): Likewise.
12395 (find_param_candidates): Likewise.
12396 (decide_one_param_reduction): Likewise.
12397 (replace_removed_params_ssa_names): Likewise.
12398 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12399 * tree-ssa-copy.c (dump_copy_of): Likewise.
12400 (copy_prop_visit_cond_stmt): Likewise.
12401 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12402 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12403 (record_equivalences_from_stmt): Likewise.
12404 * tree-ssa-dse.c (compute_trims): Likewise.
12405 (delete_dead_call): Likewise.
12406 (delete_dead_assignment): Likewise.
12407 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12408 (forward_propagate_into_cond): Likewise.
12409 (pass_forwprop::execute): Likewise.
12410 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12411 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12413 (move_computations_worker): Likewise.
12414 (execute_sm): Likewise.
12415 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12416 (remove_exits_and_undefined_stmts): Likewise.
12417 (remove_redundant_iv_tests): Likewise.
12418 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12419 (adjust_iv_update_pos): Likewise.
12420 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12421 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12422 (value_replacement): Likewise.
12423 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12424 * tree-ssa-pre.c (print_pre_expr): Likewise.
12425 (get_representative_for): Likewise.
12426 (create_expression_by_pieces): Likewise.
12427 (insert_into_preds_of_block): Likewise.
12428 (eliminate_insert): Likewise.
12429 (eliminate_dom_walker::before_dom_children): Likewise.
12430 (eliminate): Likewise.
12431 (remove_dead_inserted_code): Likewise.
12432 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12433 * tree-ssa-reassoc.c (get_rank): Likewise.
12434 (eliminate_duplicate_pair): Likewise.
12435 (eliminate_plus_minus_pair): Likewise.
12436 (eliminate_not_pairs): Likewise.
12437 (undistribute_ops_list): Likewise.
12438 (eliminate_redundant_comparison): Likewise.
12439 (update_range_test): Likewise.
12440 (optimize_range_tests_var_bound): Likewise.
12441 (optimize_vec_cond_expr): Likewise.
12442 (rewrite_expr_tree): Likewise.
12443 (rewrite_expr_tree_parallel): Likewise.
12444 (linearize_expr): Likewise.
12445 (break_up_subtract): Likewise.
12446 (linearize_expr_tree): Likewise.
12447 (attempt_builtin_powi): Likewise.
12448 (attempt_builtin_copysign): Likewise.
12449 (transform_stmt_to_copy): Likewise.
12450 (transform_stmt_to_multiply): Likewise.
12451 (dump_ops_vector): Likewise.
12452 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12453 (print_scc): Likewise.
12454 (set_ssa_val_to): Likewise.
12455 (visit_reference_op_store): Likewise.
12456 (visit_use): Likewise.
12457 (sccvn_dom_walker::before_dom_children): Likewise.
12458 (run_scc_vn): Likewise.
12459 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12461 (expr_hash_elt::print): Likewise.
12462 (const_and_copies::pop_to_marker): Likewise.
12463 (const_and_copies::record_const_or_copy_raw): Likewise.
12464 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12465 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12466 (dump_predicates): Likewise.
12467 (find_uninit_use): Likewise.
12468 (warn_uninitialized_phi): Likewise.
12469 (pass_late_warn_uninitialized::execute): Likewise.
12470 * tree-ssa.c (verify_vssa): Likewise.
12471 (verify_ssa): Likewise.
12472 (maybe_optimize_var): Likewise.
12473 * tree-vrp.c (dump_value_range): Likewise.
12474 (dump_all_value_ranges): Likewise.
12475 (dump_asserts_for): Likewise.
12476 (register_edge_assert_for_2): Likewise.
12477 (vrp_visit_cond_stmt): Likewise.
12478 (vrp_visit_switch_stmt): Likewise.
12479 (vrp_visit_stmt): Likewise.
12480 (vrp_visit_phi_node): Likewise.
12481 (simplify_cond_using_ranges_1): Likewise.
12482 (fold_predicate_in): Likewise.
12483 (evrp_dom_walker::before_dom_children): Likewise.
12484 (evrp_dom_walker::push_value_range): Likewise.
12485 (evrp_dom_walker::pop_value_range): Likewise.
12486 (execute_early_vrp): Likewise.
12488 2017-05-16 Richard Biener <rguenther@suse.de>
12490 * dwarf2out.c (loc_list_from_tree_1): Do not create
12491 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12493 2017-05-16 Richard Biener <rguenther@suse.de>
12495 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12497 (note_variable_value_in_expr): If we resolved the decl ref
12498 do not push to the stack.
12500 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12502 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12503 operations in fast-math mode.
12504 (vaddq_f16): Likewise.
12505 (vmul_f16): Likewise.
12506 (vmulq_f16): Likewise.
12507 (vsub_f16): Likewise.
12508 (vsubq_f16): Likewise.
12509 * config/arm/neon.md (add<mode>3): New.
12511 (fma:<VH:mode>3): New. Also remove outdated comment.
12514 2017-05-16 Martin Liska <mliska@suse.cz>
12518 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12519 (odr_types_equivalent_p): Likewise.
12521 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12523 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12525 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12528 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12529 non-interunit SSE move alternatives with '?'.
12530 (zero-extendsidi peephole2): New peephole to skip intermediate
12531 general register in SSE zero-extend sequence.
12533 2017-05-15 Jeff Law <law@redhat.com>
12535 * reorg.c (relax_delay_slots): Create a new variable to hold
12536 the temporary target rather than clobbering TARGET_LABEL.
12538 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12539 missing argument to extract_bit_field call.
12540 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12542 2017-05-15 Martin Liska <mliska@suse.cz>
12545 * gcc.c (process_command): Do not allow empty argument of -o option.
12547 2017-05-15 Renlin Li <renlin.li@arm.com>
12549 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12550 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12551 * config/aarch64/constraints.md (Usf): Add long call check.
12552 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12553 (call_value): Likewise.
12554 (sibcall): Likewise.
12555 (sibcall_value): Likewise.
12557 (call_value_insn): New.
12558 (sibcall_insn): Update rtx pattern.
12559 (sibcall_value_insn): Likewise.
12560 (call_internal): Remove.
12561 (call_value_internal): Likewise.
12562 (sibcall_internal): Likewise.
12563 (sibcall_value_internal): Likewise.
12564 (call_reg): Likewise.
12565 (call_symbol): Likewise.
12566 (call_value_reg): Likewise.
12567 (call_value_symbol): Likewise.
12569 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12572 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12574 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12576 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12577 compatible with CCGOCmode and with CCZmode.
12579 2017-05-14 Martin Sebor <msebor@redhat.com>
12581 PR middle-end/77671
12582 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12583 (gimple_fold_builtin_snprintf): Same.
12584 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12585 (gimple_fold_builtin_snprintf): Same.
12586 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12587 of character types.
12588 (is_call_safe): New function.
12589 (try_substitute_return_value): Call it.
12590 (try_simplify_call): New function.
12591 (pass_sprintf_length::handle_gimple_call): Call it.
12593 2017-05-14 Martin Sebor <msebor@redhat.com>
12595 PR middle-end/80669
12596 * builtins.c (expand_builtin_stpncpy): Simplify.
12598 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12600 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12601 * config/i386/i386.h
12602 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12603 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12604 (struct machine_function): Add new members call_ms2sysv,
12605 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12606 (struct machine_frame_state): New fields sp_realigned and
12607 sp_realigned_offset.
12608 * config/i386/i386.c
12609 (enum xlogue_stub): New enum.
12610 (enum xlogue_stub_sets): New enum.
12611 (class xlogue_layout): New class.
12612 (struct ix86_frame): New fields stack_realign_allocate_offset,
12613 stack_realign_offset and outlined_save_offset. Modify comments to
12614 detail stack layout when using out-of-line stubs.
12615 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12616 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12617 -mcall-ms2sysv-xlogues.
12618 (stub_managed_regs): New static variable.
12619 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12620 registers managed by out-of-line stub.
12621 (disable_call_ms2sysv_xlogues): New function.
12622 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12623 m->call_ms2sysv when appropriate and compute frame layout for
12625 (sp_valid_at, fp_valid_at): New inline functions.
12626 (choose_basereg): New function.
12627 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12629 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12630 Use align parameter of choose_baseaddr to generated aligned SSE movs
12632 (pro_epilogue_adjust_stack): Modify to track
12633 machine_frame_state::sp_realigned.
12634 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12635 (ix86_nsaved_sseregs): Likewise.
12636 (ix86_emit_save_regs): Likewise.
12637 (ix86_emit_save_regs_using_mov): Likewise.
12638 (ix86_emit_save_sse_regs_using_mov): Likewise.
12639 (get_scratch_register_on_entry): Likewise.
12640 (gen_frame_set): New function.
12641 (gen_frame_load): Likewise.
12642 (gen_frame_store): Likewise.
12643 (emit_outlined_ms2sysv_save): Likewise.
12644 (emit_outlined_ms2sysv_restore): Likewise.
12645 (ix86_expand_prologue): Modify stack re-alignment code and call
12646 emit_outlined_ms2sysv_save when appropriate.
12647 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12648 parameter rtx_insn *insn, which allows the function to be used to only
12649 generate the notes.
12650 (ix86_expand_epilogue): Modify validity checks of frame and stack
12651 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12652 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12653 * config/i386/predicates.md
12654 (save_multiple): New predicate.
12655 (restore_multiple): Likewise.
12656 * config/i386/sse.md
12657 (save_multiple<mode>): New pattern.
12658 (save_multiple_realign<mode>): Likewise.
12659 (restore_multiple<mode>): Likewise.
12660 (restore_multiple_and_return<mode>): Likewise.
12661 (restore_multiple_leave_return<mode>): Likewise.
12662 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12664 2017-05-14 Julia Koval <julia.koval@intel.com>
12666 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12667 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12668 (__builtin_ia32_xsetbv): New builtins.
12669 * config/i386/i386.c (ix86_expand_special_args_builtin):
12671 (ix86_expand_builtin): Special expand for new intrinsics.
12672 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12673 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12674 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12678 * cfganal.c (inverted_post_order_compute): Change argument type
12680 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12681 * df-core.c (rest_of_handle_df_initialize): Adjust.
12682 (rest_of_handle_df_finish): Likewise.
12683 (df_analyze_1): Likewise.
12684 (df_analyze): Likewise.
12685 (loop_inverted_post_order_compute): Change argument to be a vec *.
12686 (df_analyze_loop): Adjust.
12687 (df_get_n_blocks): Likewise.
12688 (df_get_postorder): Likewise.
12689 * df.h (struct df_d): Change field to be a vec.
12690 * lcm.c (compute_laterin): Adjust.
12691 (compute_available): Likewise.
12692 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12693 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12694 * tree-ssa-pre.c (compute_antic): Likewise.
12696 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12698 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12699 (depth_first_search::depth_first_search): Change structure init
12700 function to this constructor.
12701 (depth_first_search::add_bb): Rename function to this member.
12702 (depth_first_search::execute): Likewise.
12703 (flow_dfs_compute_reverse_finish): Adjust.
12705 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12707 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12708 (longest_simple_path): Likewise.
12709 * shrink-wrap.c (spread_components): Likewise.
12710 (disqualify_problematic_components): Likewise.
12711 (emit_common_heads_for_components): Likewise.
12712 (emit_common_tails_for_components): Likewise.
12713 (insert_prologue_epilogue_for_components): Likewise.
12715 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12717 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12720 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12722 * df-core.c (df_set_blocks): Start using auto_bitmap.
12723 (df_compact_blocks): Likewise.
12724 * df-problems.c (df_rd_confluence_n): Likewise.
12725 * df-scan.c (df_insn_rescan_all): Likewise.
12726 (df_process_deferred_rescans): Likewise.
12727 (df_update_entry_block_defs): Likewise.
12728 (df_update_exit_block_uses): Likewise.
12729 (df_entry_block_bitmap_verify): Likewise.
12730 (df_exit_block_bitmap_verify): Likewise.
12731 (df_scan_verify): Likewise.
12732 * lra-constraints.c (lra_constraints): Likewise.
12733 (undo_optional_reloads): Likewise.
12734 (lra_undo_inheritance): Likewise.
12735 * lra-remat.c (calculate_gen_cands): Likewise.
12736 (do_remat): Likewise.
12737 * lra-spills.c (assign_spill_hard_regs): Likewise.
12738 (spill_pseudos): Likewise.
12739 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12740 (bitmap_set_subtract_values): Likewise.
12742 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12744 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12745 management with auto_bitmap.
12746 (fix_inter_tick): Likewise.
12747 (fix_recovery_deps): Likewise.
12748 * ira.c (add_store_equivs): Likewise.
12749 (find_moveable_pseudos): Likewise.
12750 (split_live_ranges_for_shrink_wrap): Likewise.
12751 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12752 (rtx_reuse_manager::seen_def_p): Likewise.
12753 (rtx_reuse_manager::set_seen_def): Likewise.
12754 * print-rtl.h (class rtx_reuse_manager): Likewise.
12756 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12758 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12760 (migrate_btr_def): Likewise.
12761 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12762 * df-core.c (loop_post_order_compute): Likewise.
12763 (loop_inverted_post_order_compute): Likewise.
12764 * hsa-common.h: Likewise.
12765 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12766 * init-regs.c (initialize_uninitialized_regs): Likewise.
12767 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12768 (inline_small_functions): Likewise.
12769 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12770 * ira.c (combine_and_move_insns): Likewise.
12771 (build_insn_chain): Likewise.
12772 * loop-invariant.c (find_invariants): Likewise.
12773 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12774 * predict.c (tree_predict_by_opcode): Likewise.
12775 (predict_paths_leading_to): Likewise.
12776 (predict_paths_leading_to_edge): Likewise.
12777 (estimate_loops_at_level): Likewise.
12778 (estimate_loops): Likewise.
12779 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12780 (spread_components): Likewise.
12781 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12782 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12783 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12784 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12785 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12786 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12787 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12788 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12789 (mark_threaded_blocks): Likewise.
12790 (thread_through_all_blocks): Likewise.
12791 * tree-ssa.c (verify_ssa): Likewise.
12792 (execute_update_addresses_taken): Likewise.
12793 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12795 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12797 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12799 (post_order_compute): Likewise.
12800 (inverted_post_order_compute): Likewise.
12801 (pre_and_rev_post_order_compute_fn): Likewise.
12803 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12805 * genrecog.c (int_set::int_set): Explicitly construct our
12806 auto_vec base class.
12807 * vec.h (auto_vec::auto_vec): New constructor.
12809 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12811 * bitmap.h (class auto_bitmap): New constructor taking
12812 bitmap_obstack * argument.
12814 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12816 * bitmap.h (class auto_bitmap): Change type of m_bits to
12817 bitmap_head, and adjust ctor / dtor and member operators.
12819 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12821 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12822 when returned register mode doesn't match original mode.
12824 2017-05-12 Jeff Law <law@redhat.com>
12825 Jakub Jelinek <jakub@redhat.com>
12827 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
12828 we look for cc setter after the compare-elim changes.
12829 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
12830 within the vector to match what compare-elim now expects.
12831 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
12832 (xorsi3_flags, one_cmplsi2_flags): Likewise.
12834 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
12835 after the compare-elim changes.
12836 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
12837 the vector to match what compare-elim now expects.
12838 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
12839 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
12840 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
12841 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
12842 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
12844 * config/visium/visium.c (single_set_and_flags): Fix where
12845 we look for cc setter after the compare-elim changes.
12846 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
12847 with the vector to match what compare-elim now expects.
12848 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
12849 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
12850 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
12851 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
12852 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
12853 (neg<mode>2_insn_set_overflow): Likewise.
12855 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
12857 PR middle-end/79794
12858 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
12859 maybe_expand_insn call, set ops[0].target. If still set after call,
12860 set alt_rtl. Add extra arg to recursive calls.
12861 (extract_bit_field): Add alt_rtl argument. Pass to
12863 * expmed.h (extract_bit_field): Fix prototype.
12864 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
12865 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
12866 to extract_bit_field_calls.
12867 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
12868 Pass alt_rtl to extract_bit_field calls.
12869 * calls.c (store_unaligned_arguments_into_psuedos)
12870 load_register_parameters): Pass extra NULL to extract_bit_field calls.
12871 * optabs.c (maybe_legitimize_operand): Clear op->target when call
12873 * optabs.h (struct expand_operand): Add target bitfield.
12875 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12877 * compare-elim.c (try_eliminate_compare): Canonicalize
12878 operation with embedded compare to
12879 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
12880 (set (reg) (operation)].
12882 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
12884 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12887 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
12888 cost of adding a carry flag for ADC instruction.
12889 [case MINUS]: Ignore the cost of subtracting a carry flag
12890 for SBB instruction.
12892 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
12894 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
12896 * config/rs6000/bmiintrin.h: New file.
12897 * config/rs6000/bmi2intrin.h: New file.
12898 * config/rs6000/x86intrin.h: New file.
12900 2017-05-12 Jeff Law <law@redhat.com>
12902 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
12905 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
12907 PR middle-end/80707
12908 * tree-cfg.c: Remove cfg edges of unreachable case statements.
12910 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12912 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12913 early expansion of vector divide builtins.
12914 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
12915 builtins identified as having unsigned arguments.
12917 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12919 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
12920 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
12921 expansion of vector logical operations (and, andc, or, xor,
12924 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12926 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
12927 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
12929 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12931 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12932 early GIMPLE expansion of vector multiplies.
12934 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12936 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
12937 TARGET_HAVE_MOVT conditional.
12938 (movt splitter): Likewise.
12940 2017-05-12 Richard Biener <rguenther@suse.de>
12942 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
12943 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12944 Fold all stmts not inplace.
12946 2017-05-12 Richard Biener <rguenther@suse.de>
12948 PR tree-optimization/80713
12949 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
12950 inserted_exprs bit for not removed stmts.
12952 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
12954 PR middle-end/69921
12955 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
12956 parallelized" attribute for parallelized OpenACC kernels.
12957 * omp-offload.c (execute_oacc_device_lower): Use it.
12959 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
12960 Set "oacc kernels" attribute.
12961 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
12962 parameter. Adjust all users.
12963 (oacc_fn_attrib_kernels_p): Remove function.
12964 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
12965 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
12966 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
12967 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
12968 assert "oacc kernels" attribute is set.
12970 2017-05-11 Carl Love <cel@us.ibm.com>
12972 * config/rs6000/rs6000-c: Add support for built-in functions
12973 vector unsigned char vec_popcnt (vector signed char)
12974 vector unsigned char vec_popcnt (vector unsigned char)
12975 vector unsigned short vec_popcnt (vector signed short)
12976 vector unsigned short vec_popcnt (vector unsigned short)
12977 vector unsigned int vec_popcnt (vector signed int)
12978 vector unsigned int vec_popcnt (vector unsigned int)
12979 vector unsigned long long vec_popcnt (vector signed long long)
12980 vector unsigned long long vec_popcnt (vector unsigned long long)
12981 vector signed long long vec_slo (vector signed long long,
12982 vector signed char)
12983 vector signed long long vec_slo (vector signed long long,
12984 vector unsigned char)
12985 vector unsigned long long vec_slo (vector unsigned long long,
12986 vector signed char)
12987 vector unsigned long long vec_slo (vector unsigned long long,
12988 vector unsigned char)
12989 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
12990 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
12991 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
12992 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
12993 * doc/extend.texi: Update the built-in documentation file for the
12994 new built-in functions.
12996 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12998 * attribs.h (sorted_attr_string): Move machine independent
12999 functions for target clone support from the i386 port to common
13000 code. Rename ix86_function_versions to common_function_versions.
13001 Rename make_name to make_unique_name.
13002 (common_function_versions): Likewise.
13003 (make_unique_name): Likewise.
13004 (make_dispatcher_decl): Likewise.
13005 (is_function_default_version): Likewise.
13006 * attribs.c (attr_strcmp): Likewise.
13007 (sorted_attr_string): Likewise.
13008 (common_function_versions): Likewise.
13009 (make_unique_name): Likewise.
13010 (make_dispatcher_decl): Likewise.
13011 (is_function_default_version): Likewise.
13012 * config/i386/i386.c (attr_strcmp): Likewise.
13013 (sorted_attr_string): Likewise.
13014 (ix86_function_versions): Likewise.
13015 (make_name): Likewise.
13016 (make_dispatcher_decl): Likewise.
13017 (is_function_default_version): Likewise.
13018 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13020 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13023 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13024 Account for direct move costs for vec_construct of integer
13027 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13030 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13031 (UNSPEC_STX_ATOMIC): Ditto.
13032 (loaddi_via_sse): New insn.
13033 (storedi_via_sse): Ditto.
13034 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13035 Update corresponding peephole2 patterns.
13036 (atomic_storedi_fpu): Ditto.
13038 2017-05-11 Julia Koval <julia.koval@intel.com>
13040 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13041 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13043 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13044 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13045 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13047 2017-05-11 Nathan Sidwell <nathan@acm.org>
13049 * graphite-poly.c: Include dumpfile.h.
13051 * dumpfle.h (dump_function): Declare here ...
13052 * tree-dump.h (dump_function): ... not here.
13053 * dumpfile.c: #include tree-cfg.h.
13054 (dump_function): Move here from ...
13055 * tree-dump.c (dump_function): ... here.
13056 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13057 * graphite-poly.c: Don't include tree-dump.h.
13058 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13059 * print-tree.c: Likewise.
13060 * stor-layout.c: Likewise.
13061 * tree-nested.c: Likewise.
13063 * dumpfile.c (dump_start): Use TDF_FLAGS.
13064 (dump_enable_all): Fix TDF_KIND check thinko.
13066 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13068 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13069 array entries to represent two legal parameterizations of the
13070 overloaded __builtin_cmpb function, as represented by the
13071 P6_OV_BUILTIN_CMPB constant.
13072 (altivec_resolve_overloaded_builtin): Add special case handling
13073 for the __builtin_cmpb function, as represented by the
13074 P6_OV_BUILTIN_CMPB constant.
13075 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13076 (BU_P6_64BIT_2): New macro.
13077 (BU_P6_OVERLOAD_2): New macro
13078 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13079 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13080 (CMPB): Add overload support to represent both 32-bit and 64-bit
13081 compare-bytes function.
13082 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13083 support for TARGET_CMPB.
13084 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13085 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13086 documentation of the __builtin_cmpb overloaded built-in function.
13088 2017-05-11 Richard Biener <rguenther@suse.de>
13090 PR tree-optimization/80705
13091 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13092 bases are not vectorizable.
13094 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13096 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13097 when counting register pressure.
13099 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13101 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13102 (iv_ca_more_deps): Renamed to ...
13103 (iv_ca_compare_deps): ... this.
13104 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13106 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13108 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13110 (determine_group_iv_costs): ... here.
13111 (find_inv_vars_cb): Record inv var if it's not recorded before.
13113 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13115 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13116 (get_shiftadd_cost): Ditto.
13118 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13120 * tree-ssa-address.c: Include header file.
13121 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13123 (add_to_parts): Refactor.
13124 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13125 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13128 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13130 PR tree-optimization/53090
13131 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13133 (extract_cond_operands): Detect condition with IV on both sides
13134 and return COMP_IV_EXPR_2.
13135 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13136 (rewrite_use_compare): Simplify by removing call to function
13137 extract_cond_operands.
13139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13141 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13142 (extract_cond_operands): Detect condition comparing against non-
13143 invariant bound and return appropriate enum value.
13144 (find_interesting_uses_cond): Update use of extract_cond_operands.
13145 Handle its return value accordingly.
13146 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13148 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13150 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13151 nonlinear iv_use computation in loop invariant sensitive way.
13153 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13155 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13156 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13158 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13160 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13161 (dump_cand): Support iv_cand.inv_exprs.
13162 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13164 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13167 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13169 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13171 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13172 as local function. Include necessary header files.
13173 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13175 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13177 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13179 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13181 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13182 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13183 RSHIFT_EXPR and BIT_NOT_EXPR.
13185 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13187 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13188 (adjust_setup_cost): New parameter supporting round up adjustment.
13189 (struct address_cost_data): Delete.
13190 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13191 (split_address_cost, ptr_difference_cost): Delete.
13192 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13193 (struct ainc_cost_data): New struct.
13194 (get_address_cost_ainc): New function.
13195 (get_address_cost, get_computation_cost): Reimplement.
13196 (determine_group_iv_cost_address): Record inv_expr for all uses of
13198 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13199 (iv_ca_has_deps): Reimplemented to ...
13200 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13202 (iv_ca_extend): Call iv_ca_more_deps.
13204 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13206 * tree-ssa-address.c (struct mem_address): Move to header file.
13207 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13208 * tree-ssa-address.h (struct mem_address): Move from C file.
13209 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13211 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13213 * tree-affine.h (aff_combination_type): New interface.
13214 (aff_combination_zero_p): Remove static.
13215 (aff_combination_const_p): New interface.
13216 (aff_combination_singleton_var_p): New interfaces.
13218 2017-05-11 Richard Biener <rguenther@suse.de>
13220 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13221 Skip unreachable blocks and destinations.
13222 (eliminate): Move stmt removal and fixup ...
13223 (fini_eliminate): ... here. Skip inserted exprs.
13224 (pass_pre::execute): Move fini_pre after fini_eliminate.
13225 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13226 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13227 PRE to get rid of dead code that has invalid SSA form and
13228 split critical edges again.
13230 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13232 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13234 2017-05-11 Richard Biener <rguenther@suse.de>
13236 * passes.c (execute_function_todo): Verify loops if they are
13237 said to be up-to-date.
13238 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13239 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13241 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13244 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13245 handle calling assemble_external ourself.
13248 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13249 modes with zero size. Enhance comment.
13251 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13253 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13254 built-ins for vec_xl and vec_xst with short and char pointer
13257 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13259 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13260 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13261 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13262 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13263 (_mm_maskz_min_round_ss): New intrinsics.
13264 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13265 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13266 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13267 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13268 (__builtin_ia32_minss_mask_round): New builtins.
13269 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13270 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13271 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13273 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13274 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13276 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13279 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13281 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13282 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13283 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13284 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13285 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13286 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13287 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13288 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13289 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13290 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13291 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13292 (__builtin_ia32_mulss_mask_round): New builtins.
13293 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13294 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13295 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13297 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13298 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13300 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13303 2017-05-10 Julia Koval <julia.koval@intel.com>
13305 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13306 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13307 (_mm256_setr_m128i): New intrinsics.
13309 2017-05-10 Julia Koval <julia.koval@intel.com>
13311 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13312 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13313 (_mm_maskz_rcp14_ss): New intrinsics.
13314 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13315 (__builtin_ia32_rcp14ss_mask): New builtins.
13316 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13318 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13320 PR tree-optimization/51513
13321 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13322 (assert_unreachable_fallthru_edge_p): Use it.
13323 (group_case_labels_stmt): Likewise.
13324 * tree-cfg.h: Prototype it.
13325 * stmt.c: Include cfghooks.h and tree-cfg.h.
13326 (emit_case_dispatch_table) <gap_label>: New local variable.
13327 Use it to fill dispatch table gaps.
13328 Test for default_label before updating probabilities.
13329 (expand_case) <default_label>: Remove unneeded initialization.
13330 Test for unreachable default case statement and remove its edge.
13331 Set default_label accordingly.
13332 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13334 2017-05-10 Carl Love <cel@us.ibm.com>
13336 * config/rs6000/rs6000-c: Add support for built-in functions
13337 vector signed char vec_neg (vector signed char)
13338 vector signed short int vec_neg (vector short int)
13339 vector signed int vec_neg (vector signed int)
13340 vector signed long long vec_neg (vector signed long long)
13341 vector float vec_neg (vector float)
13342 vector double vec_neg (vector double)
13343 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13345 * config/rs6000/altivec.h: Add define for vec_neg
13346 * doc/extend.texi: Update the built-in documentation for the
13347 new built-in functions.
13349 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13351 PR tree-optimization/77644
13352 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13354 2017-05-10 Nathan Sidwell <nathan@acm.org>
13356 * dumpfile.h (TDI_lang_all): New.
13357 (TDF_KIND): New. Renumber others
13358 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13360 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13362 (get_dump_file_name): Adjust suffix generation.
13363 (dump_enable_all): Use TDF_KIND.
13364 * doc/invoke.texi (-fdump-lang-all): Document.
13366 * dumpfile.h: Tabify.
13368 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13371 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13372 Move member access before delete.
13374 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13376 * tree-inline.c (expand_call_inline): Split block at stmt
13379 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13382 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13383 are now unused after splitting mov{sf,sd}_hardfloat.
13384 (f32_lr2): Likewise.
13385 (f32_lm): Likewise.
13386 (f32_lm2): Likewise.
13387 (f32_li): Likewise.
13388 (f32_li2): Likewise.
13389 (f32_lv): Likewise.
13390 (f32_sr): Likewise.
13391 (f32_sr2): Likewise.
13392 (f32_sm): Likewise.
13393 (f32_sm2): Likewise.
13394 (f32_si): Likewise.
13395 (f32_si2): Likewise.
13396 (f32_sv): Likewise.
13397 (f32_dm): Likewise.
13398 (f32_vsx): Likewise.
13399 (f32_av): Likewise.
13400 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13401 For movsf, order stores so the VSX stores occur before the GPR
13402 store which encourages the register allocator to use a traditional
13403 FPR instead of a GPR. For movsd, order the stores so that the GPR
13404 store comes before the VSX stores to allow the power6 to work.
13405 This is due to the power6 not having a 32-bit integer store
13406 instruction from a FPR.
13407 (movsf_hardfloat): Likewise.
13408 (movsd_hardfloat): Likewise.
13410 2017-05-09 Martin Sebor <msebor@redhat.com>
13412 PR translation/80280
13413 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13416 PR translation/80280
13417 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13418 data member added in r247778.
13419 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13421 2017-05-09 Nathan Sidwell <nathan@acm.org>
13423 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13425 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13428 2017-05-09 Marek Polacek <polacek@redhat.com>
13430 * doc/invoke.texi: Fix typo.
13432 2017-05-09 Richard Biener <rguenther@suse.de>
13434 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13435 (vrp_val_is_min): Likewise.
13436 (set_value_range_to_value): Likewise.
13437 (set_value_range_to_nonnegative): Likewise.
13438 (gimple_assign_nonzero_p): Likewise.
13439 (gimple_stmt_nonzero_p): Likewise.
13440 (vrp_int_const_binop): Likewise. Remove unreachable case.
13441 (adjust_range_with_scev): Adjust comments.
13442 (compare_range_with_value): Likewise.
13443 (extract_range_from_phi_node): Likewise.
13444 (test_for_singularity): Likewise.
13446 2017-05-09 Richard Biener <rguenther@suse.de>
13448 * tree-vrp.c (get_single_symbol): Add assert that we don't
13449 get overflowed constants as invariant part.
13450 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13451 checks. Use wi::cmp instead of recursing for integer constants.
13452 (compare_values): Just ignore whether we assumed undefined
13453 overflow instead of failing the compare.
13454 (extract_range_for_var_from_comparison_expr): Add comment before the
13455 TREE_NO_WARNING sets.
13456 (test_for_singularity): Likewise.
13457 (extract_range_from_comparison): Do not disable optimization
13458 when we assumed undefined overflow.
13459 (extract_range_basic): Remove init of unused var.
13461 2017-05-09 Richard Biener <rguenther@suse.de>
13463 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13464 (extract_range_from_multiplicative_op_1): Adjust.
13465 (extract_range_from_binary_expr_1): Use int_const_binop.
13467 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13470 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13471 rs6000_store_data_bypass_p in seven define_bypass directives and
13472 in several comments.
13473 * config/rs6000/rs6000-protos.h: Add prototype for
13474 rs6000_store_data_bypass_p function.
13475 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13476 function implements slightly different (rs6000-specific) semantics
13477 than store_data_bypass_p, returning false rather than aborting
13478 with assertion error when arguments do not satisfy the
13479 requirements of store data bypass.
13480 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13481 rs6000_store_data_bypass_p.
13483 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13485 * config/xtensa/xtensa-protos.h
13486 (xtensa_initial_elimination_offset): New declaration.
13487 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13488 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13489 macro definition, add case for FRAME_POINTER_REGNUM when
13490 FRAME_GROWS_DOWNWARD.
13491 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13492 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13493 xtensa_initial_elimination_offset.
13495 2017-05-08 Nathan Sidwell <nathan@acm.org>
13497 * doc/invoke.texi: Alphabetize -fdump options.
13499 2017-05-08 Martin Sebor <msebor@redhat.com>
13501 PR translation/80280
13502 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13504 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13506 * target.def (compute_frame_layout): New optional target hook.
13507 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13508 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13509 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13511 * reload1.c (verify_initial_elim_offsets): Likewise.
13512 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13513 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13514 (arm_get_frame_offsets): Split up into this ...
13515 (arm_compute_frame_layout): ... and this function.
13517 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13519 * config/aarch64/constraints.md (Usa): New constraint.
13520 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13522 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13524 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13525 with_multilib_list after it has been checked.
13527 2017-05-08 Richard Biener <rguenther@suse.de>
13529 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13530 (bitmap_set_subtract_values): Likewise.
13532 2017-05-08 Richard Biener <rguenther@suse.de>
13534 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13535 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13537 (gimple_stmt_nonzero_warnv_p): Rename to ...
13538 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13540 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13541 (extract_range_basic): Adjust, do not disable propagation on
13542 strict overflow sensitive simplification.
13543 (vrp_visit_cond_stmt): Likewise.
13545 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13547 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13548 body size unconditionally.
13550 2017-05-07 Jeff Law <law@redhat.com>
13553 2017-05-06 Jeff Law <law@redhat.com>
13554 PR tree-optimization/78496
13555 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13558 PR tree-optimization/78496
13559 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13560 (simplify_stmt_using_ranges): Call it.
13561 (vrp_dom_walker::before_dom_children): Extract equivalences
13562 from an ASSERT_EXPR with an equality comparison against a
13565 2017-05-06 Jeff Law <law@redhat.com>
13567 PR tree-optimization/78496
13568 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13571 PR tree-optimization/78496
13572 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13573 (simplify_stmt_using_ranges): Call it.
13574 (vrp_dom_walker::before_dom_children): Extract equivalences
13575 from an ASSERT_EXPR with an equality comparison against a
13578 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13580 * lra-constraints.c (lra_copy_reg_equiv): New function.
13581 (split_reg): Use it to copy equivalence information from the
13582 original register to the spill register.
13584 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13586 PR rtl-optimization/75964
13587 * simplify-rtx.c (simplify_const_relational_operation): Remove
13588 invalid handling of comparisons of integer ABS.
13590 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13592 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13593 initialize to zero.
13594 (init_regs): Remove declaration.
13595 (function_arg_advance_32): Initialize error_p as boolean variable.
13597 2017-05-05 Nathan Sidwell <nathan@acm.org>
13599 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13600 lines. Use for (;;).
13602 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13604 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13605 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13606 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13607 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13608 VF=2 that require versioning.
13610 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13612 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13615 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13617 * diagnostic.h (diagnostic_override_option_index): Convert from
13618 macro to inline function.
13620 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13622 * diagnostic.c (last_module_changed_p): New function.
13623 (set_last_module): New function.
13624 (diagnostic_report_current_module): Convert macro usage to
13625 the above functions.
13626 * diagnostic.h (diagnostic_context::last_module): Strengthen
13627 from const line_map * to const line_map_ordinary *.
13628 (diagnostic_last_module_changed): Delete macro.
13629 (diagnostic_set_last_module): Delete macro.
13631 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13633 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13634 with diagnostic_report_diagnostic.
13635 (diagnostic_n_impl_richloc): Likewise.
13636 * diagnostic.h (report_diagnostic): Delete macro.
13637 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13638 with diagnostic_report_diagnostic.
13639 * substring-locations.c (format_warning_va): Likewise.
13641 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13643 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13644 save/restor of format_spec. Move option-printing code to...
13645 (print_option_information): ...this new function, and
13646 reimplement by simply printing to the pretty_printer,
13647 rather than appending to the format string.
13649 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13651 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13652 handling logic into...
13653 (update_effective_level_from_pragmas): ...this new function.
13655 2017-05-04 Andrew Waterman <andrew@sifive.com>
13657 * config/riscv/riscv.opt (mstrict-align): New option.
