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 * simplify-rtx.c (simplify_binary_operation_1): Use
6 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9 Alan Hayward <alan.hayward@arm.com>
10 David Sherwood <david.sherwood@arm.com>
12 * wide-int.h (int_traits<unsigned char>) New class.
13 (int_traits<unsigned short>) Likewise.
14 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17 * combine.c (set_nonzero_bits_and_sign_copies): Use
18 is_a <scalar_int_mode>.
19 (find_split_point): Likewise.
20 (combine_simplify_rtx): Likewise.
21 (simplify_logical): Likewise.
22 (expand_compound_operation): Likewise.
23 (expand_field_assignment): Likewise.
24 (make_compound_operation): Likewise.
25 (extended_count): Likewise.
26 (change_zero_ext): Likewise.
27 (simplify_comparison): Likewise.
28 * dwarf2out.c (scompare_loc_descriptor): Likewise.
29 (ucompare_loc_descriptor): Likewise.
30 (minmax_loc_descriptor): Likewise.
31 (mem_loc_descriptor): Likewise.
32 (loc_descriptor): Likewise.
33 * expmed.c (init_expmed_one_mode): Likewise.
34 * lra-constraints.c (lra_constraint_offset): Likewise.
35 * optabs.c (prepare_libcall_arg): Likewise.
36 * postreload.c (move2add_note_store): Likewise.
37 * reload.c (operands_match_p): Likewise.
38 * rtl.h (load_extend_op): Likewise.
39 * rtlhooks.c (gen_lowpart_general): Likewise.
40 * simplify-rtx.c (simplify_truncation): Likewise.
41 (simplify_unary_operation_1): Likewise.
42 (simplify_binary_operation_1): Likewise.
43 (simplify_const_binary_operation): Likewise.
44 (simplify_const_relational_operation): Likewise.
45 (simplify_subreg): Likewise.
46 * stor-layout.c (bitwise_mode_for_mode): Likewise.
47 * var-tracking.c (adjust_mems): Likewise.
48 (prepare_call_arguments): Likewise.
50 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
51 Alan Hayward <alan.hayward@arm.com>
52 David Sherwood <david.sherwood@arm.com>
54 * machmode.h (is_int_mode): New fuction.
55 * combine.c (find_split_point): Use it.
56 (combine_simplify_rtx): Likewise.
57 (simplify_if_then_else): Likewise.
58 (simplify_set): Likewise.
59 (simplify_shift_const_1): Likewise.
60 (simplify_comparison): Likewise.
61 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
62 * cse.c (notreg_cost): Likewise.
64 * cselib.c (cselib_lookup_1): Likewise.
65 * dojump.c (do_jump_1): Likewise.
66 (do_compare_rtx_and_jump): Likewise.
67 * dse.c (get_call_args): Likewise.
68 * dwarf2out.c (rtl_for_decl_init): Likewise.
69 (native_encode_initializer): Likewise.
70 * expmed.c (emit_store_flag_1): Likewise.
71 (emit_store_flag): Likewise.
72 * expr.c (convert_modes): Likewise.
73 (store_field): Likewise.
74 (expand_expr_real_1): Likewise.
75 * fold-const.c (fold_read_from_constant_string): Likewise.
76 * gimple-ssa-sprintf.c (get_format_string): Likewise.
77 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
78 * optabs.c (expand_binop): Likewise.
79 (expand_unop): Likewise.
80 (expand_abs_nojump): Likewise.
81 (expand_one_cmpl_abs_nojump): Likewise.
82 * simplify-rtx.c (mode_signbit_p): Likewise.
83 (val_signbit_p): Likewise.
84 (val_signbit_known_set_p): Likewise.
85 (val_signbit_known_clear_p): Likewise.
86 (simplify_relational_operation_1): Likewise.
87 * tree.c (vector_type_mode): Likewise.
89 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
90 Alan Hayward <alan.hayward@arm.com>
91 David Sherwood <david.sherwood@arm.com>
93 * machmode.h (smallest_mode_for_size): Fix formatting.
94 (smallest_int_mode_for_size): New function.
95 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
96 instead of smallest_mode_for_size.
97 * combine.c (make_extraction): Likewise.
98 * config/arc/arc.c (arc_expand_movmem): Likewise.
99 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
100 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
101 * config/s390/s390.c (s390_expand_insv): Likewise.
102 * config/sparc/sparc.c (assign_int_registers): Likewise.
103 * config/spu/spu.c (spu_function_value): Likewise.
104 (spu_function_arg): Likewise.
105 * coverage.c (get_gcov_type): Likewise.
106 (get_gcov_unsigned_t): Likewise.
107 * dse.c (find_shift_sequence): Likewise.
108 * expmed.c (store_bit_field_1): Likewise.
109 * expr.c (convert_move): Likewise.
110 (store_field): Likewise.
111 * internal-fn.c (expand_arith_overflow): Likewise.
112 * optabs-query.c (get_best_extraction_insn): Likewise.
113 * optabs.c (expand_twoval_binop_libfunc): Likewise.
114 * stor-layout.c (layout_type): Likewise.
115 (initialize_sizetypes): Likewise.
116 * targhooks.c (default_get_mask_mode): Likewise.
117 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
120 Alan Hayward <alan.hayward@arm.com>
121 David Sherwood <david.sherwood@arm.com>
123 * machmode.h (opt_mode::else_blk): New function.
124 (int_mode_for_mode): Declare.
125 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
126 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
128 * cfgexpand.c (expand_debug_expr): Likewise.
129 * combine.c (gen_lowpart_or_truncate): Likewise.
130 (gen_lowpart_for_combine): Likewise.
131 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
132 * config/avr/avr.c (avr_to_int_mode): Likewise.
133 (avr_out_plus_1): Likewise.
134 (avr_out_plus): Likewise.
135 (avr_out_round): Likewise.
136 * config/i386/i386.c (ix86_split_to_parts): Likewise.
137 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
138 (s390_expand_vcond): Likewise.
139 * config/spu/spu.c (spu_split_immediate): Likewise.
140 (spu_expand_mov): Likewise.
141 * dse.c (get_stored_val): Likewise.
142 * expmed.c (store_bit_field_1): Likewise.
143 (convert_extracted_bit_field): Use int_mode_for_mode instead of
145 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
146 (extract_low_bits): Likewise.
147 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
148 handling rather than repeating the check.
149 (emit_group_store): Likewise.
150 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
151 * optabs.c (expand_absneg_bit): Likewise.
152 (expand_copysign_absneg): Likewise.
153 (expand_copysign_bit): Likewise.
154 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
155 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
156 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
157 * var-tracking.c (prepare_call_arguments): Likewise.
158 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
159 int_mode_for_mode instead of mode_for_size.
160 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * machmode.h (int_mode_for_size): New function.
167 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
168 instead of mode_for_size.
169 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
171 * combine.c (expand_field_assignment): Use int_mode_for_size
172 instead of mode_for_size.
173 (make_extraction): Likewise.
174 (simplify_shift_const_1): Likewise.
175 (simplify_comparison): Likewise.
176 * dojump.c (do_jump): Likewise.
177 * dwarf2out.c (mem_loc_descriptor): Likewise.
178 * emit-rtl.c (init_derived_machine_modes): Likewise.
179 * expmed.c (flip_storage_order): Likewise.
180 (convert_extracted_bit_field): Likewise.
181 * expr.c (copy_blkmode_from_reg): Likewise.
182 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
183 * internal-fn.c (expand_mul_overflow): Likewise.
184 * lower-subreg.c (simple_move): Likewise.
185 * optabs-libfuncs.c (init_optabs): Likewise.
186 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
187 * tree.c (vector_type_mode): Likewise.
188 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
189 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
190 * tree-vect-generic.c (expand_vector_parallel): Likewise.
191 * tree-vect-stmts.c (vectorizable_load): Likewise.
192 (vectorizable_store): Likewise.
194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
195 Alan Hayward <alan.hayward@arm.com>
196 David Sherwood <david.sherwood@arm.com>
198 * coretypes.h (pod_mode): New type.
199 (scalar_int_mode_pod): New typedef.
200 * machmode.h (pod_mode): New class.
201 (int_n_data_t::m): Change type to scalar_int_mode_pod.
202 * genmodes.c (emit_mode_int_n): Update accordingly.
203 * lower-subreg.h (target_lower_subreg): Change type to
205 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
209 Alan Hayward <alan.hayward@arm.com>
210 David Sherwood <david.sherwood@arm.com>
212 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
213 machine_mode to scalar_int_mode.
214 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
215 (rs6000_option_override_internal): Remove cast to int.
216 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
217 machine_mode to scalar_int_mode.
218 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
219 (rs6000_option_override_internal): Remove cast to int.
220 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
221 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
223 * config/s390/s390.c (s390_expand_builtin): Likewise.
224 * coretypes.h (scalar_int_mode): New type.
225 (opt_scalar_int_mode): New typedef.
226 * machmode.h (scalar_int_mode): New class.
227 (scalar_int_mode::includes_p): New function.
228 (byte_mode): Change type to scalar_int_mode.
229 (word_mode): Likewise.
230 (ptr_mode): Likewise.
231 * emit-rtl.c (byte_mode): Likewise.
232 (word_mode): Likewise.
233 (ptr_mode): Likewise.
234 (init_derived_machine_modes): Update accordingly.
235 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
236 and MODE_PARTIAL_INT.
237 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
241 Alan Hayward <alan.hayward@arm.com>
242 David Sherwood <david.sherwood@arm.com>
244 * target.def (libgcc_floating_mode_supported_p): Take a
246 * doc/tm.texi: Regenerate.
247 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
249 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
250 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
254 Alan Hayward <alan.hayward@arm.com>
255 David Sherwood <david.sherwood@arm.com>
257 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
258 * doc/tm.texi: Regenerate.
259 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
260 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
261 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
262 * targhooks.h (default_floatn_mode): Likewise.
263 * targhooks.c (default_floatn_mode): Likewise.
264 * tree.c (build_common_tree_nodes): Update accordingly.
266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
267 Alan Hayward <alan.hayward@arm.com>
268 David Sherwood <david.sherwood@arm.com>
270 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
271 (mode_iterator::iterate_p): Likewise.
272 (mode_iterator::get_wider): Likewise.
273 * expr.c (init_expr_target): Use opt_scalar_float_mode.
275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
276 Alan Hayward <alan.hayward@arm.com>
277 David Sherwood <david.sherwood@arm.com>
279 * coretypes.h (opt_scalar_float_mode): New typedef.
280 * machmode.h (float_mode_for_size): New function.
281 * emit-rtl.c (double_mode): Delete.
282 (init_emit_once): Use float_mode_for_size.
283 * stor-layout.c (layout_type): Likewise.
284 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
287 Alan Hayward <alan.hayward@arm.com>
288 David Sherwood <david.sherwood@arm.com>
290 * output.h (assemble_real): Take a scalar_float_mode.
291 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
292 * config/arm/arm.md (consttable_4): Likewise.
293 (consttable_8): Likewise.
294 (consttable_16): Likewise.
295 * config/mips/mips.md (consttable_float): Likewise.
296 * config/s390/s390.c (s390_output_pool_entry): Likewise.
297 * varasm.c (assemble_real): Take a scalar_float_mode.
298 (output_constant_pool_2): Update accordingly.
299 (output_constant): Likewise.
301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
302 Alan Hayward <alan.hayward@arm.com>
303 David Sherwood <david.sherwood@arm.com>
305 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
306 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
307 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
308 (native_encode_real): Likewise.
309 (native_interpret_real): Likewise.
310 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
311 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
314 Alan Hayward <alan.hayward@arm.com>
315 David Sherwood <david.sherwood@arm.com>
317 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
318 <scalar_float_mode>. Simplify.
319 (gen_extend_conv_libfunc): Likewise.
321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
322 Alan Hayward <alan.hayward@arm.com>
323 David Sherwood <david.sherwood@arm.com>
325 * coretypes.h (scalar_float_mode): New type.
326 * machmode.h (mode_traits::from_int): Use machine_mode if
327 USE_ENUM_MODES is defined.
328 (is_a): New function.
330 (dyn_cast): Likewise.
331 (scalar_float_mode): New class.
332 (scalar_float_mode::includes_p): New function.
333 (is_float_mode): Likewise.
334 * gdbhooks.py (MachineModePrinter): New class.
335 (build_pretty_printer): Use it for scalar_float_mode.
336 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
337 (format_helper::format_helper): Turn into a template.
338 * genmodes.c (get_mode_class): New function.
339 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
340 or machine_mode if none.
341 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
342 as_a <scalar_float_mode>.
343 * dwarf2out.c (mem_loc_descriptor): Likewise.
344 (insert_float): Likewise.
345 (add_const_value_attribute): Likewise.
346 * simplify-rtx.c (simplify_immed_subreg): Likewise.
347 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
348 (expand_unop): Update accordingly.
349 (expand_abs_nojump): Likewise.
350 (expand_copysign_absneg): Take a scalar_float_mode.
351 (expand_copysign_bit): Likewise.
352 (expand_copysign): Update accordingly.
354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
355 Alan Hayward <alan.hayward@arm.com>
356 David Sherwood <david.sherwood@arm.com>
358 * coretypes.h (opt_mode): New class.
359 * machmode.h (opt_mode): Likewise.
360 (opt_mode::else_void): New function.
361 (opt_mode::require): Likewise.
362 (opt_mode::exists): Likewise.
363 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
364 (GET_MODE_2XWIDER_MODE): Likewise.
365 (mode_iterator::get_wider): Update accordingly.
366 (mode_iterator::get_2xwider): Likewise.
367 (mode_iterator::get_known_wider): Likewise, turning into a template.
368 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
369 forcing a wider mode to exist.
370 * config/cr16/cr16.h (LONG_REG_P): Likewise.
371 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
372 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
373 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
374 * lower-subreg.c (init_lower_subreg): Likewise.
375 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
376 on the final iteration.
377 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
378 a wider mode exists before asking for a move pattern.
379 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
380 forcing a wider mode to exist.
381 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
382 returning false if no such mode exists.
383 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
384 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
385 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
386 Avoid checking for a MODE_INT if we already know the mode is not a
388 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
389 forcing a wider mode to exist.
390 (expmed_mult_highpart_optab): Likewise.
391 (expmed_mult_highpart): Likewise.
392 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
394 * lto-streamer-in.c (lto_input_mode_table): Likewise.
395 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
396 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
397 * internal-fn.c (expand_mul_overflow): Update use of
398 GET_MODE_2XWIDER_MODE.
399 * omp-low.c (omp_clause_aligned_alignment): Likewise.
400 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
402 (convert_plusminus_to_widen): Likewise.
403 * tree-switch-conversion.c (array_value_type): Likewise.
404 * var-tracking.c (emit_note_insn_var_location): Likewise.
405 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
406 Return false inside rather than outside the loop if no wider mode
408 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
409 and GET_MODE_2XWIDER_MODE
410 (can_compare_p): Use else_void.
411 * gdbhooks.py (OptMachineModePrinter): New class.
412 (build_pretty_printer): Use it for opt_mode.
414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
415 Alan Hayward <alan.hayward@arm.com>
416 David Sherwood <david.sherwood@arm.com>
418 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
419 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
422 Alan Hayward <alan.hayward@arm.com>
423 David Sherwood <david.sherwood@arm.com>
425 * machmode.h (mode_traits): New structure.
426 (get_narrowest_mode): New function.
427 (mode_iterator::start): Likewise.
428 (mode_iterator::iterate_p): Likewise.
429 (mode_iterator::get_wider): Likewise.
430 (mode_iterator::get_known_wider): Likewise.
431 (mode_iterator::get_2xwider): Likewise.
432 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
433 (FOR_EACH_MODE): Likewise.
434 (FOR_EACH_MODE_FROM): Likewise.