13658 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13659 (SLOW_UNALIGNED_ACCESS): Define.
13660 (riscv_slow_unaligned_access): Declare.
13661 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13663 (riscv_slow_unaligned_access): New variable.
13664 (rocket_tune_info): Set slow_unaligned_access to true.
13665 (optimize_size_tune_info): Set slow_unaligned_access to false.
13666 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13667 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13668 (riscv_option_override): Set riscv_slow_unaligned_access.
13669 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13671 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13673 * config/riscv/riscv.md: Unify indentation.
13675 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13680 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13682 (extendsi<mode>2): Add support for doing sign extension via
13683 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13684 don't have ISA 3.0 instructions.
13685 (extendsi<mode>2 splitter): Likewise.
13686 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13687 generate the normal insns since SImode can now go in vector
13688 registers. Disallow the special UNSPECs needed for previous
13689 machines to hide SImode being used. Add new insns
13690 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13691 (fix_trunc<mode>si2_stfiwx): Likewise.
13692 (fix_trunc<mode>si2_internal): Likewise.
13693 (fixuns_trunc<mode>si2): Likewise.
13694 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13695 (fctiw<u>z_<mode>_smallint): Likewise.
13696 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13697 of floating point to 32-bit integer from doing a direct move to
13698 the GPR registers to do a store.
13699 (fctiwz_<mode>): Break long line.
13701 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13703 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13704 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13705 (addr_list, addr_offset_valid_p): New.
13706 (split_address_groups): Check offset validity with above function.
13707 (gt-tree-ssa-loop-ivopts.h): Include header file.
13709 2017-05-05 Nathan Sidwell <nathan@acm.org>
13711 * config.gcc (arm*-*-*): Add missing 'fi'.
13713 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13715 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13716 included in -fopt-info does not matter.
13717 * doc/optinfo.texi (-fopt-info): Fix description of default
13718 behavour. Explicitly say order of options included in -fopt-info
13721 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13723 * config.gcc: Allow combinations of aprofile and rmprofile values for
13724 --with-multilib-list.
13725 * config/arm/t-multilib: New file.
13726 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13727 variables. Remove setting of ISA and floating-point ABI in
13728 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13729 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13730 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13731 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13733 * config/arm/t-rmprofile: Likewise except for the matches changes.
13734 * doc/install.texi (--with-multilib-list): Document the combination of
13735 aprofile and rmprofile values and warn about pitfalls in doing that.
13737 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13739 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13740 (movdi_aarch64): Likewise.
13742 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13744 PR tree-optimization/80632
13745 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13747 (build_arrays): Initialize it for virtual phis.
13748 (fix_phi_nodes): Use it for virtual phis.
13750 PR tree-optimization/80558
13751 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13752 [x, y] op z into [x op, y op z] for op & or | if conditions
13755 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13756 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13759 * config/arm/arm.md (use_literal_pool): Remove.
13760 (64-bit immediate split): No longer takes cost into consideration
13761 if arm_disable_literal_pool is enabled.
13762 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13763 used when arm_disable_literal_pool is enabled.
13764 (arm_max_const_double_inline_cost): Remove use of
13765 arm_disable_literal_pool.
13766 (push_minipool_fix): Add assert.
13767 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13768 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13769 (no_literal_pool_sf_immediate): New.
13771 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13773 PR tree-optimization/80613
13774 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13775 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13777 2017-05-05 Richard Biener <rguenther@suse.de>
13779 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13781 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13783 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13784 of this flag from insn conditions due to removal from r247495.
13786 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13788 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13790 (arm_early_store_addr_dep_ptr): Likewise.
13791 * config/arm/aarch-common-protos.h
13792 (arm_early_load_addr_dep_ptr): Add prototype.
13793 (arm_early_store_addr_dep_ptr): Likewise.
13794 * config/arm/cortex-a53.md: Add new bypasses.
13796 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13798 * tree.c (next_type_uid): Change type to unsigned.
13799 (type_hash_canon): Decrement back next_type_uid if
13800 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13801 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13804 2017-05-04 Martin Sebor <msebor@redhat.com>
13806 * builtins.c: Fix a trivial typo in a comment.
13808 PR middle-end/79234
13809 * builtins.c (check_sizes): Adjust to handle reading past the end.
13810 Avoid printing excessive upper bound of ranges. Use %E to print
13811 tree nodes instead of converting them to %wu.
13812 (expand_builtin_memchr): New function.
13813 (compute_dest_size): Rename...
13814 (compute_objsize): ...to this.
13815 (expand_builtin_memcpy): Adjust.
13816 (expand_builtin_mempcpy): Adjust.
13817 (expand_builtin_strcat): Adjust.
13818 (expand_builtin_strcpy): Adjust.
13819 (check_strncat_sizes): Adjust.
13820 (expand_builtin_strncat): Adjust.
13821 (expand_builtin_strncpy): Adjust and simplify.
13822 (expand_builtin_memset): Adjust.
13823 (expand_builtin_bzero): Adjust.
13824 (expand_builtin_memcmp): Adjust.
13825 (expand_builtin): Handle memcmp.
13826 (maybe_emit_chk_warning): Check strncat just once.
13828 2017-05-04 Martin Sebor <msebor@redhat.com>
13830 PR preprocessor/79214
13831 PR middle-end/79222
13832 PR middle-end/79223
13833 * builtins.c (check_sizes): Add inlining context and issue
13834 warnings even when -Wno-system-headers is set.
13835 (check_strncat_sizes): Same.
13836 (expand_builtin_strncat): Same.
13837 (expand_builtin_memmove): New function.
13838 (expand_builtin_stpncpy): Same.
13839 (expand_builtin): Handle memmove and stpncpy.
13841 2017-05-04 Bin Cheng <bin.cheng@arm.com>
13843 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
13844 which is not used any more.
13846 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13848 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
13850 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13852 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
13853 (cortexa53_tunings): Likewise.
13854 (cortexa57_tunings): Likewise.
13855 (cortexa72_tunings): Likewise.
13856 (cortexa73_tunings): Likewise.
13858 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13860 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
13861 Set loop alignment to 8.
13863 2017-05-04 Martin Sebor <msebor@redhat.com>
13865 PR translation/80280
13866 * builtins.c (expand_builtin_object_size): Add missing quoting to
13867 %D and like directives.
13868 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
13869 (hsa_type_for_tree_type): Same.
13870 (verify_function_arguments): Same.
13871 * symtab.c (symbol_table::change_decl_assembler_name): Same.
13872 * varasm.c (get_section): Same.
13875 2017-05-04 Martin Sebor <msebor@redhat.com>
13877 PR translation/80280
13878 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
13880 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13882 * config/aarch64/aarch64.c (generic_addrcost_table):
13883 Change HI/TI mode setting.
13885 2017-05-04 Martin Jambor <mjambor@suse.cz>
13887 PR tree-optimization/80622
13888 * tree-sra.c (comes_initialized_p): New function.
13889 (build_accesses_from_assign): Only set write lazily when
13890 comes_initialized_p is false.
13891 (analyze_access_subtree): Use comes_initialized_p.
13892 (propagate_subaccesses_across_link): Assert !comes_initialized_p
13893 instead of testing for PARM_DECL.
13895 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13897 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
13898 constraint on operand 0 to allow more general addressing modes.
13899 Adjust output template.
13900 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
13902 * config/aarch64/aarch64-protos.h
13903 (aarch64_address_valid_for_prefetch_p): Declare prototype.
13904 * config/aarch64/constraints.md (Dp): New address constraint.
13905 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
13908 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
13910 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
13911 update use of estimate_ipcp_clone_size_and_time.
13912 (estimate_local_effects): Update use of
13913 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
13914 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
13915 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
13916 Return nonspecialized time.
13918 2017-05-04 Richard Biener <rguenther@suse.de>
13920 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
13921 for the last VUSE which def dominates the PHI. Directly call
13923 (get_continuation_for_phi_1): Remove.
13925 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
13927 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
13928 to explain the use of truncating division. Cap the number of
13929 iterations to the maximum given by nb_iterations_upper_bound,
13932 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13934 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
13935 * configure: Regenerate.
13936 * config.in: Regenerate.
13937 * config/i386/driver-mingw32.c: new file.
13938 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
13939 * config.host: Link driver-mingw32.o on MinGW host.
13940 * doc/install.texi: Document new --enable-mingw-wildcard configure
13943 2017-05-04 Marek Polacek <polacek@redhat.com>
13945 PR tree-optimization/80612
13946 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
13948 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13949 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
13951 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
13952 (movt splitter): Likewise.
13953 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
13954 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
13955 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
13956 block for Thumb-1 with MOVT.
13957 (thumb2_legitimate_address_p): Move code block ...
13958 (can_avoid_literal_pool_for_label_p): ... into this new function.
13959 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
13961 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
13962 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
13963 "M-profile targets with the MOVT instruction".
13965 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13967 * config/arm/arm-builtins.c (arm_init_builtins): Rename
13968 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
13969 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
13971 2017-05-04 Martin Liska <mliska@suse.cz>
13973 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
13974 variable cond_code.
13976 2017-05-04 Richard Biener <rguenther@suse.de>
13978 * tree.c (array_at_struct_end_p): Handle arrays at struct
13979 end with flexarrays more conservatively. Refactor and treat
13980 arrays of arrays or aggregates more strict. Fix
13981 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
13982 * tree.c (array_at_struct_end_p): Adjust prototype.
13983 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13984 * gimple-fold.c (get_range_strlen): Likewise.
13985 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
13987 2017-05-04 Richard Biener <rguenther@suse.de>
13989 PR tree-optimization/31130
13990 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
13992 (supports_overflow_infinity): Likewise.
13993 (is_negative_overflow_infinity): Likewise.
13994 (is_positive_overflow_infinity): Likewise.
13995 (is_overflow_infinity): Likewise.
13996 (stmt_overflow_infinity): Likewise.
13997 (overflow_infinity_range_p): Likewise.
13998 (usable_range_p): Remove as always returning true.
13999 (make_overflow_infinity): Remove.
14000 (negative_overflow_infinity): Likewise.
14001 (positive_overflow_infinity): Likewise.
14002 (avoid_overflow_infinity): Likewise.
14003 (set_value_range): Adjust accordingly.
14004 (set_value_range_to_nonnegative): Likewise, remove now unused
14005 overflow_infinity arg.
14006 (vrp_operand_equal_p): Adjust.
14007 (update_value_range): Likewise.
14008 (range_int_cst_singleton_p): Likewise.
14009 (operand_less_p): Likewise.
14010 (compare_values_warnv): Likewise.
14011 (extract_range_for_var_from_comparison_expr): Likewise.
14012 (vrp_int_const_binop): Likewise.
14013 (zero_nonzero_bits_from_vr): Likewise.
14014 (extract_range_from_multiplicative_op_1): Likewise.
14015 (extract_range_from_binary_expr_1): Likewise.
14016 (extract_range_from_unary_expr): Likewise.
14017 (extract_range_from_comparison): Likewise.
14018 (extract_range_basic): Likewise.
14019 (adjust_range_with_scev): Likewise.
14020 (compare_ranges): Likewise.
14021 (compare_range_with_value): Likewise.
14022 (dump_value_range): Likewise.
14023 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14025 (simplify_cond_using_ranges): Adjust.
14027 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14029 * brig-builtins.def: Added a builtin for class_f64.
14030 * builtin-types.def: Added a builtin type needed by class_f64.
14032 2017-05-03 Jason Merrill <jason@redhat.com>
14034 * timevar.def: Add TV_CONSTEXPR.
14036 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14038 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14040 2017-05-03 Martin Jambor <mjambor@suse.cz>
14042 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14043 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14044 * ipa-cp.c (ipcp_propagate_stage): Do not call
14045 ipa_update_after_lto_read.
14046 * ipa-inline.c (ipa_inline): Likewise.
14048 2017-05-03 Martin Jambor <mjambor@suse.cz>
14050 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14051 tag. Added a default constructor and a destructor.
14052 (ipa_edge_args_sum_t): New class;
14053 (ipa_edge_args_sum): Declare.
14054 (ipa_edge_args_vector): Remove declaration.
14055 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14056 (ipa_free_edge_args_substructures): Remove declaration.
14057 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14058 (ipa_edge_args_info_available_for_edge_p): Likewise.
14059 * ipa-prop.c (ipa_edge_args_vector): Removed.
14060 (edge_removal_hook_holder): Likewise.
14061 (edge_duplication_hook_holder): Likewise.
14062 (ipa_edge_args_sum): New variable.
14063 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14064 ipa_edge_args_vector.
14065 (ipa_free_edge_args_substructures): Likewise.
14066 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14067 ipa_edge_args_vector.
14068 (ipa_edge_removal_hook): Turned into method
14069 ipa_edge_args_sum_t::remove.
14070 (ipa_edge_duplication_hook): Turned into method
14071 ipa_edge_args_sum_t::duplicate.
14072 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14073 registering edge hooks.
14074 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14075 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14076 ipa_edge_args_sum instead of ipa_edge_args_vector.
14077 * ipa-profile.c (ipa_profile): Likewise.
14079 2017-05-03 Martin Jambor <mjambor@suse.cz>
14081 * symbol-summary.h (function_summary): New method exists.
14082 (function_summary::symtab_removal): Deallocate through release.
14083 (call_summary): New class.
14084 (gt_ggc_mx): New overload.
14085 (gt_pch_nx): Likewise.
14086 (gt_pch_nx): Likewise.
14088 2017-05-03 Jeff Law <law@redhat.com>
14090 PR tree-optimization/78496
14091 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14092 from simplify_cond_using_ranges. Split off code to walk
14093 backwards through casts into ...
14094 (simplify_cond_using_ranges_2): New function.
14095 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14096 (execute_vrp): After identifying jump threads, call
14097 simplify_cond_using_ranges_2.
14099 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14102 * ipa-inline.h (inline_summary): Add ctor.
14103 (create_ggc): Do not use ggc_cleared_alloc.
14105 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14106 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14108 * gcc.c (handle_braces): Support escaping in switch matching
14110 * doc/invoke.texi (Spec Files): Document it.
14111 Remove superfluous @code markup in items.
14113 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14115 * diagnostic-show-locus.c (struct column_range): New struct.
14116 (get_affected_columns): New function.
14117 (get_printed_columns): New function.
14118 (struct correction): New struct.
14119 (correction::ensure_capacity): New function.
14120 (correction::ensure_terminated): New function.
14121 (struct line_corrections): New struct.
14122 (line_corrections::~line_corrections): New dtor.
14123 (line_corrections::add_hint): New function.
14124 (layout::print_trailing_fixits): Reimplement in terms of the new
14126 (selftest::test_overlapped_fixit_printing): New function.
14127 (selftest::diagnostic_show_locus_c_tests): Call it.
14129 2017-05-03 Nathan Sidwell <nathan@acm.org>
14131 Canonicalize canonical type hashing
14132 * tree.h (type_hash_canon_hash): Declare.
14133 * tree.c (type_hash_list, attribute_hash_list): Move into
14134 type_hash_canon_hash.
14135 (build_type_attribute_qual_variant): Break out hash code calc into
14136 type_hash_canon_hash.
14137 (type_hash_canon_hash): New. Generic type hash computation.
14138 (build_range_type_1, build_array_type_1, build_function_type,
14139 build_method_type_directly, build_offset_type, build_complex_type,
14140 make_vector_type): Call it.
14142 2017-05-03 Richard Biener <rguenther@suse.de>
14144 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14145 When all DRs have unknown misaligned do not always peel
14146 when there is a store but apply the same costing model as if
14147 there were only loads.
14149 2017-05-03 Richard Biener <rguenther@suse.de>
14152 PR tree-optimization/80492
14153 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14154 compare_base_decls returning dont-know properly.
14156 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14158 * config/arm/iterators.md (CCSI): New mode iterator.
14159 (arch): New mode attribute.
14160 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14161 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14162 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14163 code iterator for success result mode.
14164 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14165 the corresponding new insn generators.
14167 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14171 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14173 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14175 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14176 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14177 (DDR_A): Wrap DDR argument in brackets.
14178 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14179 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14180 (DDR_REVERSED_P): Likewise.
14182 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14184 PR tree-optimization/79472
14185 * tree-switch-conversion.c (struct switch_conv_info): Add
14186 contiguous_range and default_case_nonstandard fields.
14187 (collect_switch_conv_info): Compute contiguous_range and
14188 default_case_nonstandard fields, don't clear final_bb if
14189 contiguous_range and only the default case doesn't have the required
14191 (check_all_empty_except_final): Set default_case_nonstandard instead
14192 of failing if contiguous_range and the default case doesn't have empty
14194 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14195 and only the default case doesn't have the required constants. Skip
14197 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14198 if default_case_nonstandard.
14199 (build_constructors): Build constant 1 just once. Assert that default
14200 values aren't inserted in between cases if contiguous_range. Skip
14202 (build_arrays): Skip virtual phis.
14203 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14204 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14205 Handle virtual phis.
14206 (gen_inbound_check): Handle default_case_nonstandard case.
14207 (process_switch): Adjust check_final_bb caller. Call
14208 gather_default_values with the first non-default case instead of
14209 default case if default_case_nonstandard.
14211 2017-05-02 Nathan Sidwell <nathan@acm.org>
14213 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14214 check. Fix formatting.
14216 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14218 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14219 errors when comparing specialized and unspecialized times.
14221 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14223 * diagnostic-show-locus.c
14224 (layout::should_print_annotation_line_p): Make private.
14225 (layout::print_annotation_line): Make private.
14226 (layout::annotation_line_showed_range_p): Make private.
14227 (layout::show_ruler): Make private.
14228 (layout::print_source_line): Make private. Pass in line and
14229 line_width, rather than calling location_get_source_line. Drop
14231 (layout::print_leading_fixits): New method.
14232 (layout::print_any_fixits): Rename to...
14233 (layout::print_trailing_fixits): ...this, and make private.
14234 Don't print newline fixits.
14235 (diagnostic_show_locus): Move logic for printing one row into...
14236 (layout::print_line): ...this new function. Move the
14237 location_get_source_line call and error-handling from
14238 print_source_line to here. Call print_leading_fixits, and rename
14239 print_any_fixits to print_trailing_fixits.
14240 (selftest::test_fixit_insert_containing_newline): Update now that
14241 newlines are partially supported.
14242 (selftest::test_fixit_insert_containing_newline_2): New test.
14243 (selftest::test_fixit_replace_containing_newline): Update comments.
14244 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14245 * edit-context.c (class added_line): New class.
14246 (class edited_line): Describe newline handling in comment.
14247 (edited_line::actually_edited_p): New method.
14248 (edited_line::print_content): Delete redundant decl.
14249 (edited_line::m_predecessors): New field.
14250 (edited_file::print_content): Call edited_line::print_content.
14251 (edited_file::print_diff): Update to support newlines.
14252 (edited_file::print_diff_hunk): Likewise.
14253 (edited_file::print_run_of_changed_lines): New function.
14254 (edited_file::print_diff_line): Convert to...
14255 (print_diff_line): ...this.
14256 (edited_file::get_effective_line_count): New function.
14257 (edited_line::edited_line): Initialize new field m_predecessors.
14258 (edited_line::~edited_line): Clean up m_predecessors.
14259 (edited_line::apply_fixit): Handle newlines.
14260 (edited_line::get_effective_line_count): New function.
14261 (edited_line::print_content): New function.
14262 (edited_line::print_diff_lines): New function.
14263 (selftest::test_applying_fixits_insert_containing_newline): New
14265 (selftest::test_applying_fixits_replace_containing_newline): New
14267 (selftest::insert_line): New function.
14268 (selftest::test_applying_fixits_multiple_lines): Add example of
14270 (selftest::edit_context_c_tests): Call the new tests.
14272 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14274 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14275 parameter cand. Update dump information.
14276 (get_computation_cost): Update uses.
14278 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14280 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14281 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14282 (get_computation_at, rewrite_use_address): Update use of
14283 get_computation_aff.
14285 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14287 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14288 (get_computation): Delete.
14289 (get_computation_cost): Implement like get_computation_cost_at.
14290 Use get_computation_at.
14291 (get_computation_cost_at): Delete.
14292 (rewrite_use_nonlinear_expr): Use get_computation_at.
14293 (rewrite_use_compare, remove_unused_ivs): Ditto.
14295 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14297 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14299 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14301 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14302 (ivopts_global_cost_for_size): Rename parameter and update uses.
14303 (iv_ca_recount_cost): Update uses.
14304 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14305 candidates seperately in n_invs and n_cands.
14306 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14308 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14310 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14311 (find_inv_vars_cb): New.
14312 (find_depends): Renamed to ...
14313 (find_inv_vars): ... this.
14314 (add_candidate_1, force_var_cost): Call find_inv_vars.
14315 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14317 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14319 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14320 inv_vars. Add inv_exprs.
14321 (struct iv_cand): Rename depends_on to inv_vars.
14322 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14323 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14324 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14325 (dump_cand): Dump inv_vars.
14326 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14327 (record_invariant, find_depends, add_candidate_1): Ditto.
14328 (set_group_iv_cost, force_var_cost): Ditto.
14329 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14330 (get_computation_cost_at, get_computation_cost): Ditto.
14331 (determine_group_iv_cost_generic): Ditto.
14332 (determine_group_iv_cost_address): Ditto.
14333 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14334 (determine_group_iv_costs): Ditto.
14335 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14336 (iv_ca_set_remove_invariants): Renamed to ...
14337 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14338 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14339 (iv_ca_set_add_invariants): Renamed to ...
14340 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14341 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14342 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14343 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14344 (create_new_ivs): Remove useless dump.
14346 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14348 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14350 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14351 (iv_ca_set_no_cp, create_new_iv): Ditto.
14353 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14355 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14357 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14359 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14360 function tree_check2.
14362 2017-05-02 Martin Liska <mliska@suse.cz>
14364 * doc/gcov.texi: Add missing preposition.
14365 * gcov.c (function_info::function_info): Properly fill up
14366 all member variables.
14368 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14370 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14372 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14374 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14376 2017-05-02 Martin Liska <mliska@suse.cz>
14379 * lto-streamer-in.c (lto_read_tree_1): Remove
14380 LTO_STREAMER_DEBUG.
14381 * lto-streamer.c (struct tree_hash_entry): Likewise.
14382 (struct tree_entry_hasher): Likewise.
14383 (tree_entry_hasher::hash): Likewise.
14384 (tree_entry_hasher::equal): Likewise.
14385 (lto_streamer_init): Likewise.
14386 (lto_orig_address_map): Likewise.
14387 (lto_orig_address_get): Likewise.
14388 (lto_orig_address_remove): Likewise.
14389 * lto-streamer.h: Likewise.
14390 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14391 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14393 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14395 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14396 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14397 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14398 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14399 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14400 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14401 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14402 (mm_maskz_sub_ss): New intrinsics.
14403 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14404 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14405 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14406 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14407 (__builtin_ia32_subss_mask_round): New builtins.
14408 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14409 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14410 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14412 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14413 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14415 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14418 2017-05-02 Martin Jambor <mjambor@suse.cz>
14420 PR tree-optimization/78687
14421 * tree-sra.c (access): New field parent.
14422 (process_subtree_disqualification): New function.
14423 (disqualify_candidate): Call it.
14424 (build_accesses_from_assign): Reset write flag if creating an
14426 (build_access_subtree): Fill in parent field and also prpagate
14427 down grp_write flag.
14428 (create_artificial_child_access): New parameter set_grp_write, set
14429 grp_write to its value.
14430 (propagate_subaccesses_across_link): Also propagate grp_write flag
14432 (propagate_all_subaccesses): Push the closest parent back to work
14433 queue if add_access_to_work_queue returned true.
14435 2017-05-02 Richard Biener <rguenther@suse.de>
14437 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14438 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14439 -fstrict-overflow documentation.
14440 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14441 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14442 flag_strict_overflow.
14443 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14444 * lto-opts.c (lto_write_options): Do not stream it.
14445 * lto-wrapper.c (merge_and_complain): Do not handle it.
14446 * opts.c (default_options_table): Do not set -fstrict-overflow.
14447 (finish_options): Likewise do not clear it when sanitizing.
14448 * simplify-rtx.c (simplify_const_relational_operation): Do not
14449 test flag_strict_overflow.
14451 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14453 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14454 using enabled attribute.
14455 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14456 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14457 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14458 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14459 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14460 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14461 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14462 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14463 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14464 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14466 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14468 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14470 2017-05-02 Richard Biener <rguenther@suse.de>
14472 PR tree-optimization/80591
14474 2017-04-10 Richard Biener <rguenther@suse.de>
14476 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14479 2017-05-02 Richard Biener <rguenther@suse.de>
14481 PR tree-optimization/80549
14482 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14483 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14484 headers if they do not have a preheader.
14486 2017-05-02 Martin Liska <mliska@suse.cz>
14489 * common.opt: Fix typo.
14490 * doc/invoke.texi: Likewise.
14492 2017-05-01 Jan Beulich <jbeulich@suse.com>
14494 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14495 swapping, add (x,x,m,x,n) alternative.
14497 2017-05-01 Nathan Sidwell <nathan@acm.org>
14499 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14500 unnecessary unadjusted_alignment check.
14502 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14505 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14507 * gimplify.c (gimplify_cilk_detach): New function.
14508 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14509 * tree-core.h: Document EXPR_CILK_SPAWN.
14510 * tree.h (EXPR_CILK_SPAWN): Define.
14512 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14514 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14515 to use new fixit_hint representation, using the "replace" logic.
14516 (get_line_span_for_fixit_hint): Likewise.
14517 (layout::print_any_fixits): Likewise.
14518 (selftest::test_one_liner_many_fixits): Rename to...
14519 (selftest::test_one_liner_many_fixits_1): ...this, and update
14520 comment and expected output to reflect that the multiple fix-it
14521 hints are now consolidated into one insertion.
14522 (selftest::test_one_liner_many_fixits_2): New test.
14523 (selftest::test_diagnostic_show_locus_one_liner): Update for
14525 (selftest::test_fixit_consolidation): Update for fix-it API
14527 * diagnostic.c (print_parseable_fixits): Likewise.
14528 * edit-context.c (edited_line::m_line_events): Convert from
14529 auto_vec <line_event *> to auto_vec <line_event>.
14530 (class line_event): Convert from abstract base class to a concrete
14531 class, taking over the role of replace_event.
14532 (class insert_event): Delete.
14533 (class replace_event): Rename to class line_event. Convert to
14535 (edit_context::add_fixits): Reimplement.
14536 (edit_context::apply_insert): Delete.
14537 (edit_context::apply_replace): Rename to...
14538 (edit_context::apply_fixit): ...this. Convert to half-open range.
14539 (edited_file::apply_insert): Delete.
14540 (edited_file::apply_replace): Rename to...
14541 (edited_file::apply_fixit): ...this.
14542 (edited_line::~edited_line): Drop deletion of events.
14543 (edited_line::apply_insert): Delete.
14544 (edited_line::apply_replace): Rename to...
14545 (edited_line::apply_fixit): ...this. Convert to half-open range.
14546 Update for change to type of m_line_events.
14547 * edit-context.h (edit_context::apply_insert): Delete.
14548 (edit_context::apply_replace): Rename to...
14549 (edit_context::apply_fixit): ...this.
14551 2017-05-01 Martin Sebor <msebor@redhat.com>
14553 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14556 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14559 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14560 __get_cpuid_max returns 0.
14561 (__get_cpuid_count): Ditto.
14563 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14565 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14566 replacement expression is another instance of one of its arguments.
14568 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14571 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14572 check for stack push/pop autoinc.
14573 * config/i386/i386.c (ix86_agi_dependent): Return false
14574 if the only reason why modified_in_p returned true is that
14575 addr is SP based and set_insn is a push or pop.
14577 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14579 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14582 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14585 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14586 (account_size_time): Use two predicates - exec_pred and
14588 (evaluate_conditions_for_known_args): Compute both clause and
14590 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14591 (inline_summary_t::duplicate): Update.
14592 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14594 (compute_inline_parameters): Likewise.
14595 (estimate_edge_size_and_time): Update caluclation of time.
14596 (estimate_node_size_and_time): Compute both time and nonspecialized
14598 (estimate_ipcp_clone_size_and_time): Update.
14599 (inline_merge_summary): Update.
14600 (do_estimate_edge_time): Update.
14601 (do_estimate_edge_size): Update.
14602 (do_estimate_edge_hints): Update.
14603 (inline_read_section, inline_write_summary): Stream both new predicates.
14604 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14606 (compute_inlined_call_time): Cleanup.
14607 (big_speedup_p): Update.
14608 (edge_badness): Update.
14609 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14610 (size_time_entry): Replace predicate by exec_predicate and
14611 nonconst_predicate.
14612 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14613 (estimate_edge_time): Return also nonspec_time.
14614 (reset_edge_growth_cache): Update.
14616 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14618 PR rtl-optimization/80491
14619 * ifcvt.c (noce_process_if_block): When looking for x setter
14620 with missing else_bb, don't check only the insn right before
14621 cond_earliest, but look for the last insn that x is modified in
14622 within the same bb.
14624 PR rtl-optimization/80491
14625 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14627 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14629 PR tree-optimization/80487
14630 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14632 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14634 PR tree-optimization/79697
14635 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14636 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14637 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14639 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14640 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14642 2017-04-28 Martin Sebor <msebor@redhat.com>
14644 PR tree-optimization/80523
14645 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14646 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14648 (maybe_warn, format_directive, parse_directive): Use new functions.
14649 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14651 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14653 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14655 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14657 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14658 target_header_dir): Set correctly.
14659 * configure: Regenerated.
14660 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14661 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14662 instead of SYSTEM_HEADER_DIR.
14664 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14666 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14667 (estimate_local_effects): Likewise.
14668 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14669 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14670 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14671 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14672 do_estimate_edge_time, estimate_edge_time): Likewise.
14673 * ipa-inline-analysis.c (estimate_node_size_and_time,
14674 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14675 (estimate_time_after_inlining): Remove.
14677 2017-04-28 Martin Liska <mliska@suse.cz>
14679 * doc/gcov.texi: Enhance documentation of gcov.
14681 2017-04-28 Martin Liska <mliska@suse.cz>
14683 * doc/gcov.texi: Sort options in alphabetic order.
14684 * doc/gcov-dump.texi: Likewise.
14685 * doc/gcov-tool.texi: Likewise.
14686 * gcov.c (print_usage): Likewise.
14687 * gcov-dump.c (print_usage): Likewise.
14688 * gcov-tool.c (print_merge_usage_message): Likewise.
14689 (print_rewrite_usage_message): Likewise.
14690 (print_overlap_usage_message): Likewise.
14692 2017-04-28 Martin Liska <mliska@suse.cz>
14694 PR gcov-profile/53915
14695 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14697 2017-04-28 Martin Liska <mliska@suse.cz>
14699 PR gcov-profile/79891
14700 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14701 is marked by compiler as living on a line.
14702 (get_cycles_count): Remove usage of the union.
14703 (output_intermediate_file): Likewise.
14704 (find_source): Fix GNU coding style.
14705 (accumulate_line_counts): Remove old non-all block mode.
14706 (output_lines): Remove usage of the union.
14707 * profile.c (output_location): Include all BBs, even if
14708 belonging to a same line (and file) as a previous BB.
14710 2017-04-28 Martin Liska <mliska@suse.cz>
14712 * gcov.c (process_args): Handle new argument 'w'.
14713 (read_graph_file): Assign ID to BBs.
14714 (output_branch_count): Display BB # if verbose flag is set.
14715 (output_lines): Likewise for arcs.
14716 (print_usage): Add '--verbose' option help.
14717 * doc/gcov.texi: Document --verbose (-w) option.
14719 2017-04-28 Martin Liska <mliska@suse.cz>
14721 * gcov.c (struct block_location_info): New struct.
14722 (process_file): Fill up the new structure.
14723 (read_graph_file): Replace usage of encoding by the newly added
14725 (add_line_counts): Likewise.
14726 (accumulate_line_counts): Remove usage of the union.
14727 (function_info::function_info): New function.
14728 (function_info::~function_info): Likewise.
14729 (process_file): Call delete instead of release_function.
14730 (release_function): Release the function.
14731 (release_structures): Call delete instead of release_function.
14732 (solve_flow_graph): Replace usage of num_blocks.
14733 (find_exception_blocks): Likewise.
14734 (output_lines): Fix GNU coding style.
14736 2017-04-28 Martin Liska <mliska@suse.cz>
14739 * coverage.c (coverage_remove_note_file): New function.
14740 * coverage.h: Declare the function.
14741 * toplev.c (finalize): Clean if an error has been seen.
14743 2017-04-28 Martin Liska <mliska@suse.cz>
14745 PR gcov-profile/80031
14746 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14747 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14748 * gcov.c (read_graph_file): Read just number of blocks.
14749 * profile.c (branch_prob): Do not stream 0 flags per a basic
14752 2017-04-28 Martin Liska <mliska@suse.cz>
14754 * gcov-dump.c (tag_*): Add new argument to declarations.
14755 (dump_gcov_file): Likewise.
14756 (tag_blocks): Add and use new argument depth.
14757 (tag_arcs): Likewise.
14758 (tag_lines): Likewise.
14759 (tag_counters): Likewise.
14760 (tag_summary): Likewise.
14761 (dump_working_sets): Use depth to do a proper indentation.
14763 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14766 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14767 * symtab.c (symtab_node::debug_symtab): Move definition here.
14769 2017-04-28 Richard Biener <rguenther@suse.de>
14771 * lto-streamer.h (LTO_major_version): Bump to 7.
14773 2017-04-28 Richard Biener <rguenther@suse.de>
14775 * tree-vrp.c (assert_info): New struct.
14776 (add_assert_info): New helper.
14777 (register_edge_assert_for_2): Refactor to add asserts to a vector
14779 (register_edge_assert_for_1): Likewise.
14780 (register_edge_assert_for): Likewise.
14781 (finish_register_edge_assert_for): New helper actually registering
14782 asserts where live on edge.
14783 (find_conditional_asserts): Adjust.
14784 (find_switch_asserts): Likewise.
14785 (evrp_dom_walker::try_find_new_range): Generalize.
14786 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14788 2017-04-27 Marek Polacek <polacek@redhat.com>
14791 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14792 arg10 and arg11 to itype.
14794 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14796 * doc/extend.texi (Object Size Checking): Improve grammar.
14798 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14801 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14802 that the logic for permitting reciprocal estimates matches that
14805 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14808 * tree.c (type_cache_hasher::equal): Only compare
14809 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14810 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14811 non-aggregate element types.
14812 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14813 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14815 2017-04-27 Richard Biener <rguenther@suse.de>
14817 PR middle-end/80533
14818 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14819 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14820 keeping a reference to a trailing array.
14822 2017-04-27 Richard Biener <rguenther@suse.de>
14824 PR middle-end/80539
14825 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
14826 being in loop-closed SSA form conservatively.
14827 (chrec_fold_multiply_poly_poly): Likewise.
14829 2017-04-27 Tamar Christina <tamar.christina@arm.com>
14831 PR middle-end/79665
14832 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
14833 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
14835 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14838 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
14839 (aarch64_function_arg_alignment): Return unsigned int again, but still
14840 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
14841 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14842 Don't emit -Wpsabi note.
14843 (aarch64_function_arg_boundary): Likewise.
14844 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14847 2017-04-26 Nathan Sidwell <nathan@acm.org>
14849 * tree.h (crc32_unsigned_n): Declare.
14850 (crc32_unsigned, crc32_unsigned): Make inline.
14851 * tree.c (crc32_unsigned_bits): Replace with ...
14852 (crc32_unsigned_n): ... this.
14853 (crc32_unsigned, crc32_byte): Remove.
14854 (crc32_string): Remove unnecessary braces.
14856 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14858 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
14859 * ipa-inline-analysis.c (MAX_TIME): Remove.
14860 (account_size_time): Use sreal for time.
14861 (dump_inline_summary): Update.
14862 (estimate_function_body_sizes): Update.
14863 (estimate_edge_size_and_time): Update.
14864 (estimate_calls_size_and_time): Update.
14865 (estimate_node_size_and_time): Update.
14866 (inline_merge_summary): Update.
14867 (inline_update_overall_summary): Update.
14868 (estimate_time_after_inlining): Update.
14869 (inline_read_section): Update.
14870 (inline_write_summary): Update.
14871 * ipa-inline.c (compute_uninlined_call_time): Update.
14872 (compute_inlined_call_time): Update.
14873 (recursive_inlining): Update.
14874 (inline_small_functions): Update.
14875 (dump_overall_stats): Update.
14876 * ipa-inline.h: Include sreal.h.
14877 (size_time_entry): Turn time to sreal.
14878 (inline_summary): Turn self_time nad time to sreal.
14880 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14882 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
14884 (sreal::stream_out, sreal::stream_in): New.
14885 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
14887 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14889 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
14892 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
14895 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
14897 (dimode_scalar_chain::compute_convert_gain): Ditto.
14898 (dimode_scalar_chain::make_vector_copies): Ditto.
14899 (dimode_scalar_chain::convert_reg): Ditto.
14900 (dimode_scalar_chain::convert_insn): Ditto.
14901 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
14902 (VI248_AVX512BW_1): New mode iterator.
14903 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
14904 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
14907 2017-04-25 Martin Sebor <msebor@redhat.com>
14909 PR tree-optimization/80497
14910 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
14911 constants are representable in HOST_WIDE_INT.
14912 (parse_directive): Ditto.
14914 2017-04-25 Martin Sebor <msebor@redhat.com>
14917 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
14918 (new_zero_array): Adjust signature.
14919 (dom_info::dom_init): Used unsigned rather that size_t.
14920 (dom_info::dom_info): Same.
14922 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14923 Jakub Jelinek <jakub@redhat.com>
14926 * config/arm/arm.c: Include gimple.h.
14927 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14928 returns negative, increment ncrn only if it returned positive.
14929 (arm_needs_doubleword_align): Return int instead of bool,
14930 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
14931 members, but if there is any such non-FIELD_DECL
14932 > PARM_BOUNDARY aligned decl, return -1 instead of false.
14933 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14934 returns negative, increment nregs only if it returned positive.
14935 (arm_setup_incoming_varargs): Likewise.
14936 (arm_function_arg_boundary): Emit -Wpsabi note if
14937 arm_needs_doubleword_align returns negative, return
14938 DOUBLEWORD_ALIGNMENT only if it returned positive.
14940 2017-04-25 Marek Polacek <polacek@redhat.com>
14943 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
14944 first argument to type.
14946 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
14949 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
14950 type checks to test for compatibility instead of equality.
14952 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14953 Jakub Jelinek <jakub@redhat.com>
14956 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
14958 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
14959 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
14960 the alignment computation, but return their maximum in warn_alignment.
14961 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14962 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
14964 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
14965 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14968 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14970 * config/arc/simdext.md (dmpyh): Fix typo.
14972 2017-04-25 Richard Biener <rguenther@suse.de>
14974 PR tree-optimization/80492
14975 * alias.c (compare_base_decls): Handle registers with asm
14976 specification conservatively.
14977 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14978 compare_base_decls returning dont-know properly.
14980 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14982 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
14983 (legitimate_offset_address_p): New function.
14984 (arc_legitimate_address_p): Use above function.
14986 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14988 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
14990 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14992 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
14993 ACCH registers whenever they are available.
14995 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14997 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
14998 double regs fix when not used.
15000 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15002 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15004 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15005 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15007 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15009 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15010 when emitting short ADD instructions.
15012 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15014 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15016 (cmpsi_cc_c_insn): Likewise.
15017 (cbranchsi4_scratch): Compute proper instruction length using
15018 compact_hreg_operand.
15019 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15021 2017-04-25 Richard Biener <rguenther@suse.de>
15023 PR middle-end/80509
15024 * passes.c (pass_manager::pass_manager): Initialize
15025 m_name_to_pass_map.
15027 2017-04-25 Richard Biener <rguenther@suse.de>
15029 PR tree-optimization/79201
15030 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15032 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15035 * config/s390/vector.md: Split MEM->GPR vector moves for
15036 non-s_operand addresses.
15038 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15041 * config/s390/predicates.md (reload_const_wide_int_operand): New
15043 * config/s390/s390.md ("movti"): Remove d/P alternative.
15044 ("movti_bigconst"): New pattern definition.
15046 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15049 * s390-protos.h (s390_expand_cs_hqi): Removed.
15050 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15051 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15052 modes as well as CCZ1mode and CCZmode.
15053 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15054 signature of s390_emit_compare_and_swap.
15055 (s390_expand_cs_hqi): Likewise, make static.
15056 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15057 compare-and-swap, in some cases.
15058 (s390_expand_cs): Wrapper function.
15059 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15061 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15062 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15063 patterns for small and large integers. Forbid symref memory operands.
15064 Move expander to s390.c. Require cc register.
15065 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15066 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15067 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15068 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15069 symref memory operands. Remove CC mode and call s390_match_ccmode
15071 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15073 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15075 * config/s390/s390.md (define_peephole2): New peephole to help
15076 combining the load-and-test pattern with volatile memory.
15078 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15080 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15081 with CCZmode for TARGET_Z196.
15083 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15085 PR rtl-optimization/80501
15086 * combine.c (make_compound_operation_int): Set subreg_code to SET
15087 even for AND with mask of the sign bit of mode.
15089 PR rtl-optimization/80500
15090 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15091 sum's initial value.
15093 2017-04-25 Julian Brown <julian@codesourcery.com>
15094 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15096 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15098 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15100 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15102 2017-04-25 Julian Brown <julian@codesourcery.com>
15103 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15105 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15106 (thunderx2t99_sha): New Reservation.
15108 2017-04-25 Julian Brown <julian@codesourcery.com>
15109 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15111 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15112 type for 1-element load.
15114 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15116 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15118 2017-04-24 Martin Jambor <mjambor@suse.cz>
15120 PR tree-optimization/80293
15121 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15122 char arrays not totally scalarizable if it is false.
15123 (analyze_all_variable_accesses): Pass correct value in the new
15124 parameter. Add a statistics counter.
15126 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15128 PR middle-end/79931
15129 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15131 2017-04-24 Richard Biener <rguenther@suse.de>
15133 PR tree-optimization/80494
15134 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15135 out for complex types.
15137 2017-04-24 Richard Biener <rguenther@suse.de>
15139 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15140 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15141 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15142 (DFS): Adjust and never fail.
15143 (sccvn_dom_walker::fail): Remove.
15144 (sccvn_dom_walker::before_dom_children): Adjust.
15145 (run_scc_vn): Likewise and never fail.
15146 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15147 (pass_fre::execute): Likewise.
15149 2017-04-24 Richard Biener <rguenther@suse.de>
15151 PR tree-optimization/79725
15152 * tree-ssa-sink.c (statement_sink_location): Return whether
15153 failure reason was zero uses. Move that check later.
15154 (sink_code_in_bb): Deal with zero uses by removing the stmt
15157 2017-04-24 Richard Biener <rguenther@suse.de>
15160 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15161 pointer-based references.
15163 2017-04-24 Richard Biener <rguenther@suse.de>
15166 * pass_manager.h (pass_manager::operator new): Remove.
15167 (pass_manager::operator delete): Likewise.
15168 * passes.c (pass_manager::operator new): Remove.
15169 (pass_manager::operator delete): Likewise.
15170 (pass_manager::pass_manager): Zero individual pass members.
15172 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15175 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15176 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15177 Check "XEXP (src, 1)" operand here.
15178 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15179 Check "XEXP (src, 1)" operand here.
15180 (dimode_scalar_chain::make_vector_copies): Detect count register
15181 of a shift instruction. Zero extend count register from QImode
15182 to DImode to satisfy vector shift pattern count operand predicate.
15183 Substitute vector shift count operand with a DImode copy.
15184 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15187 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15189 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15190 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15191 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15192 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15193 (UNSPEC_NOREX_MEM): Remove definition.
15195 2017-04-21 Richard Biener <rguenther@suse.de>
15197 PR tree-optimization/79547
15198 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15199 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15200 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15201 without any constraints.
15203 2017-04-21 Richard Biener <rguenther@suse.de>
15205 PR tree-optimization/78847
15206 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15208 2017-04-21 Richard Biener <rguenther@suse.de>
15210 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15211 (build_distinct_type_copy): Likewise.
15212 (build_variant_type_copy): Likewise.
15213 * tree.c (build_qualified_type): Pass down mem-stat info.
15214 (build_distinct_type_copy): Likewise.
15215 (build_variant_type_copy): Likewise.
15217 2017-04-21 Richard Biener <rguenther@suse.de>
15219 PR tree-optimization/80237
15220 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15222 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15223 for a simplified result.
15225 2016-04-21 Richard Biener <rguenther@suse.de>
15227 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15228 sth as strict as a simple_iv but a chrec without symbols and an
15229 operand defined in the loop we are peeling (and not some subloop).
15230 (propagate_constants_for_unrolling): Propagate all constants.
15232 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15235 * config/i386/i386.c (print_reg): Remove assert for disalowed
15236 regno values, call output_operand_lossage instead.
15238 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15241 * config/i386/constraints.md (Yc): New register constraint.
15242 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15243 Use Yc constraint for alternative 2 of operand 0. Remove
15244 preferred_for_speed attribute.
15246 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15248 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15249 lastprivate clauses in SIMT case.
15251 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15253 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15255 2017-04-20 Richard Biener <rguenther@suse.de>
15257 PR tree-optimization/57796
15258 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15259 as N scalar stores.
15260 (vect_model_load_cost): Cost gathers as N scalar loads.
15262 2017-04-20 Richard Biener <rguenther@suse.de>
15264 * ggc-page.c (ggc_allocated_p): Rename to ...
15265 (safe_lookup_page_table_entry): ... this and return the lookup
15267 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15269 2017-04-20 Richard Biener <rguenther@suse.de>
15271 PR tree-optimization/80453
15272 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15273 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15274 from the conditions.
15275 (vn_phi_eq): Pass them down.
15276 (vn_phi_lookup): Record them.
15277 (vn_phi_insert): Likewise.
15279 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15281 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15282 uninitialized variable warning to avoid buffer overrun.
15284 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15287 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15288 is suppressed for '{ 0 }' in C.
15290 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15292 * BASE-VER: Set to 8.0.0.
15294 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15296 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15297 priority .init_array and .fini_array section with SECTION_NOTYPE
15300 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15302 PR middle-end/80423
15303 * tree.h (build_array_type): Add typeless_storage default argument.
15304 * tree.c (type_cache_hasher::equal): Also compare
15305 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15306 (build_array_type): Add typeless_storage argument, set
15307 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15309 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15310 (build_array_type): Likewise. Add typeless_storage argument.
15312 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15313 Jakub Jelinek <jakub@redhat.com>
15315 PR tree-optimization/80426
15316 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15317 operation on symbolic operands, also compute the overflow for the
15318 invariant part when the operation degenerates into a negation.
15320 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15323 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15324 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15327 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15329 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15332 * config/avr/avr.c (tree.h): Include it.
15333 (cgraph.h): Include it.
15334 (avr_encode_section_info): Don't warn for uninitialized progmem
15335 variable if it's just an alias.
15337 2017-04-19 Richard Biener <rguenther@suse.de>
15340 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15341 when needed by AutoPGO.
15343 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15346 * doc/lto.texi: Remove an extra 'that'.
15348 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15350 PR rtl-optimization/80429
15351 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15352 are only used in debug insns.
15354 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15355 Vladimir Makarov <vmakarov@redhat.com>
15357 * config/sparc/predicates.md (input_operand): Add comment. Return
15358 true for any memory operand when LRA is in progress.
15359 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15361 2017-04-18 Jeff Law <law@redhat.com>
15364 * mips.md ({return,simple_return}_internal): Do not overwrite
15367 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15369 PR tree-optimization/80443
15370 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15371 instead of adding 1, subtract -1 and similarly instead of subtracting
15374 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15376 PR rtl-optimization/80357
15377 * haifa-sched.c (tmp_bitmap): New variable.
15378 (model_recompute): Handle duplicate use records.
15379 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15380 (free_global_sched_pressure_data): Free it.
15382 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15385 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15386 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15387 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15388 instead of SYSTEM_HEADER_DIR.
15390 2017-04-18 Jeff Law <law@redhat.com>
15392 PR middle-end/80422
15393 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15394 predecessors after walking up the insn chain.
15396 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15399 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15400 sizetype type into debug info.
15402 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15405 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15406 unneeded test for TARGET_UPPER_REGS_SF.
15407 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15409 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15412 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15413 instead of gsi_after_labels.
15415 2017-04-18 Jeff Law <law@redhat.com>
15417 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15421 2017-04-13 Jeff Law <law@redhat.com>
15422 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15423 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15425 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15428 * config/avr/avr.c (intl.h): Include it.
15429 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15431 2017-04-18 Martin Liska <mliska@suse.cz>
15433 PR gcov-profile/78783
15434 * gcov-tool.c (gcov_output_files): Validate that destination
15435 file is either removed by the tool or by a user.
15437 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15438 Guy Benyei <guybe@mellanox.com>
15440 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15441 block, and do not negate it, the stored id is already negative.
15443 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15445 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15447 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15450 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15451 masks of options that should be turned off if the VSX vector
15452 options are turned off.
15453 (OTHER_P8_VECTOR_MASKS): Likewise.
15454 (OTHER_VSX_VECTOR_MASKS): Likewise.
15455 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15456 rs6000_disable_incompatible_switches to validate no type switches
15458 (rs6000_incompatible_switch): New function to disallow turning on
15459 other vector options if -mno-vsx, -mno-power8-vector, or
15460 -mno-power9-vector are specified.
15462 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15464 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15466 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15468 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15469 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15470 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15471 (ARG_POINTER_CFA_OFFSET): Likewise.
15473 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15475 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15476 conditions to take advantage of various optimizations.
15478 2017-04-13 Jeff Law <law@redhat.com>
15480 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15481 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15482 (zero_extendsidi2_dext): Likewise.
15484 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15487 * fold-const.c (fold_ternary_loc): Revert
15488 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15491 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15493 PR rtl-optimization/80343
15494 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15495 new scratch pseudo.
15497 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15500 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15501 to ubsan_encode_value.
15503 2017-04-13 Jeff Law <law@redhat.com>
15505 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15506 appearing in DEBUG_INSNs.
15508 2017-04-13 Martin Liska <mliska@suse.cz>
15510 PR gcov-profile/80413
15511 * gcov-io.c (gcov_write_string): Copy to buffer just when
15512 allocated size is greater than zero.
15514 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15517 * dwarf2out.c (decls_for_scope): Ignore declarations of
15518 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15520 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15523 * ipa-visibility.c (non_local_p): Fix typos.
15524 (localize_node): When localizing symbol in same comdat group,
15525 dissolve the group only when we know external symbols are going
15527 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15529 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15531 PR tree-optimization/79390
15532 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15533 order does not result in usable sequence, retry with reversed operand
15539 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15540 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15541 op0 instead of fold_convert_loc (loc, type, arg0).
15543 2017-04-12 Jeff Law <law@redhat.com>
15545 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15546 has a delay slot in the generated code.
15548 * config/cris/cris.md (cris_preferred_reload_class): Return
15549 GENNONACR_REGS rather than GENERAL_REGS.
15551 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15554 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15555 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15556 signedness of the result type.
15558 2017-04-12 Richard Biener <rguenther@suse.de>
15559 Jeff Law <law@redhat.com>
15561 PR tree-optimization/80359
15562 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15563 trim stores to TARGET_MEM_REFs.
15565 2017-04-12 Richard Biener <rguenther@suse.de>
15567 PR tree-optimization/79390
15568 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15569 threading case even more.
15571 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15574 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15575 for quad_address_p for TImode, instead of just not indexed_address.
15577 2017-04-12 Richard Biener <rguenther@suse.de>
15578 Bernd Edlinger <bernd.edlinger@hotmail.de>
15580 PR middle-end/79671
15581 * alias.c (component_uses_parent_alias_set_from): Handle
15582 TYPE_TYPELESS_STORAGE.
15583 (get_alias_set): Likewise.
15584 * tree-core.h (tree_type_common): Add typeless_storage flag.
15585 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15586 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15587 for types containing members with TYPE_TYPELESS_STORAGE.
15588 (place_field): Likewise.
15589 (layout_type): Likewise for ARRAY_TYPE.
15590 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15591 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15592 TYPE_TYPELESS_STORAGE.
15593 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15595 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15598 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15599 first argument to type.
15601 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15605 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15606 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15607 (rs6000_expand_binop_builtin): Likewise.
15608 (rs6000_expand_ternop_builtin): Likewise; also add missing
15609 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15610 vshasigma built-ins.
15611 * doc/extend.texi: Document that vec_xxpermdi's third argument
15612 must be a constant.
15614 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15616 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15617 Use shift_const cost parameter when calculating gain of STV shifts.
15619 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15621 PR rtl-optimization/70478
15622 * lra-constraints.c (process_alt_operands): Check memory for
15623 disfavoring memory insn operand.
15625 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15627 PR middle-end/80100
15628 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15629 left shift in unsigned HOST_WIDE_INT type.
15631 PR rtl-optimization/80385
15632 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15633 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15636 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15637 if they have any depend clauses.
15639 2017-04-11 Martin Liska <mliska@suse.cz>
15642 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15643 * ipa-split.c (split_function): Create a local comdat symbol
15644 if caller is in a comdat group.
15646 2017-04-11 Martin Liska <mliska@suse.cz>
15649 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15652 2017-04-11 Martin Sebor <msebor@redhat.com>
15654 PR middle-end/80364
15655 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15656 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15658 (directive::set_width, directive::set_precision, format_character):
15660 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15663 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15666 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15667 conflict, set target->arch_name instead of target->cpu_name.
15669 2017-04-11 Richard Biener <rguenther@suse.de>
15671 PR tree-optimization/80374
15672 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15673 build_zero_cst, remove fold_convertible_p check again.
15675 2017-04-11 Martin Liska <mliska@suse.cz>
15678 * ubsan.c (instrument_object_size): Do not instrument register
15681 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15684 * config/i386/i386-builtin-types.def
15685 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15686 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15687 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15688 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15689 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15690 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15691 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15692 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15693 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15694 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15695 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15696 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15697 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15698 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15699 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15700 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15701 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15702 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15703 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15704 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15705 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15706 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15707 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15708 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15709 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15710 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15711 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15712 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15713 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15714 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15715 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15716 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15717 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15718 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15719 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15720 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15721 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15722 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15723 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15724 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15725 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15726 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15727 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15728 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15729 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15730 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15732 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15733 flag to second_arg_count, handle 4 argument function type _COUNT
15734 aliases, handle second_arg_count on second argument rather than last.
15736 2017-04-10 Jeff Law <law@redhat.com>
15738 PR tree-optimization/80374
15739 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15740 record anything if we can not convert integer_zero_node to the
15743 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15746 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15747 Enhance special handling given to the TARGET_P9_MINMAX option in
15748 relation to certain other options.
15750 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15752 PR tree-optimization/80153
15753 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15754 remove POINTER_PLUS_EXPR's base part directly, rather than through
15757 2017-04-10 Richard Biener <rguenther@suse.de>
15758 Bin Cheng <bin.cheng@arm.com>
15760 PR tree-optimization/80153
15761 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15762 the first element of pointer type aff_tree. Build result expr in
15764 (add_elt_to_tree): Convert to type unconditionally. Remove other
15765 fold_convert calls.
15766 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15767 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15769 2017-04-10 Richard Biener <rguenther@suse.de>
15771 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15774 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15776 PR rtl-optimization/70478
15777 * lra-constraints.c (curr_small_class_check): New.
15778 (update_and_check_small_class_inputs): New.
15779 (process_alt_operands): Update curr_small_class_check. Disfavor
15780 alternative insn memory operands. Check available regs for small
15783 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15786 * config/mips/mips.opt (-mvirt): Update description.
15787 * doc/invoke.texi (-mvirt): Likewise.
15789 2017-04-10 Richard Biener <rguenther@suse.de>
15791 PR middle-end/80362
15792 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15793 looking for NEGATE_EXPR in -A / -B to A / B folding.
15795 2017-04-10 Martin Liska <mliska@suse.cz>
15797 PR gcov-profile/80224
15798 * gcov.c (print_usage): Fix usage string.
15799 (get_gcov_intermediate_filename): Remove.
15800 (output_gcov_file): Use both for normal and intermediate format.
15801 (generate_results): Do not initialize special file for
15802 intermediate format.
15804 2017-04-10 Richard Biener <rguenther@suse.de>
15806 PR tree-optimization/80304
15807 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15810 2017-04-10 Nathan Sidwell <nathan@acm.org>
15813 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15814 (rs6000_init_builtins): Use it.
15816 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15818 * config/arm/arm.md (<mrc>): Add mode to SET source.
15819 (<mrrc>): Likewise.
15821 2017-04-10 Richard Biener <rguenther@suse.de>
15823 PR middle-end/80344
15824 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
15826 2017-04-10 Jakub Jelinek <jakub@redhat.com>
15829 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
15830 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
15831 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
15832 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
15833 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
15834 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
15835 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
15836 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
15837 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
15838 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
15839 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
15840 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
15841 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
15842 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
15843 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
15844 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
15845 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
15846 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
15847 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
15848 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
15849 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
15850 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
15851 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
15853 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
15855 PR rtl-optimization/70478
15856 * lra-constraints.c: Reverse the last patch.
15858 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
15860 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
15861 Add comment for WCHAR_T.
15863 2017-04-08 Martin Liska <mliska@suse.cz>
15866 2017-04-07 Martin Liska <mliska@suse.cz>
15869 * ipa-split.c (split_function): Add function part to a same comdat
15872 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15875 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
15877 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
15879 * rs6000/rs6000.c (vec_load_pendulum): Rename...
15880 (vec_pairing): ...to this.
15881 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
15882 (rs6000_sched_init): Adjust for name change.
15883 (struct rs6000_sched_context): Likewise.
15884 (rs6000_init_sched_context): Likewise.
15885 (rs6000_set_sched_context): Likewise.
15887 2017-04-07 Jakub Jelinek <jakub@redhat.com>
15893 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
15895 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
15896 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
15897 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
15899 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
15901 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
15903 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15905 PR rtl-optimization/70703
15906 * ira-color.c (update_conflict_hard_regno_costs): Use
15907 int64_t instead of HOST_WIDE_INT.
15909 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15911 PR rtl-optimization/70478
15912 * lra-constraints.c (process_alt_operands): Disfavor alternative
15913 insn memory operands.
15915 2017-04-07 Jeff Law <law@redhat.com>
15917 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
15918 CALL and NOTE_INSN_CALL_ARG_LOCATION.
15920 2017-04-07 Martin Liska <mliska@suse.cz>
15923 * config/aarch64/aarch64.c (aarch64_process_target_attr):
15924 Show error message instead of an ICE.
15926 2017-04-07 Martin Liska <mliska@suse.cz>
15929 * ipa-split.c (split_function): Add function part to a same comdat
15932 2017-04-07 Richard Biener <rguenther@suse.de>
15934 PR middle-end/80341
15935 * tree.c (get_unwidened): Also handle ! for_type case for
15937 * convert.c (do_narrow): Split out from ...
15938 (convert_to_integer_1): ... here. Do not pass final truncation
15939 type to get_unwidened for TRUNC_DIV_EXPR.
15941 2017-04-07 Richard Biener <rguenther@suse.de>
15943 * tree-affine.c (wide_int_ext_for_comb): Take type rather
15945 (aff_combination_const): Adjust.
15946 (aff_combination_scale): Likewise.
15947 (aff_combination_add_elt): Likewise.
15948 (aff_combination_add_cst): Likewise.
15949 (aff_combination_convert): Likewise.
15950 (add_elt_to_tree): Likewise. Remove unused argument.
15951 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
15953 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
15955 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
15957 * config/arm/arm.c (arm_default_short_enums): Use
15958 ARM_DEFAULT_SHORT_ENUMS.
15959 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
15961 2017-04-06 Jakub Jelinek <jakub@redhat.com>
15964 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
15965 members with redundant out-of-class redeclaration.
15967 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15970 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
15971 * config/i386/i386.md (*zero_extendsidi2):
15972 Add (?*x,*x) and (?*v,*v) alternatives.
15974 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15977 * config/i386/i386.c (ix86_expand_builtin)
15978 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
15979 mode from insn data. Convert operands to insn operand mode.
15980 Copy operands that don't satisfy insn predicate to a register.
15982 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
15984 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
15987 2017-04-06 Richard Biener <rguenther@suse.de>
15989 PR tree-optimization/80334
15990 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
15991 preserve alignment of accesses.
15993 2017-04-06 Richard Biener <rguenther@suse.de>
15995 PR tree-optimization/80262
15996 * tree-sra.c (build_ref_for_offset): Preserve address-space
15998 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
15999 Drop useless address-space information on MEM_REF offsets.
16001 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16003 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16005 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16007 PR rtl-optimization/70703
16008 * ira-color.c (update_conflict_hard_regno_costs): Use
16009 HOST_WIDE_INT instead of long.
16011 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16014 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16015 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16017 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16018 for x86_64 target. Handle -m3dnowa option.
16020 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16022 PR rtl-optimization/70703
16023 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16024 (update_conflict_hard_regno_costs): Use long instead of unsigned
16025 arithmetic for cost calculation.
16027 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16028 Bernd Edlinger <bernd.edlinger@hotmail.de>
16031 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16034 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16037 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16038 ptr_mode with Pmode throughout.
16039 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16040 into probe_stack_range and use DImode.
16042 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16045 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16046 call_eh_return is true.
16048 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16050 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16051 Initialize last_match_fntype_index.
16053 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16056 * tree-nvr.c: Include internal-fn.h.
16057 (pass_return_slot::execute): Ignore internal calls without
16060 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16061 Richard Biener <rguenther@suse.de>
16064 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16065 captures used multiple times, except for the last use.
16066 * generic-match-head.c: Include gimplify.h.
16068 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16070 PR tree-optimization/79390
16071 * target.h (struct noce_if_info): Declare.
16072 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16073 * target.def (noce_conversion_profitable_p): New target hook.
16074 * ifcvt.h (struct noce_if_info): New type, moved from ...
16075 * ifcvt.c (struct noce_if_info): ... here.
16076 (noce_conversion_profitable_p): Renamed to ...
16077 (default_noce_conversion_profitable_p): ... this. No longer
16079 (noce_try_store_flag_constants, noce_try_addcc,
16080 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16081 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16082 instead of noce_conversion_profitable_p.
16083 * config/i386/i386.c: Include ifcvt.h.
16084 (ix86_option_override_internal): Don't override
16085 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16086 (ix86_noce_conversion_profitable_p): New function.
16087 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16088 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16089 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16090 * doc/tm.texi: Regenerated.
16092 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16094 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16097 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16100 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16101 instructions for small multiply cores.
16103 2017-04-04 Jeff Law <law@redhat.com>
16105 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16107 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16108 that are not set by the loop over the elements.
16110 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16113 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16114 int mode, convert_modes it to mode as unsigned, otherwise use
16115 lowpart_subreg to mode rather than SImode.
16116 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16117 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16118 Use DImode instead of SImode for the shift count operand.
16119 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16122 2017-04-04 Richard Biener <rguenther@suse.de>
16124 PR middle-end/80281
16125 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16126 arithmetic done for the negate or the plus. Simplify.
16127 (A - (-B) -> A + B): Likewise.
16128 * fold-const.c (split_tree): Make sure to not negate pointers.
16130 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16132 PR rtl-optimization/60818
16133 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16134 a compare of comparisons with the thing compared if this results
16135 in a different machine mode.
16137 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16139 * alias.c (base_alias_check): Fix typo in comment.
16140 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16141 * cgraphunit.c (symbol_table::compile): Likewise.
16142 * collect2.c (maybe_run_lto_and_relink): Likewise.
16143 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16144 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16145 * config/avr/avr.c (avr_map_op_t): Likewise.
16146 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16147 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16148 * config/epiphany/epiphany.md (movcc): Likewise.
16149 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16150 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16152 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16153 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16154 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16155 * config/sh/sh.c (sh_rtx_costs): Likewise.
16156 * fold-const.c (fold_truth_andor): Likewise.
16157 * genautomata.c (collapse_flag): Likewise.
16158 * gengtype.h (struct type::u::s): Likewise.
16159 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16160 * input.c (FORMAT_AMOUNT): Likewise.
16161 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16162 (known_aggs_to_agg_replacement_list): Likewise.
16163 * ipa-inline-analysis.c: Likewise.
16164 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16165 * ipa-polymorphic-call.c
16166 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16167 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16168 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16170 * modulo-sched.c (apply_reg_moves): Likewise.
16171 * omp-expand.c (build_omp_regions_1): Likewise.
16172 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16173 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16174 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16175 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16176 * value-prof.c: Likewise.
16177 * var-tracking.c (val_reset): Likewise.
16179 2017-04-03 Richard Biener <rguenther@suse.de>
16181 PR tree-optimization/80275
16182 * fold-const.c (split_address_to_core_and_offset): Handle
16185 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16187 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16188 descriptors is at least equal to that of functions.
16190 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16192 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16194 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16197 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16198 (mov<IMOD4:mode>): New expander.
16199 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16201 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16203 PR rtl-optimization/79405
16204 * fwprop.c (propagations_left): New variable.
16205 (forward_propagate_into): Decrement it.
16206 (fwprop_init): Initialize it.
16207 (fw_prop): If the variable has reached zero, stop propagating.
16208 (fwprop_addr): Ditto.
16210 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16213 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16214 a FUNCTION_DECL, pass it as decl instead of origin to
16217 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16219 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16222 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16225 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16226 TARGET_VSX_SMALL_INTEGER.
16228 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16230 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16231 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16233 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16235 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16236 extraction from odd-numbered MSA register.
16238 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16240 PR middle-end/80173
16241 * expmed.c (store_bit_field_1): Don't attempt to create
16242 a word subreg out of hard registers wider than word if they
16243 have HARD_REGNO_NREGS of 1 for their mode.
16245 PR middle-end/80163
16246 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16247 conversions to integer types wider than word and pointer.
16250 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16251 (rtx_equal_for_cselib_p): Pass 0 to it.
16252 * cselib.c (cselib_hasher::equal): Likewise.
16253 (rtx_equal_for_cselib_1): Add depth argument. If depth
16254 is 128, don't look up VALUE locs and punt. Increment
16255 depth in recursive calls when walking VALUE locs.
16257 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16259 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16260 (make_gcov_file_name): Use the canonical path name for generating
16262 (read_line): Fix handling of files with ascii null bytes.
16264 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16266 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16267 to initialise a vector register instead
16268 of using a const_int.
16270 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16272 PR translation/80189
16273 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16274 diagnostic messages.
16276 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16279 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16280 (dfp_diex_<mode>): Update mode of operand 1.
16281 * doc/extend.texi (dxex, dxexq): Document change to return type.
16282 (diex, diexq): Document change to argument type.
16284 2017-03-30 Martin Jambor <mjambor@suse.cz>
16287 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16288 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16289 it reflects the signature changes performed at the callee side.
16290 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16291 to cgraph_build_function_type_skip_args.
16292 (build_function_decl_skip_args): Adjust call to the above function.
16294 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16297 * config/i386/sse.md
16298 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16299 register as dest whenever it is a MEM not rtx_equal_p to the
16300 corresponding dup operand, and when forcing into reg move the
16301 reg into the memory afterwards.
16302 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16303 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16304 for the force_reg mode.
16305 (avx512vl_vextractf128<mode>): Use register as dest either
16306 always when a MEM, or when it is a MEM not rtx_equal_p to the
16307 corresponding dup operand, or even not when it is a CONST_VECTOR
16308 depending on the mode and lo vs. hi.
16309 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16311 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16312 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16313 Likewise. Require that operands[2] is even.
16314 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16315 Remove extraneous parens. Require that operands[2] is a multiple
16317 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16318 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16319 disallow memory then.
16321 2017-03-30 Richard Biener <rguenther@suse.de>
16323 PR tree-optimization/77498
16324 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16325 to non-constants over backedges.
16327 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16329 PR rtl-optimization/80233
16330 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16331 as last_combined_insn. Do not test for BARRIER_P separately.
16333 2017-03-29 Andreas Schwab <schwab@suse.de>
16336 * calls.c (prepare_call_address): Convert funexp to Pmode before
16337 copying to temp reg.
16339 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16341 PR tree-optimization/80158
16342 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16343 Handle possible future case of more than one alternate
16345 (replace_rhs_if_not_dup): Likewise.
16346 (replace_one_candidate): Likewise.
16348 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16350 PR rtl-optimization/80193
16351 * ira.c (ira): Do not check allocation for LRA.
16353 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16355 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16356 (nvptx_output_simt_exit): Declare.
16357 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16358 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16359 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16360 (nvptx_declare_function_name): ...here. Emit declaration of local
16361 memory space buffer for omp_simt_enter insn.
16362 (nvptx_output_unisimt_switch): New.
16363 (nvptx_output_softstack_switch): New.
16364 (nvptx_output_simt_enter): New.
16365 (nvptx_output_simt_exit): New.
16366 * config/nvptx/nvptx.h (struct machine_function): New fields
16367 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16368 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16369 (UNSPECV_SIMT_EXIT): Ditto.
16370 (omp_simt_enter_insn): New insn.
16371 (omp_simt_enter): New expansion.
16372 (omp_simt_exit): New insn.
16373 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16375 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16376 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16377 (expand_GOMP_SIMT_EXIT): New.
16378 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16379 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16380 (GOMP_SIMT_EXIT): Ditto.
16381 * target-insns.def (omp_simt_enter): New insn.
16382 (omp_simt_exit): Ditto.
16383 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16385 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16386 (lower_rec_input_clauses): Likewise.
16387 (lower_lastprivate_clauses): Handle SIMT privatization.
16389 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16390 (ompdevlow_adjust_simt_enter): New.
16391 (find_simtpriv_var_op): New.
16392 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16393 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16395 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16396 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16397 (copy_decl_for_dup_finish): Ditto.
16399 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16401 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16404 * config/i386/i386.c (ix86_option_override_internal): Always
16405 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16407 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16409 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16411 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16413 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16414 mark new edge's irreducible flag accordign to it.
16415 (vect_do_peeling): Check loop preheader edge's irreducible flag
16416 and pass it to function slpeel_add_loop_guard.
16418 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16420 PR tree-optimization/80218
16421 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16422 Update block frequencies and counts.
16424 2017-03-28 Richard Biener <rguenther@suse.de>
16426 PR tree-optimization/78644
16427 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16428 of a simplification result we may not use it at all.
16430 2017-03-28 Richard Biener <rguenther@suse.de>
16433 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16434 without arguments, generate default definition of a SSA name.
16436 2017-03-28 Richard Biener <rguenther@suse.de>
16438 PR middle-end/80222
16439 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16440 TYPE_REF_CAN_ALIAS_ALL references.
16441 * fold-const.c (fold_indirect_ref_1): Likewise.
16443 2017-03-28 Martin Liska <mliska@suse.cz>
16446 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16447 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16449 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16450 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16452 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16453 (EXTRA_SPECS): Define.
16454 (SUBTARGET_EXTRA_SPECS): Likewise.
16455 (SUBTARGET_CPP_SPEC): Likewise.
16456 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16457 SUBTARGET_EXTRA_SPECS.
16458 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16460 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16462 * config/arc/simdext.md (vst64_insn): Update pattern.
16463 (vld32wh_insn): Likewise.
16464 (vld32wl_insn): Likewise.
16465 (vld64_insn): Likewise.
16466 (vld32_insn): Likewise.
16468 2017-03-28 Marek Polacek <polacek@redhat.com>
16471 * fold-const.c (fold_comparison): Use protected_set_expr_location
16472 instead of SET_EXPR_LOCATION.
16474 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16476 * tree.c (add_expr): Avoid name lookup warning.
16478 2017-03-27 Jeff Law <law@redhat.com>
16480 PR tree-optimization/80216
16481 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16482 function name. Limit recursion depth.
16483 (record_temporary_equivalences): Corresponding changes.
16485 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16487 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16490 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16493 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16495 * cfgcleanup.c (reg_note_cfa_p): New array.
16496 (insns_have_identical_cfa_notes): New function.
16497 (old_insns_match_p): Don't cross-jump in between /f
16498 and non-/f instructions. If both i1 and i2 are frame related,
16499 verify all CFA notes, their order and content.