435 (FOR_EACH_MODE_UNTIL): Likewise.
436 (FOR_EACH_WIDER_MODE): Likewise.
437 (FOR_EACH_2XWIDER_MODE): Likewise.
438 * builtins.c (expand_builtin_strlen): Use new mode iterators.
439 * combine.c (simplify_comparison): Likewise
440 * config/i386/i386.c (type_natural_mode): Likewise.
441 * cse.c (cse_insn): Likewise.
442 * dse.c (find_shift_sequence): Likewise.
443 * emit-rtl.c (init_derived_machine_modes): Likewise.
444 (init_emit_once): Likewise.
445 * explow.c (hard_function_value): Likewise.
446 * expmed.c (extract_fixed_bit_field_1): Likewise.
447 (extract_bit_field_1): Likewise.
448 (expand_divmod): Likewise.
449 (emit_store_flag_1): Likewise.
450 * expr.c (init_expr_target): Likewise.
451 (convert_move): Likewise.
452 (alignment_for_piecewise_move): Likewise.
453 (widest_int_mode_for_size): Likewise.
454 (emit_block_move_via_movmem): Likewise.
455 (copy_blkmode_to_reg): Likewise.
456 (set_storage_via_setmem): Likewise.
457 (compress_float_constant): Likewise.
458 * omp-low.c (omp_clause_aligned_alignment): Likewise.
459 * optabs-query.c (get_best_extraction_insn): Likewise.
460 * optabs.c (expand_binop): Likewise.
461 (expand_twoval_unop): Likewise.
462 (expand_twoval_binop): Likewise.
463 (widen_leading): Likewise.
464 (widen_bswap): Likewise.
465 (expand_parity): Likewise.
466 (expand_unop): Likewise.
467 (prepare_cmp_insn): Likewise.
468 (prepare_float_lib_cmp): Likewise.
469 (expand_float): Likewise.
470 (expand_fix): Likewise.
471 (expand_sfix_optab): Likewise.
472 * postreload.c (move2add_use_add2_insn): Likewise.
473 * reg-stack.c (reg_to_stack): Likewise.
474 * reginfo.c (choose_hard_reg_mode): Likewise.
475 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
476 * stor-layout.c (mode_for_size): Likewise.
477 (smallest_mode_for_size): Likewise.
478 (mode_for_vector): Likewise.
479 (finish_bitfield_representative): Likewise.
480 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
481 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
482 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
483 * var-tracking.c (prepare_call_arguments): Likewise.
485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
486 Alan Hayward <alan.hayward@arm.com>
487 David Sherwood <david.sherwood@arm.com>
489 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
490 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
491 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
492 * machmode.h (mode_size): Move earlier in file.
493 (mode_precision): Likewise.
494 (mode_inner): Likewise.
495 (mode_nunits): Likewise.
496 (mode_unit_size): Likewise.
497 (unit_unit_precision): Likewise.
498 (mode_wider): Likewise.
499 (mode_2xwider): Likewise.
500 (machine_mode): New class.
501 (mode_to_bytes): New function.
502 (mode_to_bits): Likewise.
503 (mode_to_precision): Likewise.
504 (mode_to_inner): Likewise.
505 (mode_to_unit_size): Likewise.
506 (mode_to_unit_precision): Likewise.
507 (mode_to_nunits): Likewise.
508 (GET_MODE_SIZE): Use mode_to_bytes.
509 (GET_MODE_BITSIZE): Use mode_to_bits.
510 (GET_MODE_PRECISION): Use mode_to_precision.
511 (GET_MODE_INNER): Use mode_to_inner.
512 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
513 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
514 (GET_MODE_NUNITS): Use mode_to_nunits.
515 * system.h (ALWAYS_INLINE): New macro.
516 * config/powerpcspe/powerpcspe-c.c
517 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
518 int for arg1_mode and arg2_mode.
520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
521 Alan Hayward <alan.hayward@arm.com>
522 David Sherwood <david.sherwood@arm.com>
524 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
525 Prefix mode names with E_ in case statements.
526 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
527 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
528 (aarch64_split_simd_move): Likewise.
529 (aarch64_gen_storewb_pair): Likewise.
530 (aarch64_gen_loadwb_pair): Likewise.
531 (aarch64_gen_store_pair): Likewise.
532 (aarch64_gen_load_pair): Likewise.
533 (aarch64_get_condition_code_1): Likewise.
534 (aarch64_constant_pool_reload_icode): Likewise.
535 (get_rsqrte_type): Likewise.
536 (get_rsqrts_type): Likewise.
537 (get_recpe_type): Likewise.
538 (get_recps_type): Likewise.
539 (aarch64_gimplify_va_arg_expr): Likewise.
540 (aarch64_simd_container_mode): Likewise.
541 (aarch64_emit_load_exclusive): Likewise.
542 (aarch64_emit_store_exclusive): Likewise.
543 (aarch64_expand_compare_and_swap): Likewise.
544 (aarch64_gen_atomic_cas): Likewise.
545 (aarch64_emit_bic): Likewise.
546 (aarch64_emit_atomic_swap): Likewise.
547 (aarch64_emit_atomic_load_op): Likewise.
548 (aarch64_evpc_trn): Likewise.
549 (aarch64_evpc_uzp): Likewise.
550 (aarch64_evpc_zip): Likewise.
551 (aarch64_evpc_ext): Likewise.
552 (aarch64_evpc_rev): Likewise.
553 (aarch64_evpc_dup): Likewise.
554 (aarch64_gen_ccmp_first): Likewise.
555 (aarch64_gen_ccmp_next): Likewise.
556 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
557 (alpha_emit_xfloating_libcall): Likewise.
558 (emit_insxl): Likewise.
559 (alpha_arg_type): Likewise.
560 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
561 (arc_preferred_simd_mode): Likewise.
562 (arc_secondary_reload): Likewise.
563 (get_arc_condition_code): Likewise.
564 (arc_print_operand): Likewise.
565 (arc_legitimate_constant_p): Likewise.
566 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
567 * config/arc/arc.md (casesi_load): Likewise.
568 (casesi_compact_jump): Likewise.
569 * config/arc/predicates.md (proper_comparison_operator): Likewise.
570 (cc_use_register): Likewise.
571 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
572 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
573 (arm_init_iwmmxt_builtins): Likewise.
574 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
575 (neon_expand_vector_init): Likewise.
576 (arm_attr_length_move_neon): Likewise.
577 (maybe_get_arm_condition_code): Likewise.
578 (arm_emit_vector_const): Likewise.
579 (arm_preferred_simd_mode): Likewise.
580 (arm_output_iwmmxt_tinsr): Likewise.
581 (thumb1_output_casesi): Likewise.
582 (thumb2_output_casesi): Likewise.
583 (arm_emit_load_exclusive): Likewise.
584 (arm_emit_store_exclusive): Likewise.
585 (arm_expand_compare_and_swap): Likewise.
586 (arm_evpc_neon_vuzp): Likewise.
587 (arm_evpc_neon_vzip): Likewise.
588 (arm_evpc_neon_vrev): Likewise.
589 (arm_evpc_neon_vtrn): Likewise.
590 (arm_evpc_neon_vext): Likewise.
591 (arm_validize_comparison): Likewise.
592 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
593 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
594 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
595 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
596 (c6x_preferred_simd_mode): Likewise.
597 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
598 (epiphany_rtx_costs): Likewise.
599 * config/epiphany/predicates.md (proper_comparison_operator):
601 * config/frv/frv.c (condexec_memory_operand): Likewise.
602 (frv_emit_move): Likewise.
603 (output_move_single): Likewise.
604 (output_condmove_single): Likewise.
605 (frv_hard_regno_mode_ok): Likewise.
606 (frv_matching_accg_mode): Likewise.
607 * config/h8300/h8300.c (split_adds_subs): Likewise.
608 (h8300_rtx_costs): Likewise.
609 (h8300_print_operand): Likewise.
610 (compute_mov_length): Likewise.
611 (output_logical_op): Likewise.
612 (compute_logical_op_length): Likewise.
613 (compute_logical_op_cc): Likewise.
614 (h8300_shift_needs_scratch_p): Likewise.
615 (output_a_shift): Likewise.
616 (compute_a_shift_length): Likewise.
617 (compute_a_shift_cc): Likewise.
618 (expand_a_rotate): Likewise.
619 (output_a_rotate): Likewise.
620 * config/i386/i386.c (classify_argument): Likewise.
621 (function_arg_advance_32): Likewise.
622 (function_arg_32): Likewise.
623 (function_arg_64): Likewise.
624 (function_value_64): Likewise.
625 (ix86_gimplify_va_arg): Likewise.
626 (ix86_legitimate_constant_p): Likewise.
627 (put_condition_code): Likewise.
628 (split_double_mode): Likewise.
629 (ix86_avx256_split_vector_move_misalign): Likewise.
630 (ix86_expand_vector_logical_operator): Likewise.
631 (ix86_split_idivmod): Likewise.
632 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
633 (ix86_build_const_vector): Likewise.
634 (ix86_build_signbit_mask): Likewise.
635 (ix86_match_ccmode): Likewise.
636 (ix86_cc_modes_compatible): Likewise.
637 (ix86_expand_branch): Likewise.
638 (ix86_expand_sse_cmp): Likewise.
639 (ix86_expand_sse_movcc): Likewise.
640 (ix86_expand_int_sse_cmp): Likewise.
641 (ix86_expand_vec_perm_vpermi2): Likewise.
642 (ix86_expand_vec_perm): Likewise.
643 (ix86_expand_sse_unpack): Likewise.
644 (ix86_expand_int_addcc): Likewise.
645 (ix86_split_to_parts): Likewise.
646 (ix86_vectorize_builtin_gather): Likewise.
647 (ix86_vectorize_builtin_scatter): Likewise.
648 (avx_vpermilp_parallel): Likewise.
649 (inline_memory_move_cost): Likewise.
650 (ix86_tieable_integer_mode_p): Likewise.
651 (x86_maybe_negate_const_int): Likewise.
652 (ix86_expand_vector_init_duplicate): Likewise.
653 (ix86_expand_vector_init_one_nonzero): Likewise.
654 (ix86_expand_vector_init_one_var): Likewise.
655 (ix86_expand_vector_init_concat): Likewise.
656 (ix86_expand_vector_init_interleave): Likewise.
657 (ix86_expand_vector_init_general): Likewise.
658 (ix86_expand_vector_set): Likewise.
659 (ix86_expand_vector_extract): Likewise.
660 (emit_reduc_half): Likewise.
661 (ix86_emit_i387_round): Likewise.
662 (ix86_mangle_type): Likewise.
663 (ix86_expand_round_sse4): Likewise.
664 (expand_vec_perm_blend): Likewise.
665 (canonicalize_vector_int_perm): Likewise.
666 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
667 (expand_vec_perm_1): Likewise.
668 (expand_vec_perm_interleave3): Likewise.
669 (expand_vec_perm_even_odd_pack): Likewise.
670 (expand_vec_perm_even_odd_1): Likewise.
671 (expand_vec_perm_broadcast_1): Likewise.
672 (ix86_vectorize_vec_perm_const_ok): Likewise.
673 (ix86_expand_vecop_qihi): Likewise.
674 (ix86_expand_mul_widen_hilo): Likewise.
675 (ix86_expand_sse2_abs): Likewise.
676 (ix86_expand_pextr): Likewise.
677 (ix86_expand_pinsr): Likewise.
678 (ix86_preferred_simd_mode): Likewise.
679 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
680 * config/i386/sse.md (*andnot<mode>3): Likewise.
681 (<mask_codefor><code><mode>3<mask_name>): Likewise.
682 (*<code><mode>3): Likewise.
683 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
684 (ia64_expand_atomic_op): Likewise.
685 (ia64_arg_type): Likewise.
686 (ia64_mode_to_int): Likewise.
687 (ia64_scalar_mode_supported_p): Likewise.
688 (ia64_vector_mode_supported_p): Likewise.
689 (expand_vec_perm_broadcast): Likewise.
690 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
691 (iq2000_function_arg_advance): Likewise.
692 (iq2000_function_arg): Likewise.
693 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
694 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
695 (m68k_libcall_value): Likewise.
696 (m68k_function_value): Likewise.
697 (sched_attr_op_type): Likewise.
698 * config/mcore/mcore.c (mcore_output_move): Likewise.
699 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
701 (microblaze_function_arg): Likewise.
702 * config/mips/mips.c (mips16_build_call_stub): Likewise.
703 (mips_print_operand): Likewise.
704 (mips_mode_ok_for_mov_fmt_p): Likewise.
705 (mips_vector_mode_supported_p): Likewise.
706 (mips_preferred_simd_mode): Likewise.
707 (mips_expand_vpc_loongson_even_odd): Likewise.
708 (mips_expand_vec_unpack): Likewise.
709 (mips_expand_vi_broadcast): Likewise.
710 (mips_expand_vector_init): Likewise.
711 (mips_expand_vec_reduc): Likewise.
712 (mips_expand_msa_cmp): Likewise.
713 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
714 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
715 (cc_flags_for_mode): Likewise.
716 * config/msp430/msp430.c (msp430_print_operand): Likewise.
717 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
718 (nds32_output_casesi_pc_relative): Likewise.
719 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
720 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
721 (nvptx_gen_unpack): Likewise.
722 (nvptx_gen_pack): Likewise.
723 (nvptx_gen_shuffle): Likewise.
724 (nvptx_gen_wcast): Likewise.
725 (nvptx_preferred_simd_mode): Likewise.
726 * config/pa/pa.c (pa_secondary_reload): Likewise.
727 * config/pa/predicates.md (base14_operand): Likewise.
728 * config/powerpcspe/powerpcspe-c.c
729 (altivec_resolve_overloaded_builtin): Likewise.
730 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
732 (rs6000_preferred_simd_mode): Likewise.
733 (output_vec_const_move): Likewise.
734 (rs6000_expand_vector_extract): Likewise.
735 (rs6000_split_vec_extract_var): Likewise.
736 (reg_offset_addressing_ok_p): Likewise.
737 (rs6000_legitimate_offset_address_p): Likewise.
738 (rs6000_legitimize_address): Likewise.
739 (rs6000_emit_set_const): Likewise.
740 (rs6000_const_vec): Likewise.
741 (rs6000_emit_move): Likewise.
742 (spe_build_register_parallel): Likewise.
743 (rs6000_darwin64_record_arg_recurse): Likewise.
744 (swap_selector_for_mode): Likewise.
745 (spe_init_builtins): Likewise.
746 (paired_init_builtins): Likewise.
747 (altivec_init_builtins): Likewise.
748 (do_load_for_compare): Likewise.
749 (rs6000_generate_compare): Likewise.
750 (rs6000_expand_float128_convert): Likewise.
751 (emit_load_locked): Likewise.
752 (emit_store_conditional): Likewise.
753 (rs6000_output_function_epilogue): Likewise.
754 (rs6000_handle_altivec_attribute): Likewise.
755 (rs6000_function_value): Likewise.
756 (emit_fusion_gpr_load): Likewise.
757 (emit_fusion_p9_load): Likewise.
758 (emit_fusion_p9_store): Likewise.
759 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
760 (fusion_gpr_mem_load): Likewise.
761 (fusion_addis_mem_combo_load): Likewise.
762 (fusion_addis_mem_combo_store): Likewise.
763 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
764 (fusion_gpr_mem_load): Likewise.
765 (fusion_addis_mem_combo_load): Likewise.
766 (fusion_addis_mem_combo_store): Likewise.
767 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
769 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
770 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
771 (rs6000_preferred_simd_mode): Likewise.
772 (output_vec_const_move): Likewise.
773 (rs6000_expand_vector_extract): Likewise.
774 (rs6000_split_vec_extract_var): Likewise.
775 (reg_offset_addressing_ok_p): Likewise.
776 (rs6000_legitimate_offset_address_p): Likewise.
777 (rs6000_legitimize_address): Likewise.
778 (rs6000_emit_set_const): Likewise.
779 (rs6000_const_vec): Likewise.
780 (rs6000_emit_move): Likewise.
781 (rs6000_darwin64_record_arg_recurse): Likewise.
782 (swap_selector_for_mode): Likewise.
783 (paired_init_builtins): Likewise.
784 (altivec_init_builtins): Likewise.
785 (rs6000_expand_float128_convert): Likewise.
786 (emit_load_locked): Likewise.
787 (emit_store_conditional): Likewise.
788 (rs6000_output_function_epilogue): Likewise.
789 (rs6000_handle_altivec_attribute): Likewise.
790 (rs6000_function_value): Likewise.
791 (emit_fusion_gpr_load): Likewise.
792 (emit_fusion_p9_load): Likewise.
793 (emit_fusion_p9_store): Likewise.
794 * config/rx/rx.c (rx_gen_move_template): Likewise.
795 (flags_from_mode): Likewise.
796 * config/s390/predicates.md (s390_alc_comparison): Likewise.
797 (s390_slb_comparison): Likewise.
798 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
799 (s390_vector_mode_supported_p): Likewise.
800 (s390_cc_modes_compatible): Likewise.
801 (s390_match_ccmode_set): Likewise.
802 (s390_canonicalize_comparison): Likewise.
803 (s390_emit_compare_and_swap): Likewise.
804 (s390_branch_condition_mask): Likewise.
805 (s390_rtx_costs): Likewise.
806 (s390_secondary_reload): Likewise.
807 (__SECONDARY_RELOAD_CASE): Likewise.
808 (s390_expand_cs): Likewise.
809 (s390_preferred_simd_mode): Likewise.
810 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
811 * config/sh/sh.c (sh_print_operand): Likewise.
812 (dump_table): Likewise.
813 (sh_secondary_reload): Likewise.
814 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
815 * config/sh/sh.md (casesi_worker_1): Likewise.
816 (casesi_worker_2): Likewise.
817 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
818 (fcc_comparison_operator): Likewise.
819 * config/sparc/sparc.c (sparc_expand_move): Likewise.
820 (emit_soft_tfmode_cvt): Likewise.
821 (sparc_preferred_simd_mode): Likewise.
822 (output_cbranch): Likewise.
823 (sparc_print_operand): Likewise.
824 (sparc_expand_vec_perm_bmask): Likewise.
825 (vector_init_bshuffle): Likewise.
826 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
827 (spu_vector_mode_supported_p): Likewise.
828 (spu_expand_insv): Likewise.
829 (spu_emit_branch_or_set): Likewise.
830 (spu_handle_vector_attribute): Likewise.
831 (spu_builtin_splats): Likewise.
832 (spu_builtin_extract): Likewise.
833 (spu_builtin_promote): Likewise.
834 (spu_expand_sign_extend): Likewise.
835 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
836 (tilegx_simd_int): Likewise.
837 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
838 (tilepro_simd_int): Likewise.
839 * config/v850/v850.c (const_double_split): Likewise.
840 (v850_print_operand): Likewise.
841 (ep_memory_offset): Likewise.
842 * config/vax/vax.c (vax_rtx_costs): Likewise.
843 (vax_output_int_move): Likewise.
844 (vax_output_int_add): Likewise.
845 (vax_output_int_subtract): Likewise.
846 * config/visium/predicates.md (visium_branch_operator): Likewise.
847 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
848 (visium_print_operand_address): Likewise.
849 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
850 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
851 (xtensa_expand_conditional_branch): Likewise.
852 (xtensa_copy_incoming_a7): Likewise.
853 (xtensa_output_literal): Likewise.
854 * dfp.c (decimal_real_maxval): Likewise.
855 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
858 Alan Hayward <alan.hayward@arm.com>
859 David Sherwood <david.sherwood@arm.com>
861 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
862 (mode_nunits_inline): Likewise.
863 (mode_inner_inline): Likewise.
864 (mode_unit_size_inline): Likewise.
865 (mode_unit_precision_inline): Likewise.
866 (emit_insn_modes_h): Likewise. Also emit a #define of the
868 (emit_mode_wider): Add an E_ prefix to mode names.
869 (emit_mode_complex): Likewise.
870 (emit_mode_inner): Likewise.
871 (emit_mode_adjustments): Likewise.
872 (emit_mode_int_n): Likewise.
873 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
874 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
875 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
876 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
877 (CRC32_BUILTIN, ENTRY): Likewise.
878 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
879 (aarch64_pop_regs): Likewise.
880 (aarch64_process_components): Likewise.
881 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
882 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
883 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
884 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
885 * config/arm/arm.c (arm_init_libfuncs): Likewise.
886 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
888 * config/i386/i386-builtin.def (pcmpestr): Likewise.
889 (pcmpistr): Likewise.
890 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
891 * config/mmix/mmix.c (mmix_output_condition): Likewise.
892 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
894 * config/rl78/rl78.c (mduc_regs): Likewise.
895 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
896 (htm_expand_builtin): Likewise.
897 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
898 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
899 * config/xtensa/xtensa.c (print_operand): Likewise.
900 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
901 (NUM_MODE_VECTOR_INT): Likewise.
902 * genoutput.c (null_operand): Likewise.
903 (output_operand_data): Likewise.
904 * genrecog.c (print_parameter_value): Likewise.
905 * lra.c (debug_operand_data): Likewise.
907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
909 * dwarf2out.c (scompare_loc_descriptor_wide)
910 (scompare_loc_descriptor_narrow): New functions, split out from...
911 (scompare_loc_descriptor): ...here.
912 * expmed.c (emit_store_flag_int): New function, split out from...
913 (emit_store_flag): ...here.
915 2017-08-30 Richard Biener <rguenther@suse.de>
917 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
918 (dwarf2out_early_finish): Move setting of AT_pubnames from
919 early debug output to early finish.
921 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
923 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
924 and -mdata-region to the assembler.
926 2017-08-30 Richard Biener <rguenther@suse.de>
928 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
930 (gen_subprogram_die): Add DW_AT_object_pointer only early.
931 (dwarf2out_early_global_decl): Only generate a DIE for the
932 abstract origin if it doesn't already exist or is a declaration DIE.
933 (resolve_addr): Do not add the linkage name twice when
934 generating a stub DIE for the DW_TAG_GNU_call_site target.
936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
938 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
939 Use machine_mode rather than int for arg1_mode.
941 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
944 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
945 that the second argument of the built-in functions to unpack
946 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
947 switch statement instead a lot of if statements.
948 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
949 Allow 64-bit values to be in Altivec registers as well as
950 traditional floating point registers.
951 (pack<mode>, FMOVE128_VSX iterator): Likewise.
953 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
955 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
956 MAX_REGS_PER_ADDRESS == 1.
958 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
960 * config/i386/i386.opt (flag_fentry): Do not init to -1.
961 * config/i386/i386.c (ix86_option_override_internal): Simplify
962 setting of opts->x_flag_entry.
964 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
965 Jakub Jelinek <jakub@redhat.com>
966 Richard Biener <rguenther@suse.de>
968 PR tree-optimization/81503
969 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
970 folded constant fits in the target type; reorder tests for clarity.
972 2017-08-29 Martin Liska <mliska@suse.cz>
974 * passes.def: Include pass_lower_switch.
975 * stmt.c (dump_case_nodes): Remove and move to
976 tree-switch-conversion.
977 (case_values_threshold): Likewise.
978 (expand_switch_as_decision_tree_p): Likewise.
979 (emit_case_decision_tree): Likewise.
980 (expand_case): Likewise.
981 (balance_case_nodes): Likewise.
982 (node_has_low_bound): Likewise.
983 (node_has_high_bound): Likewise.
984 (node_is_bounded): Likewise.
985 (emit_case_nodes): Likewise.
986 (struct simple_case_node): New struct.
987 (add_case_node): Remove.
988 (emit_case_dispatch_table): Use vector instead of case_list.
989 (reset_out_edges_aux): Remove.
990 (compute_cases_per_edge): Likewise.
991 (expand_case): Build list of simple_case_node.
992 (expand_sjlj_dispatch_table): Use it.
993 * tree-switch-conversion.c (struct case_node): Moved from
995 (emit_case_nodes): Likewise.
996 (node_has_low_bound): Likewise.
997 (node_has_high_bound): Likewise.
998 (node_is_bounded): Likewise.
999 (case_values_threshold): Likewise.
1000 (reset_out_edges_aux): Likewise.
1001 (compute_cases_per_edge): Likewise.
1002 (add_case_node): Likewise.
1003 (dump_case_nodes): Likewise.
1004 (balance_case_nodes): Likewise.
1005 (expand_switch_as_decision_tree_p): Likewise.
1006 (emit_jump): Likewise.
1007 (emit_case_decision_tree): Likewise.
1008 (try_switch_expansion): Likewise.
1009 (do_jump_if_equal): Likewise.
1010 (emit_cmp_and_jump_insns): Likewise.
1011 (fix_phi_operands_for_edge): New function.
1012 (record_phi_operand_mapping): Likewise.
1013 (class pass_lower_switch): New pass.
1014 (pass_lower_switch::execute): New function.
1015 (make_pass_lower_switch): Likewise.
1016 (conditional_probability):
1017 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1018 * tree-pass.h: Add make_pass_lower_switch.
1020 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1023 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1026 2017-08-29 Richard Biener <rguenther@suse.de>
1028 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1029 we do not add a DW_AT_inline attribute twice.
1030 (gen_subprogram_die): Remove code setting DW_AT_inline on
1031 DECL_ABSTRACT_P nodes.
1033 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1035 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1036 calls to internal functions.
1037 (gimplify_modify_expr): Likewise.
1038 * tree-call-cdce.c (use_internal_fn): Likewise.
1039 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1040 (convert_to_divmod): Set the nothrow flag.
1041 * tree-if-conv.c (predicate_mem_writes): Likewise.
1042 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1043 (vectorizable_call): Likewise.
1044 (vectorizable_store): Likewise.
1045 (vectorizable_load): Likewise.
1046 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1047 (vect_recog_mask_conversion_pattern): Likewise.
1049 2017-08-29 Martin Liska <mliska@suse.cz>
1052 * gcc.c (driver_handle_option): Add new argument.
1053 * opts-common.c (handle_option): Pass
1054 target_option_override_hook.
1055 * opts-global.c (lang_handle_option): Add new option.
1056 (set_default_handlers): Add new argument.
1057 (decode_options): Likewise.
1058 * opts.c (target_handle_option): Likewise.
1059 (common_handle_option): Call target_option_override_hook.
1060 * opts.h (struct cl_option_handler_func): Add hook for
1061 target option override.
1062 (struct cl_option_handlers): Likewise.
1063 (set_default_handlers): Add new argument.
1064 (decode_options): Likewise.
1065 (common_handle_option): Likewise.
1066 (target_handle_option): Likewise.
1067 * toplev.c (toplev::main): Pass targetm.target_option.override
1070 2017-08-29 Richard Biener <rguenther@suse.de>
1071 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1073 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1074 life to the active subtree.
1076 2017-08-28 Jeff Law <law@redhat.com>
1078 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1079 derive_equivalences.
1080 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1082 (edge_info::derive_equivalences): New private member function
1084 * tree-ssa-dom.c (class edge_info): Changed from a struct
1085 to a class. Add ctor/dtor, methods and data members.
1086 (edge_info::edge_info): Renamed from allocate_edge_info.
1087 Initialize additional members.
1088 (edge_info::~edge_info): New.
1089 (free_dom_edge_info): Delete the edge info.
1090 (record_edge_info): Use new class & associated member functions.
1091 Tighten forms for testing for edge equivalences.
1092 (record_temporary_equivalences): Iterate over the simple
1093 equivalences rather than assuming there's only one per edge.
1094 (cprop_into_successor_phis): Iterate over the simple
1095 equivalences rather than assuming there's only one per edge.
1096 (optimize_stmt): Use operand_equal_p rather than pointer
1097 equality for mini-DSE code.
1099 2017-08-28 Nathan Sidwell <nathan@acm.org>
1101 * gcc.c (execute): Fold SIGPIPE handling into switch
1102 statement. Adjust internal error message.
1104 2017-08-28 Richard Biener <rguenther@suse.de>
1107 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1108 Do nothing for removed DIEs.
1110 2017-08-28 Richard Biener <rguenther@suse.de>
1112 PR tree-optimization/81977
1113 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1116 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1119 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1120 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1121 using targetm.gen_mem_thread_fence.
1123 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1126 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1127 predicate to register_operand. Reorder operands.
1128 (*btr<mode>): Ditto.
1129 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1130 (*btr<mode>_mask): Ditto.
1132 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1134 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1135 * config/rs6000/xmmintrin.h: New file.
1136 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1138 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1141 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1142 parameter and_insn and return it.
1143 (recombine_lvx_pattern): Insert a copy to ensure availability of
1144 the base register of the copied masking operation at the point of
1145 the instruction replacement.
1146 (recombine_stvx_pattern): Likewise.
1148 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1150 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1151 undocumented switches.
1152 (-mpower9-dform-vector): Likewise.
1153 (-mpower9-dform): Likewise.
1154 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1155 comments to delete references to -mpower9-dform* switches.
1156 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1157 Delete reference to -mpower9-dform* switches, test for
1158 -mpower9-vector instead.
1159 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1160 (OTHER_P9_VECTOR_MASKS): Likewise.
1161 (POWERPC_MASKS): Likewise.
1162 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1163 tests against -mpower9-dform* to -mpower9-vector. Delete code
1164 that checked for -mpower9-dform* consistancy with other options.
1165 Add test for -mpower9-misc to enable other power9 switches.
1166 (rs6000_init_hard_regno_mode_ok): Likewise.
1167 (rs6000_option_override_internal): Likewise.
1168 (rs6000_emit_prologue): Likewise.
1169 (rs6000_emit_epilogue): Likewise.
1170 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1171 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1172 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1174 (emit_fusion_p9_store): Likewise.
1175 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1176 resetting these macros if the assembler does not support ISA 3.0
1178 (TARGET_P9_DFORM_VECTOR): Likewise.
1179 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1180 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1182 2017-08-25 Alan Modra <amodra@gmail.com>
1184 PR rtl-optimization/81747
1185 * cse.c (cse_extended_basic_block): Don't attempt to record
1186 equivalences for degenerate conditional branches that branch
1187 to their fall-through.
1189 2017-08-24 Martin Sebor <msebor@redhat.com>
1192 * gimple-fold.c (size_must_be_zero_p): New function.
1193 (gimple_fold_builtin_memory_op): Call it.
1195 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1197 * config/rs6000/mm_malloc.h: New file.
1199 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1201 PR tree-optimization/81913
1202 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1203 analysis when either IVs in condition can wrap.
1205 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1207 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1208 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1210 2017-08-24 Richard Biener <rguenther@suse.de>
1213 * targhooks.c (default_target_can_inline_p): Properly
1214 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1215 is present and always compare.
1216 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1217 infer -mfpmath=sse from TARGET_SSE_P.
1218 (ix86_can_inline_p): Properly use target_option_default_node when
1219 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1221 2017-08-24 Richard Biener <rguenther@suse.de>
1224 * dwarf2out.c (output_die): Handle flag_generate_offload like
1226 (output_comp_unit): Likewise.
1227 (gen_array_type_die): Likewise.
1228 (dwarf2out_early_finish): Likewise.
1229 (note_variable_value_in_expr): Likewise.
1230 (dwarf2out_finish): Likewise. Adjust assert.
1231 * cgraphunit.c (symbol_table::compile): Move setting of
1232 flag_generate_offload earlier ...
1233 (symbol_table::finalize_compilation_unit): ... here, before
1234 early debug finalization.