16501 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16504 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16505 HImode and SImode with zero extend to DImode to one insn.
16506 (bswap<mode>2_extenddi): Likewise.
16507 (bswapsi2_extenddi): Likewise.
16508 (bswaphi2_extendsi): Likewise.
16509 (bswaphi2): Combine bswap HImode and SImode into one insn.
16510 Separate memory insns from swapping register.
16511 (bswapsi2): Likewise.
16512 (bswap<mode>2): Likewise.
16513 (bswaphi2_internal): Delete, no longer used.
16514 (bswapsi2_internal): Likewise.
16515 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16516 store, and gpr<-gpr swap insns.
16517 (bswap<mode>2_store): Likewise.
16518 (bswaphi2_reg): Register only splitter, combine with the splitter.
16519 (bswaphi2 splitter): Likewise.
16520 (bswapsi2_reg): Likewise.
16521 (bswapsi2 splitter): Likewise.
16522 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16523 the insns into load, store, and register/register insns.
16524 (bswapdi2_ldbrx): Likewise.
16525 (bswapdi2_load): Likewise.
16526 (bswapdi2_store): Likewise.
16527 (bswapdi2_reg): Likewise.
16529 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16531 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16532 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16534 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16537 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16539 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16540 special handling for target option conflicts between dform
16541 options (-mpower9-dform, -mpower9-dform-vector,
16542 -mpower9-dform-scalar) and -mno-direct-move.
16544 2017-03-27 Richard Biener <rguenther@suse.de>
16546 PR tree-optimization/80181
16547 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16549 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16551 * config/arc/predicates.md (move_double_src_operand): Replace the
16552 call to move_double_src_operand with a call to address_operand.
16554 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16556 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16557 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16558 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16560 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16562 * config/arc/predicates.md (long_immediate_loadstore_operand):
16563 Consider scaled addresses cases.
16565 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16567 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16568 restored when in interrupt.
16569 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16570 doesn't have delay slot.
16572 2017-03-27 Richard Biener <rguenther@suse.de>
16575 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16576 inlined thunk clones.
16578 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16581 * asan.c (instrument_derefs): Copy over last operand from
16582 original COMPONENT_REF to the new COMPONENT_REF with
16583 DECL_BIT_FIELD_REPRESENTATIVE.
16584 * ubsan.c (instrument_object_size): Likewise.
16586 2017-03-27 Richard Biener <rguenther@suse.de>
16588 PR tree-optimization/80170
16589 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16590 sure DR/SCEV didnt fold in constants we do not see when looking
16591 at the reference base alignment.
16593 2017-03-27 Richard Biener <rguenther@suse.de>
16595 PR middle-end/80171
16596 * gimple-fold.c (fold_ctor_reference): Properly guard against
16597 NULL return value from canonicalize_constructor_val.
16599 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16602 * config/i386/i386.c (ix86_expand_builtin)
16603 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16604 flags reg setting and flags reg using instructions.
16605 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16606 clobbering instructions to zero extend op2.
16608 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16610 * doc/install.texi (Configuration) <--with-aix-soname>:
16611 Update link to AIX ld.
16613 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16615 PR rtl-optimization/80160
16616 PR rtl-optimization/80159
16617 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16618 reg_alternate_class into account.
16620 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16623 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16624 to consider in curr_insn_transform.
16626 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16628 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16629 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16630 and emit_mode_inner.
16632 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16634 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16635 argument to the overloaded builtin variants. Use the new flag to
16636 deprecate certain builtin variants.
16637 * config/s390/s390-builtin-types.def: Add new builtin types.
16638 * config/s390/s390-builtins.h: Support new flags field for
16639 overloaded builtins.
16640 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16641 (s390_macro_to_expand): Enable vector float data type.
16642 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16643 builtins by incrementing the __VEC__ version number.
16644 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16646 (s390_resolve_overloaded_builtin): Emit error messages depending
16647 on the builtin flags.
16648 * config/s390/s390.c (s390_expand_builtin): Support additional
16649 flags argument. Change error message to match the messages
16650 emitted in s390-c.c.
16651 * config/s390/s390.md: New UNSPEC_* constants.
16652 (op_type): Add new instruction types.
16653 * config/s390/vecintrin.h: Add new builtins and test data class
16655 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16656 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16657 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16658 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16659 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16660 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16662 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16663 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16664 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16665 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16667 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16668 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16669 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16670 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16672 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16673 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16674 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16675 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16676 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16677 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16678 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16680 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16681 ("vec_scatter_element<V_HW_4:mode>_DI")
16682 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16683 ("vec_fpint<mode>", "vflls")
16684 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16685 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16686 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16687 ("*vec_cmphe<mode>_cc"): ... these.
16689 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16690 mode constant instead of magic value.
16692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16694 * config/s390/s390.c (s390_expand_vec_compare): Support other
16695 vector floating point modes than just V2DF.
16696 (s390_expand_vcond): Likewise.
16697 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16698 (s390_cannot_change_mode_class): Prevent mode changes between TF
16699 and V1TF in vector registers.
16700 * config/s390/s390.md (DF, SF): New mode attributes.
16701 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16702 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16703 SFmode support for VRs.
16704 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16706 (VFT, VF_HW): New mode iterators.
16707 (vw, sdx): New mode attributes.
16708 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16709 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16710 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16711 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16712 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16713 also the new vector floating point modes. Renaming to ...
16715 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16716 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16717 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16718 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16719 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16720 ("vec_unordered<mode>"): ... these.
16722 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16723 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16724 ("*vec_extendv2df"): New insn definitions.
16726 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16728 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16729 ("mulditi3_2", "*muldi3_sign"): New patterns.
16730 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16731 rename the pattern definition.
16733 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16735 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16737 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16739 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16741 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16742 instruction if possible.
16743 * config/s390/vector.md (vec_halfnumelts): New mode
16745 ("*vec_vllezlf<mode>"): New pattern.
16747 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16749 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16750 ("popcountv4si2", "popcountv2di2"): Rename to ...
16751 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16752 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16754 ("popcount<mode>2_vxe"): New pattern.
16756 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16758 * common/config/s390/s390-common.c (processor_flags_table): Add
16760 * config.gcc: Add arch12.
16761 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16762 Default to arch12 for unknown CPU model numbers.
16763 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16764 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16765 PROCESSOR_max sanity check.
16766 * config/s390/s390-opts.h (enum processor_type): Add
16768 * config/s390/s390.c (processor_table): Add arch12.
16769 (s390_expand_builtin): Add check for B_VXE flag.
16770 (s390_issue_rate): Add PROCESSOR_ARCH12.
16771 (s390_get_sched_attrmask): Likewise.
16772 (s390_get_unit_mask): Likewise.
16773 (s390_sched_score): Enable z13 scheduling for arch12.
16774 (s390_sched_reorder): Likewise.
16775 (s390_sched_variable_issue): Likewise.
16776 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16778 (s390_tune_attr): Use z13 scheduling also for arch12.
16779 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16780 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16781 (TARGET_VXE_P): New macros.
16782 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16783 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16784 * config/s390/s390.opt: Add arch12 as processor_type.
16786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16788 * config/s390/s390.md
16789 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16790 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16791 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16793 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16794 Rename expanders to ...
16796 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16797 ("fixuns_truncdddi2_emu"): ... these.
16799 ("fixuns_trunc<mode>si2_emu"): New expander.
16801 ("*fixuns_truncdfdi2_z13"): Rename to ...
16802 ("*fixuns_truncdfdi2_vx"): ... this.
16804 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16806 * config/s390/2964.md: Remove the single element vector compare
16807 instructions which are no longer used.
16808 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16810 (s390_canonicalize_comparison): Remove handling of DFmode
16812 (s390_expand_vec_compare_scalar): Remove function.
16813 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16814 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16816 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16818 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16820 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16822 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
16823 will anyway by matched by mov<mode>_64dfp.
16825 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16827 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
16828 vlef/vstef. Add missing operand to vleif.
16830 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16832 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
16833 pair for all vector types with 64 bit elements.
16834 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
16835 * config/s390/vector.md (V_HW_64): ... here.
16836 (V_128_NOSINGLE): New mode iterator.
16837 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
16838 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
16839 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
16840 ("*vec_load_pairv2di"): Change to ...
16841 ("*vec_load_pair<mode>"): ... this one.
16843 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16845 * config/s390/constraints.md: Add comments.
16846 (jKK): Reject element sizes > 8 bytes.
16847 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
16849 * config/s390/s390.md: Add the s_operand checks formerly in
16850 s390_split_ok_p to various splitters where they are still
16852 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
16853 for 128 bit vectors. Plus two splitters.
16855 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16857 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
16860 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16863 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
16864 error if the boundary argument is not constant.
16866 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16868 PR rtl-optimization/80112
16869 * loop-doloop.c (doloop_condition_get): Don't check condition
16870 if cmp isn't SET with IF_THEN_ELSE src.
16872 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16874 PR tree-optimization/80158
16875 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
16876 replacing a candidate statement, also replace it for the
16877 candidate's alternate interpretation.
16878 (replace_rhs_if_not_dup): Likewise.
16879 (replace_one_candidate): Likewise.
16881 2017-03-24 Richard Biener <rguenther@suse.de>
16883 PR tree-optimization/80167
16884 * graphite-isl-ast-to-gimple.c
16885 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
16887 (translate_isl_ast_to_gimple::get_rename): Likewise.
16889 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16891 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
16892 handling of certain combinations of target options, including the
16893 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
16894 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
16896 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16899 * config/arm/arm.md (*load_multiple): Add reload_completed to
16900 matching condition.
16902 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16903 Richard Biener <rguenth@suse.de>
16905 PR tree-optimization/79908
16906 PR tree-optimization/80136
16907 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16908 been cast away, gimplify_and_add suffices.
16910 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
16912 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
16914 2017-03-23 Richard Biener <rguenther@suse.de>
16916 PR tree-optimization/80032
16917 * gimplify.c (gimple_push_cleanup): Forced unconditional
16918 cleanups still have to go to the conditional_cleanups
16921 2017-03-22 Jakub Jelinek <jakub@redhat.com>
16923 PR tree-optimization/80072
16924 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
16926 (next_operand_entry_id): Change type to unsigned int.
16927 (sort_by_operand_rank): Make sure to return the right return value
16928 even if unsigned fields are bigger than INT_MAX.
16929 (struct oecount): Change cnt and id type to unsigned int.
16930 (oecount_hasher::equal): Formatting fix.
16931 (oecount_cmp): Make sure to return the right return value
16932 even if unsigned fields are bigger than INT_MAX.
16933 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
16936 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
16937 TREE_READONLY on result if writing it more than once.
16940 * doc/invoke.texi (-fsanitize=thread): Document that with
16941 -fnon-call-exceptions atomics are not able to throw
16945 * tsan.c: Include tree-eh.h.
16946 (instrument_builtin_call): Call maybe_clean_eh_stmt or
16947 maybe_clean_or_replace_eh_stmt where needed.
16948 (instrument_memory_accesses): Add cfg_changed argument.
16949 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
16950 if it returned true.
16951 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
16953 PR rtl-optimization/63191
16954 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
16955 wrapper function, moved the whole old content into ...
16956 (ix86_delegitimize_address_1): ... this. New inline function.
16957 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
16958 true as last argument instead of ix86_delegitimize_address.
16960 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16962 * config/aarch64/aarch64.c (generic_branch_cost): Copy
16963 cortexa57_branch_cost.
16965 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16967 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
16969 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16972 * doc/md.texi (Constraints): Document wA constraint.
16973 * config/rs6000/constraints.md (wA): New.
16974 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
16975 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
16976 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
16977 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
16979 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
16982 * gimplify.c (is_oacc_declared): New function.
16983 (oacc_default_clause): Use it to set default flags for acc declared
16984 variables inside parallel regions.
16985 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
16986 declared variables.
16987 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
16988 declare attribute to any decl as necessary.
16990 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16993 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
16994 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
16995 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
16996 (arm_arch_lpae): This.
16997 * config/arm/arm.c (arm_arch7ve): Rename into ...
16998 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
16999 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17002 2017-03-22 Martin Liska <mliska@suse.cz>
17005 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17006 error message instead of an ICE.
17008 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17010 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17012 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17014 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17019 2017-03-21 Martin Sebor <msebor@redhat.com>
17021 * doc/extend.texi: Use "cannot" instead of "can't."
17022 * doc/hostconfig.texi: Same.
17023 * doc/install.texi: Same.
17024 * doc/invoke.texi: Same.
17025 * doc/loop.texi: Same.
17026 * doc/md.texi: Same.
17027 * doc/objc.texi: Same.
17028 * doc/rtl.texi: Same.
17029 * doc/tm.texi: Same.
17030 * doc/tm.texi.in: Same.
17031 * doc/trouble.texi: Same.
17033 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17036 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17037 (collect_checksum_attributes): Set it.
17038 (die_checksum_ordered): Use it.
17040 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17042 PR tree-optimization/79908
17043 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17044 change: For a VA_ARG whose LHS has been cast away, use
17045 force_gimple_operand to construct the side effects.
17047 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17049 PR translation/80001
17050 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17051 more amenable to translation.
17052 (oacc_loop_auto_partitions): Likewise.
17054 2017-03-21 Marek Polacek <polacek@redhat.com>
17055 Martin Sebor <msebor@redhat.com>
17057 PR tree-optimization/80109
17058 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17059 on INTEGRAL_TYPE_P.
17061 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17062 Segher Boessenkool <segher@kernel.crashing.org>
17065 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17066 check reg_used_between_p between insn and one of succ or succ2
17067 depending on if succ is artificial insn not inserted into insn
17070 2017-03-21 Martin Liska <mliska@suse.cz>
17072 PR gcov-profile/80081
17073 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17074 * doc/gcc.texi: Include gcov-dump stuff.
17075 * doc/gcov-dump.texi: New file.
17077 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17079 PR rtl-optimization/79150
17080 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17081 conditional jump, if the jump is the last insn of the loop.
17083 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17084 Richard Biener <rguenth@suse.de>
17086 PR tree-optimization/79908
17087 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17088 been cast away, use force_gimple_operand to construct the side
17091 2017-03-21 Martin Liska <mliska@suse.cz>
17093 PR libfortran/79956
17094 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17097 2017-03-21 Brad Spengler <spender@grsecurity.net>
17100 * plugin.c (htab_hash_plugin): New function.
17101 (add_new_plugin): Use it and adjust.
17102 (parse_plugin_arg_opt): Adjust.
17103 (init_one_plugin): Likewise.
17105 2017-03-21 Richard Biener <rguenther@suse.de>
17107 PR tree-optimization/80032
17108 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17109 if set force the cleanup to happen unconditionally.
17110 (gimplify_target_expr): Push inserted clobbers with force_uncond
17111 to avoid them being removed by control-dependent DCE.
17113 2017-03-21 Richard Biener <rguenther@suse.de>
17115 PR tree-optimization/80122
17116 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17117 va_arg_pack_len when the inlined call stmt requires pack
17119 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17121 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17124 * tsan.c (instrument_builtin_call): If the memory model argument
17125 is not a constant, assume it is valid.
17128 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17131 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17133 PR rtl-optimization/79910
17134 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17135 if its dest is used by an insn before I2 (other than the combined
17136 insns themselves, which are properly handled already).
17138 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17141 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17143 * combine.c (record_used_regs): New static function.
17144 (try_combine): Handle situations where there is an additional
17145 instruction between I2 and I3 which needs to have a LOG_LINK
17149 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17151 * combine.c (try_combine): Delete redundant i1 test. Call
17152 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17154 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17157 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17158 alternatives 13/14.
17160 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17162 PR tree-optimization/80054
17163 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17164 the optimization if a PHI or any of its arguments is not dominated
17165 by the candidate's basis. Use gphi* rather than gimple* as
17167 (replace_profitable_candidates): Clean up a gimple* variable that
17168 should be a gphi* variable.
17170 2017-03-20 Martin Sebor <msebor@redhat.com>
17173 * doc/extend.texi (attribute constructor): Document present limitation.
17175 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17178 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17179 __POWER9_VECTOR__ #ifdef control, change template definition to
17180 use Power9-specific built-in function.
17181 (vec_any_eq): Likewise.
17182 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17183 to control outcomes from this test.
17184 (vector_ae_<mode>p): For VEC_F modes, likewise.
17186 2017-03-20 Ian Lance Taylor <iant@google.com>
17188 * config/i386/i386.c (ix86_function_regparm): Save an extra
17189 register for -fsplit-stack with DECL_STATIC_CHAIN.
17191 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17194 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17195 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17197 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17199 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17201 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17204 2017-03-20 Marek Polacek <polacek@redhat.com>
17207 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17209 2017-03-20 Richard Biener <rguenther@suse.de>
17211 PR tree-optimization/80113
17212 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17213 allocate extra SSA name for PHI def.
17214 (add_close_phis_to_outer_loops): Likewise.
17215 (add_close_phis_to_merge_points): Likewise.
17216 (copy_loop_close_phi_args): Likewise.
17217 (copy_cond_phi_nodes): Likewise.
17219 2017-03-20 Martin Liska <mliska@suse.cz>
17221 PR middle-end/79753
17222 * tree-chkp.c (chkp_build_returned_bound): Do not build
17223 returned bounds for a LHS that's not a BOUNDED_P type.
17225 2017-03-20 Martin Liska <mliska@suse.cz>
17229 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17230 COMPLEX_CST and VECTOR_CST.
17232 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17235 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17236 target operand. A new splitter adds the clobber statement in case
17237 the target operand is dead anyway.
17239 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17241 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17242 to age-old versions of binutils and glibc.
17244 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17246 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17248 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17250 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17252 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17254 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17255 requirement for binutils 2.13.
17257 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17259 * combine.c (try_combine): Delete redundant i1 test. Call
17260 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17262 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17264 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17265 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17267 <riscv64-*-elf>: Re-arrange section
17268 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17269 <riscv32-*-linux>: Likewise.
17270 <riscv64-*-elf>: Likewise
17271 <riscv64-*-linux>: Likewise.
17273 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17276 * aarch64.opt(verbose-cost-dump): Fix typo.
17278 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17281 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17282 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17284 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17286 * reload.c (find_reloads): When reloading a nonoffsettable address,
17287 use RELOAD_OTHER for it and its address reloads.
17289 PR rtl-optimization/79910
17290 * combine.c (record_used_regs): New static function.
17291 (try_combine): Handle situations where there is an additional
17292 instruction between I2 and I3 which needs to have a LOG_LINK
17295 2017-03-17 Jeff Law <law@redhat.com>
17297 PR tree-optimization/71437
17298 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17299 conditional in the hash table first.
17300 (vrp_dom_walker::before_dom_children): Extract condition from
17301 ASSERT_EXPR. Record condition, its inverion and any implied
17302 conditions as well.
17304 2017-03-17 Marek Polacek <polacek@redhat.com>
17305 Markus Trippelsdorf <markus@trippelsdorf.de>
17307 PR tree-optimization/80079
17308 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17311 2017-03-17 Richard Biener <rguenther@suse.de>
17313 PR middle-end/80075
17314 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17315 Properly verify the LHS before the RHS possibly claims to be
17317 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17320 2017-03-17 Martin Jambor <mjambor@suse.cz>
17322 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17323 (List of -O2 options): Likewise.
17324 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17327 2017-03-17 Tom de Vries <tom@codesourcery.com>
17329 * gcov-dump.c (print_usage): Print bug_report_url.
17331 2017-03-17 Richard Biener <rguenther@suse.de>
17333 PR middle-end/80050
17334 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17335 (parser::peek): Likewise.
17337 2017-03-17 Richard Biener <rguenther@suse.de>
17339 PR tree-optimization/80048
17340 * sese.c (free_sese_info): Properly release rename_map and
17341 copied_bb_map elements.
17343 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17345 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17346 Add linked-list forward and backlinks. Insert on
17347 construction, remove on destruction.
17348 (class pass_store_merging): Add m_stores_head field.
17349 (pass_store_merging::terminate_and_process_all_chains):
17350 Iterate over m_stores_head list.
17351 (pass_store_merging::terminate_all_aliasing_chains):
17353 (pass_store_merging::execute): Check for debug stmts first.
17354 Push new chains onto the m_stores_head stack.
17356 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17359 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17360 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17361 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17363 2017-03-16 Jeff Law <law@redhat.com>
17365 PR tree-optimization/71437
17366 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17367 member function. Implementation moved into after_dom_children
17368 member function and into the threader's thread_outgoing_edges
17370 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17371 some code into new thread_outgoing_edges.
17372 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17373 definition. Simplify marker handling (do it here). Assume we always
17374 have the available expression and the const/copies tables.
17375 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17377 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17378 * tree-vrp.c (equiv_stack): No longer file scoped.
17379 (vrp_dom_walker): New class.
17380 (vrp_dom_walker::before_dom_children): New member function.
17381 (vrp_dom_walker::after_dom_children): Likewise.
17382 (identify_jump_threads): Setup domwalker. Use it rather than
17383 walking edges in a random order by hand. Simplify setup/finalization.
17384 (finalize_jump_threads): Remove.
17385 (vrp_finalize): Do not call identify_jump_threads here.
17386 (execute_vrp): Do it here instead and call thread_through_all_blocks
17389 PR tree-optimization/71437
17390 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17392 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17394 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17395 (dom_opt_dom_walker::thread_across_edge): Remove
17396 handle_dominating_asserts argument. All callers changed.
17397 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17398 changes. Remove calls to lhs_of_dominating_assert. Other
17399 uses of handle_dominating_asserts turn into unconditional code
17400 (simplify_control_stmt_condition_1): Likewise.
17401 (simplify_control_stmt_condition): Likewise.
17402 (thread_through_normal_block, thread_across_edge): Likewise.
17403 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17404 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17405 object if it is not an SSA_NAME.
17406 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17407 before calling into the VRP specific simplifiers.
17408 (identify_jump_threads): Remove handle_dominating_asserts
17411 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17414 * tree-diagnostic.c (default_tree_printer): No longer static.
17415 * tree-diagnostic.h (default_tree_printer): New prototype.
17417 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17419 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17420 Change ins into fmov.
17422 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17424 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17425 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17426 Use h_con constraint for operand 1.
17427 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17428 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17430 2017-03-15 Jeff Law <law@redhat.com>
17432 PR tree-optimization/71437
17433 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17434 (record_temporary_equivalences): Use it.
17436 PR tree-optimization/71437
17437 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17438 tree-ssa-scopedtables.
17439 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17440 (record_conditions, record_cond, vuse_eq): Likewise.
17441 (record_edge_info): Adjust to API tweak of record_conditions.
17442 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17443 (record_temporary_equivalences, optimize_stmt): Likewise.
17444 (eliminate_redundant_computations): Likewise.
17445 (record_equivalences_from_stmt): Likewise.
17446 * tree-ssa-scopedtables.c: Include options.h and params.h.
17447 (vuse_eq): New function, moved from tree-ssa-dom.c
17448 (build_and_record_new_cond): Likewise.
17449 (record_conditions): Likewise. Accept vector of conditions rather
17450 than edge_equivalence structure for first argument.
17451 for the first argument.
17452 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17453 from tree-ssa-dom.c.
17454 (avail_exprs_stack::record_cond): Likewise.
17455 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17456 from tree-ssa-dom.c.
17457 (avail_exprs_stack): Add new member functions lookup_avail_expr
17459 (record_conditions): Declare.
17461 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17464 * lra-constraints.c (process_alt_operands): Increase reject for
17465 reloading an input/output operand.
17467 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17470 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17471 insns to convert from signed/unsigned char/short to IEEE 128-bit
17473 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17475 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17478 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17479 subreg of inner mode for values already in registers.
17481 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17483 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17484 iteration reg is used after the loop.
17486 2017-03-14 Martin Sebor <msebor@redhat.com>
17488 PR tree-optimization/79800
17489 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17490 precision in negative-positive range.
17491 (format_floating): Call non-const overload with adjusted precision.
17493 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17496 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17499 2017-03-14 Martin Sebor <msebor@redhat.com>
17501 PR middle-end/80020
17502 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17503 * builtins.def (aligned_alloc): Use it.
17506 * Makefile.in (GTFILES): Add calls.c.
17507 * calls.c: Include "gt-calls.h".
17509 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17511 PR rtl-optimization/79728
17512 * regs.h (struct target_regs): New field
17513 x_contains_allocatable_regs_of_mode.
17514 (contains_allocatable_regs_of_mode): New macro.
17515 * reginfo.c (init_reg_sets_1): Initialize it, and change
17516 contains_reg_of_mode so it includes global regs as well.
17517 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17518 rather than contains_regs_of_mode.
17520 2017-03-14 Martin Liska <mliska@suse.cz>
17522 * doc/invoke.texi: Document options that can't be combined with
17523 -fcheck-pointer-bounds.
17525 2017-03-14 Martin Liska <mliska@suse.cz>
17527 PR middle-end/79831
17528 * doc/invoke.texi (-Wchkp): Document the option.
17530 2017-03-14 Martin Liska <mliska@suse.cz>
17532 * Makefile.in: Install gcov-dump.
17534 2017-03-14 Martin Liska <mliska@suse.cz>
17536 * multiple_target.c (expand_target_clones): Bail out for
17537 an invalid attribute.
17539 2017-03-14 Richard Biener <rguenther@suse.de>
17541 * alias.c (struct alias_set_entry): Pack properly.
17542 * cfgloop.h (struct loop): Likewise.
17543 * cse.c (struct set): Likewise.
17544 * ipa-utils.c (struct searchc_env): Likewise.
17545 * loop-invariant.c (struct invariant): Likewise.
17546 * lra-remat.c (struct cand): Likewise.
17547 * recog.c (struct change_t): Likewise.
17548 * rtl.h (struct address_info): Likewise.
17549 * symbol-summary.h (function_summary): Likewise.
17550 * tree-loop-distribution.c (struct partition): Likewise.
17551 * tree-object-size.c (struct object_size_info): Likewise.
17552 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17553 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17554 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17555 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17556 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17557 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17558 (struct _stmt_vec_info): Likewise.
17560 2017-03-14 Martin Liska <mliska@suse.cz>
17563 * multiple_target.c (create_dispatcher_calls): Check that
17564 a target can create a function dispatcher.
17566 2017-03-14 Martin Liska <mliska@suse.cz>
17569 * multiple_target.c (expand_target_clones): Drop local.local
17570 flag for default implementation.
17572 2017-03-14 Richard Biener <rguenther@suse.de>
17574 PR tree-optimization/80030
17575 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17577 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17579 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17580 gcc_fallthrough() instead of __attribute__((fallthrough));
17582 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17584 * doc/gcc.texi: Remove "up" link to (DIR).
17585 * doc/gccint.texi: Ditto.
17587 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17589 * doc/install.texi (Specific) <avr>: Remove reference to
17592 2017-03-13 Jeff Law <law@redhat.com>
17594 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17595 attribute rather than comments.
17597 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17598 match_scratch operand is highest.
17600 2017-03-13 Martin Liska <mliska@suse.cz>
17602 PR middle-end/78339
17603 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17604 is a CHKP clone, use original declaration.
17606 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17608 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17609 (arc_conditional_register_usage): Use a different allocation order
17610 when optimizing for size.
17611 * common/config/arc/arc-common.c (arc_option_optimization_table):
17612 Section anchors default on when optimizing for size.
17614 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17616 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17618 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17620 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17621 * config/arc/arc.md (cpu_facility): Add cd variant.
17622 (*movqi_insn): Add code density variant.
17623 (*movhi_insn): Likewise.
17624 (*movqi_insn): Likewise.
17625 (*addsi3_mixed): Likewise.
17626 (subsi3_insn): Likewise.
17628 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17630 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17632 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17634 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17635 expressions with MINUS and UNARY ops.
17637 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17640 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17642 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17643 between vec_select and vector argument.
17644 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17645 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17646 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17647 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17648 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17649 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17651 2017-03-13 Richard Biener <rguenther@suse.de>
17654 * params.def (vect-max-peeling-for-alignment): Fix typo.
17656 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17658 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17659 issue that only occurred with binutils below 2.18.
17661 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17663 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17664 refer to binutils 2.11/2.12 minimum.
17666 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17668 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17669 ftp.kernel.org and simplify binutils requirement.
17671 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17673 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17675 (Optimize Options): Ditto. Also remove redundancy.
17677 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17679 PR translation/79848
17680 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17682 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17683 to G_ to avoid double translation.
17685 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17687 PR translation/79923
17688 * auto-profile.c (get_combined_location): Convert leading
17689 character of diagnostics to lower case and remove trailing period.
17690 (read_profile): Likewise for various diagnostics.
17691 * config/arm/arm.c (arm_option_override): Remove trailing period
17692 from various diagnostics.
17693 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17694 (msp430_expand_delay_cycles): Likewise.
17696 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17699 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17700 full command-line argument, rather than just "str".
17701 (aarch64_validate_march): Likewise.
17702 (aarch64_validate_mtune): Likewise.
17704 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17706 PR rtl-optimization/78911
17707 * lra-assigns.c (must_not_spill_p): New function.
17708 (spill_for): Use it.
17710 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17712 PR tree-optimization/79981
17713 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17714 ATOMIC_COMPARE_EXCHANGE ifn result.
17715 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17716 IFN_ATOMIC_COMPARE_EXCHANGE.
17718 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17721 * opts.c (parse_sanitizer_options): Add missing question mark to
17722 "did you mean" message.
17724 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17726 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17728 (VMULEUH_UNS): Likewise.
17729 (VMULOUB_UNS): Likewise.
17730 (VMULOUH_UNS): Likewise.
17731 * config/rs6000/rs6000.c (builtin_function_type): Remove
17732 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17734 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17737 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17738 x with result of extra_parsing_for_operand_code_0.
17739 (function_reader::extra_parsing_for_operand_code_0): Convert
17740 return type from void to rtx, returning x. When reading
17741 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17742 larger size containing struct block_symbol.
17744 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17746 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17747 -mfloat128-hardware without -m64.
17749 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17752 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17753 entries to the case statement that marks unsigned arguments to
17754 overloaded functions.
17756 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17758 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17759 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17761 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17764 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17765 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17767 2017-03-10 Martin Liska <mliska@suse.cz>
17771 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17772 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17775 2017-03-10 Olivier Hainque <hainque@adacore.com>
17777 * tree-switch-conversion (array_value_type): Start by resetting
17778 candidate type to it's main variant.
17780 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17782 PR rtl-optimization/79909
17783 * combine.c (try_combine): Use simplify_replace_rtx on individual
17784 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17785 of the whole CALL_INSN_FUNCTION_USAGE.
17787 PR tree-optimization/79972
17788 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17789 get_range_info on SSA_NAMEs. Formatting fixes.
17791 2017-03-10 Richard Biener <rguenther@suse.de>
17792 Jakub Jelinek <jakub@redhat.com>
17794 PR tree-optimization/77975
17795 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17796 edge to be constant.
17797 (get_val_for): For constant x return it. Formatting fix.
17798 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17799 would use the same bases as the current one.
17801 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17803 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17804 instead of vec_select for V1TImode.
17805 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17807 (VSX_LE_128): Add V1TI to this mode iterator.
17808 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17809 (*vsx_le_perm_store_<mode>): Likewise.
17810 (pre-reload splitter for VSX stores): Likewise.
17811 (post-reload splitter for VSX stores): Likewise.
17812 (*vsx_xxpermdi2_le_<mode>): Likewise.
17813 (*vsx_lxvd2x2_le_<mode>): Likewise.
17814 (*vsx_stxvd2x2_le_<mode>): Likewise.
17816 2017-03-09 Michael Eager <eager@eagercon.com>
17818 Correct failures with --enable-checking=yes,rtl.
17820 * config/microblaze/microblaze.c (microblaze_expand_shift):
17821 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17822 test for const0_rtx.
17823 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
17824 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
17826 2017-03-09 Richard Biener <rguenther@suse.de>
17828 PR tree-optimization/79977
17829 * graphite-scop-detection.c (scop_detection::merge_sese):
17830 Handle the case of extra exits to blocks dominating the entry.
17832 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
17834 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17837 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
17839 PR rtl-optimization/79949
17840 * lra-constraints.c (process_alt_operands): Check memory when
17841 trying to predict a cycle. Print about the overall increase.
17843 2017-03-09 Richard Biener <rguenther@suse.de>
17845 PR middle-end/79971
17846 * gimple-expr.c (useless_type_conversion_p): Preserve
17847 TYPE_SATURATING for fixed-point types.
17849 2017-03-09 Richard Biener <rguenther@suse.de>
17852 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
17853 alignment of BLKmode params.
17855 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17858 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
17859 (VALL_NO_V2Q): Likewise.
17861 * config/aarch64/aarch64-simd.md
17862 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
17864 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
17865 VALL_NO_V2Q mode iterator.
17866 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
17868 2017-03-09 Martin Liska <mliska@suse.cz>
17870 PR tree-optimization/79631
17871 * tree-chkp-opt.c (chkp_is_constant_addr): Call
17872 tree_int_cst_sign_bit just for INTEGER constants.
17874 2017-03-09 Martin Liska <mliska@suse.cz>
17878 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
17881 2017-03-09 Marek Polacek <polacek@redhat.com>
17884 * tree.c (inchash::add_expr): Handle TREE_VEC.
17886 2017-03-09 Martin Liska <mliska@suse.cz>
17889 (chkp_narrow_size_and_offset): New function.
17890 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
17891 (void chkp_parse_bit_field_ref): New function.
17892 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
17893 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
17895 2017-03-09 Martin Liska <mliska@suse.cz>
17898 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
17899 (chkp_find_bounds_1): Remove gcc_unreachable.
17901 2017-03-09 Jakub Jelinek <jakub@redhat.com>
17904 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
17905 BUILT_IN_SYNC*, determine the access type from the size suffix and
17906 always build a MEM_REF with that type. Handle forgotten
17907 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
17910 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
17911 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
17912 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
17913 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
17914 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
17915 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
17916 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
17917 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
17918 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
17919 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
17920 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
17921 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
17922 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
17923 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
17924 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
17925 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
17926 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
17927 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
17928 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
17929 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
17930 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
17931 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
17932 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
17933 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
17934 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
17935 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
17936 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
17937 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
17938 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
17939 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
17940 definitions outside of __OPTIMIZE__ guarded section.
17943 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
17944 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
17945 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
17946 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
17949 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17951 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
17952 ("vfenez<mode>"): Add missing constraints.
17954 2017-03-08 Martin Sebor <msebor@redhat.com>
17957 * config/nds32/nds32.c (nds32_option_override):
17958 Fix misspelled diagnostic.
17960 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17963 * gimplify.c (gimplify_omp_for): Replace index var in outer
17964 taskloop statement with an artificial variable and add
17965 OMP_CLAUSE_PRIVATE clause for it.
17967 2017-03-08 Richard Biener <rguenther@suse.de>
17969 PR tree-optimization/79955
17970 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
17971 for accesses that are completely outside of the variable.
17973 2017-03-08 Andrew Haley <aph@redhat.com>
17975 PR tree-optimization/79943
17976 * tree-ssa-loop-split.c (compute_new_first_bound): When
17977 calculating the new upper bound, (END-BEG) should be added, not
17980 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17982 * config/avr/avr.md (setmemhi): Make sure match_dup
17983 operand number comes before match_scratch.
17985 2017-03-08 Richard Biener <rguenther@suse.de>
17987 PR tree-optimization/79920
17988 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
17989 with ncopies == 1 to ...
17990 (vect_transform_slp_perm_load): ... here. Properly compute
17991 all element loads by iterating VF times over the group. Do
17992 not handle ncopies (computed in a broken way) in
17993 vect_create_mask_and_perm.
17995 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17998 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
17999 is a uniform vector, use uniform_vector_p return value instead of
18000 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18002 2017-03-07 Marek Polacek <polacek@redhat.com>
18004 PR middle-end/79809
18005 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18006 (alloca_call_type): Likewise.
18008 2017-03-07 Martin Liska <mliska@suse.cz>
18010 * gcov.c (process_args): Put comment to correct location.
18012 2017-03-07 Martin Liska <mliska@suse.cz>
18014 PR middle-end/68270
18015 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18016 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18017 (chkp_narrow_bounds_for_field): Likewise.