1236 2017-08-24 Richard Biener <rguenther@suse.de>
1238 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1239 and ipa-fnsummary.h.
1240 (ix86_can_inline_p): When ix86_fpmath flags do not match
1241 check whether the callee uses FP math at all.
1243 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1246 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1247 instead of TYPE_PRECISION.
1249 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1250 Alan Hayward <alan.hayward@arm.com>
1251 David Sherwood <david.sherwood@arm.com>
1253 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1255 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1257 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1259 (ix86_frame::stack_realign_allocate): New field.
1260 (struct machine_frame_state): Modify comments.
1261 (machine_frame_state::sp_realigned_fp_end): New field.
1262 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1264 (sp_valid_at): Add assertion to assure no attempt to access invalid
1265 offset of a realigned stack.
1266 (fp_valid_at): Likewise.
1267 (choose_baseaddr): Modify comments.
1268 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1269 ix86_expand_prologue.
1270 (ix86_expand_prologue): Modify stack realignment and allocation.
1271 (ix86_expand_epilogue): Modify comments.
1272 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1273 avx2_runtime, avx512f, and avx512f_runtime.
1275 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1277 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1278 (mstackrealign): Do not init to -1.
1279 * config/i386/i386.c (ix86_option_override_internal):
1280 Check opts_set, not opts when setting default value of
1281 opts->x_ix86_force_align_arg_pointer.
1283 2017-08-23 Richard Biener <rguenther@suse.de>
1285 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1286 lang_hooks.decl_printable_name.
1287 * print-rtl-function.c (print_rtx_function): Likewise.
1288 * tree-pretty-print.c (dump_function_header): Likewise.
1290 2017-08-23 Richard Biener <rguenther@suse.de>
1293 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1294 -g0 at compile-time.
1296 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1299 * doc/sourcebuild.texi (Other hardware attributes):
1302 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1305 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1306 Fix single-use check.
1308 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1310 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1312 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1314 * doc/install.texi: Modify to add more details on running selected
1317 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1319 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1320 is combined with -mabi=ms.
1321 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1324 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1326 PR tree-optimization/81488
1327 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1328 and cached_basis fields.
1329 (MAX_SPREAD): New constant.
1330 (alloc_cand_and_find_basis): Initialize new fields.
1331 (clear_visited): New function.
1332 (create_phi_basis_1): Rename from create_phi_basis, set visited
1333 and cached_basis fields.
1334 (create_phi_basis): New wrapper function.
1335 (phi_add_costs_1): Rename from phi_add_costs, add spread
1336 parameter, set visited field, short-circuit when limits reached.
1337 (phi_add_costs): New wrapper function.
1338 (record_phi_increments_1): Rename from record_phi_increments, set
1340 (record_phi_increments): New wrapper function.
1341 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1342 (phi_incr_cost): New wrapper function.
1343 (all_phi_incrs_profitable_1): Rename from
1344 all_phi_incrs_profitable, set visited field.
1345 (all_phi_incrs_profitable): New wrapper function.
1347 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1348 Alan Hayward <alan.hayward@arm.com>
1349 David Sherwood <david.sherwood@arm.com>
1351 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1352 that takes the outer and inner modes.
1353 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1354 comparison as the canonical test for a paradoxical subreg.
1355 * combine.c (simplify_set): Use paradoxical_subreg_p.
1356 (make_extraction): Likewise.
1357 (force_to_mode): Likewise.
1358 (rtx_equal_for_field_assignment_p): Likewise.
1359 (gen_lowpart_for_combine): Likewise.
1360 (simplify_comparison): Likewise.
1361 * cse.c (equiv_constant): Likewise.
1362 * expmed.c (store_bit_field_1): Likewise.
1363 * final.c (alter_subreg): Likewise.
1364 * fwprop.c (propagate_rtx): Likewise.
1365 (forward_propagate_subreg): Likewise.
1366 * ira-conflicts.c (ira_build_conflicts): Likewise.
1367 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1368 * lra-constraints.c (curr_insn_transform): Likewise.
1369 (split_reg): Likewise.
1370 * lra-eliminations.c (move_plus_up): Likewise.
1371 (lra_eliminate_regs_1): Likewise.
1372 * recog.c (general_operand): Likewise.
1373 * ree.c (combine_reaching_defs): Likewise.
1374 * reload.c (push_reload): Likewise.
1375 (find_reloads): Likewise.
1376 * reload1.c (elimination_effects): Likewise.
1377 (compute_reload_subreg_offset): Likewise.
1378 (choose_reload_regs): Likewise.
1379 * rtlanal.c (subreg_lsb_1): Likewise.
1380 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1381 (simplify_subreg): Likewise.
1382 * var-tracking.c (track_loc_p): Likewise.
1383 * emit-rtl.c (byte_lowpart_offset): Likewise.
1384 (paradoxical_subreg_p): Delete out-of-line definition.
1386 2017-08-22 Jeff Law <law@redhat.com>
1388 PR tree-optimization/81741
1389 PR tree-optimization/71947
1390 * tree-ssa-dom.c: Include tree-inline.h.
1391 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1392 equivalences if one is more expensive to compute than the other.
1393 * tree-ssa-scopedtables.h (class const_or_copies): Make
1394 record_const_or_copy_raw method private.
1395 (class avail_exprs_stack): New method simplify_binary_operation.
1396 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1397 avail_exprs_stack::simplify_binary_operation as needed.
1398 (avail_exprs_stack::simplify_binary_operation): New function.
1400 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
1402 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
1403 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
1404 (DOT_SYMBOLS): Likewise.
1405 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
1406 (RELOCATABLE_NEEDS_FIXUP): Likewise.
1407 (RS6000_ABI_NAME): Likewise.
1408 (TARGET_CMODEL): Likewise.
1409 (TOC_SECTION_ASM_OP): Likewise.
1410 (SET_CMODEL): New macro.
1411 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
1413 2017-08-22 Richard Biener <rguenther@suse.de>
1415 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
1416 to allow for free-lang-data replacements similar to verify_type_variant.
1418 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1419 Alan Hayward <alan.hayward@arm.com>
1420 David Sherwood <david.sherwood@arm.com>
1422 * config/aarch64/aarch64.md (casesi): Use DImode rather than
1423 VOIDmode for the LABEL_REF.
1425 2017-08-22 Richard Biener <rguenther@suse.de>
1427 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
1429 2017-08-22 Richard Biener <rguenther@suse.de>
1431 * common.opt (feliminate-dwarf2-dups): Ignore.
1432 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
1433 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
1434 same_die_p_wrap, compute_section_prefix,
1435 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
1436 (comdat_symbol_id, comdat_symbol_number): Likewise.
1437 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
1439 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
1440 (output_die): Mark unreachable path unreachable.
1441 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
1442 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
1443 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
1444 (dwarf2out_early_finish): Likewise.
1446 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
1448 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
1450 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
1453 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
1454 not VAR_P. Filter attribute warnings with OPT_Wattributes.
1455 (avr_attribute_table) <io, io_low, address>: Initialize
1456 .decl_required with true.
1458 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
1460 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
1461 undocumented debugging options.
1462 (-mvsx-scalar-double): Likewise.
1463 (-mallow-df-permute): Likewise.
1464 (-mvectorize-builtins): Likewise.
1465 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1466 (rs6000_builtin_vectorized_function): Likewise.
1467 (rs6000_builtin_md_vectorized_function): Likewise.
1468 (rs6000_opt_vars): Likewise.
1470 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
1473 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
1474 (*btrq_imm): Rename from *btrq.
1475 (*btcq_imm): Rename from *btcq.
1476 (btsc): New code attribute.
1477 (*<btsc><mode>): New insn pattern.
1478 (*btr<mode>): Ditto.
1479 (*<btsc><mode>_mask): New insn_and_split pattern.
1480 (*btr<mode>_mask): Ditto.
1482 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1483 Alan Hayward <alan.hayward@arm.com>
1484 David Sherwood <david.sherwood@arm.com>
1486 * function.c (pad_below): Simplify padding calculation.
1488 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1489 Alan Hayward <alan.hayward@arm.com>
1490 David Sherwood <david.sherwood@arm.com>
1492 * target.def (function_prologue): Remove frame size argument.
1493 (function_epilogue): Likewise.
1494 * doc/tm.texi: Regenerate.
1495 * final.c (final_start_function): Update call to function_prologue.
1496 (final_end_function): Update call to function_epilogue.
1497 (default_function_pro_epilogue): Remove frame size argument.
1498 * output.h (default_function_pro_epilogue): Likewise.
1499 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
1500 (arm_output_function_prologue): Likewise.
1501 * config/frv/frv.c (frv_function_prologue): Likewise.
1502 (frv_function_epilogue): Likewise.
1503 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
1504 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
1505 (ia64_output_function_epilogue): Likewise.
1506 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
1507 (m32r_output_function_epilogue): Likewise.
1508 * config/microblaze/microblaze.c (microblaze_function_prologue)
1509 (microblaze_function_epilogue): Likewise.
1510 * config/mips/mips.c (mips_output_function_prologue): Likewise.
1511 (mips_output_function_epilogue): Likewise.
1512 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
1513 (mmix_target_asm_function_epilogue): Likewise.
1514 * config/msp430/msp430.c (msp430_start_function): Likewise.
1515 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
1516 (nds32_asm_function_epilogue): Likewise.
1517 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
1518 * config/pa/pa.c (pa_output_function_prologue): Likewise.
1519 (pa_output_function_epilogue): Likewise.
1520 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
1521 (rs6000_output_function_epilogue): Likewise.
1522 * config/rl78/rl78.c (rl78_start_function): Likewise.
1523 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
1524 (rs6000_output_function_epilogue): Likewise.
1525 * config/rx/rx.c (rx_output_function_prologue): Likewise.
1526 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
1527 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
1528 (sparc_asm_function_epilogue): Likewise.
1530 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1532 * tree.h (type_has_mode_precision_p): New function.
1533 * convert.c (convert_to_integer_1): Use it.
1534 * expr.c (expand_expr_real_2): Likewise.
1535 (expand_expr_real_1): Likewise.
1536 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
1537 * match.pd: Likewise.
1538 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
1539 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
1540 * tree-tailcall.c (process_assignment): Likewise.
1541 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1542 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
1543 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
1544 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1545 (vectorizable_assignment): Likewise.
1546 (vectorizable_shift): Likewise.
1547 (vectorizable_operation): Likewise.
1548 * tree-vrp.c (register_edge_assert_for_2): Likewise.
1550 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
1552 * match.pd: Add pow (C, x) simplification.
1554 2017-08-21 Richard Biener <rguenther@suse.de>
1556 PR tree-optimization/81900
1557 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
1558 for blocks with abnormal predecessors.
1559 (compute_antic): Do not set visited flag prematurely.
1561 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
1564 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
1566 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1568 * stor-layout.h (vector_type_mode): Move to...
1569 * tree.h (vector_type_mode): ...here.
1570 * stor-layout.c (vector_type_mode): Move to...
1571 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
1573 2017-08-21 Richard Biener <rguenther@suse.de>
1575 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
1576 register_external_die hooks.
1577 (debug_false_tree_charstarstar_uhwistar): Declare.
1578 (debug_nothing_tree_charstar_uhwi): Likewise.
1579 * debug.c (do_nothing_debug_hooks): Adjust.
1580 (debug_false_tree_charstarstar_uhwistar): New do nothing.
1581 (debug_nothing_tree_charstar_uhwi): Likewise.
1582 * dbxout.c (dbx_debug_hooks): Adjust.
1583 (xcoff_debug_hooks): Likewise.
1584 * sdbout.c (sdb_debug_hooks): Likewise.
1585 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1586 * dwarf2out.c (macinfo_label_base): New global.
1587 (dwarf2out_register_external_die): New function for the
1588 register_external_die hook.
1589 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
1590 (dwarf2_debug_hooks): Use them.
1591 (dwarf2_lineno_debug_hooks): Adjust.
1592 (struct die_struct): Add with_offset flag.
1593 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
1594 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
1595 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
1596 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
1597 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
1598 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
1599 defining section names for the early LTO debug variants.
1600 (reset_indirect_string): New helper.
1601 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
1602 (print_dw_val): Add support for offsetted symbol references.
1603 (get_ultimate_context): Split out from is_cxx.
1604 (is_cxx): Use get_ultimate_context.
1605 (is_fortran): Add decl overload.
1606 (compute_comp_unit_symbol): Split out worker from
1607 compute_section_prefix.
1608 (compute_section_prefix): Call compute_comp_unit_symbol and
1609 set comdat_type_p here.
1610 (output_die): Skip DIE symbol output for the LTO added one.
1611 Handle DIE symbol references with offset.
1612 (output_comp_unit): Guard section name mangling properly.
1613 For LTO debug sections emit a symbol at the section beginning
1614 which we use to refer to its DIEs.
1615 (add_abstract_origin_attribute): For DIEs registered via
1616 dwarf2out_register_external_die directly refer to the early
1617 DIE rather than indirectly through the shadow one we created.
1618 Remove obsolete call to dwarf2out_abstract_function for
1619 non-function/block origins.
1620 (gen_array_type_die): When generating early LTO debug do
1621 not emit DW_AT_string_length.
1622 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
1623 late when in LTO. As suggested place a gcc_unreachable for
1624 the DECL_ABSTRACT_P case.
1625 (gen_subprogram_die): Avoid another specification DIE
1626 for early built declarations/definitions for the late LTO case.
1627 (gen_variable_die): Add type references for late duplicated VLA dies
1629 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
1630 we have the abstract instance already.
1631 (process_scope_var): Adjust decl DIE contexts in LTO which
1632 first puts them in limbo.
1633 (gen_decl_die): Do not generate type DIEs late apart from
1634 types for VLAs or for decls we do not yet have a DIE. Do not
1635 call dwarf2out_abstract_function late.
1636 (dwarf2out_early_global_decl): Make sure to create DIEs
1637 for abstract instances of a decl first.
1638 (dwarf2out_late_global_decl): Adjust comment.
1639 (output_macinfo_op): With multiple macro sections use
1640 macinfo_label_base to distinguish labels.
1641 (output_macinfo): Likewise. Update macinfo_label_base.
1642 Pass in the line info label.
1643 (note_variable_value_in_expr): When generating LTO resolve
1644 all variable values here by generating DIEs as needed.
1645 (init_sections_and_labels): Add early LTO debug flag parameter
1646 and generate different sections and names if set. Add generation
1647 counter for the labels so we can have multiple of them.
1648 (reset_dies): Helper to allow DIEs to be output multiple times.
1649 (dwarf2out_finish): When outputting DIEs to the fat part of an
1650 LTO object first reset DIEs.
1651 (dwarf2out_early_finish): Output early DIEs when generating LTO.
1652 (modified_type_die): Check for decl_ultimate_origin being self
1654 (gen_type_die_with_usage): Likewise.
1655 (gen_typedef_die): Allow decl_ultimate_origin being self.
1656 (set_decl_abstract_flags): Remove.
1657 (set_block_abstract_flags): Likewise.
1658 (dwarf2out_abstract_function): Treat the early generated DIEs
1659 as the abstract copy and only add DW_AT_inline and
1660 DW_AT_artificial here and call set_decl_origin_self.
1661 If the DIE has an abstract origin don't do anything.
1662 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
1663 if we have none yet (Go fails to build one, PR78628).
1664 (variably_modified_type_p): Prevent endless recursion for Ada
1665 cyclic pointer types.
1666 * lto-streamer-in.c: Include debug.h.
1667 (dref_queue): New global.
1668 (lto_read_tree_1): Stream in DIE references.
1669 (lto_input_tree): Register DIE references.
1670 (input_function): Stream DECL_DEBUG_ARGS.
1671 * lto-streamer-out.c: Include debug.h.
1672 (lto_write_tree_1): Output DIE references.
1673 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
1674 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
1675 (output_function): Stream DECL_DEBUG_ARGS.
1676 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
1677 Stream DECL_ABSTRACT_ORIGIN.
1678 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
1679 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
1680 DECL_CONTEXT for file-scope decls.
1681 * lto-streamer.h (struct dref_entry): Declare.
1682 (dref_queue): Likewise.
1683 * cfgexpand.c (pass_expand::execute): Do not call the
1684 outlining_inline_function hook here.
1685 * lto-wrapper.c (debug_obj): New global.
1686 (tool_cleanup): Unlink it if required.
1687 (debug_objcopy): New function.
1688 (run_gcc): Handle early debug sections in the IL files by
1689 extracting them to separate files, partially linkin them and
1690 feeding the result back as result to the linker.
1691 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
1692 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
1693 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
1694 sections into a separate segment.
1695 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
1697 (darwin_asm_dwarf_section): Likewise.
1698 (darwin_asm_output_dwarf_offset): Likewise.
1699 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
1701 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1702 Alan Hayward <alan.hayward@arm.com>
1703 David Sherwood <david.sherwood@arm.com>
1705 * read-md.h (md_reader::record_potential_iterator_use): Replace
1706 pointer argument with an rtx and an index.
1707 * read-rtl.c (iterator_group::apply_iterator): Likewise.
1708 (apply_mode_iterator): Likewise.
1709 (apply_code_iterator): Likewise.
1710 (apply_int_iterator): Likewise.
1711 (apply_subst_iterator): Likewise.
1712 (record_iterator_use): Likewise.
1713 (record_attribute_use): Likewise.
1714 (md_reader::record_potential_iterator_use): Likewise. Update calls
1715 to record_iterator_use and apply_iterator.
1716 (iterator_use): Replace ptr with x and index.
1717 (attribute_use): Likewise.
1718 (apply_attribute_uses): Update calls to apply_iterator.
1719 (apply_iterators): Likewise. Update initialization of iterator_use.
1720 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
1721 and record_potential_iterator_use.
1722 (rtx_reader::read_rtx_operand): Likewise.
1724 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1725 Alan Hayward <alan.hayward@arm.com>
1726 David Sherwood <david.sherwood@arm.com>
1728 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
1731 2017-08-21 Richard Biener <rguenther@suse.de>
1734 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
1735 at struct end conservatively when comparing common bases.
1737 2017-08-21 Richard Biener <rguenther@suse.de>
1739 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
1740 (mem_ref_in_stmt): Remove.
1741 (determine_max_movement): Use ref index to get at the reference.
1742 (invariantness_dom_walker::before_dom_children): Deal with
1743 lim data already initialized.
1744 (gather_mem_refs_stmt): Initialize lim data and record ref index.
1746 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
1748 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
1749 (TARGET_ISA_ROUND): Ditto.
1750 (TARGET_ROUND): Ditto.
1751 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
1752 * config/i386/i386.md: Ditto.
1753 * config/i386/sse.md: Ditto.
1754 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
1755 with OPTION_MASK_ISA_SSE4_1.
1757 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
1760 * doc/extend.texi (x86 Built-in Functions): Correct the name of
1761 __builtin_ia32_lzcnt_u16.
1763 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1766 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
1767 (rs6000_set_current_function): Rewrite function to use it.
1769 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
1772 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
1773 and TYPE_WARN_IF_NOT_ALIGN.
1774 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
1775 (handle_warn_if_not_align): New.
1776 (place_union_field): Call handle_warn_if_not_align.
1777 (place_field): Call handle_warn_if_not_align.
1778 Copy TYPE_WARN_IF_NOT_ALIGN.
1779 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
1780 (layout_type): Likewise.
1781 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
1783 (tree_decl_common): Add warn_if_not_align.
1784 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
1785 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
1786 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
1787 (DECL_WARN_IF_NOT_ALIGN): Likewise.
1788 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
1789 * doc/extend.texi: Document warn_if_not_aligned attribute.
1790 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
1792 2017-08-17 Martin Liska <mliska@suse.cz>
1795 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
1796 (get_data_dependence): Use it as pointer type.
1797 (distribute_loop): Likewise.
1799 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1801 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
1802 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
1803 (p8_vmrgow_<mode>_direct): New define_insn.
1804 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
1805 handle endianness for vmrgew and vmrgow permute patterns.
1807 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
1809 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
1810 * config/rs6000/rs6000-cpus.def: Remove comment.
1811 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
1812 (POWERPC_MASKS): Likewise.
1813 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
1814 use of TARGET_VSX_TIMODE.
1815 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
1816 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
1817 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
1818 (rs6000_option_override_internal): Remove dead code.
1819 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
1820 (rs6000_legitimize_reload_address): Likewise.
1821 (rs6000_legitimate_address_p): Likewise.
1822 (rs6000_opt_masks): Delete "vsx-timode".
1823 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
1824 from function comment.
1825 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
1826 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
1827 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
1829 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
1830 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
1831 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
1834 (define_peephole2): Likewise.
1836 2017-08-17 Martin Sebor <msebor@redhat.com>
1839 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
1840 past the end of an array.
1841 (test_pp_format): Add test cases.
1843 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
1845 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
1846 missing ECF_NOTHROW flags.
1848 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
1851 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
1852 operands residing in integer registers.
1853 (*vsx_le_perm_load_<mode>): Likewise.
1854 (*vsx_le_perm_store_<mode>): Likewise.
1855 (define_peephole2): Add peepholes to optimize the above.
1857 2017-08-17 Marek Polacek <polacek@redhat.com>
1860 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
1861 to mimic what shorten_compare did. Change the return type to bool.
1862 (fold_cond_expr_with_comparison): Update call to
1863 operand_equal_for_comparison_p.
1864 (fold_ternary_loc): Likewise.
1866 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
1868 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
1870 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
1872 2017-08-17 Richard Biener <rguenther@suse.de>
1874 * tree-ssa-structalias.c (solve_graph): When propagating
1875 to successors update the graphs succ edges and avoid duplicate work.
1877 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
1880 * config/i386/i386.c (ix86_option_override_internal): Save target
1881 specific options after ix86_stack_protector_guard_reg was changed.
1883 2017-08-17 Richard Biener <rguenther@suse.de>
1885 PR tree-optimization/81827
1886 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
1887 (new_var_info): Initialize it conservatively.
1888 (get_call_vi): Mark register vars.
1889 (new_scalar_tmp_constraint_exp): Likewise.
1890 (handle_rhs_call): Likewise.
1891 (handle_const_call): Likewise.
1892 (create_function_info_for): Likewise.
1893 (solve_constraints): Sort varinfos to separate register from
1894 non-register vars to pack points-to solution bitmaps during
1897 2017-08-17 Marek Polacek <polacek@redhat.com>
1899 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
1901 2017-08-17 Richard Biener <rguenther@suse.de>
1903 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
1904 to true when overflow is undefined and we saturated the result.
1906 2017-08-17 Alan Modra <amodra@gmail.com>
1909 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
1910 Don't use store multiple if only one reg needs saving.
1911 (interesting_frame_related_regno): New function.
1912 (rs6000_frame_related): Don't emit frame info for regs that
1914 (rs6000_emit_epilogue): Likewise.
1916 2017-08-16 Nathan Sidwell <nathan@acm.org>
1918 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
1919 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
1920 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
1921 * tree.c (free_lang_data_in_type): Use else-if chain. Always
1922 clear TYPE_LANG_1. Remove obsolete member-function stripping.
1923 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
1924 (verify_type): Adjust for TYPE_BINFO move.
1925 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
1926 process TYPE_BINFO directly.
1927 (hash_tree): Likewise.
1928 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1930 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1933 2017-08-16 David Malcolm <dmalcolm@redhat.com>
1935 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
1937 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
1940 * config/i386/i386.md (*anddi_1_btr): Change predicates of
1941 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
1942 Add ix86_binary_operator_ok to insn constraint.
1943 (*iordi_1_bts): Ditto.
1944 (*xordi_1_btc): Ditto.
1945 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
1946 Update corresponding peephole2 pattern.
1950 2017-08-16 Bin Cheng <bin.cheng@arm.com>
1952 PR tree-optimization/81832
1953 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
1954 copy loop header which has IFN_LOOP_DIST_ALIAS call.
1956 2017-08-16 Marek Polacek <polacek@redhat.com>
1959 * fold-const.c (fold_indirect_ref_1): Restore original behavior
1960 regarding size_zero_node.
1962 2017-08-16 Martin Liska <mliska@suse.cz>
1965 * config.gcc: Respect previously set extra_objs in case
1968 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
1970 PR tree-optimization/81835
1971 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
1972 the phi SSA_NAME. Check that the condition in a COND_EXPR does
1973 not depend on the phi.
1975 2017-08-16 Alan Modra <amodra@gmail.com>
1977 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
1980 2017-08-16 Alan Modra <amodra@gmail.com>
1982 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
1983 (save_reg_p): ..into this. Update all callers.
1984 (first_reg_to_save): Simplify.
1986 2017-08-16 Alan Modra <amodra@gmail.com>
1988 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
1991 2017-08-15 Joseph Myers <joseph@codesourcery.com>
1995 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
1996 constant count if that count is less than 32.
1998 2017-08-15 Nathan Sidwell <nathan@acm.org>
2000 * gcc.c (execute): Emit friendlier message if inferior is killed
2001 by an external cause.
2003 2017-08-15 Richard Biener <rguenther@suse.de>
2005 PR tree-optimization/81790
2006 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2007 CONSTRUCTORs from simplifying and VN.
2009 2017-08-14 Martin Sebor <msebor@redhat.com>
2011 * builtin-attrs.def: Add comments.
2013 2017-08-14 Martin Sebor <msebor@redhat.com>
2016 * doc/extend.texi (attribute nonstring): Document new attribute.
2018 2017-08-14 Martin Sebor <msebor@redhat.com>
2021 * tree-diagnostic.c (default_tree_printer): Handle %G.
2022 * gimple-pretty-print.h (percent_G_format): Declare new function.
2023 * gimple-pretty-print.c (percent_G_format): Define.
2024 * tree-pretty-print.c (percent_K_format): Add argument.
2026 2017-08-14 Martin Sebor <msebor@redhat.com>
2028 PR translation/79998
2029 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2030 Remove a stray space.
2032 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2035 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2036 (*iordi_1_bts): Ditto.
2037 (*xordi_1_btc): Ditto.
2039 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2042 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2043 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2045 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2046 quoted strings, and make more translator-friendly.
2047 (darwin_rs6000_override_options): Likewise.
2048 (rs6000_option_override_internal): Likewise.
2049 (rs6000_return_in_memory): Fix overlong line.
2050 (init_cmulative_args): Use quoted strings, and make more
2051 translator-friendly.
2052 (rs6000_pass_by_reference): Fix overlong line.
2053 (def_builtin): Use quoted strings.
2054 (altivec_expand_predicate_builtin): Use quoted strings, and make
2055 more translator-friendly.
2056 (htm_expand_builtin): Use quoted strings.
2057 (cpu_expand_builtin): Use quoted strings, and make more
2058 translator-friendly.
2059 (altivec_expand_builtin): Likewise.
2060 (paired_expand_predicate_builtin): Likewise.
2061 (rs6000_invalid_builtin): Likewise.
2062 (builtin_function_type): Use quoted strings.
2063 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2064 more translator-friendly.
2065 (rs6000_trampoline_init): Likewise.
2066 (rs6000_handle_altivec_attribute): Likewise.
2067 (rs6000_inner_target_options): Use quoted strings.
2068 (rs6000_disable_incompatible_switches): Likewise.
2069 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2070 strings, and make more translator-friendly.
2071 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2073 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2075 PR tree-optimization/81799
2076 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2077 cond_expr to simple gimple operand.
2079 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2082 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2084 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2088 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2089 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2090 TARGET_GASISR_PROLOGUES.
2091 * config/avr/avr.c (avr_option_override): Same.
2092 (avr_pass_pre_proep::execute): Same.
2094 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2097 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2098 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2100 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2102 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2103 define_insn_and_split. Split to a memory load from 0 in
2104 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2105 using PTR mode iterator.
2106 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2107 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2108 (*add_tp_<mode>): Redefine as define_insn_and_split.
2109 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2110 address space. Merge with *add_tp_x32 using PTR mode iterator.
2111 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2112 Split to an add with a memory load from 0 in
2113 DEFAULT_TLS_SEG_REG address space.
2115 2017-08-12 Andrew Pinski <apinski@cavium.com>
2117 * config/aarch64/aarch64-option-extensions.def (rdma):
2118 Fix feature string to what Linux prints out in /proc/cpuinfo.
2120 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2123 * dwarf2out.c (modified_type_die): For C typedef types that have
2124 an ultimate origin, process the ultimate origin instead of the
2126 (gen_typedef_die): Assert that input DECLs have no ultimate
2128 (gen_type_die_with_usage): For typedef variants that have an
2129 ultimate origin, just call gen_decl_die on the original DECL.
2130 (process_scope_var): Avoid creating DIEs for local typedefs and
2131 concrete static variables.
2133 2017-08-12 Alan Modra <amodra@gmail.com>
2137 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2138 match gnu-user.h startfile.
2139 (ENDFILE_LINUX_SPEC): Similarly.
2141 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2144 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2146 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2148 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2149 * config/aarch64/aarch64.md (mov<mode>): Change.
2150 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2151 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2152 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2154 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2156 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2157 for storage order barriers.
2159 2017-08-11 Martin Liska <mliska@suse.cz>
2162 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2163 variables of void type.
2165 2017-08-11 Martin Liska <mliska@suse.cz>
2167 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2168 TARGET_SUPPORTS_ALIASES.
2169 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2170 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2171 (optimize_weakref): Likewise.
2172 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2173 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2174 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2176 2017-08-11 Martin Liska <mliska@suse.cz>
2179 * config/i386/i386.c (make_resolver_func): Do complete
2180 refactoring of the function.
2182 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2185 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2186 * config/i386/i386.c (ix86_stack_protect_guard): Use
2187 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2188 * doc/invoke.texi (x86 Options): Document
2189 -mstack-protector-guard-symbol= option.
2191 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2193 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2194 * config/i386/i386.c (ix86_split_stack_guard): New function.
2195 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2196 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2197 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2198 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2199 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2200 (split_stack_space_check): Call ix86_split_stack_guard.
2202 2017-08-10 Martin Sebor <msebor@redhat.com>
2204 * print-tree.c (print_node): Print location using the established
2206 Replace spaces with colons.
2207 (debug_raw, debug): Ditto.
2209 2017-08-10 Martin Sebor <msebor@redhat.com>
2212 * pretty-print.c (pp_format): Correct the handling of %s precision.
2214 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2217 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2219 (ix86_finalize_stack_frame_flags): This. Also clear
2220 frame_pointer_needed if -fno-omit-frame-pointer is used without
2222 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2223 with ix86_finalize_stack_frame_flags.
2224 (ix86_expand_epilogue): Likewise.
2225 (ix86_expand_split_stack_prologue): Likewise.
2226 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2228 2017-08-10 Martin Liska <mliska@suse.cz>
2231 * c-attribs.c (handle_target_attribute):
2232 Report warning for an empty string argument of target attribute.
2234 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2237 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2239 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2240 or DECL_NONLOCAL labels.
2241 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2242 or DECL_NONLOCAL labels here.
2244 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2246 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2247 to indicate when early gimple folding has been disabled.
2248 (rs6000_gimple_fold_builtin): Add debug content.
2249 (rs6000_invalid_builtin): Fix whitespace.
2250 (rs6000_expand_builtin): Fix whitespace.
2251 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2253 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2256 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2257 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2259 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2261 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2262 (qdf24xx): Likewise.
2263 * config/aarch64/aarch64.md: Include falkor.md.
2264 * config/aarch64/falkor.md: New.