18018 (chkp_parse_array_and_component_ref): Pass one more argument to
18021 2017-03-07 Richard Biener <rguenther@suse.de>
18023 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18026 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18028 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18029 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18031 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18034 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18035 to end of description.
18036 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18038 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18040 PR rtl-optimization/79901
18041 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18043 (*avx512f_<code><mode>3<mask_name>): ... this.
18044 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18045 iterator instead of VI8_AVX2_AVX512BW.
18047 PR rtl-optimization/79901
18048 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18049 min/max expander, expand it using expand_vec_cond_expr.
18052 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18055 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18058 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18059 to void * for PCH reasons.
18060 * dwarf2out.c (output_loc_operands, output_die): Cast
18061 v.val_vec.array to unsigned char *.
18063 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18066 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18069 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18071 PR rtl-optimization/79571
18072 * lra-constraints.c (process_alt_operands): Calculate static
18073 reject and subtract it from overall when only addresses will be
18076 2017-03-06 Julia Koval <julia.koval@intel.com>
18079 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18080 incoming stack boundary to 128 for 64-bit targets.
18082 2017-03-06 Richard Biener <rguenther@suse.de>
18084 PR tree-optimization/79894
18085 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18086 to NULL after folding it.
18088 2017-03-06 Richard Biener <rguenther@suse.de>
18090 PR tree-optimization/79824
18091 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18092 check disabling peeling for gaps.
18094 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18096 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18097 attributes): Document gettimeofday.
18099 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18101 * config/s390/s390.c (s390_option_override_internal): Set
18102 PARAM_MIN_VECT_LOOP_BOUND
18104 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18106 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18107 * config/s390/s390.md: Likewise.
18109 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18112 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18113 (<avx2_avx512>_perm<mode>): Rename to ...
18114 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18116 (<avx512>_perm<mode>_mask): Rename to ...
18117 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18119 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18120 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18121 instead of VI8F_256_512.
18122 (avx512f_perm<mode>): New define_expand.
18123 (avx512f_perm<mode>_mask): Likewise.
18124 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18125 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18127 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18129 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18130 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18132 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18134 2017-03-06 Martin Liska <mliska@suse.cz>
18137 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18138 when having a SSA NAME w/o VAR_DECL assigned to it.
18140 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18142 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18143 msa_dpsub_<su>_d): Fix MODE for vec_select.
18145 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18147 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18149 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18151 2017-03-06 Richard Biener <rguenther@suse.de>
18153 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18154 * plugin.c (register_plugin_info): Likewise.
18155 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18157 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18159 * config/i386/sse.md (sse_storehps, sse_storelps,
18160 avx_<castmode><avxsizesuffix>_<castmode>,
18161 avx512f_<castmode><avxsizesuffix>_<castmode>,
18162 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18163 in condition that at least one operand is not a MEM.
18165 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18167 PR middle-end/79805
18168 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18169 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18171 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18172 gimple_call_nothrow_p flag based on whether original builtin can throw.
18173 If it can, emit following stmts on the fallthrough edge.
18174 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18175 don't create new bb if inserting just debug stmts on the edge, try to
18176 insert them on the fallthru bb or just reset debug stmts.
18178 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18181 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18182 restore recog_data (including the operand rtxes inside it) around
18183 the call to get_insn_template.
18185 2017-03-03 Martin Sebor <msebor@redhat.com>
18187 PR tree-optimization/79699
18188 * context.c (context::~context): Free MPFR caches to avoid
18189 a memory leak on program exit.
18191 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18193 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18194 Use wide_int::ulow () instead of .elt (0).
18196 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18198 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18199 (*pushxf): Limit oF constraint to 32bit targets and add oC
18200 constraint for 64bit targets.
18201 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18202 (*pushdf): Change rmF constraint to rmC.
18204 2017-03-03 Martin Liska <mliska@suse.cz>
18206 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18207 Remove unused variable.
18209 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18212 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18213 is a memory operand, increase num_memory.
18214 (ix86_expand_args_builtin): Likewise.
18216 2017-03-03 Jan Hubicka <jh@suse.cz>
18219 * ipa-devirt.c (maybe_record_node): Properly handle
18220 __cxa_pure_virtual visibility.
18222 2017-03-03 Martin Liska <mliska@suse.cz>
18224 PR tree-optimization/79803
18225 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18227 (pass_loop_prefetch::execute): Disabled optimization if an
18228 assumption about L1 cache size is not met.
18230 2017-03-03 Martin Liska <mliska@suse.cz>
18232 PR rtl-optimization/79574
18233 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18234 (hash_scan_set): Likewise.
18235 (dump_hash_table): Likewise.
18236 (hoist_code): Likewise.
18238 2017-03-03 Richard Biener <rguenther@suse.de>
18240 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18241 in place of uhigh ().
18242 (fixed_convert_from_real): Likewise.
18244 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18247 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18249 2017-03-03 Richard Biener <rguenther@suse.de>
18251 PR middle-end/79818
18252 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18253 TYPE_OVERFLOW_UNDEFINED check.
18255 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18257 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18259 (vector_ae_<mode>_p): Likewise.
18260 (vector_nez_<mode>_p): Likewise.
18261 (vector_ne_v2di_p): Likewise.
18262 (vector_ae_v2di_p): Likewise.
18263 (vector_ne_<mode>_p): Likewise.
18264 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18266 (vsx_tsqrt<mode>2_fe): Likewise.
18268 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18271 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18273 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18275 PR rtl-optimization/79780
18276 * cprop.c (one_cprop_pass): When second and further conditional trap
18277 in a single basic block is turned into an unconditional trap, turn it
18278 into a deleted note to avoid RTL verification failures.
18280 2017-03-02 Richard Biener <rguenther@suse.de>
18282 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18284 2017-03-02 Richard Biener <rguenther@suse.de>
18286 PR tree-optimization/79345
18288 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18289 param and abort the walk, returning -1 if it is hit.
18290 (walk_aliased_vdefs): Take a limit param and pass it on.
18291 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18292 defaulting to 0 and return a signed int.
18293 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18294 (check_defs): New helper.
18295 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18296 about uninitialized memory.
18297 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18298 bogus uninitialized warning.
18299 (fixed_convert_from_real): Likewise.
18301 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18303 PR tree-optimization/66768
18304 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18305 iv_use if base object can't be determined.
18307 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18309 PR tree-optimization/79345
18310 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18311 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18312 (get_pattern_stats): Initialize it.
18313 * genemit.c (gen_expand): Verify match_scratch numbers come after
18314 match_operand/match_dup numbers.
18315 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18316 match_scratch numbers.
18317 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18319 * config/s390/s390.md (trunctdsd2): Likewise.
18321 2017-03-02 Richard Biener <rguenther@suse.de>
18323 * wide-int.h (wide_int_storage::operator=): Implement in terms
18326 2017-03-02 Richard Biener <rguenther@suse.de>
18328 PR tree-optimization/79777
18329 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18330 the to insert expression to sth existing.
18332 2017-03-01 Martin Sebor <msebor@redhat.com>
18334 PR middle-end/79692
18335 * gimple-ssa-sprintf.c
18336 (directive::known_width_and_precision): New function.
18337 (format_integer): Use it.
18338 (get_mpfr_format_length): Consider the full range of precision
18339 when computing %g output with the # flag. Set the likely byte
18340 count to 3 rather than 1 when precision is indeterminate.
18341 (format_floating): Correct the lower bound of precision.
18343 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18345 * doc/invoke.texi: Document default code model for 64-bit Linux.
18347 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18350 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18351 udiv rather than div since input pattern is unsigned.
18353 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18355 * config/i386/i386.c (print_reg): Warn for values of
18356 unsupported size in integer register.
18358 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18361 * config/rs6000/predicates.md (current_file_function_operand): Do
18362 not allow self calls to be local if the function is replaceable.
18364 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18367 * config/rs6000/altivec.h (vec_ctz and others): Change the
18368 preprocessor macro that controls conditional compilation from
18369 _ARCH_PWR9 to __POWER9_VECTOR__.
18370 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18371 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18372 control (instead of _ARCH_PWR9 control) so that template
18373 definition uses power9-specific function.
18374 (vec_any_eq): Likewise.
18375 (vec_all_ne): Change macro definition to use a power9-specific
18376 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18377 _ARCH_PWR9 control).
18378 (vec_any_eq) Likewise.
18379 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18380 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18381 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18382 support for xvcmpnedp instruction.
18383 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18384 macro expansion so that Power9 implementation of vec_all_ne does
18385 not use the AltiVec predicate framework.
18386 (VCMPNEH_P): Likewise.
18387 (VCMPNEW_P): Likewise.
18388 (VCMPNED_P): Likewise.
18389 (VCMPNEFP_P): Likewise.
18390 (VCMPNEDP_P): Likewise.
18391 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18392 implementation of vec_any_eq to not use AltiVec predicate
18394 (VCMPAEH_P): Likewise.
18395 (VCMPAEW_P): Likewise.
18396 (VCMPAED_P): Likewise.
18397 (VCMPAEFP_P): Likewise.
18398 (VCMPAEDP_P): Likewise.
18399 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18400 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18401 not use the AltiVec predicate framework.
18402 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18403 of vec_any_eq to not use AltiVec predicate framework.
18404 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18405 support for predefined __POWER9_VECTOR__ macro to indicate that
18406 Power9 instruction selection is enabled.
18407 (altivec_overloaded_builtins): Remove extraneous
18408 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18409 function argument types RS6000_BTI_bool_V16QI and
18410 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18411 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18412 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18413 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18414 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18415 Power9 for implementations of vec_cmpne. Change the signature for
18416 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18417 (representing vec_all_ne) to remove the previously described first
18418 argument of type RS6000_BTI_INTSI, as this was an artifact of
18419 reliance on the AltiVec predicate framework, which is no longer
18420 used in the implementation of these functions. Add
18421 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18422 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18423 since, unlike the AltiVec predicate framework implementation, we
18424 do not share function descriptors between vec_alle and vec_anyeq.
18425 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18426 set of modes that receive special treatment even when
18427 TARGET_P9_VECTOR is true. The special treatment emits code that
18428 does not depend on Power9 instructions.
18429 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18430 define_expand to not rely on AltiVec predicate framework.
18431 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18433 (vector_ne_v2di_p): Change this define_expand to not rely on
18434 AltiVec predicate framework.
18435 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18437 (vector_ne_<mode>_p): Change this define_expand to not rely on
18438 AltiVec predicate framework.
18439 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18441 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18442 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18443 define_insn pattern.
18444 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18445 define_insn pattern because the xvcmpne<VSs>. instruction is not
18447 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18448 instruction is not supported.
18450 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18452 * config/nvptx/nvptx.c: Include intl.h.
18454 2017-03-01 Martin Jambor <mjambor@suse.cz>
18457 * ipa-prop.h (ipa_bits): Removed field known.
18458 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18459 to pointers. Adjusted their comments to warn about their sharing.
18460 (ipcp_transformation_summary): Change bits to a vector of pointers.
18461 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18462 (ipa_get_ipa_bits_for_value): Declare.
18463 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18464 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18465 (ipa_bits_hash_table): Likewise.
18466 (ipa_vr_ggc_hash_traits): Likewise.
18467 (ipa_vr_hash_table): Likewise.
18468 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18469 being pointers and vr_known being removed.
18470 (ipa_set_jf_unknown): Likewise.
18471 (ipa_get_ipa_bits_for_value): New function.
18472 (ipa_set_jfunc_bits): Likewise.
18473 (ipa_get_value_range): New overloaded functions.
18474 (ipa_set_jfunc_vr): Likewise.
18475 (ipa_compute_jump_functions_for_edge): Use the above functions to
18476 construct bits and vr parts of jump functions.
18477 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18478 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18480 (ipcp_grow_transformations_if_necessary): Also allocate
18481 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18483 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18484 them. Fix too long lines.
18485 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18486 vr_known being removed.
18487 (ipa_read_jump_function): Use new setter functions to construct bits
18488 and vr parts of jump functions or set them to NULL.
18489 (write_ipcp_transformation_info): Adjust for bits being pointers.
18490 (read_ipcp_transformation_info): Likewise.
18491 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18493 Include gt-ipa-prop.h.
18494 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18496 (ipcp_store_bits_results): Likewise.
18497 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18498 Do not write to existing jump functions but use a temporary instead.
18500 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18503 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18504 attempt to use its first operand as BIT_FIELD_REF base.
18506 2017-03-01 Richard Biener <rguenther@suse.de>
18508 PR middle-end/79721
18509 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18510 interpolating formula in wrapping arithmetic.
18511 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18513 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18515 PR tree-optimization/79734
18516 * tree-vect-generic.c (expand_vector_condition): Optimize
18517 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18518 Handle VEC_COND_EXPR where comparison has different inner width from
18519 type's inner width.
18521 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18523 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18524 markup, and similar issues. Remove @opindex entries for things
18525 that aren't options. Add missing -mmpy-option entries.
18527 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18529 PR tree-optimization/79737
18530 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18531 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18532 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18533 instead of byte_size. Formatting fix.
18534 (shift_bytes_in_array_right): Formatting fix.
18536 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18539 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18540 condition on optimize for the leaf function test.
18542 2017-02-28 Martin Liska <mliska@suse.cz>
18545 * read-rtl-function.c (function_reader::handle_unknown_directive):
18546 Bail out when one uses -flto.
18548 2017-02-28 Martin Liska <mliska@suse.cz>
18550 * common.opt: Replace space with tabular for options of <number>
18552 * config/i386/i386.opt: Show <number> value for
18553 -mlarge-data-threshold.
18554 * opts.c (print_filtered_help): Do not display number in hexadecimal
18557 2017-02-28 Martin Liska <mliska@suse.cz>
18559 * common.opt: Fix --help=option -Q for options which are of
18562 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18564 * config/i386/i386.c (print_reg): Error out for values
18565 of 8-bit size in invalid integer register.
18567 2017-02-28 Martin Sebor <msebor@redhat.com>
18569 PR tree-optimization/79691
18570 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18572 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18575 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18576 gcc_unreachable with output_operand_lossage.
18578 2017-02-28 Richard Biener <rguenther@suse.de>
18580 PR tree-optimization/79740
18581 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18583 (visit_nary_op): Insert the nary into the hashtable if we
18584 pattern-matched sth.
18585 * tree-ssa-pre.c (eliminate_insert): Robustify.
18587 2017-02-28 Richard Biener <rguenther@suse.de>
18589 PR middle-end/79731
18590 * fold-const.c (decode_field_reference): Reject out-of-bound
18593 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18595 * config/i386/i386.c: Include intl.h.
18596 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18597 instead of just cond ? "..." : "...".
18598 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18599 * coverage.c (read_counts_file): Likewise.
18600 * omp-offload.c: Include intl.h.
18601 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18602 of just cond ? "..." : "...".
18603 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18604 of just cond ? "..." : "...".
18606 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18609 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18611 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18612 'tune for' CPU name.
18613 * config/arm/arm-cpu-data.h: Regenerated.
18615 2017-02-28 Richard Biener <rguenther@suse.de>
18617 PR tree-optimization/79732
18618 * tree-inline.c (expand_call_inline): Do not shadow var.
18620 2017-02-28 Richard Biener <rguenther@suse.de>
18622 PR tree-optimization/79723
18623 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18624 address-space properly.
18626 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18628 * doc/optinfo.texi (Optimization groups): Fix option used for
18630 * doc/invoke.texi (-fopt-info): Document "omp".
18631 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18632 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18633 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18634 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18635 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18637 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18639 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18640 OPTGROUP_OMP with "omp".
18642 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18645 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18646 for arithmetic shift of unsigned V2DI.
18648 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18650 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18651 arc/linux.h headers.
18652 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18653 (LINK_SPEC): Likewise.
18654 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18655 (EXTRA_SPECS): Likewise.
18656 (STARTFILE_SPEC): Likewise.
18657 (ENDFILE_SPEC): Likewise.
18658 (LIB_SPEC): Likewise.
18659 (TARGET_SDATA_DEFAULT): Likewise.
18660 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18661 (MULTILIB_DEFAULTS): Likewise.
18662 (DWARF2_UNWIND_INFO): Likewise.
18663 * config/arc/big.h: New file.
18664 * config/arc/elf.h: Likewise.
18665 * config/arc/linux.h: Likewise.
18666 * config/arc/t-uClibc: Remove.
18668 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18670 PR tree-optimization/77536
18671 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18672 (tree_transform_and_unroll_loop): Use above function to compute the
18673 estimated niter of unrolled loop and use it when scaling profile.
18674 Also use count info rather than frequency if it's non-zero.
18675 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18676 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18677 (vect_transform_loop): Call above function.
18679 2017-02-27 Richard Biener <rguenther@suse.de>
18681 PR tree-optimization/45397
18682 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18683 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18684 (visit_nary_op): Add pattern matching for CSEing sign-changed
18685 or truncated operations with wider ones.
18687 2017-02-27 Richard Biener <rguenther@suse.de>
18689 PR tree-optimization/79690
18690 * tree-vect-stmts.c (vectorizable_store): Use vector type
18691 built from the DR with address-space.
18693 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18695 * doc/invoke.texi (Optimize Options): Refine the description
18696 of asan-use-after-return.
18698 2017-02-25 Alan Modra <amodra@gmail.com>
18700 PR rtl-optimization/79584
18701 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18702 base, not ad->base_term, the reg within base. Remove assertion
18703 that ad->base == ad->base_term. Replace gen_int_mode using
18704 bogus mode with const0_rtx.
18706 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18708 PR middle-end/79396
18709 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18710 FMA_EXPR like tcc_binary or tcc_unary.
18712 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18715 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18717 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18718 (output_loc_operands): Handle DW_OP_call_ref and
18719 DW_OP_GNU_variable_value.
18720 (struct variable_value_struct): New type.
18721 (struct variable_value_hasher): Likewise.
18722 (variable_value_hash): New variable.
18723 (string_types): Remove.
18724 (copy_loc_descr): New function.
18725 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18726 (prepend_loc_descr_to_each): New function.
18727 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18728 instead of add_loc_descr_to_each if the first argument is single
18729 location list and the second has multiple.
18730 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18731 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18732 when looking for variable value which doesn't have other location info.
18733 (loc_list_from_tree): Formatting fix.
18734 (gen_array_type_die): Simplify DW_AT_string_length handling.
18735 (adjust_string_types): Remove.
18736 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18737 string_types. Call resolve_variable_values.
18738 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18739 (resolve_addr_in_expr): Likewise. Add A argument.
18740 (copy_deref_exprloc): Remove deref argument. Adjust for the
18741 original expression being DW_OP_GNU_variable_value with optionally
18742 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18743 optionally after it.
18744 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18745 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18746 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18747 (variable_value_hasher::hash, variable_value_hasher::equal): New
18749 (resolve_variable_value_in_expr, resolve_variable_value,
18750 resolve_variable_values, note_variable_value_in_expr,
18751 note_variable_value): New functions.
18752 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18755 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18758 * opts.h (handle_generated_option): Add GENERATED_P argument.
18759 * opts-common.c (handle_option): Adjust function comment.
18760 (handle_generated_option): Add GENERATED_P argument, pass it to
18762 (control_warning_option): Pass false to handle_generated_option
18764 * opts.c (maybe_default_option): Pass true to handle_generated_option
18766 * optc-gen.awk: Likewise.
18768 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18770 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18771 a REG, look at the REG it is a SUBREG of.
18772 (splitter for cmpeqsi_t): Ditto.
18774 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18776 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18777 the special USEs with the pattern of the insn, not the insn itself.
18779 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18782 * doc/invoke.texi: Document -mload-store-pairs.
18784 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18785 Sandra Loosemore <sandra@codesourcery.com>
18787 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18788 argument isn't a CONST_INT.
18789 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18790 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18791 (nios2_validate_compare): Bypass alternate compare logic if *op2
18792 is not a CONST_INT.
18793 (ldstwm_operation_p): Return false if first_base is not a REG or
18794 if first_offset is not a CONST_INT.
18796 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18798 * config/cris/cris.md: Use correct operand in a define_peephole2.
18800 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18802 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18804 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18806 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18807 this_insn if it is an INSN or JUMP_INSN.
18808 (force_offsettable): Look at base, not at addr.
18809 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18810 on things that aren't necessarily CONST_INTs.
18812 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18814 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18815 -mfpmath=sse is the default also for x86-32 targets with SSE2
18816 instruction set when @option{-ffast-math} is enabled
18818 2017-02-24 Jeff Law <law@redhat.com>
18820 PR rtl-optimizatoin/79286
18821 * ira.c (update_equiv_regs): Drop may_trap_p exception to
18824 2017-02-24 Richard Biener <rguenther@suse.de>
18826 PR tree-optimization/79389
18827 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
18830 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
18832 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
18833 function comment to reflect reality.
18834 (loop_exits_before_overflow): Fix typo in function description.
18836 2017-02-24 Richard Biener <rguenther@suse.de>
18838 PR tree-optimization/79389
18839 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
18840 properly that a threading opportunity exists. Detect conditional
18841 copy/constant propagation opportunities.
18843 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
18845 * config/visium/visium.md (type): Add trap.
18846 (b): New mode attribute.
18847 (*btst): Rename into...
18848 (*btst<mode>): ...this and adjust.
18849 (*cbranchsi4_btst_insn): Rename into...
18850 (*cbranch<mode>4_btst_insn): ...this and adjust.
18851 (trap): New define_insn.
18853 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18855 PR tree-optimization/79389
18856 * ifcvt.c (struct noce_if_info): Add rev_cond field.
18857 (noce_reversed_cond_code): New function.
18858 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
18859 reversed_comparison_code. Formatting fix.
18860 (noce_try_store_flag): Test rev_cond != NULL in addition to
18861 reversed_comparison_code.
18862 (noce_try_store_flag_constants): Likewise.
18863 (noce_try_store_flag_mask): Likewise.
18864 (noce_try_addcc): Use rev_cond if non-NULL instead of
18865 reversed_comparison_code.
18866 (noce_try_cmove_arith): Likewise. Formatting fixes.
18867 (noce_try_minmax, noce_try_abs): Clear rev_cond.
18868 (noce_find_if_block): Initialize rev_cond.
18869 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
18870 instead of false as last argument never attempt to reverse it
18873 2017-02-23 Bin Cheng <bin.cheng@arm.com>
18875 PR tree-optimization/79663
18876 * tree-predcom.c (combine_chains): Process refs in reverse order
18877 only for ZERO length chains, and add explaining comment.
18879 2017-02-23 Jeff Law <law@redhat.com>
18881 PR tree-optimization/79578
18882 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
18883 in call to operand_equal_p.
18885 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
18888 * config/i386/cpuid.h: Fix another undefined behavior.
18890 2017-02-23 Richard Biener <rguenther@suse.de>
18892 PR tree-optimization/79683
18893 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
18894 vector types for data-refs.
18896 2017-02-23 Martin Liska <mliska@suse.cz>
18898 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
18900 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18902 PR middle-end/79665
18903 * internal-fn.c (get_range_pos_neg): Moved to ...
18904 * tree.c (get_range_pos_neg): ... here. No longer static.
18905 * tree.h (get_range_pos_neg): New prototype.
18906 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
18907 are known to be in between 0 and signed maximum inclusive, try to
18908 expand both unsigned and signed divmod and use the cheaper one from
18911 2017-02-22 Jeff Law <law@redhat.com>
18913 PR tree-optimization/79578
18914 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
18915 to compare base operands.
18917 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
18920 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
18921 gpc_reg_operand instead of fpr_reg_operand.
18923 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
18925 * config/mips/mips.c (mips_return_in_memory): Force FP
18926 vector types to be returned in memory for o32 ABI.
18928 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18930 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
18931 instead of DW_TAG_member for static data member declarations and don't
18932 set no_linkage_name for static inline data members.
18933 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
18936 2017-02-22 Martin Liska <mliska@suse.cz>
18938 * doc/invoke.texi: Replace inequality signs with square brackets
18941 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18943 PR tree-optimization/68644
18944 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
18946 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
18949 * lra-constraints.c (simplify_operand_subreg): Handle
18950 WORD_REGISTER_OPERATIONS targets.
18952 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18955 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
18956 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
18957 elimination by swapping fld*.
18959 2017-02-22 Richard Biener <rguenther@suse.de>
18961 PR tree-optimization/79673
18962 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
18963 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
18964 irrelevant address-space qualifiers and avoiding a
18965 ADDR_SPACE_CONVERT_EXPR from fold_convert.
18967 2017-02-22 Richard Biener <rguenther@suse.de>
18969 PR tree-optimization/79666
18970 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
18971 to not symbolically negate if that may introduce undefined
18974 2017-02-22 Martin Liska <mliska@suse.cz>
18977 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
18978 * data-streamer-out.c (streamer_write_gcov_count_stream):
18980 * value-prof.c (stream_out_histogram_value): Make assert more
18981 precise based on type of counter.
18983 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
18986 * config/i386/i386.md (standard_x87sse_constant_load splitter):
18987 Use nonimmediate_operand instead of memory_operand for operand 1.
18988 (float-extend standard_x87sse_constant_load splitter): Ditto.
18990 2017-02-21 Jeff Law <law@redhat.com>
18992 PR tree-optimization/79621
18993 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
18994 blocks with edges to themselves.
18996 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18999 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19000 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19001 Use gimple_call_builtin_p.
19004 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19005 on temporarily removed DEBUG_INSNs.
19007 PR tree-optimization/79649
19008 * tree-loop-distribution.c (classify_partition): Give up on
19009 non-generic address space loads/stores.
19011 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19013 * doc/loop.texi (Loop manipulation): Remove nonexistent
19014 tree_ssa_loop_version from the documentation.
19015 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19017 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19020 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19021 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19022 * config/rs6000/rs6000.c: Include except.h.
19023 (rs6000_expand_split_stack_prologue): Call
19024 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19026 2017-02-21 Martin Jambor <mjambor@suse.cz>
19029 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19030 have been analyzed.
19032 2017-02-21 Martin Jambor <mjambor@suse.cz>
19034 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19035 for backward compatibility only.
19036 * doc/invoke.texi (Option Summary): Remove all references to
19037 -fipa-cp-alignment.
19039 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19043 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19045 * lra-constraints.c (curr_insn_transform): Handle
19046 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19048 2017-02-21 Martin Liska <mliska@suse.cz>
19050 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19052 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19054 PR translation/79638
19055 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19057 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19060 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19061 (arm_function_ok_for_sibcall): Return false for an indirect call by
19062 descriptor if all the argument registers are used.
19063 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19064 alignment of the function.
19066 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19068 PR tree-optimization/61441
19069 * simplify-rtx.c (simplify_const_unary_operation): For
19070 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19071 the sNaN unmodified.
19073 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19075 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19076 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19077 instead of SYSTEM_HEADER_DIR.
19079 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19080 Martin Liška <mliska@suse.cz>
19082 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19083 Fix typos and grammar, use active voice, and clarify.
19085 2017-02-20 Marek Polacek <polacek@redhat.com>
19087 PR middle-end/79537
19088 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19091 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19093 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19096 * config/i386/i386.c (ix86_expand_builtin): Handle
19097 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19098 ix86_builtins_isa[fcode].isa as a requirement of those
19099 flags and any other flag in the bitmask.
19100 (ix86_init_mmx_sse_builtins): Use 0 instead of
19101 ~OPTION_MASK_ISA_64BIT as mask.
19102 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19103 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19104 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19105 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19107 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19110 * lra-constraints.c (split_reg): Check requested split mode
19111 is supported by the register.
19113 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19115 * lra-constraints.c (simplify_operand_subreg): Remove early
19118 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19121 * lra-constraints.c (curr_insn_transform): Tighten condition
19122 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19124 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19127 * lra-constraints.c (curr_insn_transform): Handle
19128 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19130 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19133 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19135 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19137 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19140 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19143 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19145 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19148 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19150 * final.c (last_columnnum, override_columnnum): New variables.
19151 (final_start_function): Set last_columnnum, pass it to begin_prologue
19152 hook and pass 0 to dwarf2out_begin_prologue.
19153 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19154 to source_line debug hook.
19155 (notice_source_line): Compute last_columnnum and for debug_column_info
19156 return true on column changes.
19157 * debug.h (struct gcc_debug_hooks): Add column argument to
19158 source_line and begin_prologue hooks.
19159 (debug_nothing_int_charstar_int_bool): Remove prototype.
19160 (debug_nothing_int_int_charstar,
19161 debug_nothing_int_int_charstar_int_bool): New prototypes.
19162 (dwarf2out_begin_prologue): Add column argument.
19163 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19164 begin_prologue hooks.
19165 (debug_nothing_int_charstar_int_bool): Remove.
19166 (debug_nothing_int_int_charstar,
19167 debug_nothing_int_int_charstar_int_bool): New functions.
19168 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19169 through to dwarf2out_source_line.
19170 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19171 (dwarf2out_source_line): Add column argument, emit it if requested.
19172 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19174 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19175 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19176 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19177 through to dwarf2out_begin_prologue.
19178 (vmsdbgout_source_line): Add column argument, pass it through to
19179 dwarf2out_source_line.
19180 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19181 dbxout_source_line caller.
19182 (dbxout_source_line): Add column argument.
19184 * common.opt (gno-column-info, gcolumn-info): New options.
19185 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19186 (check_die): Also test for multiple DW_AT_decl_column attributes.
19187 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19188 DW_AT_decl_column if requested.
19189 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19191 (gen_variable_die): Likewise.
19192 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19193 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19196 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19197 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19198 (ix86_handle_option): Handle OPT_m3dnowa.
19199 * doc/invoke.texi (-m3dnowa): Document.
19200 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19201 -m3dnowa instead of -m3dnow -march=athlon.
19204 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19205 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19207 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19210 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19211 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19212 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19213 generator for vsx_xxpermdi_<mode>_be.
19214 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19215 force big-endian semantics.
19216 (vsx_xxpermdi_<mode>_be): New define_expand with same
19217 implementation as previous version of vsx_xxpermdi_<mode>.
19219 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19221 PR tree-optimization/79327
19222 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19223 variable, its initialization and use.
19225 2017-02-17 Julia Koval <julia.koval@intel.com>
19227 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19228 (OPTION_MASK_ISA_PKU_UNSET): New.
19229 (ix86_handle_option): Handle -mrdpid.
19230 * config/i386/cpuid.h (bit_RDPID): New.
19231 * config/i386/driver-i386.c (host_detect_local_cpu):
19232 Detect RDPID feature.
19233 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19234 * config/i386/i386-c.c (ix86_target_macros_internal):
19236 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19237 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19238 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19239 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19240 * config/i386/i386.md (define_insn "rdpid"): New.
19241 * config/i386/i386.opt Add -mrdpid.
19242 * config/i386/immintrin.h (_rdpid_u32): New.
19244 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19246 PR rtl-optimization/79541
19247 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19248 instead of transforming it into USE.
19250 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19252 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19253 If HONOR_SNANS (SFmode) force the input to a register.
19254 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19255 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19256 an frsp or similar insn.
19258 2017-02-17 Martin Liska <mliska@suse.cz>
19260 PR rtl-optimization/79577
19261 * params.def (selsched-max-sched-times): Increase minimum to 1.
19263 2017-02-17 Martin Liska <mliska@suse.cz>
19265 PR rtl-optimization/79574
19266 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19268 2017-02-17 Martin Liska <mliska@suse.cz>
19270 PR tree-optimization/79529
19271 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19272 ssa_defined_default_def_p to handle cases which are implicitly
19274 * tree-ssa.c (ssa_defined_default_def_p): New function.
19275 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19276 which are implicitly defined.
19277 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19279 2017-02-17 Richard Biener <rguenther@suse.de>
19281 PR middle-end/79576
19282 * params.def (max-ssa-name-query-depth): Limit to 10.
19284 2017-02-17 Richard Biener <rguenther@suse.de>
19286 PR tree-optimization/79552
19287 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19290 2017-02-17 Richard Biener <rguenther@suse.de>
19293 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19295 2017-02-17 Marek Polacek <polacek@redhat.com>
19297 PR middle-end/79536
19298 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19299 (fold_negate_expr): New wrapper.
19301 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19303 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19304 Correct terminology and de-emphasize pre-standard behavior.
19306 2017-02-16 Alan Modra <amodra@gmail.com>
19308 PR rtl-optimization/79286
19309 * ira.c (def_dominates_uses): New function.
19310 (update_equiv_regs): Don't create an equivalence for insns that
19311 may trap where the register def does not dominate the use.
19313 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19315 PR rtl-optimization/78127
19316 * lra.c (lra): Call lra_eliminate before finish the loop after
19319 2017-02-16 Richard Biener <rguenther@suse.de>
19321 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19323 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19324 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19325 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19326 (isl_val_int_from_wi): New function.
19327 (extract_affine_gmp): Rename to ...
19328 (extract_affine_wi): ... this, take a widest_int.
19329 (extract_affine_int): Just wrap extract_affine_wi.
19330 (add_param_constraints): Use isl_val_int_from_wi.
19331 (add_loop_constraints): Likewise, and extract_affine_wi.
19333 2017-02-15 Jeff Law <law@redhat.com>
19335 PR middle-end/79521
19336 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19337 ira_init_register_move_cost_if_necessary.
19339 2017-02-15 Martin Sebor <msebor@redhat.com>
19341 PR middle-end/32003
19342 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19343 removed in a prior commit.
19345 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19347 PR tree-optimization/79347
19348 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19349 counters during peeling.
19351 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19353 * Makefile.in (site.exp): Remove "set ISLVER".
19355 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19358 * real.c (real_from_integer): Call real_convert even for decimal.
19360 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19363 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19365 2017-02-14 Andrew Pinski <apinski@cavium.com>
19367 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19368 cores and change the partno/implementer to be correct.
19369 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19370 the 'B" as the implementer.
19371 * config/aarch64/aarch64-tune.md: Regenerate.
19373 2017-02-14 Carl Love <cel@us.ibm.com>
19375 * config/rs6000/rs6000.c: Add case statement entry to make the
19376 xvcvuxdsp built-in argument unsigned.
19377 * config/rs6000/vsx.md: Fix the source and return operand types so they
19378 match the instruction definitions from the ISA document. Fix typo
19379 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19382 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19385 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19386 member early_clobber_alts.
19387 * lra-lives.c (reg_early_clobber_p): New.
19388 (process_bb_lives): Use it.
19389 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19390 (debug_operand_data): Initialize early_clobber_alts.
19391 (setup_operand_alternative): Set up early_clobber_alts.
19392 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19393 alternatives to new_insn_reg.
19394 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19396 (lra_update_insn_regno_info): Pass the new arg.
19398 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19400 PR middle-end/79505
19401 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19402 (new_oacc_loop_raw): Don't clear already cleared fields.
19405 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19406 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19407 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19409 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19412 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19414 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19417 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19418 the extra instruction to the right place to store 128-bit constant
19421 2017-02-14 Martin Sebor <msebor@redhat.com>
19423 PR middle-end/79448
19424 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19425 warning for strings of unknown length.
19427 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19429 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19431 2017-02-14 Jeff Law <law@redhat.com>
19434 * ira-costs.c (scan_one_insn): Initialize register move costs
19435 for pseudos seen in USE/CLOBBER insns.
19437 PR tree-optimization/79095
19438 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19439 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19440 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19441 if the operands are known to be not equal, then the resulting range
19443 (intersect_ranges): If the new range is ~[0,0] and the old range is
19444 wide, then prefer ~[0,0].
19445 * tree-vrp.c (overflow_comparison_p_1): New function.
19446 (overflow_comparison_p): New function.
19447 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19448 if NAME is used in an overflow test.
19449 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19450 overflow check that can be expressed as an equality test, then adjust
19451 ops to be that equality test.
19453 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19455 * config/s390/s390-builtin-types.def: Remove flags argument.
19456 * config/s390/s390.c (s390_init_builtins): Likewise.
19458 2017-02-14 Martin Liska <mliska@suse.cz>
19460 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19461 vector. Fix trailing white spaces.
19463 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19465 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19468 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19470 PR rtl-optimization/68664
19471 * config/arm/arm.c (arm_sched_can_speculate_insn):
19472 New function. Declare prototype.
19473 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19475 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19477 PR rtl-optimization/68664
19478 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19480 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19482 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19484 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19485 max skip bytes for function, loop and jump.