2266 2017-08-09 Marek Polacek <polacek@redhat.com>
2269 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2270 * diagnostic.c (emit_diagnostic): Add a comment.
2271 (emit_diagnostic_valist): New function.
2273 2017-08-09 Marek Polacek <polacek@redhat.com>
2276 * input.c (make_location): New overload.
2277 * input.h (make_location): Declare.
2279 2017-08-08 Alan Modra <amodra@gmail.com>
2280 H.J. Lu <hongjiu.lu@intel.com>
2283 * gcc.c (NO_PIE_SPEC): Delete.
2284 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2286 (LINK_PIE_SPEC): ..to here.
2287 (LINK_COMMAND_SPEC): Support -no-pie.
2288 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2289 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2291 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2292 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2293 (ENDFILE_CRTEND_SPEC): Similarly.
2295 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2298 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2299 (mstack-protector-guard-offset=): Ditto.
2300 * config/i386/i386.c (ix86_option_override): Handle
2301 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2303 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2304 ix86_stack_protect_guard_offset variables.
2305 (TARGET_STACK_PROTECT_GUARD): Always define.
2306 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2307 and -mstack-protector-guard-offset= options.
2309 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2311 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2312 boundary case for the last candidate.
2314 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2316 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2317 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2319 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2322 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2323 * config/aarch64/aarch64-builtins.c
2324 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2325 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2326 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2328 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2329 Andrew Pinski <pinskia@gmail.com>
2332 * internal-fn.def (XORSIGN): New.
2333 * optabs.def (xorsign_optab): New.
2334 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2335 (convert_expand_mult_copysign): New.
2336 (pass_optimize_widening_mul::execute): Call
2337 convert_expand_mult_copysign.
2339 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2341 PR tree-optimization/81354
2342 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2343 Insert on edges rather than explicitly creating landing pads.
2344 (analyze_candidates_and_replace): Commit edge inserts.
2346 2017-08-08 Richard Biener <rguenther@suse.de>
2349 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2350 (expand_simple_operations): Also look through ADDR_EXPRs with
2351 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2353 2017-08-08 Richard Biener <rguenther@suse.de>
2355 PR tree-optimization/81723
2356 * tree-vect-slp.c (struct bst_traits): New hash traits.
2357 (bst_fail): New global.
2358 (vect_build_slp_tree_2): New worker, split out from ...
2359 (vect_build_slp_tree): ... this now wrapping it with using
2360 bst_fail set to cache SLP tree build fails. Properly handle
2362 (vect_analyze_slp_instance): Allocate and free bst_fail.
2364 2017-08-08 Martin Liska <mliska@suse.cz>
2367 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2368 LABEL_DECLs that can be from a different function.
2370 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2372 PR tree-optimization/81744
2373 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2374 loop's number of iterations.
2376 2017-08-08 Martin Liska <mliska@suse.cz>
2378 * asan.c: Include header files.
2379 * attribs.c (build_decl_attribute_variant): New function moved
2381 (build_type_attribute_qual_variant): Likewise.
2382 (cmp_attrib_identifiers): Likewise.
2383 (simple_cst_list_equal): Likewise.
2384 (omp_declare_simd_clauses_equal): Likewise.
2385 (attribute_value_equal): Likewise.
2386 (comp_type_attributes): Likewise.
2387 (build_type_attribute_variant): Likewise.
2388 (lookup_ident_attribute): Likewise.
2389 (remove_attribute): Likewise.
2390 (merge_attributes): Likewise.
2391 (merge_type_attributes): Likewise.
2392 (merge_decl_attributes): Likewise.
2393 (merge_dllimport_decl_attributes): Likewise.
2394 (handle_dll_attribute): Likewise.
2395 (attribute_list_equal): Likewise.
2396 (attribute_list_contained): Likewise.
2397 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2398 (lookup_attribute_by_prefix): Likewise.
2399 * bb-reorder.c: Include header files.
2400 * builtins.c: Likewise.
2401 * calls.c: Likewise.
2402 * cfgexpand.c: Likewise.
2403 * cgraph.c: Likewise.
2404 * cgraphunit.c: Likewise.
2405 * convert.c: Likewise.
2406 * dwarf2out.c: Likewise.
2407 * final.c: Likewise.
2408 * fold-const.c: Likewise.
2409 * function.c: Likewise.
2410 * gimple-expr.c: Likewise.
2411 * gimple-fold.c: Likewise.
2412 * gimple-pretty-print.c: Likewise.
2413 * gimple.c: Likewise.
2414 * gimplify.c: Likewise.
2415 * hsa-common.c: Likewise.
2416 * hsa-gen.c: Likewise.
2417 * internal-fn.c: Likewise.
2418 * ipa-chkp.c: Likewise.
2419 * ipa-cp.c: Likewise.
2420 * ipa-devirt.c: Likewise.
2421 * ipa-fnsummary.c: Likewise.
2422 * ipa-inline.c: Likewise.
2423 * ipa-visibility.c: Likewise.
2425 * lto-cgraph.c: Likewise.
2426 * omp-expand.c: Likewise.
2427 * omp-general.c: Likewise.
2428 * omp-low.c: Likewise.
2429 * omp-offload.c: Likewise.
2430 * omp-simd-clone.c: Likewise.
2431 * opts-global.c: Likewise.
2432 * passes.c: Likewise.
2433 * predict.c: Likewise.
2434 * sancov.c: Likewise.
2435 * sanopt.c: Likewise.
2436 * symtab.c: Likewise.
2437 * toplev.c: Likewise.
2438 * trans-mem.c: Likewise.
2439 * tree-chkp.c: Likewise.
2440 * tree-eh.c: Likewise.
2441 * tree-into-ssa.c: Likewise.
2442 * tree-object-size.c: Likewise.
2443 * tree-parloops.c: Likewise.
2444 * tree-profile.c: Likewise.
2445 * tree-ssa-ccp.c: Likewise.
2446 * tree-ssa-live.c: Likewise.
2447 * tree-ssa-loop.c: Likewise.
2448 * tree-ssa-sccvn.c: Likewise.
2449 * tree-ssa-structalias.c: Likewise.
2450 * tree-ssa.c: Likewise.
2451 * tree-streamer-in.c: Likewise.
2452 * tree-vectorizer.c: Likewise.
2453 * tree-vrp.c: Likewise.
2455 * ubsan.c: Likewise.
2456 * varasm.c: Likewise.
2457 * varpool.c: Likewise.
2458 * tree.c: Remove functions moved to attribs.[ch].
2460 * config/aarch64/aarch64.c: Add attrs.h header file.
2461 * config/alpha/alpha.c: Likewise.
2462 * config/arc/arc.c: Likewise.
2463 * config/arm/arm.c: Likewise.
2464 * config/avr/avr.c: Likewise.
2465 * config/bfin/bfin.c: Likewise.
2466 * config/c6x/c6x.c: Likewise.
2467 * config/cr16/cr16.c: Likewise.
2468 * config/cris/cris.c: Likewise.
2469 * config/darwin.c: Likewise.
2470 * config/epiphany/epiphany.c: Likewise.
2471 * config/fr30/fr30.c: Likewise.
2472 * config/frv/frv.c: Likewise.
2473 * config/ft32/ft32.c: Likewise.
2474 * config/h8300/h8300.c: Likewise.
2475 * config/i386/winnt.c: Likewise.
2476 * config/ia64/ia64.c: Likewise.
2477 * config/iq2000/iq2000.c: Likewise.
2478 * config/lm32/lm32.c: Likewise.
2479 * config/m32c/m32c.c: Likewise.
2480 * config/m32r/m32r.c: Likewise.
2481 * config/m68k/m68k.c: Likewise.
2482 * config/mcore/mcore.c: Likewise.
2483 * config/microblaze/microblaze.c: Likewise.
2484 * config/mips/mips.c: Likewise.
2485 * config/mmix/mmix.c: Likewise.
2486 * config/mn10300/mn10300.c: Likewise.
2487 * config/moxie/moxie.c: Likewise.
2488 * config/msp430/msp430.c: Likewise.
2489 * config/nds32/nds32-isr.c: Likewise.
2490 * config/nds32/nds32.c: Likewise.
2491 * config/nios2/nios2.c: Likewise.
2492 * config/nvptx/nvptx.c: Likewise.
2493 * config/pa/pa.c: Likewise.
2494 * config/pdp11/pdp11.c: Likewise.
2495 * config/powerpcspe/powerpcspe.c: Likewise.
2496 * config/riscv/riscv.c: Likewise.
2497 * config/rl78/rl78.c: Likewise.
2498 * config/rx/rx.c: Likewise.
2499 * config/s390/s390.c: Likewise.
2500 * config/sh/sh.c: Likewise.
2501 * config/sol2.c: Likewise.
2502 * config/sparc/sparc.c: Likewise.
2503 * config/spu/spu.c: Likewise.
2504 * config/stormy16/stormy16.c: Likewise.
2505 * config/tilegx/tilegx.c: Likewise.
2506 * config/tilepro/tilepro.c: Likewise.
2507 * config/v850/v850.c: Likewise.
2508 * config/vax/vax.c: Likewise.
2509 * config/visium/visium.c: Likewise.
2510 * config/xtensa/xtensa.c: Likewise.
2512 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2515 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
2516 constraints since the -mupper-regs-* switches have been
2518 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
2519 into a vector from a double word element that was extracted from
2520 another vector, and eliminate extra XXPERMDI instructions.
2521 (vsx_concat_<mode>_2): Likewise.
2522 (vsx_concat_<mode>_3): Likewise.
2523 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
2524 concat to allow optimizing inserts from previous extracts.
2526 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
2528 * config/i386/i386.c (ix86_stack_protect_guard): Generate
2529 memory reference to a SSP offset in TLS address space.
2530 (ix86_print_operand) <case '@'>: Remove.
2531 (ix86_print_operand_punct_valid_p): Remove '@' code.
2532 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
2534 (stack_tls_protect_set_<mode>): Remove.
2535 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
2536 (stack_tls_protect_test_<mode>): Remove.
2537 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
2539 2017-08-07 Olivier Hainque <hainque@adacore.com>
2542 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
2544 2017-08-07 Douglas Rupp <rupp@adacore.com>
2546 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
2547 variable was referenced as multidir in command.
2549 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2552 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
2555 2017-08-07 Martin Liska <mliska@suse.cz>
2557 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
2558 * config/rl78/rl78.c: Add include of attribs.h.
2559 * config/sh/sh.c: Likewise.
2560 * config/v850/v850.c: Likewise.
2562 2017-08-07 Tom de Vries <tom@codesourcery.com>
2565 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
2567 2017-08-07 Martin Liska <mliska@suse.cz>
2569 * config/mips/mips.c: Include attribs.h.
2571 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
2574 * doc/invoke.texi: Document change in behvaior for -Ofast for
2577 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
2579 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
2581 (aarch64_pop_regs): Likewise.
2582 (aarch64_gen_load_pair): Likewise.
2583 (aarch64_save_callee_saves): Likewise.
2584 (aarch64_restore_callee_saves): Likewise.
2586 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2588 * config/i386/i386.c: Revert the last change.
2590 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2593 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2595 (ix86_finalize_stack_frame_flags): This. Also clear
2596 frame_pointer_needed if -fno-omit-frame-pointer is used without
2598 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2599 with ix86_finalize_stack_frame_flags.
2600 (ix86_expand_epilogue): Likewise.
2601 (ix86_expand_split_stack_prologue): Likewise.
2603 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2606 * config/i386/i386.c (get_builtin_code_for_version): Set priority
2607 to P_AES for Westmere.
2609 2017-08-07 Jonathan Yong <10walls@gmail.com>
2611 * config/i386/mingw.opt (fset-stack-executable): Removed.
2612 * config/i386/cygming.opt (fset-stack-executable): Moved
2614 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
2616 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
2618 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
2620 2017-08-07 Marek Polacek <polacek@redhat.com>
2623 * fold-const.c (fold_indirect_ref_1): Check type_domain.
2625 2017-08-07 Martin Liska <mliska@suse.cz>
2627 * attribs.h (canonicalize_attr_name): New function.
2628 (cmp_attribs): Move from c-format.c and adjusted.
2629 (is_attribute_p): Moved from tree.h.
2630 * tree-inline.c: Add new includes.
2631 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
2632 (private_is_attribute_p): Remove.
2633 (private_lookup_attribute): Likewise.
2634 (private_lookup_attribute_by_prefix): Simplify.
2635 (remove_attribute): Use is_attribute_p.
2636 * tree.h: Remove removed declarations.
2638 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2641 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
2642 instead of computing it in the function. Formatting fix.
2643 (expand_case): Don't rely on default_edge being the first edge,
2644 clear it if removing it, pass default_edge to
2645 emit_case_dispatch_table.
2646 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
2649 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
2651 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
2652 insn in the function, emit NOP after the insn.
2654 2017-08-06 Tom de Vries <tom@codesourcery.com>
2656 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
2659 2017-08-06 Tom de Vries <tom@codesourcery.com>
2661 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
2664 2017-08-04 Yury Gribov <tetra2005@gmail.com>
2666 PR tree-optimization/57371
2667 * match.pd: New pattern.
2669 2017-08-04 Marek Polacek <polacek@redhat.com>
2672 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
2673 perform the computation in offset_int.
2675 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2677 PR tree-optimization/81136
2678 * tree-vectorizer.h: Include tree-hash-traits.h.
2679 (vec_base_alignments): New typedef.
2680 (vec_info): Add a base_alignments field.
2681 (vect_record_base_alignments): Declare.
2682 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
2684 (DR_IS_CONDITIONAL_IN_STMT): New macro.
2685 (create_data_ref): Add an is_conditional_in_stmt argument.
2686 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
2687 the is_conditional_in_stmt field.
2688 (data_ref_loc): Add an is_conditional_in_stmt field.
2689 (get_references_in_stmt): Set the is_conditional_in_stmt field.
2690 (find_data_references_in_stmt): Update call to create_data_ref.
2691 (graphite_find_data_references_in_stmt): Likewise.
2692 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
2693 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
2694 (vect_record_base_alignment): New function.
2695 (vect_record_base_alignments): Likewise.
2696 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
2697 for nested statements even if we fail to compute a misalignment.
2698 Use pooled base alignments for unconditional references.
2699 (vect_find_same_alignment_drs): Compare base addresses instead
2701 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
2702 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
2704 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2706 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
2707 Add an explicit name for the enum. Use auto_vec for slp_instances
2709 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
2711 (_bb_vec_info): Add a constructor and destructor.
2712 (vinfo_for_stmt): Return NULL for uids of -1 as well.
2713 (destroy_loop_vec_info): Delete.
2714 (vect_destroy_datarefs): Likewise.
2715 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
2716 (vec_info::vec_info): New function.
2717 (vec_info::~vec_info): Likewise.
2718 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
2719 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
2720 destroy_loop_vec_info.
2721 * tree-vect-loop.c (new_loop_vec_info): Replace with...
2722 (_loop_vec_info::_loop_vec_info): ...this.
2723 (destroy_loop_vec_info): Replace with...
2724 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
2725 the stmt_vec_infos. Leave handling of vec_info information to its
2726 destructor. Remove explicit vector releases.
2727 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
2728 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
2729 * tree-vect-slp.c (new_bb_vec_info): Replace with...
2730 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
2731 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
2732 (destroy_bb_vec_info): Replace with...
2733 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
2734 information to its destructor.
2735 (vect_slp_analyze_bb_1): Use new and delete instead of
2736 new_bb_vec_info and destroy_bb_vec_info.
2737 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
2740 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2742 * tree-data-ref.h (subscript): Add access_fn field.
2743 (data_dependence_relation): Add could_be_independent_p.
2744 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
2745 (same_access_functions): Move to tree-data-ref.c.
2746 * tree-data-ref.c (ref_contains_union_access_p): New function.
2747 (access_fn_component_p): Likewise.
2748 (access_fn_components_comparable_p): Likewise.
2749 (dr_analyze_indices): Add a reference to access_fn_component_p.
2750 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
2752 (constant_access_functions): Likewise.
2753 (add_other_self_distances): Likewise.
2754 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
2755 (initialize_data_dependence_relation): Use XCNEW and remove
2756 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
2757 of access functions that have the same type. Allow the
2758 subsequence to end with different bases in some circumstances.
2759 Record the chosen access functions in SUB_ACCESS_FN.
2760 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
2761 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
2762 (subscript_dependence_tester_1): Likewise dra and drb.
2763 (build_classic_dist_vector): Update calls accordingly.
2764 (subscript_dependence_tester): Likewise.
2765 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
2766 DDR_COULD_BE_INDEPENDENT_P.
2767 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
2768 comp_alias_ddrs instead of may_alias_ddrs.
2769 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
2771 (vect_analyze_data_ref_dependence): Use it if
2772 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
2773 distance vectors if that fails.
2774 (dependence_distance_ge_vf): New function.
2775 (vect_prune_runtime_alias_test_list): Use it. Don't clear
2776 LOOP_VINFO_MAY_ALIAS_DDRS.
2778 2017-08-04 Richard Biener <rguenther@suse.de>
2781 * fold-const.c (fold_binary_loc): Properly restrict
2782 minus_var0 && minus_var1 case when associating undefined overflow
2785 2017-08-04 Tom de Vries <tom@codesourcery.com>
2787 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
2789 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2791 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2792 Don't start diagnostic messages with a capital letter.
2793 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2795 (rs6000_invalid_builtin): Likewise.
2796 (rs6000_trampoline_init): Likewise.
2798 2017-08-03 Jakub Jelinek <jakub@redhat.com>
2801 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
2802 after setting changeable df flags.
2804 2017-08-03 Richard Biener <rguenther@suse.de>
2806 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
2807 up if the use is in USE - X.
2809 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
2811 * toplev.c (dumpfile.h): New include.
2812 (internal_error_reentered): New static function. Use it...
2813 (internal_error_function): ...here to handle reentered internal_error.
2815 2017-08-03 Richard Biener <rguenther@suse.de>
2818 * fold-const.c (split_tree): Add minus_var and minus_con
2819 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
2820 here but always use minus_*.
2821 (associate_trees): Assert we never associate with MINUS_EXPR
2822 and NULL first operand. Do not recurse for PLUS_EXPR operands
2823 when associating as MINUS_EXPR either.
2824 (fold_binary_loc): Track minus_var and minus_con.
2826 2017-08-03 Tom de Vries <tom@codesourcery.com>
2829 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
2830 ACCEL_COMPILER, apply finish_options on
2831 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
2833 2017-08-03 Tom de Vries <tom@codesourcery.com>
2836 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
2837 function_entry_patch_area_size > 0.
2839 2017-08-03 Jakub Jelinek <jakub@redhat.com>
2842 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
2843 instead of 10??LU, perform unit multiplication in wide_int,
2844 don't change alloc_object_size_limit if the limit is larger
2847 PR tree-optimization/81655
2848 PR tree-optimization/81588
2849 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
2850 the case when ranges[i].low and high are 1 for unsigned type with
2854 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
2855 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
2857 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2859 * tree-vrp.h: Add include guard.
2861 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
2864 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
2865 (ud2): New insn pattern.
2866 * config/i386/i386.c (ix86_expand_epilogue):
2867 For naked functions, generate ud2 instead of trap insn.
2869 2017-08-02 Marek Polacek <polacek@redhat.com>
2872 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
2874 2017-08-02 Tom de Vries <tom@codesourcery.com>
2875 Cesar Philippidis <cesar@codesourcery.com>
2877 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
2878 Add missing edge probabilities.
2880 2017-08-02 Tamar Christina <tamar.christina@arm.com>
2882 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
2885 2017-08-02 Jakub Jelinek <jakub@redhat.com>
2888 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
2889 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
2890 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
2891 prologue_seq sequences - if any.
2893 2017-08-02 Richard Biener <rguenther@suse.de>
2895 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
2896 via vectors if supported, integer extracts via punning if supported
2897 or otherwise vector extracts.
2899 2017-08-02 Richard Biener <rguenther@suse.de>
2901 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
2903 (bitmap_insert_into_set): ... this.
2905 2017-08-02 Richard Biener <rguenther@suse.de>
2907 PR tree-optimization/81633
2909 2015-08-17 Alan Hayward <alan.hayward@arm.com>
2911 PR tree-optimization/71752
2912 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
2914 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
2916 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
2917 (machine_function::call_ms2sysv_pad_out): Remove field.
2918 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
2919 (ix86_compute_frame_layout): Likewise.
2921 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
2924 * config/i386/i386.c (ix86_set_func_type): Disallow naked
2925 attribute with interrupt attribute.
2927 2017-08-01 Andrew Pinski <apinski@cavium.com>
2929 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
2930 BIT_INSERT_EXPR's operand 1
2931 to see if the types precision matches.
2933 2017-08-01 Martin Liska <mliska@suse.cz>
2936 * builtins.c (expand_builtin_memcpy_args): Remove.
2937 (expand_builtin_memcpy): Call newly added function
2938 expand_builtin_memory_copy_args.
2939 (expand_builtin_memcpy_with_bounds): Likewise.
2940 (expand_builtin_mempcpy): Remove last argument.
2941 (expand_builtin_mempcpy_with_bounds): Likewise.
2942 (expand_builtin_memory_copy_args): New function created from
2943 expand_builtin_mempcpy_args with small modifications.
2944 (expand_builtin_mempcpy_args): Remove.
2945 (expand_builtin_stpcpy): Remove unused argument.
2946 (expand_builtin): Likewise.
2947 (expand_builtin_with_bounds): Likewise.
2949 2017-08-01 Martin Liska <mliska@suse.cz>
2952 Make mempcpy more optimal (PR middle-end/70140).
2954 2017-08-01 Jakub Jelinek <jakub@redhat.com>
2957 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2958 __builtin_vec_cmpne verify both arguments are compatible vectors
2959 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
2960 verify arg1_type is a pointer or array type. For __builtin_vec_st,
2961 move computation of aligned to after checking the argument types.
2965 * config/rs6000/vsx.md (vextract_fp_from_shorth,
2966 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
2969 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
2970 Doug Rupp <rupp@adacore.com>
2971 Olivier Hainque <hainque@adacore.com>
2973 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
2974 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
2976 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
2977 for TARGET_OS_CPP_BUILTIN.
2978 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
2979 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
2981 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
2982 passing required abi options to the assembler for EABI configurations.
2983 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
2984 of .text.hot and .text.unlikely sections for kernel modules when
2985 using ARM style exceptions.
2986 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
2987 options. Add EXTRA_CC1_SPEC.
2988 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
2990 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
2992 (ARM_TARGET2_DWARF_FORMAT): Define.
2993 * config/arm/t-vxworks: Adjust multilib control to removal of the
2994 Diab command line options.
2996 2017-08-01 Martin Liska <mliska@suse.cz>
2998 PR gcov-profile/81561
2999 * gcov.c (unblock): Make unblocking safe as we need to preserve
3000 index correspondence of blocks and block_lists.
3002 2017-08-01 Richard Biener <rguenther@suse.de>
3004 PR tree-optimization/81181
3005 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3006 (compute_antic): ... end of iteration here.
3008 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3010 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3011 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3012 (ftree-slp-vectorize): Likewise.
3013 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3014 can no longer be set independent of flag_tree_loop_vectorize.
3015 * omp-general.c (emp_max_vf): Likewise.
3016 * opts.c (enable_fdo_optimizations): Remove references to
3017 flag_tree_vectorize, these are now implicit.
3018 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3019 and leave it for the options machinery.
3021 2017-08-01 Martin Liska <mliska@suse.cz>
3024 * builtins.c (expand_builtin_memcpy_args): Remove.
3025 (expand_builtin_memcpy): Call newly added function
3026 expand_builtin_memory_copy_args.
3027 (expand_builtin_memcpy_with_bounds): Likewise.
3028 (expand_builtin_mempcpy): Remove last argument.
3029 (expand_builtin_mempcpy_with_bounds): Likewise.
3030 (expand_builtin_memory_copy_args): New function created from
3031 expand_builtin_mempcpy_args with small modifications.
3032 (expand_builtin_mempcpy_args): Remove.
3033 (expand_builtin_stpcpy): Remove unused argument.
3034 (expand_builtin): Likewise.
3035 (expand_builtin_with_bounds): Likewise.
3037 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3040 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3041 print "ds:" only for immediates in generic address space.
3043 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3046 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3047 (ix86_function_ok_for_sibcall): Return false for naked functions.
3049 2017-08-01 Richard Biener <rguenther@suse.de>
3051 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3052 (compute_antic): Seed worklist with exit block predecessors.
3053 * cfganal.c (dfs_find_deadend): For a cycle return the source
3054 of the edge closing it.
3056 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3058 * config/aarch64/aarch64.c
3059 (aarch64_can_const_movi_rtx_p): Move 0 check.
3061 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3063 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3064 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3066 * match.pd: Ditto in address comparison pattern.
3068 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3070 PR tree-optimization/81627
3071 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3072 closed ssa form for store-store chain.
3074 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3076 PR tree-optimization/81620
3077 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3078 for store-store chain.
3080 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3082 PR tree-optimization/81588
3083 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3084 ranges[i].in_p, invert comparison code ccode. For >/>=,
3085 swap rhs1 and rhs2 and comparison code unconditionally,
3086 for </<= don't do that. Don't swap rhs1/rhs2 again if
3087 ranges[i].in_p, instead invert comparison code ccode if
3088 opcode or oe->rank is BIT_IOR_EXPR.
3091 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3092 a direct optab to conversion optab.
3093 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3094 with GET_MODE_INNER as last argument instead of optab_handler.
3095 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3096 vector extraction if possible and optab is available.
3097 * expr.c (store_constructor): Use convert_optab_handler instead
3098 of optab_handler. Use vector initialization from smaller
3099 vectors if possible and optab is available.
3100 * tree-vect-stmts.c (vectorizable_load): Likewise.
3101 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3103 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3104 of vec_init from half-sized vectors with the same element mode.
3105 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3106 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3107 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3108 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3109 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3110 after mode in gen_vec_extract* calls.
3111 (vec_extract<mode>): Renamed to ...
3112 (vec_extract<mode><ssescalarmodelower>): ... this.
3113 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3114 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3115 element mode after mode in gen_vec_init* calls.
3116 (VEC_INIT_HALF_MODE): New mode iterator.
3117 (vec_init<mode>): Renamed to ...
3118 (vec_init<mode><ssescalarmodelower>): ... this.
3119 (vec_init<mode><ssehalfvecmodelower>): New expander.
3120 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3121 (vec_extractv2sfsf): ... this.
3122 (vec_initv2sf): Renamed to ...
3123 (vec_initv2sfsf): ... this.
3124 (vec_extractv2si): Renamed to ...
3125 (vec_extractv2sisi): ... this.
3126 (vec_initv2si): Renamed to ...
3127 (vec_initv2sisi): ... this.
3128 (vec_extractv4hi): Renamed to ...
3129 (vec_extractv4hihi): ... this.
3130 (vec_initv4hi): Renamed to ...
3131 (vec_initv4hihi): ... this.
3132 (vec_extractv8qi): Renamed to ...
3133 (vec_extractv8qiqi): ... this.
3134 (vec_initv8qi): Renamed to ...
3135 (vec_initv8qiqi): ... this.
3136 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3137 (vec_init<mode>): Renamed to ...
3138 (vec_init<mode><VEC_base_l>): ... this.
3139 (vec_extract<mode>): Renamed to ...
3140 (vec_extract<mode><VEC_base_l>): ... this.
3141 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3142 (vec_initv2sfsf): ... this.
3143 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3144 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3145 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3146 element mode after mode in gen_vec_init* calls.
3147 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3148 (vec_init<mode><Vel>): ... this.
3149 (vec_extract<mode>): Renamed to ...
3150 (vec_extract<mode><Vel>): ... this.
3151 * config/aarch64/iterators.md (Vel): New mode attribute.
3152 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3153 Add element mode after mode in gen_vec_extract* calls.
3154 * config/s390/vector.md (non_vec_l): New mode attribute.
3155 (vec_extract<mode>): Renamed to ...
3156 (vec_extract<mode><non_vec_l>): ... this.
3157 (vec_init<mode>): Renamed to ...
3158 (vec_init<mode><non_vec_l>): ... this.
3159 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3160 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3162 * config/arm/iterators.md (V_elem_l): New mode attribute.
3163 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3164 (vec_extract<mode><V_elem_l>): ... this.
3165 (vec_extractv2di): Renamed to ...
3166 (vec_extractv2didi): ... this.
3167 (vec_init<mode>): Renamed to ...
3168 (vec_init<mode><V_elem_l>): ... this.
3169 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3170 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3171 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3172 Add element mode after gen_vec_extract* calls.
3173 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3174 (vec_init<mode><unitmode>): ... this.
3175 (vec_extract<mode>): Renamed to ...
3176 (vec_extract<mode><unitmode>): ... this.
3177 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3178 (vec_init<mode><unitmode>): ... this.
3179 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3180 (vec_initv2sfsf): ... this.
3181 (vec_extractv2sf): Renamed to ...
3182 (vec_extractv2sfsf): ... this.
3183 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3184 Add element mode after gen_vec_extract* calls.
3185 * config/mips/mips.md (unitmode): New mode iterator.
3186 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3187 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3188 * config/spu/spu.md (inner_l): New mode attribute.
3189 (vec_init<mode>): Renamed to ...
3190 (vec_init<mode><inner_l>): ... this.
3191 (vec_extract<mode>): Renamed to ...
3192 (vec_extract<mode><inner_l>): ... this.
3193 * config/sparc/sparc.md (veltmode): New mode iterator.
3194 (vec_init<VMALL:mode>): Renamed to ...
3195 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3196 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3197 (vec_initv2sisi): ... this.
3198 (vec_initv2sf): Renamed to ...
3199 (vec_initv2sfsf): ... this.
3200 (vec_extractv2sf): Renamed to ...
3201 (vec_extractv2sfsf): ... this.
3202 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3203 (vec_init<mode>): Renamed to ...
3204 (vec_init<mode><VEC_base_l>): ... this.
3205 (vec_extract<mode>): Renamed to ...
3206 (vec_extract<mode><VEC_base_l>): ... this.
3207 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3208 (vec_initv2sfsf): ... this.
3209 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3210 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3211 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3212 gen_vec_init* calls.
3214 2017-08-01 Richard Biener <rguenther@suse.de>
3216 PR tree-optimization/81297
3217 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3218 TREE_OVERFLOW from INTEGER_CSTs.
3220 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3222 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3224 2017-07-31 Carl Love <cel@us.ibm.com>
3226 * config/rs6000/rs6000-c: Add support for built-in functions
3227 vector signed char vec_xl_be (signed long long, signed char *);
3228 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3229 vector signed int vec_xl_be (signed long long, signed int *);
3230 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3231 vector signed long long vec_xl_be (signed long long, signed long long *);
3232 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3233 vector signed short vec_xl_be (signed long long, signed short *);
3234 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3235 vector double vec_xl_be (signed long long, double *);
3236 vector float vec_xl_be (signed long long, float *);
3237 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3238 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3239 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3241 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3242 (altivec_expand_builtin): Add switch statement to call
3243 altivec_expand_xl_be for each builtin.
3244 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3245 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3246 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3247 __builtin_vsx_le_be_v16qi.
3248 * doc/extend.texi: Update the built-in documentation file for the
3249 new built-in functions.