19487 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19489 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19490 ABS_EXPR for gimple dump.
19492 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19495 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19497 PR tree-optimization/79408
19498 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19499 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19500 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19501 also if rhs1 is INTEGER_CST.
19503 2017-02-14 Richard Biener <rguenther@suse.de>
19505 PR middle-end/79432
19506 * tree-into-ssa.c (insert_phi_nodes): When the function can
19507 have abnormal edges rewrite SSA names with broken use-def
19508 dominance out of SSA and register them for PHI insertion.
19510 2017-02-13 Martin Sebor <msebor@redhat.com>
19512 PR middle-end/79496
19513 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19514 clearing info.nowrite flag when snprintf size argument is a range.
19516 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19518 * cprop.c (cprop_jump): Add missing space in string literal.
19519 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19520 (get_constraint_for_component_ref): Likewise.
19521 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19522 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19523 * lra-constraints.c (process_alt_operands): Likewise.
19524 * ipa-inline.c (inline_small_functions): Likewise.
19525 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19526 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19527 * trans-mem.c (diagnose_tm_1_op): Likewise.
19528 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19529 (grid_parallel_clauses_gridifiable): Likewise.
19531 * config/nvptx/mkoffload.c (process): Add space in between
19534 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19535 "MOD4_SSE_REGS" and "ALL_REGS".
19537 * spellcheck.c (test_data): Add , in between "foo" and "food".
19539 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19542 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19543 boundary crossing check and subsequent code generation agree.
19545 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19547 * config/aarch64/aarch64.c (has_memory_op): Delete.
19548 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19551 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19553 PR rtl-optimization/79388
19554 PR rtl-optimization/79450
19555 * combine.c (distribute_notes): When removing TEM_INSN for which
19556 corresponding dest has last value recorded, invalidate that last
19559 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19561 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19562 of explicit '@'. Add missing assembly comment marker on branch costs
19565 2017-02-13 Nathan Sidwell <nathan@acm.org>
19567 * gengtype-lex.l (<in_struct>): Add '/'.
19569 2017-02-13 Martin Liska <mliska@suse.cz>
19572 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19574 2017-02-13 Richard Biener <rguenther@suse.de>
19576 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19578 * configure: Re-generate.
19579 * config.in: Likewise.
19580 * graphite-dependences.c: Simplify as if
19581 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19582 * graphite-isl-ast-to-gimple.c: Likewise.
19583 * graphite-optimize-isl.c: Likewise.
19584 * graphite-poly.c: Likewise.
19585 * graphite-sese-to-poly.c: Likewise.
19586 * graphite.h: Likewise.
19587 * toplev.c: Include isl/version.h and use isl_version () for
19588 printing the ISL version.
19589 * doc/install.texi: Update ISL requirement.
19591 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19593 * doc/standards.texi (Standards): Update reference to
19596 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19598 * doc/extend.texi (Named Address Spaces): sourceware.org now
19600 * doc/install.texi (Binaries): Ditto.
19603 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19605 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19606 standard terminology "stringize"/"stringizing" throughout.
19607 * doc/cppinternals.texi: Likewise.
19609 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19611 * doc/extend.texi: Fix some spelling mistakes and typos.
19612 * doc/invoke.texi: Likewise.
19614 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19617 * params.def (inline-min-speedup) Change from 10 to 8.
19619 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19621 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19624 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19627 * ipa-inline-analysis.c (get_minimal_bb): New function.
19628 (record_modified): Use it.
19629 (remap_edge_change_prob): Handle also ancestor functions.
19631 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19633 * doc/contrib.texi (Contributors): Remove broken link into
19634 the Mauve CVS repository.
19636 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19638 PR middle-end/79454
19639 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19640 result computation whenever lhs doesn't have vector mode, not
19641 just when it has BLKmode.
19643 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19645 * doc/makefile.texi (profiledbootstrap): Refer to the
19646 installation instructions only in textual form.
19648 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19651 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19653 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19655 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19656 (Specific): Update mingw-w64 reference.
19658 (Specific): Remove broken link to Renesas RX processor.
19660 2017-02-10 Richard Biener <rguenther@suse.de>
19662 * toplev.c (process_options): Do not mention obsolete graphite
19663 options when printing sorry message about missing graphite support.
19664 Mention -floop-nest-optimize.
19666 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19668 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19669 (vtst_p16): Likewise.
19670 (vtstq_p8): Likewise.
19671 (vtstq_p16): Likewise.
19673 (vtstq_p64): Likewise.
19674 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19675 (vset_lane_p64): New.
19676 (vsetq_lane_p64): New.
19678 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19680 PR tree-optimization/79411
19681 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19682 stmt operands are SSA_NAMEs used in abnormal phis.
19683 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19686 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19689 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19692 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19694 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19696 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19698 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19702 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19703 not arbitrary TREE_CONSTANT.
19706 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19707 "omp declare target link" attribute unless is_global_var.
19708 * omp-offload.c (find_link_var_op): Likewise.
19710 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19711 Chung-Lin Tang <cltang@codesourcery.com>
19713 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19715 (gimplify_adjust_omp_clauses): Don't delete TILE.
19716 (gimplify_omp_for): Deal with TILE.
19717 * internal-fn.c (expand_GOACC_TILE): New function.
19718 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19720 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19721 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19723 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19724 avoid DIV for outermost collapse var.
19725 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19726 Remove out of date comments, fix whitespace.
19727 * omp-general.c (omp_extract_for_data): Deal with tiling.
19728 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19729 adjust OLF_DIM_BASE value.
19730 (struct omp_for_data): Add tiling field.
19731 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19732 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19733 for auto loops. Remove default auto determining, moved to
19734 oacc_loop_fixed_partitions.
19735 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19736 stmts, add e_mask field.
19737 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19738 (oacc_thread_numbers): Use oacc_dim_call.
19739 (oacc_xform_tile): New.
19740 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19741 (finish_oacc_loop): Adjust for ifns vector.
19742 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19743 add case for GOACC_TILE fns.
19744 (oacc_loop_xform_loop): Delete.
19745 (oacc_loop_process): Iterate over call list directly, and add
19746 handling for GOACC_TILE fns.
19747 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19749 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19750 vector partitioning to outer loops. Assign 2 partitions to loops
19751 when available. Add TILE handling.
19752 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19753 (execite_oacc_device_lower): Process GOACC_TILE fns,
19754 ignore unknown specs.
19755 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19756 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19757 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19759 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19761 * configure.ac (ACX_BUGURL): Update.
19762 * configure: Regenerate.
19764 2017-02-09 Richard Biener <rguenther@suse.de>
19766 PR tree-optimization/69823
19767 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19768 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19770 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19772 * config/arc/arc-c.def: Add __NPS400__ definition.
19773 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19774 (TARGET_NPS400): Define.
19776 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19778 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19780 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19781 pointer, arch_info.
19782 (arc_cpu_types): Fill the arch_info field with a pointer into the
19783 arc_arch_types table.
19784 (arc_selected_cpu): Declare.
19785 * config/arc/arc.c (arc_selected_cpu): Make global.
19786 (arc_selected_arch): Delete.
19787 (arc_base_cpu): Delete.
19788 (arc_override_options): Remove references to deleted variables,
19789 update access to arch information.
19790 (ARC_OPT): Update access to arch information.
19791 (ARC_OPTX): Likewise.
19792 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19793 (TARGET_ARC600): Update access to arch information.
19794 (TARGET_ARC601): Likewise.
19795 (TARGET_ARC700): Likewise.
19796 (TARGET_EM): Likewise.
19797 (TARGET_HS): Likewise.
19798 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19801 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19804 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19805 condition/operands for integer GE/LE/GEU/LEU operations.
19807 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19809 PR translation/79397
19810 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19813 2017-02-08 Martin Jambor <mjambor@suse.cz>
19816 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19817 whether allocation happened.
19818 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19819 nothing was allocated.
19821 2017-02-08 Jakub Jelinek <jakub@redhat.com>
19823 PR tree-optimization/79408
19824 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
19825 constant, but SSA_NAME with a known integer range, use the minimum
19826 of that range instead of op1 to determine if modulo can be replaced
19827 with its first operand.
19829 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19831 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
19833 2017-02-08 Richard Biener <rguenther@suse.de>
19835 PR tree-optimization/71824
19836 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19837 Check all loops contained in the merged region.
19839 2017-02-07 Andrew Pinski <apinski@cavium.com>
19841 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
19843 2017-02-07 Andrew Pinski <apinski@cavium.com>
19845 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
19846 (thunderxt88): Likewise.
19847 (thunderxt81): Disable LSE and change v8.1 to v8.
19848 (thunderxt83): Likewise.
19850 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19851 Richard Biener <rguenther@suse.de>
19853 PR middle-end/79399
19854 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
19855 type from int to size_t.
19856 * ira-costs.c (struct_costs_size): Change type from int to size_t.
19858 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19860 PR rtl-optimization/79386
19861 * cprop.c (bypass_conditional_jumps): Initialize
19862 bypass_last_basic_block already before splitting bbs after
19863 unconditional traps...
19864 (bypass_conditional_jumps): ... rather than here.
19867 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
19868 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
19869 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
19870 fix -masm=intel patterns.
19872 2017-02-07 Richard Biener <rguenther@suse.de>
19874 PR tree-optimization/79256
19875 PR middle-end/79278
19876 * builtins.c (get_object_alignment_2): Use min_align_of_type
19877 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
19878 and ADJUST_FIELD_ALIGN.
19880 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
19882 * doc/tm.texi: Regenerate.
19883 * stor-layout.c (layout_decl): Adjust.
19884 (update_alignment_for_field): Likewise.
19885 (place_field): Likewise.
19886 (min_align_of_type): Likewise.
19887 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
19888 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
19889 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
19890 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
19891 * config/frv/frv.c (frv_adjust_field_align): Likewise.
19892 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
19893 * config/i386/i386.c (x86_field_alignment): Likewise.
19894 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
19895 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
19896 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
19897 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
19898 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
19899 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
19903 2017-01-30 Richard Biener <rguenther@suse.de>
19905 PR tree-optimization/79256
19906 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
19907 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
19910 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
19912 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
19913 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
19914 builtins to SImode and emit a zero-extend, if necessary.
19916 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19918 * docs/invoke.texi (RISC-V Options): Alphabetize.
19920 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19922 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
19925 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19927 * config/riscv/riscv.c: New file.
19928 * common/config/riscv/riscv-common.c: Likewise.
19929 * config.gcc: Likewise.
19930 * config/riscv/constraints.md: Likewise.
19931 * config/riscv/elf.h: Likewise.
19932 * config/riscv/generic.md: Likewise.
19933 * config/riscv/linux.h: Likewise.
19934 * config/riscv/multilib-generator: Likewise.
19935 * config/riscv/peephole.md: Likewise.
19936 * config/riscv/pic.md: Likewise.
19937 * config/riscv/predicates.md: Likewise.
19938 * config/riscv/riscv-builtins.c: Likewise.
19939 * config/riscv/riscv-c.c: Likewise.
19940 * config/riscv/riscv-ftypes.def: Likewise.
19941 * config/riscv/riscv-modes.def: Likewise.
19942 * config/riscv/riscv-opts.h: Likewise.
19943 * config/riscv/riscv-protos.h: Likewise.
19944 * config/riscv/riscv.h: Likewise.
19945 * config/riscv/riscv.md: Likewise.
19946 * config/riscv/riscv.opt: Likewise.
19947 * config/riscv/sync.md: Likewise.
19948 * config/riscv/t-elf-multilib: Likewise.
19949 * config/riscv/t-linux: Likewise.
19950 * config/riscv/t-linux-multilib: Likewise.
19951 * config/riscv/t-riscv: Likewise.
19952 * configure.ac: Likewise.
19953 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
19954 Waterman as RISC-V maintainers.
19955 * doc/install.texi: Add RISC-V entries.
19956 * doc/invoke.texi: Add RISC-V options section.
19957 * doc/md.texi: Add RISC-V constraints section.
19958 * configure: Regenerated.
19960 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19963 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
19964 false values to be constant vectors with all 0 or all 1 bits set.
19965 (vcondu<mode><mode>): Likewise.
19966 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
19968 (fpmask_comparison_operator): Update comment.
19969 (vecint_comparison_operator): New predicate.
19970 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
19971 vector conditionals when the true and false values are constant
19972 vectors with all 0 bits or all 1 bits set.
19974 2017-02-06 Martin Sebor <msebor@redhat.com>
19976 PR tree-optimization/79376
19977 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
19979 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
19981 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
19982 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
19983 to simplify split condition.
19985 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19987 * omp-expand.c (oxpand_omp_atomic_fetch_op,
19988 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
19991 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
19993 PR rtl-optimization/68664
19994 * target.def (can_speculate_insn): New hook.
19995 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
19996 * doc/tm.texi: Regenerate.
19997 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
19998 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
19999 (rs6000_sched_can_speculate_insn): New function.
20001 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20003 PR tree-optimization/79284
20004 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20005 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20006 vectorizable_mask_load_store, vectorizable_operation,
20007 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20008 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20009 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20010 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20011 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20012 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20013 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20014 is_gimple_assign (stmt). Replace another such test with
20015 is_gimple_assign (stmt).
20017 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20020 * config/avr/avr.c (rtl-iter.h): Include it.
20021 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20022 (avr_legitimate_combined_insn): ...and implementation.
20024 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20026 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20027 * config/s390/s390.c (s390_const_operand_ok)
20028 (s390_canonicalize_comparison, s390_extract_part)
20029 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20030 (s390_contiguous_bitmask_p, s390_rtx_costs)
20031 (legitimize_pic_address): Likewise.
20032 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20033 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20034 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20035 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20036 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20038 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20040 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20041 REGNO($0) == REGNO($1).
20043 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20045 * config/s390/linux.h(SIZE_TYPE): Add comment.
20047 2017-02-06 Julian Brown <julian@codesourcery.com>
20048 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20049 Virendra Pathak <virendra.pathak@broadcom.com>
20051 * config/aarch64/aarch64-cores.def: Change the scheduler
20053 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20054 * config/aarch64/thunderx2t99.md: New file.
20056 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20058 * doc/standards.texi (Go Language): Update link to language
20061 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20063 * tree-eh.c (lower_resx): Sanitize profile.
20064 (cleanup_empty_eh_move_lp): Likewise.
20066 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20069 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20071 * cfgloopmanip.h (loop_version): Update prototype.
20072 * modulo-sched.c (sms_schedule): Update call of loop_version.
20073 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20074 * tree-parloops.c (gen_parallel_loop): Likewise.
20075 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20076 * tree-ssa-loop-split.c (split_loop): Likewise.
20077 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20078 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20080 2017-02-05 Martin Liska <mliska@suse.cz>
20083 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20085 (print_operand_address): Initialize a struct to zero.
20087 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20089 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20090 garbage collector only in textual form.
20092 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20094 * doc/extend.texi (x86 specific memory model extensions for
20095 transactional memory): Simplify a phrase.
20097 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20100 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20101 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20102 (atomic_storedi_1): Likewise.
20104 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20106 PR tree-optimization/79338
20107 * tree-parloops.c (gather_scalar_reductions): Don't call
20108 vect_analyze_loop_form for loop->inner before destroying loop's
20111 2017-02-03 Martin Sebor <msebor@redhat.com>
20113 PR tree-optimization/79327
20114 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20115 when precision has resulted in leading zeros.
20116 (format_integer): Adjust the likely counter to assume an unknown
20117 argument that may be zero is non-zero.
20119 2017-02-03 Jason Merrill <jason@redhat.com>
20122 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20123 avoid copying non-taken branch.
20125 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20127 PR tree-optimization/79340
20128 * tree-vect-loop.c (vectorizable_reduction): Release
20129 vec_defs elements after safe_splicing them into other vectors.
20132 PR tree-optimization/79327
20133 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20134 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20136 (format_integer): Use wide_int_to_tree instead of build_int_cst
20137 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20138 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20139 of shortest and longest sequence.
20141 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20143 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20144 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20146 2017-02-03 Walter Lee <walt@tilera.com>
20149 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20150 after initial stackframe link reg save.
20151 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20153 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20156 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20157 wu for stxssp alternative.
20159 2017-02-03 Martin Sebor <msebor@redhat.com>
20161 PR tree-optimization/79352
20162 * gimple-fold.c (get_range_strlen): Add argument.
20163 (get_range_strlen): Change return type to bool.
20164 (get_maxval_strlen): Pass in a dummy argument.
20165 * gimple-fold.h (get_range_strlen): Change return type to bool.
20166 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20167 * tree.h (array_at_struct_end_p): Add argument.
20168 * tree.c (array_at_struct_end_p): Handle it.
20170 2017-02-03 Martin Liska <mliska@suse.cz>
20173 * multiple_target.c (create_dispatcher_calls): Redirect edge
20174 from a caller of a dispatcher.
20175 (expand_target_clones): Make the clones local.
20176 (ipa_target_clone): Do both target clones and resolvers.
20177 (ipa_dispatcher_calls): Remove the pass.
20178 (pass_dispatcher_calls::gate): Likewise.
20179 (make_pass_dispatcher_calls): Likewise.
20180 * passes.def (pass_target_clone): Put as very first IPA early
20183 2017-02-03 Martin Liska <mliska@suse.cz>
20185 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20186 in case of a function with ifunc attribute.
20188 2017-02-03 Martin Liska <mliska@suse.cz>
20190 * cgraph.c (cgraph_node::dump): Dump function version info.
20191 * symtab.c (symtab_node::dump_base): Add missing new line.
20193 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20195 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20196 (ifcombine_ifandif): Use it.
20198 2017-02-03 Martin Liska <mliska@suse.cz>
20200 * doc/invoke.texi: Document default value for
20201 use-after-scope-direct-emission-threshold.
20203 2017-02-03 Martin Liska <mliska@suse.cz>
20205 PR tree-optimization/79339
20206 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20207 (format_floating): Likewise.
20209 2017-02-03 Martin Liska <mliska@suse.cz>
20212 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20214 (ipa_node_params_t::remove): Likewise.
20215 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20216 initialization from removed ipa_node_params_t::insert.
20217 (ipa_node_params::~ipa_node_params): Move from removed
20218 ipa_node_params_t::release.
20219 * symbol-summary.h (symbol_summary::m_released): New member.
20220 Do not release a summary twice. Do not allow to call finalizer
20221 for types of a summary that live in GGC memory.
20223 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20225 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20228 2017-02-02 Martin Sebor <msebor@redhat.com>
20230 PR middle-end/79275
20231 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20232 (format_string): Tighten up the range of output for non-constant
20233 strings and correct the expected range for wide non-constant strings.
20235 2017-02-02 Martin Sebor <msebor@redhat.com>
20237 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20239 PR middle-end/32003
20240 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20242 (-fdump-tree-@var): Add to index and document how to come up
20243 with pass-specific option and dump file names.
20244 (-fdump-passes): Clarify where to look for output.
20246 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20248 PR middle-end/77445
20249 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20250 statistics of the analyzed path; allow threading for speed when
20251 any of BBs along the path are optimized for speed.
20253 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20255 PR middle-end/78468
20256 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20257 settings of the virtual registers.
20260 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20262 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20263 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20266 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20268 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20269 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20271 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20273 * config/s390/s390.md: Add missing comments with the expanded
20275 * config/s390/vector.md: Likewise.
20276 * config/s390/vx-builtins.md: Likewise.
20278 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20281 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20282 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20283 conditions on a single line.
20285 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20287 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20288 __S390_VX__ to __VX__.
20290 2017-02-01 Andrew Pinski <apinski@cavium.com>
20292 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20293 stmt_info to record_stmt_cost.
20294 (vect_get_known_peeling_cost): Pass stmt_info if known to
20296 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20297 cpu_vector_cost field into
20298 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20299 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20300 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20301 splitting of scalar_stmt_cost and vec_stmt_cost.
20302 (thunderx_vector_cost): Likewise.
20303 (cortexa57_vector_cost): LIkewise.
20304 (exynosm1_vector_cost): Likewise.
20305 (xgene1_vector_cost): Likewise.
20306 (thunderx2t99_vector_cost): Improve after the splitting of the two
20308 (aarch64_builtin_vectorization_cost): Update for the splitting of
20309 scalar_stmt_cost and vec_stmt_cost.
20311 2017-02-01 Torvald Riegel <triegel@redhat.com>
20312 Richard Henderson <rth@redhat.com>
20314 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20315 conditional on existance of a fast atomic load.
20316 * optabs-query.c (can_atomic_load_p): New function.
20317 * optabs-query.h (can_atomic_load_p): Declare it.
20318 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20319 no fast atomic load is available for the particular size of access.
20320 (expand_atomic_compare_and_swap): Likewise.
20321 (expand_atomic_load): Likewise.
20322 (expand_atomic_store): Likewise.
20323 (expand_atomic_fetch_op): Likewise.
20324 * testsuite/lib/target-supports.exp
20325 (check_effective_target_sync_int_128): Remove x86 because it provides
20326 no fast atomic load.
20327 (check_effective_target_sync_int_128_runtime): Likewise.
20329 2017-02-01 Richard Biener <rguenther@suse.de>
20331 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20332 (graphite_transform_loops): Provide opt-info for optimized nests.
20333 * tree-parloop.c (parallelize_loops): Provide opt-info for
20334 parallelized loops.
20336 2017-02-01 Richard Biener <rguenther@suse.de>
20338 PR middle-end/79315
20339 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20340 was not set before.
20342 2017-02-01 Richard Biener <rguenther@suse.de>
20344 PR tree-optimization/71824
20345 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20346 Verify the loops are valid in the merged SESE region.
20347 (scop_detection::can_represent_loop_1): Check analyzing the
20348 evolution of the number of iterations in the region succeeds.
20350 2017-01-31 Ian Lance Taylor <iant@golang.org>
20352 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20353 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20355 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20357 PR preprocessor/79210
20358 * input.c (get_substring_ranges_for_loc): Replace line_width
20359 assertion with error-handling.
20361 2017-01-31 Richard Biener <rguenther@suse.de>
20363 PR tree-optimization/77318
20364 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20365 (create_pw_aff_from_tree): Take loop parameter.
20366 (add_condition_to_pbb): Pass loop of the condition to
20367 create_pw_aff_from_tree.
20369 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20371 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20372 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20374 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20378 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20380 (convert_int_to_float128): Likewise.
20381 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20382 (convert_int_to_float128): Likewise.
20383 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20384 (UNSPEC_IEEE128_CONVERT): Likewise.
20385 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20386 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20387 Use local variables for IBM extended format.
20388 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20389 (fix_trunc<mode>si2_fprs): Likewise.
20390 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20391 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20392 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20393 to know that we can now have integers of all sizes in vector
20395 (fix<uns>_<mode>di2_hw): Likewise.
20396 (float<uns>_<mode>si2_hw): Likewise.
20397 (fix_<mode>si2_hw): Likewise.
20398 (fixuns_<mode>si2_hw): Likewise.
20399 (float<uns>_<mode>di2_hw): Likewise.
20400 (float_<mode>di2_hw): Likewise.
20401 (float_<mode>si2_hw): Likewise.
20402 (floatuns_<mode>di2_hw): Likewise.
20403 (floatuns_<mode>si2_hw): Likewise.
20404 (xscvqp<su>wz_<mode>): Delete, no longer used.
20405 (xscvqp<su>dz_<mode>): Likewise.
20406 (xscv<su>dqp_<mode>): Likewise.
20407 (ieee128_mfvsrd_64bit): Likewise.
20408 (ieee128_mfvsrd_32bit): Likewise.
20409 (ieee128_mfvsrwz): Likewise.
20410 (ieee128_mtvsrw): Likewise.
20411 (ieee128_mtvsrd_64bit): Likewise.
20412 (ieee128_mtvsrd_32bit): Likewise.
20414 2017-01-31 Martin Liska <mliska@suse.cz>
20417 * ipa-prop.c (ipa_free_all_node_params): Call release method
20418 instead of ~sumbol_summary to not to trigger double times
20421 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20423 PR tree-optimization/71691
20424 * bitmap.h (class auto_bitmap): New.
20425 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20426 is_maybe_undefined instead of ssa_undefined_value_p.
20428 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20430 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20431 __S390_ARCH_LEVEL__ to __ARCH__.
20433 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20435 PR tree-optimization/79267
20436 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20437 if should_remove_lhs_p is true.
20439 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20442 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20443 (add_alignment_attribute): New.
20444 (base_type_die): Add alignment attribute.
20445 (subrange_type_die): Likewise.
20446 (modified_type_die): Likewise.
20447 (gen_array_type_die): Likewise.
20448 (gen_descr_array_type_die: Likewise.
20449 (gen_enumeration_type_die): Likewise.
20450 (gen_subprogram_die): Likewise.
20451 (gen_variable_die): Likewise.
20452 (gen_field_die): Likewise.
20453 (gen_ptr_to_mbr_type_die): Likewise.
20454 (gen_struct_or_union_type_die): Likewise.
20455 (gen_subroutine_type_die): Likewise.
20456 (gen_typedef_die): Likewise.
20457 (base_type_cmp): Compare alignment attribute.
20459 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20462 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20463 (setb_unsigned) New pattern for setb with CCUNS.
20464 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20465 subfc./subfe sequence to avoid overflow problems. Generate a
20466 shorter sequence with cmpld/setb for power9.
20467 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20468 for generating subfc. instruction.
20469 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20470 now uses this instruction.
20472 2017-01-30 Ian Lance Taylor <iant@google.com>
20475 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20476 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20478 2017-01-30 Martin Sebor <msebor@redhat.com>
20480 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20481 Move constant to the right of a relational operator.
20482 (get_mpfr_format_length, format_character, format_string): Ditto.
20483 (should_warn_p, maybe_warn): Same.
20485 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20487 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20490 * asan.c (get_translation_unit_decl): Remove function.
20491 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20493 2017-01-30 Martin Liska <mliska@suse.cz>
20495 PR gcov-profile/79259
20496 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20497 -fprofile-generate.
20499 2017-01-30 Martin Liska <mliska@suse.cz>
20502 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20503 Initialize variables with NULL value.
20505 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20508 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20510 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20512 2017-01-30 Richard Biener <rguenther@suse.de>
20514 PR tree-optimization/79276
20515 * tree-vrp.c (process_assert_insertions): Properly adjust common
20516 when removing a duplicate.
20518 * gcc.dg/torture/pr79276.c: New testcase.
20520 2017-01-30 Richard Biener <rguenther@suse.de>
20522 PR tree-optimization/79256
20523 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20524 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20526 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20528 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20531 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20532 ("*r<noxa>sbg_<mode>_sll_bitmask")
20533 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20534 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20535 Use contiguous_bitmask_nowrap_operand.
20537 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20540 * config/rs6000/altivec.h (vec_xl): Revise #define.
20541 (vec_xst): Likewise.
20543 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20545 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20547 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20549 PR rtl-optimization/79194
20550 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20551 traps before call to bypass_conditional_jumps.
20553 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20555 PR tree-optimization/71374
20556 * lra-constraints.c (check_conflict_input_operands): New.
20557 (match_reload): Use it.
20559 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20562 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20563 account to calculate conflict_set.
20565 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20567 PR rtl-optimization/78559
20568 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20569 other_insn in combine.
20571 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20573 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20574 uint16_type_node for BT_UINT16.
20576 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20578 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20579 "RTL Tests" to menu.
20580 (GIMPLE Tests): New node.
20581 (RTL Tests): New node.
20583 2017-01-27 Richard Biener <rguenther@suse.de>
20585 PR tree-optimization/79245
20586 * tree-loop-distribution.c (distribute_loop): Apply cost
20587 modeling also to detected patterns.
20589 2017-01-27 Richard Biener <rguenther@suse.de>
20591 PR tree-optimization/71433
20592 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20593 (compare_assert_loc): New function.
20594 (process_assert_insertions): Sort and optimize assert locations
20595 to remove duplicates and push down identical assertions on
20596 edges to their destination block.
20598 2017-01-27 Richard Biener <rguenther@suse.de>
20600 PR tree-optimization/79244
20601 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20602 out SSA names even if abnormal.
20604 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20606 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20607 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20608 instead of MPFR_RNDN.
20610 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20613 * arm.c (arm_option_override): Don't call build_target_option_node
20614 until after doing all option overrides.
20615 (arm_valid_target_attribute_tree): Likewise.
20617 2017-01-27 Martin Liska <mliska@suse.cz>
20619 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20620 for {cd}tors and C++ {cd}tors.
20622 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20624 * config/s390/s390.md ("*setmem_long_and")
20625 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20627 2017-01-26 Martin Sebor <msebor@redhat.com>
20629 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20632 2017-01-26 Martin Sebor <msebor@redhat.com>
20634 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20635 HAVE_DFmode before using XFmode or DFmode.
20636 (parse_directive): Avoid using the z length modifier to avoid
20637 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20639 PR middle-end/78703
20640 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20641 to accept adjustment as an array.
20642 (get_int_range): New function.
20643 (struct directive): Make width and prec arrays.
20644 (directive::set_width, directive::set_precision): Call get_int_range.
20645 (format_integer, format_floating): Handle width and precision ranges.
20646 (format_string, parse_directive): Same.
20648 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20651 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20652 comdat_type_p set, just clone them, but keep the children in the
20656 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20657 which have direct callers with -fvar-tracking-assignments enabled
20659 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20660 inside of type units.
20662 2017-01-26 Martin Sebor <msebor@redhat.com>
20664 PR middle-end/78703
20665 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20667 (struct format_result): Replace number_chars, number_chars_min,
20668 and number_chars_max with a single member of struct result_range.
20670 (format_result::operator+=): Adjust.
20671 (struct fmtresult): Remove bounded. Handle likely and unlikely
20673 (fmtresult::adjust_for_width_or_precision): New function.
20674 (fmtresult:type_max_digits): New function.
20675 (bytes_remaining): Handle likely and unlikely counters.
20676 (min_bytes_remaining): Remove.
20677 (format_percent): Simplify.
20678 (format_integer, format_floating): Set likely and unlikely counters.
20679 (get_string_length, format_character, format_string): Same.
20680 (format_plain, should_warn_p): New function.
20681 (maybe_warn): Call should_warn_p. Update diagnostic messages
20682 and handle those for all directives, including plain strings.
20683 (format_directive): Handle likely and unlikely counters.
20684 Remove unnecessary quoting from diagnostics. Add an informational
20686 (add_bytes): Remove.
20687 (pass_sprintf_length::compute_format_length): Simplify.
20688 (try_substitute_return_value): Handle likely and unlikely counters.
20690 2017-01-26 Carl Love <cel@us.ibm.com>
20692 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20693 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20695 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20698 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20699 endianess for subregs into account.
20700 * lra-constraints.c (lra_constraints): Do risky transformations
20701 always on the first iteration.
20702 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20703 last_call_used_reg_set.
20704 (process_bb_lives): Define and use last_call_used_reg_set.
20705 * lra.c (lra): Always continue after lra_constraints on the first
20708 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20710 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20712 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20714 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20716 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20717 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20718 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20719 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20720 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20721 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20722 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20723 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20724 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20726 2017-01-26 Marek Polacek <polacek@redhat.com>
20729 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20730 for the third operand.
20732 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20734 PR middle-end/79236
20735 * omp-low.c (struct omp_context): Add simt_stmt field.
20736 (scan_omp_for): Return omp_context *.
20737 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20738 context to the _simt_ SIMD stmt.
20739 (lower_omp_for): For combined SIMD with sibling _simt_
20740 SIMD, make sure to use the same decls in _looptemp_
20741 clauses as in the sibling.
20743 2017-01-26 David Sherwood <david.sherwood@arm.com>
20745 PR middle-end/79212
20746 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20749 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20752 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20753 emit fld b; fld a; if possible.
20755 * brig-builtins.def: Update copyright years.
20756 * config/arm/arm_acle_builtins.def: Update copyright years.
20758 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20761 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20762 constraint instead of o for the stxsd instruction.
20764 2017-01-25 Carl Love <cel@us.ibm.com>
20766 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20767 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20769 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20771 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20773 2017-01-25 Richard Biener <rguenther@suse.de>
20775 PR tree-optimization/69264
20776 * target.def (vector_alignment_reachable): Improve documentation.
20777 * doc/tm.texi: Regenerate.
20778 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20780 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20781 earlier changes with respect to TYPE_USER_ALIGN.
20782 (vector_alignment_reachable_p): Likewise. Improve dumping.
20784 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20787 * config/arm/arm.md (xordi3): Force constant operand into a register
20790 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20792 * doc/invoke.texi (-fstore-merging): Correct default optimization
20793 levels at which it is enabled.
20794 (-O): Move -fstore-merging from list to...
20797 2017-01-25 Richard Biener <rguenther@suse.de>
20800 * omp-expand.c: Include debug.h.
20801 (expand_omp_taskreg): Make sure to generate early debug before
20802 outlining anything from a function.
20803 (expand_omp_target): Likewise.
20804 (grid_expand_target_grid_body): Likewise.
20806 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20809 * asan.c (get_translation_unit_decl): New function.
20810 (asan_add_global): Extract modules file name from globals
20811 TRANSLATION_UNIT_DECL name.
20813 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20816 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20817 for long calls with APCS frame and VFP.
20819 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20821 * cfg.c (original_copy_tables_initialized_p): New function.
20822 * cfg.h (original_copy_tables_initialized_p): New decl.
20823 * cfgrtl.c (relink_block_chain): Guard the call to
20824 free_original_copy_tables with a call to
20825 original_copy_tables_initialized_p.
20826 * cgraph.h (symtab_node::native_rtl_p): New decl.
20827 * cgraphunit.c (symtab_node::native_rtl_p): New function.
20828 (symtab_node::needed_p): Don't assert for early assembly output
20829 for __RTL functions.
20830 (cgraph_node::finalize_function): Set "force_output" for __RTL
20832 (cgraph_node::analyze): Bail out early for __RTL functions.
20833 (analyze_functions): Update assertion to support __RTL functions.
20834 (cgraph_node::expand): Bail out early for __RTL functions.
20835 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
20837 * function.h (struct function): Update comment for field
20839 * gimple-expr.c: Include "tree-pass.h".
20840 (gimple_has_body_p): Return false for __RTL functions.
20841 * Makefile.in (OBJS): Add run-rtl-passes.o.
20842 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
20844 (gcc::pass_manager::get_clean_slate): New accessor.
20845 * passes.c: Include "insn-addr.h".
20846 (should_skip_pass_p): Add logging. Update logic for running
20847 "expand" to be compatible with both __GIMPLE and __RTL. Guard
20848 property-provider override so it is only done for gimple passes.
20850 (skip_pass): New function.
20851 (execute_one_pass): Call skip_pass when skipping passes.
20852 * read-md.c (md_reader::read_char): Support filtering
20853 the input to a subset of line numbers.
20854 (md_reader::md_reader): Initialize fields
20855 m_first_line and m_last_line.
20856 (md_reader::read_file_fragment): New function.
20857 * read-md.h (md_reader::read_file_fragment): New decl.
20858 (md_reader::m_first_line): New field.
20859 (md_reader::m_last_line): New field.
20860 * read-rtl-function.c (function_reader::create_function): Only
20861 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
20862 curr_properties. Set DECL_INITIAL to a dummy block.
20863 (read_rtl_function_body_from_file_range): New function.
20864 * read-rtl-function.h (read_rtl_function_body_from_file_range):
20866 * run-rtl-passes.c: New file.
20867 * run-rtl-passes.h: New file.
20869 2017-01-24 Jeff Law <law@redhat.com>
20871 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
20874 2017-01-24 Bin Cheng <bin.cheng@arm.com>
20876 PR tree-optimization/79159
20877 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
20878 (record_nonwrapping_iv): Improve boundary using above function if no
20879 value range information.
20881 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
20882 Martin Jambor <mjambor@suse.cz>
20884 * brig-builtins.def: New file.
20885 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
20886 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
20887 (DEF_HSAIL_SAT_BUILTIN): Likewise.
20888 (DEF_HSAIL_INTR_BUILTIN): Likewise.
20889 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
20890 * builtin-types.def (BT_INT8): New.
20891 (BT_INT16): Likewise.
20892 (BT_UINT8): Likewise.
20893 (BT_UINT16): Likewise.
20894 (BT_FN_ULONG): Likewise.
20895 (BT_FN_UINT_INT): Likewise.
20896 (BT_FN_UINT_ULONG): Likewise.
20897 (BT_FN_UINT_LONG): Likewise.
20898 (BT_FN_UINT_PTR): Likewise.
20899 (BT_FN_ULONG_PTR): Likewise.
20900 (BT_FN_INT8_FLOAT): Likewise.
20901 (BT_FN_INT16_FLOAT): Likewise.
20902 (BT_FN_UINT32_FLOAT): Likewise.
20903 (BT_FN_UINT16_FLOAT): Likewise.
20904 (BT_FN_UINT8_FLOAT): Likewise.
20905 (BT_FN_UINT64_FLOAT): Likewise.
20906 (BT_FN_UINT16_UINT32): Likewise.
20907 (BT_FN_UINT32_UINT16): Likewise.
20908 (BT_FN_UINT16_UINT16_UINT16): Likewise.
20909 (BT_FN_INT_PTR_INT): Likewise.
20910 (BT_FN_UINT_PTR_UINT): Likewise.
20911 (BT_FN_LONG_PTR_LONG): Likewise.
20912 (BT_FN_ULONG_PTR_ULONG): Likewise.
20913 (BT_FN_VOID_UINT64_UINT64): Likewise.
20914 (BT_FN_UINT8_UINT8_UINT8): Likewise.
20915 (BT_FN_INT8_INT8_INT8): Likewise.
20916 (BT_FN_INT16_INT16_INT16): Likewise.
20917 (BT_FN_INT_INT_INT): Likewise.
20918 (BT_FN_UINT_FLOAT_UINT): Likewise.
20919 (BT_FN_FLOAT_UINT_UINT): Likewise.
20920 (BT_FN_ULONG_UINT_UINT): Likewise.
20921 (BT_FN_ULONG_UINT_PTR): Likewise.
20922 (BT_FN_ULONG_ULONG_ULONG): Likewise.
20923 (BT_FN_UINT_UINT_UINT): Likewise.
20924 (BT_FN_VOID_UINT_PTR): Likewise.
20925 (BT_FN_UINT_UINT_PTR: Likewise.
20926 (BT_FN_UINT32_UINT64_PTR): Likewise.
20927 (BT_FN_INT_INT_UINT_UINT): Likewise.
20928 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
20929 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
20930 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
20931 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
20932 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
20933 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
20934 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
20935 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
20936 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
20937 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
20938 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
20939 * doc/frontends.texi: List BRIG FE.
20940 * doc/install.texi (Testing): Add BRIG tesring requirements.
20941 * doc/invoke.texi (Overall Options): Mention BRIG.
20942 * doc/standards.texi (Standards): Doucment BRIG HSA version.
20944 2017-01-24 Richard Biener <rguenther@suse.de>
20946 PR translation/79208
20947 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
20949 2017-01-24 Martin Jambor <mjambor@suse.cz>
20952 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
20953 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
20954 and known_contexts.
20956 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
20958 PR middle-end/79123
20959 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
20960 casts from signed to unsigned really don't have a range.
20962 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
20964 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
20965 GMP_RNDx for compatiblity.
20967 2017-01-24 Martin Liska <mliska@suse.cz>
20970 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
20971 that would prevent us to call alloca with -1 as argument.
20973 2017-01-24 Jakub Jelinek <jakub@redhat.com>
20975 * dwarf2out.c (output_compilation_unit_header, output_file_names):
20976 Avoid -Wformat-security warning.
20978 2017-01-23 Andrew Pinski <apinski@cavium.com>
20980 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
20983 2017-01-23 Martin Sebor <msebor@redhat.com>
20985 PR middle-end/78703
20986 * gimple-ssa-sprintf.c (warn_level): New global.
20987 (format_integer): Use it here and throughout the rest of the file.
20988 Use the same switch to compute sign as base.
20989 (maybe_warn): New function.
20990 (format_directive): Factor out warnings into maybe_warn.
20991 Add debugging output. Use warn_level.
20992 (add_bytes): Use warn_level.
20993 (pass_sprintf_length::compute_format_length): Add debugging output.
20994 (try_substitute_return_value): Same.
20995 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
20997 PR middle-end/78703
20998 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
20999 (struct fmtresult, format_integer, format_floating): Adjust.
21000 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21001 (get_string_length, format_string,format_directive): Same.
21002 (pass_sprintf_length::compute_format_length): Same.
21003 (try_substitute_return_value): Simplify slightly.
21005 PR middle-end/78703
21006 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21007 (fmtresult::operator+=): Outlined.
21008 (struct fmtresult): Add ctors.
21009 (struct conversion_spec): Rename...
21010 (struct directive): ...to this. Add and remove data members.
21011 (directive::set_width, directive::set_precision): New functions.
21012 (format_percent): Use fmtresult ctor.
21013 (get_width_and_precision): Remove.
21014 (format_integer): Make naming changes. Avoid computing width and
21016 (format_floating): Same. Adjust indentation.
21017 (format_character, format_none): New functions.
21018 (format_string): Moved character handling to format_character.
21019 (format_directive): Remove arguments, change return type.
21020 (parse_directive): New function.
21021 (pass_sprintf_length::compute_format_length): Move directive
21022 parsing to parse_directive.
21024 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21026 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21027 (assign_assembler_name_if_needed): ... this.
21028 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21029 (assign_assembler_name_if_needed): ... this.
21030 (free_lang_data_in_cgraph): Adjust callers.
21031 * cgraphunit.c (cgraph_node::analyze): Likewise.
21032 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21035 2017-01-23 Richard Biener <rguenther@suse.de>
21037 PR tree-optimization/79088
21038 PR tree-optimization/79188
21039 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21040 resetting loop bounds after last path deletion. Reset loop
21041 bounds of the target loop, make code match the comments.
21042 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21043 Make sure loops need no fixups.
21045 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21047 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21048 exponent support with double type for first argument.
21049 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21050 type returned by __builtin_vec_extract_sig,
21051 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21052 functions from "vector int" to "vector unsigned int" or from
21053 "vector long long int" to "vector unsigned long long int".
21054 Changed type returned by __builtin_vec_extract_exp,
21055 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21056 functions from "vector int" to "vector unsigned int" or from
21057 "vector long long int" to "vector unsigned long long int".
21058 Changed return type of __builtin_vec_test_data_class,
21059 __builtin_vec_test_data_class_sp, and
21060 __builtin_vec_test_data_class_dp from "vector int" to
21061 "vector bool int" or from "vector long long int" to "vector bool
21062 long long int" and changed second argument type from "unsigned
21063 int" to "int". Added new overloaded function forms "vector float
21064 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21065 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21066 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21067 double, vector unsigned long long int)" and "vector double
21068 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21069 long int)". Changed return type of
21070 __builtin_scalar_test_data_class and
21071 __builtin_scalar_test_data_class_sp and
21072 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21073 int" and changed second argument from "unsigned int" to "int".
21074 Changed type returned by __builtin_scalar_test_neg,
21075 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21076 from "int" to "bool int". Added new overloaded function form
21077 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21078 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21079 exponent double-precision with floating point first argument.
21080 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21081 documentation of scalar_test_data_class, scalar_test_neg,
21082 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21083 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21084 vec_test_data_class built-in functions to reflect refinements in
21085 their type signatures.
21087 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21089 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21091 (aarch64_elf_asm_destructor): Likewise.
21093 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21095 PR rtl-optimization/78634
21096 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21097 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21098 * ifcvt.c (noce_try_cmove): Add missing cost check.
21100 PR rtl-optimization/71724
21101 * combine.c (if_then_else_cond): Look for situations where it is
21102 beneficial to undo the work of one of the recursive calls.
21104 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21106 PR tree-optimization/70754
21107 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21108 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21109 combined stmt before it if not NULL.
21110 (combine_chains): Process refs reversely and compute dominance point
21113 2017-01-23 Martin Liska <mliska@suse.cz>
21115 PR tree-optimization/79196
21116 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21117 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21119 (strlen_optimize_stmt): Call the renamed function.
21121 2017-01-23 Michael Matz <matz@suse.de>
21123 PR tree-optimization/78384
21124 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21126 2017-01-23 Richard Biener <rguenther@suse.de>
21128 PR tree-optimization/79186
21129 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21130 both incoming edges before moving an assert.
21132 2017-01-23 Martin Jambor <mjambor@suse.cz>
21134 * ipa-prop.c (load_from_param_1): Removed.
21135 (load_from_unmodified_param): Bits from load_from_param_1 put back
21137 (load_from_param): Removed.
21138 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21139 with stmt. Reverted back to use of load_from_unmodified_param.
21141 2017-01-23 Martin Jambor <mjambor@suse.cz>
21144 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21145 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21146 field a pointer to garbage collected vector, mark lattices and
21147 ipcp_orig_node with GTY((skip)).
21148 (ipa_get_param_count): Adjust to descriptors being a pointer.
21149 (ipa_get_param): Likewise.
21150 (ipa_get_type): Likewise.
21151 (ipa_get_param_move_cost): Likewise.
21152 (ipa_set_param_used): Likewise.
21153 (ipa_get_controlled_uses): Likewise.
21154 (ipa_set_controlled_uses): Likewise.
21155 (ipa_is_param_used): Likewise.
21156 (ipa_node_params_t): Move into garbage collector. New methods insert
21158 (ipa_node_params_sum): Annotate wth GTY(()).
21159 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21161 (ipa_load_from_parm_agg): Adjust declaration.
21162 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21163 * ipa-profile.c (ipa_profile): Likewise.
21164 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21165 (ipa_populate_param_decls): Make descriptors parameter garbage
21167 (ipa_dump_param): Adjust to descriptors being a pointer.
21168 (ipa_alloc_node_params): Likewise.
21169 (ipa_initialize_node_params): Likewise.
21170 (load_from_param_1): Make descriptors parameter garbage collected.
21171 (load_from_unmodified_param): Likewise.
21172 (load_from_param): Likewise.
21173 (ipa_load_from_parm_agg): Likewise.
21174 (ipa_node_params::~ipa_node_params): Removed.
21175 (ipa_free_all_node_params): Remove call to delete operator.
21176 (ipa_node_params_t::insert): New.
21177 (ipa_node_params_t::remove): Likewise.
21178 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21179 copy known_csts and known_contexts vectors.
21180 (ipa_read_node_info): Adjust to descriptors being a pointer.
21181 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21183 (ipcp_transform_function): Make descriptors variable garbage
21186 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21188 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21189 * config/i386/avx512dqintrin.h: Ditto.
21190 * config/i386/avx512fintrin.h: Ditto.
21191 * config/i386/i386.c: Handle new builtins.
21192 * config/i386/i386-builtin.def: Add new builtins.
21193 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21194 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21196 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21197 Martin Liska <mliska@suse.cz>
21199 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21200 * asan.c (asan_expand_poison_ifn): Support stores and use
21201 appropriate ASAN report function.
21202 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21203 * internal-fn.def (ASAN_POISON_USE): Declare.
21204 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21205 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21206 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21207 ASAN_POISON calls w/o LHS.
21208 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21209 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21210 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21211 * gimplify.c (asan_poison_variables): Add attribute
21212 use_after_scope_memory to variables that really needs to live
21214 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21215 having the attribute.
21217 2017-01-23 Martin Liska <mliska@suse.cz>
21219 * asan.c (create_asan_shadow_var): New function.
21220 (asan_expand_poison_ifn): Likewise.
21221 * asan.h (asan_expand_poison_ifn): New declaration.
21222 * internal-fn.c (expand_ASAN_POISON): Likewise.
21223 * internal-fn.def (ASAN_POISON): New builtin.
21224 * sanopt.c (pass_sanopt::execute): Expand
21225 asan_expand_poison_ifn.
21226 * tree-inline.c (copy_decl_for_dup_finish): Make function
21228 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21229 * tree-ssa.c (is_asan_mark_p): New function.
21230 (execute_update_addresses_taken): Rewrite local variables
21231 (identified just by use-after-scope as addressable) into SSA.
21233 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21235 * doc/install.texi (Specific): opensource.apple.com uses https
21236 now. Remove trailing slash.
21238 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21240 * README.Portability: Remove note on an Irix compatibility issue.
21242 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21244 * gcov.c (INCLUDE_ALGORITHM): Define.
21245 (INCLUDE_VECTOR): Define.
21246 No longer include <vector> and <algorithm> directly.
21248 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21250 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21252 * doc/invoke.texi (Code Gen Options): Ditto.
21254 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21257 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21259 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21261 rtl-optimization/79125
21262 * cprop.c (local_cprop_pass): Handle cases where we make an
21263 unconditional trap.
21265 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21269 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21270 read from, for big endian.
21272 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21274 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21275 register pauth builtins for LP64 only.
21277 2017-01-20 Marek Polacek <polacek@redhat.com>
21280 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21283 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21285 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21287 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21288 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21289 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21291 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21294 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21295 in big-endian mode when they are not a single duplicated value.
21297 2017-01-20 Richard Biener <rguenther@suse.de>
21299 * BASE-VER: Bump to 7.0.1.
21301 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21303 * omp-low.c (omplow_simd_context): New struct. Use it...
21304 (lower_rec_simd_input_clauses): ...here and...
21305 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21306 references to idx, lane, max_vf, is_simt.
21308 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21310 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21313 2017-01-20 Martin Jambor <mjambor@suse.cz>
21315 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21316 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21318 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21319 (GTFILES): Rename hsa.c to hsa-common.c.
21320 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21321 * hsa-dump.c: Likewise.
21322 * hsa-gen.c: Likewise.
21323 * hsa-regalloc.c: Likewise.
21324 * ipa-hsa.c: Likewise.
21325 * omp-expand.c: Likewise.
21326 * omp-low.c: Likewise.
21327 * toplev.c: Likewise.
21329 2017-01-20 Marek Polacek <polacek@redhat.com>
21332 * doc/invoke.texi: Document -Wduplicated-branches.
21333 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21334 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21335 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21336 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21337 return 0 only when not OEP_LEXICOGRAPHIC.
21338 (fold_build_cleanup_point_expr): Use the expression
21339 location when building CLEANUP_POINT_EXPR.
21340 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21341 * tree.c (add_expr): Handle error_mark_node.
21343 2017-01-20 Martin Liska <mliska@suse.cz>
21346 * tree-profile.c (init_ic_make_global_vars): Do not call
21348 (gimple_init_gcov_profiler): Likewise.
21350 2017-01-20 Martin Liska <mliska@suse.cz>
21353 * cgraph.h (maybe_create_reference): Remove argument and
21355 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21357 * ipa-cp.c (create_specialized_node): Likewise.
21358 * symtab.c (symtab_node::maybe_create_reference): Handle
21359 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21361 2017-01-20 Martin Liska <mliska@suse.cz>
21363 * read-rtl-function.c (function_reader::create_function): Use
21364 build_decl instread of build_decl_stat.
21366 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21368 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21369 * config/i386/avx512dqintrin.h: Ditto.
21370 * config/i386/avx512fintrin.h: Ditto.
21371 * config/i386/i386-builtin-types.def: Add new types.
21372 * config/i386/i386.c: Handle new types.
21373 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21374 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21375 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21376 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21377 (__builtin_ia32_kshiftridi): New.
21378 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21380 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21384 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21385 define to rs6000_init_stack_protect_guard.
21386 (rs6000_init_stack_protect_guard): New function.
21388 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21389 Yunqiang Su <yunqiang.su@imgtec.com>
21391 * config.gcc (supported_defaults): Add madd4.
21392 (with_madd4): Add validation.
21393 (all_defaults): Add madd4.
21394 * config/mips/mips.opt (mmadd4): New option.
21395 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21397 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21399 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21400 (ISA_HAS_FUSED_MADD4): Likewise.
21401 * doc/invoke.texi (-mmadd4): Document the new option.
21402 * doc/install.texi (--with-madd4): Document the new option.
21404 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21406 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21407 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21408 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21409 (aarch64_init_pauth_hint_builtins): New.
21410 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21411 (aarch64_expand_builtin): Expand new builtins.
21413 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21415 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21416 * combine-stack-adj.c (no_unhandled_cfa): Handle
21417 REG_CFA_TOGGLE_RA_MANGLE.
21418 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21419 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21420 info for return address signing.
21421 (aarch64_expand_epilogue): Likewise.
21423 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21425 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21426 * config/aarch64/aarch64-protos.h
21427 (aarch64_return_address_signing_enabled): New declaration.
21428 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21430 (aarch64_expand_prologue): Sign return address before it's pushed onto
21432 (aarch64_expand_epilogue): Authenticate return address fetched from
21434 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21435 (aarch64_attributes): New function attributes for "sign-return-address".
21436 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21437 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21438 ("*do_return"): Generate combined instructions according to key index.
21439 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21440 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21442 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21443 * config/aarch64/aarch64.opt (msign-return-address=): New.
21444 * doc/extend.texi (AArch64 Function Attributes): Documents
21445 "sign-return-address=".
21446 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21448 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21450 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21451 overall option summary.
21453 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21455 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21456 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21457 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21458 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21460 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21462 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21463 -mpower9-minmax by default for -mcpu=power9.
21464 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21465 128-bit floating point.
21467 2017-01-20 Alan Modra <amodra@gmail.com>
21469 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21470 optimizing for size.
21472 2017-01-20 Alan Modra <amodra@gmail.com>
21475 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21476 for strcmp and strncmp from corresponding builtin decl.
21478 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21480 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21481 instead of i386/rtems-64.h.
21482 * config/i386/rtems-64.h: Remove.
21484 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21488 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21490 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21492 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21494 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21495 Change int to HOST_WIDE_INT.
21496 * config/aarch64/aarch64-protos.h
21497 (aarch64_simd_gen_const_vector_dup): Likewise.
21498 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21500 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21502 * langhooks-def.h (lhd_type_for_size): New decl.
21503 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21504 * langhooks.c (lhd_type_for_size): New function, taken from
21507 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21509 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21510 define_bypass for CR latency.
21511 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21512 (power9-alu2): Add define_bypass for CR latency.
21514 (power9-mul): Update insn latency.
21515 (power9-mul-compare): Update insn latency, bypass latency and remove
21518 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21520 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21522 * config/aarch64/aarch64.md
21523 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21524 aarch64_nopcrelative_literal_loads.
21525 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21527 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21529 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21530 TARGET_LOONGSON_3A.
21531 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21533 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21536 * config.gcc (supported_defaults): Add lxc1-sxc1.
21537 (with_lxc1_sxc1): Add validation.
21538 (all_defaults): Add lxc1-sxc1.
21539 * config/mips/mips.opt (mlxc1-sxc1): New option.
21540 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21542 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21543 __mips_no_lxc1_sxc1.
21544 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21545 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21546 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21548 2017-01-19 Richard Biener <rguenther@suse.de>
21550 PR tree-optimization/72488
21551 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21552 sure to restore SSA info.
21553 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21555 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21557 PR rtl-optimization/79121
21558 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21559 of the inner type when shifting an extended value.
21561 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21564 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21565 interposable aliases.
21567 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21570 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21571 Use the evmergelohi instruction.
21572 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21573 (mov_sitf_e500_subreg8_2_be): Likewise.
21574 (mov_sitf_e500_subreg12_2_le): Likewise.
21575 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21576 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21577 (mov_sitf_e500_subreg8_2_le): Likewise.
21578 (mov_sitf_e500_subreg12_2_be): Likewise.
21580 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21582 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21583 attribute from vecsimple to vecperm.
21584 (altivec_vbpermq2): Likewise.
21586 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21589 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21591 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21592 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21593 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21594 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21595 case where N arg is SIZE_MAX.
21596 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21597 (cmpstrsi): Add pattern.
21599 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21602 __builtin_vec_revb builtins.
21603 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21604 built-in functions to support generation of the ISA 3.0 XXBR<x>
21605 vector byte reverse instructions.
21606 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21607 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21608 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21609 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21610 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21611 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21612 (P9V_BUILTIN_VEC_REVB): Likewise.
21613 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21614 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21615 (p9_xxbrq_v16qi): Likewise.
21616 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21617 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21618 (p9_xxbrh_v8hi): Likewise.
21619 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21620 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21621 vec_revb built-in functions.
21623 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21625 PR rtl-optimization/78952
21626 * config/i386/i386.md (any_extract): New code iterator.
21627 (*insvqi_2): Use any_extract for source operand.
21628 (*insvqi_3): Use any_shiftrt for source operand.
21630 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21632 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21634 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21636 2017-01-18 Matthias Klose <doko@ubuntu.com>
21638 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21640 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21642 * config/rs6000/altivec.h (vec_bperm): Change #define.
21643 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21644 (altivec_vbpermq2): New define_insn.
21645 (altivec_vbpermd): Likewise.
21646 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21647 function interface.
21648 (VBPERMD): Likewise.
21649 (VBPERM): New polymorphic function interface.
21650 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21651 Add entries for P9V_BUILTIN_VEC_VBPERM.
21652 * doc/extend.texi: Add interfaces for vec_bperm.
21654 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21656 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21657 first letter of error messages.
21658 (s390_resolve_overloaded_builtin): Likewise.
21659 * config/s390/s390.c (s390_expand_builtin): Likewise.
21660 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21661 (s390_valid_target_attribute_inner_p): Likewise.
21662 * config/s390/s390.md ("tabort"): Likewise.
21664 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21666 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21667 (ISA_AVOID_DIV_HILO): New macro.
21668 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21669 (ISA_HAS_DDIV): Likewise.
21671 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21673 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21675 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21677 * doc/invoke.texi (fabi-version): Spelling fix.
21679 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21682 * doc/invoke.texi (fabi-version): Mention mangling fix for
21685 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21688 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21690 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21693 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21694 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21696 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21698 (stack_protect_test): Ditto.
21699 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21700 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21702 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21703 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21704 -mstack-protector-guard-offset=.
21705 (RS/6000 and PowerPC Options): Ditto.
21707 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21709 * config/i386/i386.h (MASK_CLASS_P): New define.
21710 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21711 there are no registers from different register sets also when
21712 mask registers are used. Update function comment.
21713 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21714 to (*k/*r) and (*k/*km) alternatives.
21716 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21718 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21719 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21720 (EH_RETURN_HANDLER_RTX): New define.
21721 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21722 Force frame pointer in EH return functions.
21723 (aarch64_expand_epilogue): Add barrier for eh_return.
21724 (aarch64_final_eh_return_addr): Remove.
21725 (aarch64_eh_return_handler_rtx): New function.
21726 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21728 (aarch64_eh_return_handler_rtx): New prototype.
21730 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21732 * config/rs6000/altivec.h (vec_rlmi): New #define.
21733 (vec_vrlnm): Likewise.
21734 (vec_rlnm): Likewise.
21735 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21736 (UNSPEC_VRLNM): Likewise.
21737 (VIlong): New mode iterator.
21738 (altivec_vrl<VI_char>mi): New define_insn.
21739 (altivec_vrl<VI_char>nm): Likewise.
21740 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21742 (VRLDNM): Likewise.
21743 (RLNM): New polymorphic function entry.
21744 (VRLWMI): New monomorphic function entry.
21745 (VRLDMI): Likewise.
21746 (RLMI): New polymorphic function entry.
21747 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21748 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21749 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21752 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21755 * dwarf2out.c (field_byte_offset): Restore the
21756 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21757 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21758 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21761 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21764 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21766 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21769 * dwarf2out.c (add_data_member_location_attribute): For constant
21770 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21771 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21772 DW_AT_byte_size attributes.
21774 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21776 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21777 after forcing to constant memory when the code model is medium.
21779 2017-01-17 Julia Koval <julia.koval@intel.com>
21782 * config/i386/avx512fintrin.h
21783 (_mm512_i32gather_ps): Change __addr type to void const*.
21784 (_mm512_mask_i32gather_ps): Ditto.
21785 (_mm512_i32gather_pd): Ditto.
21786 (_mm512_mask_i32gather_pd): Ditto.
21787 (_mm512_i64gather_ps): Ditto.
21788 (_mm512_mask_i64gather_ps): Ditto.
21789 (_mm512_i64gather_pd): Ditto.
21790 (_mm512_mask_i64gather_pd): Ditto.
21791 (_mm512_i32gather_epi32): Ditto.
21792 (_mm512_mask_i32gather_epi32): Ditto.
21793 (_mm512_i32gather_epi64): Ditto.
21794 (_mm512_mask_i32gather_epi64): Ditto.
21795 (_mm512_i64gather_epi32): Ditto.
21796 (_mm512_mask_i64gather_epi32): Ditto.
21797 (_mm512_i64gather_epi64): Ditto.
21798 (_mm512_mask_i64gather_epi64): Ditto.
21799 (_mm512_i32scatter_ps): Change __addr type to void*.
21800 (_mm512_mask_i32scatter_ps): Ditto.
21801 (_mm512_i32scatter_pd): Ditto.
21802 (_mm512_mask_i32scatter_pd): Ditto.
21803 (_mm512_i64scatter_ps): Ditto.
21804 (_mm512_mask_i64scatter_ps): Ditto.
21805 (_mm512_i64scatter_pd): Ditto.
21806 (_mm512_mask_i64scatter_pd): Ditto.
21807 (_mm512_i32scatter_epi32): Ditto.
21808 (_mm512_mask_i32scatter_epi32): Ditto.
21809 (_mm512_i32scatter_epi64): Ditto.
21810 (_mm512_mask_i32scatter_epi64): Ditto.
21811 (_mm512_i64scatter_epi32): Ditto.
21812 (_mm512_mask_i64scatter_epi32): Ditto.
21813 (_mm512_i64scatter_epi64): Ditto.
21814 (_mm512_mask_i64scatter_epi64): Ditto.
21815 * config/i386/avx512pfintrin.h
21816 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21817 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21818 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21819 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21820 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21821 (_mm512_prefetch_i32scatter_ps): Ditto.
21822 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
21823 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
21824 (_mm512_prefetch_i64scatter_pd): Ditto.
21825 (_mm512_prefetch_i64scatter_ps): Ditto.
21826 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
21827 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
21828 * config/i386/avx512vlintrin.h
21829 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
21830 (_mm_mmask_i32gather_ps): Ditto.
21831 (_mm256_mmask_i32gather_pd): Ditto.
21832 (_mm_mmask_i32gather_pd): Ditto.
21833 (_mm256_mmask_i64gather_ps): Ditto.
21834 (_mm_mmask_i64gather_ps): Ditto.
21835 (_mm256_mmask_i64gather_pd): Ditto.
21836 (_mm_mmask_i64gather_pd): Ditto.
21837 (_mm256_mmask_i32gather_epi32): Ditto.
21838 (_mm_mmask_i32gather_epi32): Ditto.
21839 (_mm256_mmask_i32gather_epi64): Ditto.
21840 (_mm_mmask_i32gather_epi64): Ditto.
21841 (_mm256_mmask_i64gather_epi32): Ditto.
21842 (_mm_mmask_i64gather_epi32): Ditto.
21843 (_mm256_mmask_i64gather_epi64): Ditto.
21844 (_mm_mmask_i64gather_epi64): Ditto.
21845 (_mm256_i32scatter_ps): Change __addr type to void*.
21846 (_mm256_mask_i32scatter_ps): Ditto.
21847 (_mm_i32scatter_ps): Ditto.
21848 (_mm_mask_i32scatter_ps): Ditto.
21849 (_mm256_i32scatter_pd): Ditto.
21850 (_mm256_mask_i32scatter_pd): Ditto.
21851 (_mm_i32scatter_pd): Ditto.
21852 (_mm_mask_i32scatter_pd): Ditto.
21853 (_mm256_i64scatter_ps): Ditto.
21854 (_mm256_mask_i64scatter_ps): Ditto.
21855 (_mm_i64scatter_ps): Ditto.
21856 (_mm_mask_i64scatter_ps): Ditto.
21857 (_mm256_i64scatter_pd): Ditto.
21858 (_mm256_mask_i64scatter_pd): Ditto.
21859 (_mm_i64scatter_pd): Ditto.
21860 (_mm_mask_i64scatter_pd): Ditto.
21861 (_mm256_i32scatter_epi32): Ditto.
21862 (_mm256_mask_i32scatter_epi32): Ditto.
21863 (_mm_i32scatter_epi32): Ditto.
21864 (_mm_mask_i32scatter_epi32): Ditto.
21865 (_mm256_i32scatter_epi64): Ditto.
21866 (_mm256_mask_i32scatter_epi64): Ditto.
21867 (_mm_i32scatter_epi64): Ditto.
21868 (_mm_mask_i32scatter_epi64): Ditto.
21869 (_mm256_i64scatter_epi32): Ditto.
21870 (_mm256_mask_i64scatter_epi32): Ditto.
21871 (_mm_i64scatter_epi32): Ditto.
21872 (_mm_mask_i64scatter_epi32): Ditto.
21873 (_mm256_i64scatter_epi64): Ditto.
21874 (_mm256_mask_i64scatter_epi64): Ditto.
21875 (_mm_i64scatter_epi64): Ditto.
21876 (_mm_mask_i64scatter_epi64): Ditto.
21877 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
21878 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
21879 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
21880 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
21881 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
21882 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
21883 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
21884 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
21885 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
21886 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
21887 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
21888 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
21889 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
21890 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
21891 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
21892 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
21893 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
21894 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
21895 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
21896 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
21897 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
21898 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
21899 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
21900 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
21901 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
21902 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
21903 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
21904 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
21905 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
21906 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
21907 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
21908 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
21909 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
21910 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
21911 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
21912 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
21913 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
21914 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
21915 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
21916 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
21917 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
21918 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
21919 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
21920 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
21921 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
21922 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
21923 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
21924 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
21925 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
21926 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
21927 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
21928 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
21929 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
21930 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
21931 definitions accordingly.
21933 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
21934 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21937 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
21940 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
21943 * ira-conflicts.c (ira_build_conflicts): Update total conflict
21944 hard regs for inner regno.
21946 2017-01-17 Martin Liska <mliska@suse.cz>
21949 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
21950 assumption and add comment.
21952 2017-01-17 Nathan Sidwell <nathan@acm.org>
21954 * ipa-visibility.c (localize_node): New function, broken out of ...
21955 (function_and_variable_visibility): ... here. Call it.
21957 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21959 PR middle-end/77445
21960 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
21961 correctly set frequency of oudgoing edge.
21962 (duplicate_thread_path): Fix profile updating.
21964 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21967 * configure.ac: Add GCC_BASE_VER.
21968 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
21969 version from BASE-VER file.
21970 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
21971 (gcc.o): Depend on $(BASEVER).
21972 * common.opt (dumpfullversion): New option.
21973 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
21974 * doc/invoke.texi: Document -dumpfullversion.
21975 * doc/install.texi: Document --with-gcc-major-version-only.
21976 * configure: Regenerated.
21978 2017-01-17 Richard Biener <rguenther@suse.de>
21980 PR tree-optimization/71433
21981 * tree-vrp.c (register_new_assert_for): Merge same asserts
21982 on all incoming edges.
21983 (process_assert_insertions_for): Handle insertions at the
21986 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
21988 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
21989 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
21991 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
21994 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
21997 2017-01-17 Alan Modra <amodra@gmail.com>
22000 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22001 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22002 symbolic stack limit when pic.
22004 2017-01-16 Martin Sebor <msebor@redhat.com>
22006 PR tree-optimization/78608
22007 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22009 2017-01-16 Jeff Law <law@redhat.com>
22012 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22013 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22014 for several include directories that may be relative to sysroot.
22015 * config/i386/x-mingw32 (gplus_includedir): Define.
22016 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22017 (native_system_includedir): Likewise.
22018 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22019 override if TARGET_SYSTEM_ROOT is defined.
22020 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22022 PR tree-optimization/79090
22023 PR tree-optimization/33562
22024 PR tree-optimization/61912
22025 PR tree-optimization/77485
22026 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22027 and computed trims into the dump file.
22029 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22031 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22033 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22036 * gimplify.c (gimplify_init_constructor): If want_value and
22037 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22041 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22042 sequence. Formatting fixes.
22043 (doloop_optimize): Formatting fixes.
22046 * gcc.c (debug_level_greater_than_spec_func): New function.
22047 (static_spec_functions): Add debug-level-gt spec function.
22048 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22050 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22051 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22052 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22053 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22054 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22055 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22057 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22059 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22060 QImode fixups to general and mask registers only.
22062 2017-01-16 Carl Love <cel@us.ibm.com>
22064 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22065 for built-in functions
22066 vector signed char vec_nabs (vector signed char)
22067 vector signed short vec_nabs (vector signed short)
22068 vector signed int vec_nabs (vector signed int)
22069 vector signed long long vec_nabs (vector signed long long)
22070 vector float vec_nabs (vector float)
22071 vector double vec_nabs (vector double)
22072 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22074 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22075 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22076 * doc/extend.texi: Update the documentation file for the new built-in
22079 2017-01-16 Martin Sebor <msebor@redhat.com>
22081 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22084 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22086 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22087 UNSPEC_VSX__XXSPLTD to require special splat handling.
22089 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22092 * system.h: Poison strndup.
22094 2017-01-16 Alan Modra <amodra@gmail.com>
22097 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22100 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22102 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22104 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22106 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22107 call recog here. Assert that INSN_CODE (insn) is non-negative.
22109 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22112 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22114 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22115 in the currently scheduled RTL fragment.
22117 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22119 PR rtl-optimization/78751
22120 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22123 2017-01-14 Jeff Law <law@redhat.com>
22125 PR tree-optimization/79090
22126 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22127 variable length stores.
22128 (compute_trims): Delete dead assignment to *trim_tail.
22129 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22132 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22134 PR rtl-optimization/78626
22135 PR rtl-optimization/78727
22136 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22137 of a block, and split such blocks after everything else is finished.
22139 2017-01-14 Alan Modra <amodra@gmail.com>
22142 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22143 target legitimate_combined_insn.
22144 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22145 (rs6000_legitimate_combined_insn): New function.
22146 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22148 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22149 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22150 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22152 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22154 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22156 2017-01-13 Jeff Law <law@redhat.com>
22158 PR tree-optimization/33562
22159 PR tree-optimization/61912
22160 PR tree-optimization/77485
22161 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22163 (delete_dead_assignment): Likewise.
22164 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22165 statement to delete_dead_call and delete_dead_assignment.
22167 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22170 * substring-locations.c (format_warning_va): Strengthen case 1 so
22171 that both endpoints of the substring must be within the format
22172 range for just the substring to be printed.
22174 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22176 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22177 * config/i386/i386.c (ix86_target_string): Add missing options
22178 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22179 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22180 flags_other and ix86_target_other to flags2_other. Display unknown
22182 (ix86_valid_target_attribute_inner_p): Add missing options and
22183 reorder options by implied ISAs, as in ix86_target_string.
22185 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22187 * hash-table.h (hash_table::too_empty_p): New function.
22188 (hash_table::expand): Use it.
22189 (hash_table::traverse): Likewise.
22190 (hash_table::empty_slot): Use sizeof (value_type) instead of
22191 sizeof (PTR) to convert bytes to elements. Shrink the table
22192 if the current size is excessive for the current number of
22195 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22197 * ira-costs.c (record_reg_classes): Break from the inner loop
22198 early once alt_fail is known to be true. Update outer loop
22199 handling accordingly.
22201 2017-01-13 Jeff Law <law@redhat.com>
22203 * tree-ssa-dse.c (decrement_count): New function.
22204 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22205 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22206 when we know the partially dead statement is a mem* function.
22208 PR tree-optimization/61912
22209 PR tree-optimization/77485
22210 * tree-ssa-dse.c: Include expr.h.
22211 (maybe_trim_constructor_store): New function.
22212 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22214 PR tree-optimization/33562
22215 PR tree-optimization/61912
22216 PR tree-optimization/77485
22217 * doc/invoke.texi: Document new dse-max-object-size param.
22218 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22219 * tree-ssa-dse.c: Include params.h.
22220 (dse_store_status): New enum.
22221 (initialize_ao_ref_for_dse): New, partially extracted from
22223 (valid_ao_ref_for_dse, normalize_ref): New.
22224 (setup_live_bytes_from_ref, compute_trims): Likewise.
22225 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22226 (maybe_trim_partially_dead_store): Likewise.
22227 (maybe_trim_complex_store): Likewise.
22228 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22229 Track what bytes live from the original store. Return tri-state
22230 for dead, partially dead or live.
22231 (dse_dom_walker): Add constructor, destructor and new private members.
22232 (delete_dead_call, delete_dead_assignment): New extracted from
22234 (dse_optimize_stmt): Make a member of dse_dom_walker.
22235 Use initialize_ao_ref_for_dse.
22237 PR tree-optimization/33562
22238 PR tree-optimization/61912
22239 PR tree-optimization/77485
22240 * sbitmap.h (bitmap_count_bits): Prototype.
22241 (bitmap_clear_range, bitmap_set_range): Likewise.
22242 * sbitmap.c (bitmap_clear_range): New function.
22243 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22245 2017-01-13 Martin Liska <mliska@suse.cz>
22248 * function.c (set_cfun): Add new argument force.
22249 * function.h (set_cfun): Likewise.
22250 * ipa-inline-transform.c (inline_call): Use the function when
22251 strict alising from is dropped for function we inline to.
22253 2017-01-13 Richard Biener <rguenther@suse.de>
22255 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22256 for dumping GIMPLE INTEGER_CSTs.
22258 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22260 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22261 to 201112L since C++17.
22263 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22266 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22267 if -fsanitize=kernel-address is present.
22269 2017-01-13 Richard Biener <rguenther@suse.de>
22271 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22272 as _Literal ( type ) number in case usual suffixes do not
22273 preserve all information.
22275 2017-01-13 Richard Biener <rguenther@suse.de>
22277 PR tree-optimization/77283
22278 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22279 and ssa-iterators.h.
22280 (is_feasible_trace): Implement a cost model based on joiner
22283 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22286 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22287 char or short to __float128/_Float128 directly.
22289 2017-01-12 Martin Sebor <msebor@redhat.com>
22291 to -Wformat-overflow.
22292 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22293 (min_bytes_remaining): Same.
22294 (get_string_length): Same.
22295 (format_string): Same.
22296 (format_directive): Same.
22298 (pass_sprintf_length::handle_gimple_call): Same.
22300 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22302 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22303 info.nowrite calls with no lhs that can't throw. Return bool
22304 whether gsi_remove has been called or not.
22305 (pass_sprintf_length::handle_gimple_call): Return bool whether
22306 try_substitute_return_value called gsi_remove. Formatting fix.
22307 (pass_sprintf_length::execute): Don't use gsi_remove if
22308 handle_gimple_call returned true.
22311 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22312 be removed due to side-effects, don't remove following barrier nor
22313 turn the successor edge into fallthru edge.
22315 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22318 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22319 element-reversing loads and stores as not swappable.
22321 2017-01-12 Nathan Sidwell <nathan@acm.org>
22322 Nicolai Stange <nicstange@gmail.com>
22324 * combine.c (try_combine): Don't ignore result of overlap checking
22325 loop. Combine overlap & asm check into single loop.
22327 2017-01-12 Richard Biener <rguenther@suse.de>
22329 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22330 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22332 2017-01-12 Richard Biener <rguenther@suse.de>
22334 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22335 and TS_TARGET_OPTION directly derive from TS_BASE.
22336 * tree-core.h (tree_optimization_option): Derive from tree_base.
22337 (tree_target_option): Likewise.
22339 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22341 * config/i386/i386.c (memory_address_length): Increase len
22342 only when rip_relative_addr_p returns false.
22344 2017-01-11 Julia Koval <julia.koval@intel.com>
22346 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22347 (OPTION_MASK_ISA_SGX_SET): New.
22348 (ix86_handle_option): Handle OPT_msgx.
22349 * config.gcc: Added sgxintrin.h.
22350 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22351 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22352 * config/i386/i386.c (ix86_target_string): Add -msgx.
22354 (ix86_option_override_internal): Handle new options.
22355 (ix86_valid_target_attribute_inner_p): Add sgx.
22356 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22357 * config/i386/i386.opt: Add msgx.
22358 * config/i386/sgxintrin.h: New file.
22359 * config/i386/x86intrin.h: Add sgxintrin.h.
22361 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22364 * fold-const.c (maybe_nonzero_address): Return 1 for function
22366 (tree_single_nonzero_warnv_p): Don't handle function local objects
22370 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22373 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22376 * opts.c: Include "spellcheck.h"
22377 (struct string_fragment): New struct.
22378 (struct edit_distance_traits<const string_fragment &>): New
22380 (get_closest_sanitizer_option): New function.
22381 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22383 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22385 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22387 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22388 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22389 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22390 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22391 for initial die_offset if dwarf_split_debug_info.
22392 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22393 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22395 (output_skeleton_debug_sections): Formatting fix. Use
22396 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22397 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22399 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22401 * config/arm/cortex-a53.md: Add bypasses for
22402 cortex_a53_r2f_cvt.
22403 (cortex_a53_r2f): Only use for transfers.
22404 (cortex_a53_f2r): Likewise.
22405 (cortex_a53_r2f_cvt): Add reservation for conversions.
22406 (cortex_a53_f2r_cvt): Likewise.
22408 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22410 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22411 to all inlined functions, change static to extern.
22413 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22416 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22418 (arm_assemble_integer): Likewise.
22420 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22422 * config.gcc: Use new awk script to check CPU, FPU and architecture
22423 parameters for --with-... options.
22424 * config/arm/parsecpu.awk: New file
22425 * config/arm/arm-cpus.in: New file.
22426 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22428 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22430 * config/arm/t-arm: Update dependency rules.
22431 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22432 of processing .def files.
22433 * config/arm/genopt.sh: Deleted.
22434 * config/arm/gentune.sh: Deleted.
22435 * config/arm/arm-cores.def: Deleted.
22436 * config/arm/arm-arches.def: Deleted.
22437 * config/arm/arm-fpus.def: Deleted.
22438 * config/arm/arm-tune.md: Regenerated.
22439 * config/arm/arm-tables.opt: Regenerated.
22440 * config/arm/arm-cpu.h: New generated file.
22441 * config/arm/arm-cpu-data.h: New generated file.
22442 * config/arm/arm-cpu-cdata.h: New generated file.
22444 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22447 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22449 (input_varpool_node): Unpack dynamically_initialized bit.
22451 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22453 PR rtl-optimization/79032
22454 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22455 the alignment of the adjusted memory reference against that of MODE,
22456 instead of the alignment of the original memory reference.
22458 2017-01-11 Martin Jambor <mjambor@suse.cz>
22460 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22462 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22463 decorated functions.
22465 2017-01-11 Richard Biener <rguenther@suse.de>
22467 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22468 set range/nonnull info for PHI results. Do not set it on
22469 stmts marked for removal.
22471 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22473 * expr.c (store_field): In the bitfield case, fetch the return value
22474 from the registers before applying a single big-endian adjustment.
22475 Always do a final load for a BLKmode value not larger than a word.
22477 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22480 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22481 that we correctly handle column numbers greater than
22482 LINE_MAP_MAX_COLUMN_NUMBER.
22484 2017-01-10 Martin Sebor <msebor@redhat.com>
22486 PR middle-end/78245
22487 * gimple-ssa-sprintf.c (get_destination_size): Call
22488 {init,fini}object_sizes.
22489 * tree-object-size.c (addr_object_size): Adjust.
22490 (pass_through_call): Adjust.
22491 (pass_object_sizes::execute): Adjust.
22492 * tree-object-size.h (fini_object_sizes): Declare.
22494 2017-01-10 Martin Sebor <msebor@redhat.com>
22496 PR tree-optimization/78775
22497 * builtins.c (get_size_range): Move...
22498 * calls.c: ...to here.
22499 (alloc_max_size): Accept zero argument.
22500 (operand_signed_p): Remove.
22501 (maybe_warn_alloc_args_overflow): Call get_size_range.
22502 * calls.h (get_size_range): Declare.
22504 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22506 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22507 from TI's devices.csv file as of September 2016.
22508 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22510 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22512 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22513 * doc/invoke.texi: Likewise.
22514 * doc/md.texi: Likewise.
22515 * doc/objc.texi: Likewise.
22517 2017-01-10 Joshua Conner <joshconner@google.com>
22519 * config/arm/fuchsia-elf.h: New file.
22520 * config/fuchsia.h: New file.
22521 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22522 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22524 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22526 2016-01-10 Richard Biener <rguenther@suse.de>
22528 PR tree-optimization/79034
22529 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22530 Propagate out degenerate PHIs in the joiner.
22532 2017-01-10 Martin Liska <mliska@suse.cz>
22534 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22535 (sort_congruence_classes_by_decl_uid): Likewise.
22536 (sort_congruence_class_groups_by_decl_uid): Likewise.
22537 (sem_item_optimizer::merge_classes): Sort class, groups in these
22538 classes and members in the groups by DECL_UID of declarations.
22539 This would make merge operations stable.
22541 2017-01-10 Martin Liska <mliska@suse.cz>
22543 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22544 usage of m_classes_vec.
22545 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22546 (sem_item_optimizer::get_group_by_hash): Likewise.
22547 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22548 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22549 (sem_item_optimizer::verify_classes): Likewise.
22550 (sem_item_optimizer::process_cong_reduction): Likewise.
22551 (sem_item_optimizer::dump_cong_classes): Likewise.
22552 (sem_item_optimizer::merge_classes): Likewise.
22553 * ipa-icf.h (congruence_class_hash): Rename from
22554 congruence_class_group_hash. Remove declaration of m_classes_vec.
22556 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22558 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22559 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22560 * config.gcc: Add avx512vpopcntdqintrin.h.
22561 * config/i386/avx512vpopcntdqintrin.h: New.
22562 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22563 * config/i386/i386-builtin-types.def: Add new types.
22564 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22565 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22566 __builtin_ia32_vpopcountq_v8di_mask): New.
22567 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22568 __AVX512VPOPCNTDQ__.
22569 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22570 (PTA_AVX512VPOPCNTDQ): Define.
22571 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22572 TARGET_AVX512VPOPCNTDQ_P): Define.
22573 * config/i386/i386.opt: Add mavx512vpopcntdq.
22574 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22575 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22577 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22579 PR middle-end/77484
22580 * predict.def (PRED_CALL): Set to 67.
22582 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22584 * expr.c (store_field): In the bitfield case, if the value comes from
22585 a function call and is of an aggregate type returned in registers, do
22586 not modify the field mode; extract the value in all cases if the mode
22587 is BLKmode and the size is not larger than a word.
22589 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22592 * config/i386/cpuid.h: Fix undefined behavior.
22594 2017-01-04 Jeff Law <law@redhat.com>
22596 PR tree-optimization/79007
22597 PR tree-optimization/67955
22598 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22599 conservative for pt.null when flag_non_call_exceptions is on.
22601 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22603 PR translation/79019
22604 PR translation/79020
22605 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22606 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22607 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22609 * config/avr/avr.opt (maccumulate-args): Likewise.
22610 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22611 * common.opt (freport-bug): Likewise.
22612 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22613 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22614 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22615 translatable string.
22616 * config/i386/i386.c (function_value_32): Likewise.
22617 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22618 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22620 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22621 * common/config/msp430/msp430-common.c (msp430_handle_option):
22623 * symtab.c (symtab_node::verify_base): Likewise.
22624 * opts.c (set_debug_level): Likewise.
22625 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22626 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22627 missing whitespace to translatable strings.
22628 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22629 * config/sh/superh.h: Likewise.
22630 * config/i386/xopintrin.h: Likewise.
22631 * config/i386/znver1.md: Likewise.
22632 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22633 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22634 * double-int.h (struct double_int): Likewise.
22635 * double-int.c (div_and_round_double): Likewise.
22636 * wide-int.cc: Likewise.
22637 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22638 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22639 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22640 (crossjumps_occurred): ... this.
22641 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22644 PR tree-optimization/78899
22645 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22646 returning bool return struct loop *, NULL for failure and the new
22648 (versionable_outer_loop_p): Don't version outer loop if it has
22649 dont_vectorized bit set.
22650 (tree_if_conversion): When versioning outer loop, ensure
22651 tree_if_conversion is performed also on the inner loop of the
22652 non-vectorizable outer loop copy.
22653 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22654 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22655 prevent vectorization of it.
22656 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22657 the outer loop vectorization of the non-scalar version is attempted
22658 before vectorization of the inner loop in scalar version. If
22659 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22660 vectorization of its inner loop.
22661 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22662 has 2 inner loops, rename also on edges from bb whose single pred
22663 is outer_loop->header. Fix typo in function comment.
22665 2017-01-09 Martin Sebor <msebor@redhat.com>
22668 * asan.c (asan_emit_stack_protection): Increase local buffer size
22669 to avoid snprintf truncation warning.
22671 2017-01-09 Andrew Pinski <apinski@cavium.com>
22673 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22674 to reference thunderx2t99 for the tuning structure
22675 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22677 (thunderx2t99_extra_costs): This.
22678 * config/aarch64/aarch64-tune.md: Regenerate.
22679 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22680 (vulcan_addrcost_table): This.
22681 (vulcan_regmove_cost): Rename to ...
22682 (thunderx2t99_regmove_cost): This.
22683 (vulcan_vector_cost): Rename to ...
22684 (thunderx2t99_vector_cost): this.
22685 (vulcan_branch_cost): Rename to ...
22686 (thunderx2t99_branch_cost): This.
22687 (vulcan_tunings): Rename to ...
22688 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22689 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22691 2017-01-09 Martin Jambor <mjambor@suse.cz>
22695 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22696 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22697 (propagate_vr_accross_jump_function): Use the above function for all
22698 value range computations for pass-through jump functions and type
22699 converasion from explicit value range values.
22700 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22701 parameters from TYPE_ARG_TYPES.
22702 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22703 (ipa_write_node_info): Stream type of the actual argument.
22704 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22706 2017-01-09 Martin Liska <mliska@suse.cz>
22709 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22710 (lookup_compiler): Do not show error message with have_E.
22712 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22714 PR tree-optimization/78938
22715 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22716 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22717 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22718 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22721 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22723 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22726 2017-01-09 Richard Biener <rguenther@suse.de>
22728 PR tree-optimization/78997
22729 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22730 name condition properly.
22732 2017-01-09 Richard Biener <rguenther@suse.de>
22735 * dwarf2out.c (is_cxx): New overload with context.
22736 (is_naming_typedef_decl): Use it.
22738 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22740 * invoke.texi (Option Summary): Correct spacing in option lists
22741 and add line breaks to fix over-long lines.
22743 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22745 PR middle-end/17660
22747 * extend.texi (Common Variable Attributes): Add xref to GCC
22748 Internals manual to explain mode attribute keywords.
22750 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22753 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22754 and Preprocessor Options.
22755 (Options for Linking): Document -pthread here....
22756 (RS/6000 and PowerPC Options): ...not here.
22757 (Solaris 2 Options): ...or here.
22758 * doc/cppopts.texi: Document -pthread.
22760 2017-01-08 Martin Sebor <msebor@redhat.com>
22762 PR middle-end/77708
22763 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22764 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22765 New member functions.
22766 (format_directive): Used them.
22768 (pass_sprintf_length::handle_gimple_call): Same.
22769 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22770 to avoid truncation for any argument.
22771 (extract_affine_mul): Same.
22772 * tree.c (get_file_function_name): Same.
22774 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22776 PR middle-end/77484
22777 * predict.def (PRED_INDIR_CALL): Set to 86.
22779 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22781 PR preprocessor/54124
22782 * doc/cppopts.texi: Reformat -d subtable to list the full name
22783 of the options. Add cross-reference to the docs for the general
22784 compiler -d options.
22785 * doc/invoke.texi (Developer Options): Add cross-reference to the
22786 preprocessor-specific -d option documentation.
22788 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22790 PR preprocessor/13498
22791 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22792 redudant material, and reflect new command-line options.
22793 (System Headers): Likewise.
22795 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22797 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22798 -isystem, and -idirafter. Copy-edit.
22799 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22800 default for -ftrack-macro-expansion. Delete obsolete and
22801 badly-formatted implementation details about -fdebug-cpp output.
22802 * doc/cppwarnopts.texi: Copy-edit.
22804 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22807 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22808 that the transition from a max line width >= 1<<10 to narrower
22809 lines works correctly.
22811 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22813 * doc/options.texi (PerFunction): New.
22814 * opt-functions.awk (switch_flags): Map both Optimization and
22815 PerFunction to CL_OPTIMIZATION.
22816 * opth-gen.awk: Test for PerFunction flag along with
22818 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22819 it only when the latter is present. Skip those that don't in
22820 the hash function generator.
22821 * common.opt (fvar-tracking): Mark as PerFunction instead of
22823 (fvar-tracking-assignments): Likewise.
22824 (fvar-tracking-assignments-toggle): Likewise.
22825 (fvar-tracking-uninit): Likewise.
22827 2017-01-07 Jakub Jelinek <jakub@redhat.com>
22829 PR translation/79018
22830 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
22833 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
22836 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
22837 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
22838 TARGET_LONG_JUMP_TABLE_OFFSETS.
22839 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
22840 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
22841 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
22842 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
22843 * config/m68k/m68k.md (tablejump expander): Likewise.
22844 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
22845 TARGET_LONG_JUMP_TABLE_OFFSETS.
22846 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
22847 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
22849 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
22850 David Holsgrove <david.holsgrove@xilinx.com>
22852 * common/config/microblaze/microblaze-common.c
22853 (TARGET_EXCEPT_UNWIND_INFO): Remove.
22854 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
22856 * config/microblaze/microblaze.c (microblaze_must_save_register)
22857 (microblaze_expand_epilogue, microblaze_return_addr): Handle
22859 (microblaze_eh_return): New function.
22860 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
22861 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
22862 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
22863 * config/microblaze/microblaze.md (eh_return): New pattern.
22865 2017-01-06 Jakub Jelinek <jakub@redhat.com>
22867 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
22868 GCC_DIAGNOSTIC_STRINGIFY): Define.
22870 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
22872 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22874 * config/arm/arm.md (<mcrr>): New.
22876 * config/arm/arm.c (arm_arch5te): New.
22877 (arm_option_override): Set arm_arch5te.
22878 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
22880 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
22881 (arm_mcrr_qualifiers): ... this. New.
22882 (MRRC_QUALIFIERS): Define to...
22883 (arm_mrrc_qualifiers): ... this. New.
22884 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
22886 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
22887 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
22888 (MRRCI, mrrc, MRRC): New.
22889 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
22890 VUNSPEC_MRRC2): New.
22892 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22894 * config/arm/arm.md (<mcr>): New.
22896 * config/arm/arm.c (arm_coproc_builtin_available): Add
22897 support for mcr, mrc, mcr2 and mrc2.
22898 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
22899 (arm_mcr_qualifiers): ... this. New.
22900 (MRC_QUALIFIERS): Define to ...
22901 (arm_mrc_qualifiers): ... this. New.
22902 (MCR_QUALIFIERS): Define to ...
22903 (arm_mcr_qualifiers): ... this. New.
22904 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
22906 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
22907 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
22908 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
22909 VUNSPEC_MRC2): New.
22911 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22913 * config/arm/arm.md (*ldc): New.
22917 * config/arm/arm.c (arm_coproc_builtin_available): Add
22918 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
22919 (arm_coproc_ldc_stc_legitimate_address): New.
22920 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
22921 'qualifier_const_pointer'.
22922 (LDC_QUALIFIERS): Define to...
22923 (arm_ldc_qualifiers): ... this. New.
22924 (STC_QUALIFIERS): Define to...
22925 (arm_stc_qualifiers): ... this. New.
22926 * config/arm/arm-protos.h
22927 (arm_coproc_ldc_stc_legitimate_address): New.
22928 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
22929 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
22930 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
22931 stc2, stcl, stc2l): New.
22932 * config/arm/constraints.md (Uz): New.
22933 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
22934 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
22935 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
22936 VUNSPEC_STC2L): New.
22938 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22940 * config/arm/arm.md (<cdp>): New.
22941 * config/arm/arm.c (neon_const_bounds): Rename this ...
22942 (arm_const_bounds): ... this.
22943 (arm_coproc_builtin_available): New.
22944 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
22945 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
22946 (CDP_QUALIFIERS): Define to...
22947 (arm_cdp_qualifiers): ... this. New.
22949 (arm_expand_builtin_args): Add case for 6 arguments.
22950 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
22951 (arm_const_bounds): ... this.
22952 (arm_coproc_builtin_available): New.
22953 * config/arm/arm_acle.h (__arm_cdp): New.
22955 * config/arm/arm_acle_builtins.def (cdp): New.
22957 * config/arm/iterators.md (CDPI,CDP,cdp): New.
22958 * config/arm/neon.md: Rename all 'neon_const_bounds' to
22959 'arm_const_bounds'.
22960 * config/arm/types.md (coproc): New.
22961 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
22962 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
22963 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
22964 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
22966 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22968 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
22969 (UBINOP_QUALIFIERS): New.
22971 (acle_builtin_data): New. Change comment.
22972 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
22973 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
22974 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
22975 arm_acle_builtins.def.
22976 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
22977 (arm_init_acle_builtins): New.
22978 (CRC32_BUILTIN): Remove.
22979 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
22980 crc32cb, crc32ch and crc32cw.
22981 (arm_init_crc32_builtins): Remove.
22982 (arm_init_builtins): Use arm_init_acle_builtins rather
22983 than arm_init_crc32_builtins.
22984 (arm_expand_acle_builtin): New.
22985 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
22986 * config/arm/arm_acle_builtins.def: New.
22988 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22990 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
22991 (arm_builtin_datum): ... this.
22992 (arm_init_neon_builtin): Rename to ...
22993 (arm_init_builtin): ... this. Add a new parameters PREFIX
22994 and USE_SIG_IN_NAME.
22995 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
22996 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
22997 'arm_builtin_datum'.
22998 (arm_init_vfp_builtins): Likewise.
22999 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23000 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23001 (arm_expand_neon_args): Rename to ...
23002 (arm_expand_builtin_args): ... this. Rename builtin_arg
23003 enum values and differentiate between ARG_BUILTIN_MEMORY
23004 and ARG_BUILTIN_NEON_MEMORY.
23005 (arm_expand_neon_builtin_1): Rename to ...
23006 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23007 values, arm_expand_builtin_args and add bool parameter NEON.
23008 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23009 (arm_expand_vfp_builtin): Likewise.
23010 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23012 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23014 PR middle-end/77484
23015 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23016 * predict.c (tree_estimate_probability_bb): Reverse direction of
23017 polymorphic call predictor.
23019 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23021 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23022 (determine_pass_name_match): ...this new function and...
23023 (should_skip_pass_p): ...this new function.
23025 2017-01-06 Nathan Sidwell <nathan@acm.org>
23027 * ipa-visibility.c (function_and_variable_visibility): Reformat
23028 comments and long lines. Remove extrneous if.
23029 * symtab.c (symtab_node::make_decl_local): Fix code format.
23030 (symtab_node::set_section_for_node): Fix comment typo.
23032 2017-01-06 Martin Liska <mliska@suse.cz>
23035 * lra-constraints.c: Rename invariant to lra_invariant.
23036 * predict.c (set_even_probabilities): Initialize e to NULL.
23038 2017-01-05 Martin Sebor <msebor@redhat.com>
23040 PR tree-optimization/78910
23041 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23042 (format_integer): Correct off-by-one error in the handling
23043 of precision with negative numbers in signed conversions..
23045 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23047 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23049 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23051 PR tree-optimization/71016
23052 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23053 factor_out_conditional_conversion. Formatting fix.
23054 (factor_out_conditional_conversion): Add cond_stmt argument.
23055 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23056 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23059 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23061 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23062 read-rtl-function.o, and selftest-rtl.o.
23063 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23064 (selftest::aarch64_test_loading_full_dump): New function.
23065 (selftest::aarch64_run_selftests): New function.
23066 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23067 selftest::aarch64_run_selftests.
23068 * config/i386/i386.c
23069 (selftest::ix86_test_loading_dump_fragment_1): New function.
23070 (selftest::ix86_test_loading_call_insn): New function.
23071 (selftest::ix86_test_loading_full_dump): New function.
23072 (selftest::ix86_test_loading_unspec): New function.
23073 (selftest::ix86_run_selftests): Call the new functions.
23074 * emit-rtl.c (maybe_set_max_label_num): New function.
23075 * emit-rtl.h (maybe_set_max_label_num): New decl.
23076 * function.c (instantiate_decls): Guard call to
23077 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23078 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23080 * gensupport.c (gen_reader::gen_reader): Pass "false"
23081 for new "compact" param of rtx_reader.
23082 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23083 rather than an empty string for NULL strings.
23084 * read-md.c: Potentially include config.h rather than bconfig.h.
23085 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23086 (have_error): New global, copied from errors.c.
23087 (md_reader::read_name): Rename to...
23088 (md_reader::read_name_1): ...this, adding "out_loc" param,
23089 and converting "missing name or number" to returning false, rather
23091 (md_reader::read_name): Reimplement in terms of read_name_1.
23092 (md_reader::read_name_or_nil): New function.
23093 (md_reader::read_string): Handle "(nil)" by returning NULL.
23094 (md_reader::md_reader): Add new param "compact".
23095 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23096 (md_reader::read_file): New method.
23097 * read-md.h (md_reader::md_reader): Add new param "compact".
23098 (md_reader::read_file): New method.
23099 (md_reader::is_compact): New accessor.
23100 (md_reader::read_name): Convert return type from void to file_location.
23101 (md_reader::read_name_or_nil): New decl.
23102 (md_reader::read_name_1): New decl.
23103 (md_reader::m_compact): New field.
23104 (noop_reader::noop_reader): Pass "false" for new "compact" param
23106 (rtx_reader::rtx_reader): Add new "compact" param.
23107 (rtx_reader::read_rtx_operand): Make virtual and convert return
23108 type from void to rtx.
23109 (rtx_reader::read_until): New decl.
23110 (rtx_reader::handle_any_trailing_information): New virtual function.
23111 (rtx_reader::postprocess): New virtual function.
23112 (rtx_reader::finalize_string): New virtual function.
23113 (rtx_reader::m_in_call_function_usage): New field.
23114 (rtx_reader::m_reuse_rtx_by_id): New field.
23115 * read-rtl-function.c: New file.
23116 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23117 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23118 (selftest::verify_three_block_rtl_cfg): New decl.
23119 * read-rtl-function.h: New file.
23120 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23121 For host, include function.h, memmodel.h, and emit-rtl.h.
23122 (one_time_initialization): New function.
23123 (struct compact_insn_name): New struct.
23124 (compact_insn_names): New array.
23125 (find_code): Handle insn codes in compact dumps.
23126 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23127 (bind_subst_iter_and_attr): Likewise.
23128 (add_condition_to_string): Likewise.
23129 (add_condition_to_rtx): Likewise.
23130 (apply_attribute_uses): Likewise.
23131 (add_current_iterators): Likewise.
23132 (apply_iterators): Likewise.
23133 (initialize_iterators): Guard usage of apply_subst_iterator with
23134 #ifdef GENERATOR_FILE.
23135 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23136 (md_reader::read_mapping): Likewise.
23137 (add_define_attr_for_define_subst): Likewise.
23138 (add_define_subst_attr): Likewise.
23139 (read_subst_mapping): Likewise.
23140 (check_code_iterator): Likewise.
23141 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23143 (one_time_initialization): New function.
23144 (rtx_reader::read_until): New method.
23145 (read_flags): New function.
23146 (parse_reg_note_name): New function.
23147 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23148 Handle reuse_rtx ids.
23149 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23150 Add parsing support for RTL dumps, mirroring the special-cases in
23151 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23152 values, and calling handle_any_trailing_information.
23153 (rtx_reader::read_rtx_operand): Convert return type from void
23154 to rtx, returning return_rtx. Handle case 'e'. Call
23155 finalize_string on XSTR and XTMPL fields.
23156 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23157 "(nil)" values were omitted. Call the postprocess vfunc on the
23159 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23160 class ctor. Initialize m_in_call_function_usage. Call
23161 one_time_initialization.
23162 * rtl-tests.c (selftest::test_uncond_jump): Call
23163 set_new_first_and_last_insn.
23164 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23165 * selftest-rtl.c: New file.
23166 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23167 (selftest::get_insn_by_uid): New decl.
23168 * selftest-run-tests.c (selftest::run_tests): Call
23169 read_rtl_function_c_tests.
23170 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23171 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23174 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23176 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23177 operands in a special way. Assert that pos+len <= mode precision.
23179 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23181 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23182 3 argument Alias with unlimited for the negative form.
23183 (fno-vect-cost-model): Removed.
23185 2017-01-05 Martin Liska <mliska@suse.cz>
23187 * hsa-gen.c (gen_hsa_divmod): New function.
23188 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23190 2017-01-05 Martin Liska <mliska@suse.cz>
23193 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23196 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23198 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23199 small constant length operands.
23201 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23203 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23204 between loop iterations.
23206 2017-01-05 Martin Liska <mliska@suse.cz>
23209 * gimplify.c (gimplify_decl_expr): Compare to
23210 asan_poisoned_variables instread of checking flags.
23211 (gimplify_target_expr): Likewise.
23212 (gimplify_expr): Likewise.
23213 (gimplify_function_tree): Conditionally initialize
23214 asan_poisoned_variables.
23216 2017-01-04 Jeff Law <law@redhat.com>
23218 PR tree-optimizatin/78812
23219 * rtl.h (contains_mem_rtx_p): Prototype.
23220 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23221 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23222 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23223 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23224 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23226 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23228 * input.c (assert_char_at_range): Default-initialize actual_range.
23230 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23232 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23233 to match the caller.
23235 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23237 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23238 insns after final jump in test to emit dummy move.
23240 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23242 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23243 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23245 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23247 * multiple_target.c (create_dispatcher_calls): Init e_next.
23248 * tree-ssa-loop-split.c (split_loop): Init border.
23249 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23252 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23257 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23258 (altivec_register_operand): Do not return true if the operand
23259 contains a SUBREG mixing SImode and SFmode.
23260 (vsx_register_operand): Likewise.
23261 (vsx_reg_sfsubreg_ok): New predicate.
23262 (vfloat_operand): Do not return true if the operand contains a
23263 SUBREG mixing SImode and SFmode.
23264 (vint_operand): Likewise.
23265 (vlogical_operand): Likewise.
23266 (gpc_reg_operand): Likewise.
23267 (int_reg_operand): Likewise.
23268 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23269 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23270 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23272 (rs6000_emit_move_si_sf_subreg): New helper function.
23273 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23274 fixup SUBREGs involving SImode and SFmode.
23275 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23276 numbers for the new peephole2 optimization.
23277 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23278 the GLIBC math library that do AND/IOR/XOR operations on single
23279 precision floating point.
23280 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23281 target macros to say whether we need to avoid SUBREGs mixing
23283 (TARGET_ALLOW_SF_SUBREG): Likewise.
23284 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23285 (UNSPEC_SI_FROM_SF): Likewise.
23286 (iorxor): Change spacing.
23287 (and_ior_xor): New iterator for AND, IOR, and XOR.
23288 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23289 (movdi_from_sf_zero_ext): Likewise.
23290 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23291 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23292 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23293 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23294 (fms<mode>4): Likewise.
23295 (fnma<mode>4): Likewise.
23296 (fnms<mode>4): Likewise.
23297 (nfma<mode>4): Likewise.
23298 (nfms<mode>4): Likewise.
23300 2017-01-04 Marek Polacek <polacek@redhat.com>
23303 * doc/invoke.texi: Document -Wpointer-compare.
23305 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23307 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23310 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23311 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23314 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23317 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23318 documentation of the powerpc_popcntb_ok attribute.
23319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23320 code to issue warning messages if a requested CPU configuration is
23321 not supported by the binary (assembler and loader) toolchain.
23322 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23323 made to define a built-in function that has been disabled.
23324 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23325 made to define a built-in function that has been disabled.
23326 (altivec_init_builtins): Add comment explaining why definition
23327 of the DST built-in functions is not preceded by an assertion
23328 check. Add assertions to prevent ICE if attempts are made to
23329 define an altivec predicate or an abs* built-in function that has
23331 (htm_init_builtins): Add comment explaining why definition of the
23332 htm built-in functions is not preceded by an assertion check.
23334 2017-01-04 Jeff Law <law@redhat.com>
23336 PR tree-optimizatin/67955
23337 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23338 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23339 the points-to solution does not include pt_null. Use DECL_PT_UID
23342 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23344 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23345 Use gen_int_mode instead of gen_lopwart for const_int operands.
23347 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23349 PR tree-optimization/71563
23350 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23351 out of range value - has low bits known to be zero.
23353 2017-01-04 Alan Modra <amodra@gmail.com>
23355 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23356 * configure: Regenerate.
23357 * config.in: Regenerate.
23359 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23362 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23363 a substring of the message, but strcmp with the whole message. Ifdef
23364 ENABLE_NLS, translate the message first using dgettext.
23366 2017-01-03 Jeff Law <law@redhat.com>
23368 PR tree-optimizatin/78856
23369 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23370 (mark_threaded_blocks): Remove code to truncate thread paths that
23371 cross multiple loop headers. Instead invalidate the cached loop
23372 iteration information and handle case of a thread path walking
23373 into an irreducible region.
23375 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23378 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23379 assertions. Add support for doing the signbit if the IEEE 128-bit
23380 floating point value is in a GPR.
23381 * config/rs6000/rs6000.md (Fsignbit): Delete.
23382 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23383 Update the length attribute if the value is in a GPR.
23384 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23385 the sign or zero extension instruction, since the value is always 0/1.
23386 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23389 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23390 extracting SImode to a GPR register so that we can generate a
23391 store, limit the vector to be in a traditional Altivec register
23392 for the vextuwrx instruction.
23394 2017-01-03 Ian Lance Taylor <iant@google.com>
23396 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23398 2017-01-03 Martin Sebor <msebor@redhat.com>
23400 PR tree-optimization/78696
23401 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23402 precision. Use MPFR for %f for greater fidelity. Correct handling
23404 (pass_sprintf_length::compute_format_length): Set width and precision
23405 specified by asrerisk to void_node for vararg functions.
23406 (try_substitute_return_value): Adjust dump output.
23408 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23410 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23412 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23414 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23415 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23416 -mlra/-mno-lra was passed to the compiler.
23418 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23420 PR rtl-optimization/65618
23421 * emit-rtl.c (try_split): Move initialization of "before" and
23422 "after" to just before the call to emit_insn_after_setloc.
23424 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23426 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23428 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23430 * dwarf2out.c (gen_enumeration_type_die): When
23431 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23433 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23435 PR tree-optimization/78965
23436 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23437 Change first argument from const call_info & to call_info &. For %n
23438 set info.nowrite to false.
23440 PR middle-end/78901
23441 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23442 possibly throwing calls.
23444 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23445 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23446 and fns handling, rather than in a separate case SSA_NAME.
23448 2017-01-02 Jeff Law <law@redhat.com>
23450 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23451 fixes for first_period and second_period variables.
23453 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23456 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23457 (*insvqi_1): New insn pattern.
23458 (*insvqi_1_mem_rex64): Ditto.
23459 (*insvqi_2): Ditto.
23460 (*insvqi_3): Rename from *insvqi.
23462 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23464 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23466 * doc/cfg.texi (Edges): Remove reference to Java.
23467 (Maintaining the CFG): Ditto.
23469 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23471 PR middle-end/77674
23472 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23473 transparent aliases.
23475 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23477 PR middle-end/77484
23478 * predict.def (PRED_CALL): Update hitrate.
23479 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23480 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23481 into direct/indirect/polymorphic variants.
23483 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23485 Update copyright years.
23487 * gcc.c (process_command): Update copyright notice dates.
23488 * gcov-dump.c (print_version): Ditto.
23489 * gcov.c (print_version): Ditto.
23490 * gcov-tool.c (print_version): Ditto.
23491 * gengtype.c (create_file): Ditto.
23492 * doc/cpp.texi: Bump @copying's copyright year.
23493 * doc/cppinternals.texi: Ditto.
23494 * doc/gcc.texi: Ditto.
23495 * doc/gccint.texi: Ditto.
23496 * doc/gcov.texi: Ditto.
23497 * doc/install.texi: Ditto.
23498 * doc/invoke.texi: Ditto.
23500 Copyright (C) 2017 Free Software Foundation, Inc.
23502 Copying and distribution of this file, with or without modification,
23503 are permitted in any medium without royalty provided the copyright
23504 notice and this notice are preserved.