3251 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3254 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3256 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3258 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3260 * config.gcc: Add z14.
3261 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3262 CPU model numbers for z13s and z14.
3263 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3265 * config/s390/s390-opts.h (enum processor_type): Rename
3266 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3267 * config/s390/s390.c (processor_table): Add field for CPU name to
3268 be passed to Binutils.
3269 (s390_asm_output_machine_for_arch): Use the new field in
3270 processor_table for Binutils.
3271 (s390_expand_builtin): Replace arch12 with z14.
3272 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3273 (s390_get_sched_attrmask): Likewise.
3274 (s390_get_unit_mask): Likewise.
3275 * config/s390/s390.opt: Add z14 to processor_type enum.
3277 2017-07-31 Martin Jambor <mjambor@suse.cz>
3280 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3281 regardless of optimization level.
3283 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3284 Martin Liska <mliska@suse.cz>
3286 * predict.def: Remove old comment and adjust probability.
3287 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3290 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3293 * config/i386/i386.c (ix86_function_naked): New function.
3294 (ix86_can_use_return_insn_p): Return false for naked functions.
3295 (ix86_expand_prologue): Skip prologue for naked functions.
3296 (ix86_expand_epilogue): Skip epilogue for naked functions
3297 and emit trap instruction.
3298 (ix86_warn_func_return): New function.
3299 (ix86_attribute_table): Add "naked" attribute specification.
3300 (TARGET_WARN_FUNC_RETURN): Define.
3301 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3303 2017-07-31 Martin Liska <mliska@suse.cz>
3305 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3306 (dump_gimple_bb_header): Always dump BB info.
3307 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3309 2017-07-31 Martin Liska <mliska@suse.cz>
3312 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3313 also with current_function_decl non-null equality.
3315 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3318 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3319 change type to the element type, instead add eltype variable and
3320 use it where we are interested in the element type.
3322 PR tree-optimization/81603
3323 * ipa-polymorphic-call.c
3324 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3325 offset arithmetic in offset_int, bail out if the resulting bit offset
3326 doesn't fit into shwi.
3328 2017-07-31 Martin Liska <mliska@suse.cz>
3330 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3331 (gimplify_save_expr): Fix comment.
3333 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3336 * config/i386/i386.c (ix86_function_arg): Update arguments for
3338 (ix86_compute_frame_layout): Set the initial stack offset to
3339 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3340 INCOMING_FRAME_SP_OFFSET.
3341 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3342 stack before exception handler returns.
3343 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3344 the 'ERROR_CODE' for exception handler.
3346 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3348 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3349 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3350 (ASM_OUTPUT_REG_POP): Ditto.
3351 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3352 instead of asm_fprintf to output pure string.
3354 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3356 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3357 to imported_module_or_decl hook.
3358 (debug_nothing_tree_tree_tree_bool): Remove.
3359 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3360 * debug.c (do_nothing_debug_hooks): Use
3361 debug_nothing_tree_tree_tree_bool_bool instead of
3362 debug_nothing_tree_tree_tree_bool.
3363 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3364 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3365 * sdbout.c (sdb_debug_hooks): Likewise.
3366 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3367 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3369 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3370 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3371 attribute, don't add anything.
3373 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3375 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3376 (fold_build2_stat_loc): Likewise.
3377 (fold_build3_stat_loc): Likewise.
3378 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3379 (fold_build1_loc): Remove macro.
3380 (fold_build2_loc): Likewise.
3381 (fold_build3_loc): Likewise.
3383 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3385 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3386 (gimple_build_debug_bind_source_stat): Likewise.
3387 * gimple.h (gimple_build_debug_bind): Remove macro.
3388 (gimple_build_debug_bind_source): Likewise.
3390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3392 * bitmap.c (bitmap_alloc): Adjust.
3393 (bitmap_gc_alloc): Likewise.
3394 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3398 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3399 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3400 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
3401 (bitmap_gc_alloc_stat): Likewise.
3402 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
3404 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3406 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
3407 * rtl.h (shallow_copy_rtx): Remove macro.
3409 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3411 * emit-rtl.c (gen_raw_REG): Adjust.
3412 * gengenrtl.c (gendef): Likewise.
3413 * rtl.c (rtx_alloc_stat): Remove _stat from name.
3414 * rtl.h (rtx_alloc): Remove macro.
3416 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3418 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
3419 (build_tree_list_stat): Likewise.
3420 * tree.h (build_tree_list): Remove macro.
3421 (build_tree_list_vec): Likewise.
3423 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3425 * tree.c (make_vector_stat): Remove _stat from name.
3426 (build_vector_stat): Likewise.
3427 * tree.h (make_vector_stat): Remove macro.
3428 (build_vector_stat): Likewise.
3430 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3432 * tree.h (build_var_debug_value): Remove prototype.
3434 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3436 * tree.c (tree_cons_stat): Remove _stat from name.
3437 * tree.h (tree_cons): Remove macro.
3439 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3441 * tree.c (build_vl_exp_stat): Remove _stat from name.
3442 * tree.h (build_vl_exp): Remove macro.
3444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3446 * tree.c (build_decl_stat): Remove _stat from name.
3447 * tree.h (build_decl): Remove macro.
3449 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3451 * gimple.c (gimple_build_with_ops_stat): Adjust.
3452 (gimple_alloc_stat): Remove _stat from name.
3453 * gimple.h (gimple_alloc): Remove macro.
3455 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3457 * tree.c (make_tree_vec_stat): Remove _stat from name.
3458 (grow_tree_vec_stat): Likewise.
3459 * tree.h (make_tree_vec_stat): Adjust prototype.
3460 (grow_tree_vec_stat): Likewise.
3461 (make_tree_vec): Remove macro.
3462 (grow_tree_vec): Likewise.
3464 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3466 * fold-const.c (fold_build1_stat_loc): Adjust.
3467 (fold_build2_stat_loc): Likewise.
3468 (fold_build3_stat_loc): Likewise.
3469 * tree.c (build0_stat): Remove _stat from name.
3470 (build1_stat): Likewise.
3471 (build2_stat): Likewise.
3472 (build3_stat): Likewise.
3473 (build4_stat): Likewise.
3474 (build5_stat): Likewise.
3475 * tree.h (build1_loc): Remove macro, and rename _stat function
3477 (build2_loc): Likewise.
3478 (build3_loc): Likewise.
3479 (build4_loc): Likewise.
3480 (build5_loc): Likewise.
3482 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3484 * tree.c (make_int_cst_stat): Remove _stat from name.
3485 * tree.h (make_int_cst_stat): Adjust prototype.
3486 (make_int_cst): Remove macro.
3488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3490 * tree.c (make_tre_binfo_stat): Remove _stat from name.
3491 * tree.h (make_tree_binfo_stat): Adjust prototype.
3492 (make_tree_binfo): Remove.
3494 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3496 * tree.c (copy_node_stat): Rename to copy_node.
3497 (build_distinct_type_copy): Adjust.
3498 * tree.h (copy_node_stat): Adjust prototype.
3499 (copy_node): Remove macro.
3501 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3503 * tree.c (make_node_stat): rename to make_node.
3504 (build_tree_list_stat): Adjust.
3505 (build0_stat): Likewise.
3506 (build2_stat): Likewise.
3507 (build3_stat): Likewise.
3508 (build4_stat): Likewise.
3509 (build5_stat): Likewise.
3510 (build_decl_stat): Likewise.
3511 * tree.h (make_node_stat): Adjust prototype.
3512 (make_node): remove macro.
3514 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
3516 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
3517 (PPC_FEATURE2_SCV): Likewise.
3518 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
3520 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3522 * config/aarch64/aarch64.c
3523 (aarch64_internal_mov_immediate): Add new special pattern.
3524 * config/aarch64/aarch64.md (*movdi_aarch64):
3525 Add reg/32bit const mov case.
3527 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3528 Richard Sandiford <richard.sandiford@linaro.org>
3530 * config/aarch64/aarch64.md (mov<mode>): Generalize.
3531 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3532 Add integer and movi cases.
3533 (movi-split-hf-df-sf split, fp16): New.
3534 (enabled): Added TARGET_FP_F16INST.
3535 * config/aarch64/iterators.md (GPF_HF): New.
3536 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
3538 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3540 * config/aarch64/aarch64.c
3541 (aarch64_simd_container_mode): Add prototype.
3542 (aarch64_expand_mov_immediate): Add HI support.
3543 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
3544 (aarch64_can_const_movi_rtx_p): New.
3545 (aarch64_preferred_reload_class):
3546 Remove restrictions of using FP registers for certain SIMD operations.
3547 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
3548 (aarch64_valid_floating_const): Add integer move validation.
3549 (aarch64_simd_imm_scalar_p): Remove.
3550 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
3551 (aarch64_legitimate_constant_p): Expand list of supported cases.
3552 * config/aarch64/aarch64-protos.h
3553 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
3554 (aarch64_reinterpret_float_as_int): New.
3555 (aarch64_simd_imm_scalar_p): Remove.
3556 * config/aarch64/constraints.md (Uvi): New.
3557 (Dd): Split into Ds and new Dd.
3558 * config/aarch64/aarch64.md (*movsi_aarch64):
3560 (*movdi_aarch64): Add SIMD mov case.
3562 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3564 * tree-predcom.c: (struct chain): Handle store-store chain in which
3565 stores for elimination only store loop invariant values.
3566 (execute_pred_commoning_chain): Ditto.
3567 (prepare_initializers_chain_store_elim): Ditto.
3568 (prepare_finalizers): Ditto.
3569 (is_inv_store_elimination_chain): New function.
3570 (initialize_root_vars_store_elim_1): New function.
3572 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3574 * tree-predcom.c: Revise general description of the pass.
3575 (enum chain_type): New enum type for store elimination.
3576 (struct chain): New field supporting store elimination.
3577 (struct component): Ditto.
3578 (dump_chain): Dump store-stores chain.
3579 (release_chain): Release resources.
3580 (split_data_refs_to_components): Compute and create component
3581 contains only stores for elimination.
3582 (get_chain_last_ref_at): New function.
3583 (make_invariant_chain): Initialization.
3584 (make_rooted_chain): Specify chain type in parameter and record it.
3585 (add_looparound_copies): Skip for store-stores chain.
3586 (determine_roots_comp): Compute type of chain and pass it to
3588 (initialize_root_vars_store_elim_2): New function.
3589 (finalize_eliminated_stores): New function.
3590 (remove_stmt): Handle store for elimination.
3591 (execute_pred_commoning_chain): Execute predictive commoning on
3593 (determine_unroll_factor): Skip unroll for store-stores chain.
3594 (prepare_initializers_chain_store_elim): New function.
3595 (prepare_initializers_chain): Hanlde store-store chain.
3596 (prepare_finalizers_chain, prepare_finalizers): New function.
3597 (tree_predictive_commoning_loop): Return integer value indicating
3598 if loop is unrolled or lcssa form is corrupted.
3599 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
3601 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3603 * tree-predcom.c (initialize_root): Delete.
3604 (execute_pred_commoning_chain): Initialize root vars and replace
3605 reference of non-combined chain directly, rather than call above
3608 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3610 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
3611 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
3613 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3615 * tree-predcom.c (struct chain): New field init_seq.
3616 (release_chain): Release init_seq.
3617 (prepare_initializers_chain): Record intialization stmts in above
3619 (insert_init_seqs): New function.
3620 (tree_predictive_commoning_loop): Call insert_init_seqs.
3622 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3624 * tree-predcom.c (determine_roots_comp): Skip trivial components.
3626 2017-07-28 Richard Biener <rguenther@suse.de>
3628 * match.pd: Remove superfluous :c.
3629 * genmatch.c (simplify::id): Add member.
3630 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
3632 (current_id): New global.
3633 (dt_node::parent): Move from ...
3634 (dt_operand::parent): ... here. Add for_id member.
3635 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
3636 (decision_tree::find_node): Relax order requirement when
3637 merging DT_TRUE nodes to ones inbetween the current simplify
3638 and the one we try to merge with. Add diagnostic whenever
3639 we need to enforce pattern order by not merging.
3640 (decision_tree::insert): Set current_id.
3641 (decision_tree::print_node): Dump parent node and for_id.
3642 (parser::last_id): Add member.
3643 (parser::push_simplify): Assign unique id.
3644 (parser::parser): Initialize last_id.
3646 2017-07-28 Martin Liska <mliska@suse.cz>
3649 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
3650 gimple_build_debug_bind.
3652 2017-07-28 Richard Biener <rguenther@suse.de>
3654 PR tree-optimization/81502
3655 * match.pd: Add pattern combining BIT_INSERT_EXPR with
3657 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
3659 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
3660 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
3661 for BIT_FIELD_REF args.
3662 * fold-const.c (make_bit_field_ref): Likewise.
3663 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3665 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3668 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
3669 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
3670 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
3671 Or it into SANITIZER_UNDEFINED.
3672 * ubsan.c: Include gimple-fold.h and varasm.h.
3673 (ubsan_expand_ptr_ifn): New function.
3674 (instrument_pointer_overflow): New function.
3675 (maybe_instrument_pointer_overflow): New function.
3676 (instrument_object_size): Formatting fix.
3677 (pass_ubsan::execute): Call instrument_pointer_overflow
3678 and maybe_instrument_pointer_overflow.
3679 * internal-fn.c (expand_UBSAN_PTR): New function.
3680 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
3681 * sanitizer.def (__ubsan_handle_pointer_overflow,
3682 __ubsan_handle_pointer_overflow_abort): New builtins.
3683 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
3684 * internal-fn.def (UBSAN_PTR): New internal function.
3685 * opts.c (sanitizer_opts): Add pointer-overflow.
3686 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
3687 * fold-const.c (build_range_check): Compute pointer range check in
3688 integral type if pointer arithmetics would be needed. Formatting
3691 2017-07-28 Martin Liska <mliska@suse.cz>
3694 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
3695 parameters that are of a variable-length.
3697 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
3699 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
3701 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
3702 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
3703 (CRT_CALL_STATIC_FUNCTION): Likewise.
3704 (ASM_DEFAULT_SPEC): New define.
3705 (ASM_SPEC32): Likewise.
3706 (ASM_SPEC64): Likewise.
3707 (ASM_SPEC_COMMON): Likewise.
3708 (ASM_SPEC): Likewise.
3709 (INVALID_64BIT): Likewise.
3710 (LINK_OS_DEFAULT_SPEC): Likewise.
3711 (LINK_OS_SPEC32): Likewise.
3712 (LINK_OS_SPEC64): Likewise.
3713 (POWERPC_LINUX): Likewise.
3714 (PTRDIFF_TYPE): Likewise.
3715 (RESTORE_FP_PREFIX): Likewise.
3716 (RESTORE_FP_SUFFIX): Likewise.
3717 (SAVE_FP_PREFIX): Likewise.
3718 (SAVE_FP_SUFFIX): Likewise.
3719 (SIZE_TYPE): Likewise.
3720 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
3721 (TARGET_64BIT): Likewise.
3722 (TARGET_64BIT): Likewise.
3723 (TARGET_AIX): Likewise.
3724 (WCHAR_TYPE_SIZE): Likewise.
3725 (WCHAR_TYPE): Undefine.
3726 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
3727 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
3728 (CPP_OS_RTEMS_SPEC): Delete.
3729 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
3730 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
3732 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
3734 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3736 PR tree-optimization/81578
3737 * tree-parloops.c (build_new_reduction): Bail out if
3738 reduction_code isn't one of the standard OpenMP reductions.
3739 Move the details printing after that decision.
3741 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3743 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
3744 related to reload_in_progress.
3745 (splat_input_operand): Likewise.
3746 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
3748 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
3750 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
3751 (TARGET_INSTANTIATE_DECLS): Likewise.
3752 (legitimate_indexed_address_p): Delete reload_in_progress code.
3753 (rs6000_debug_legitimate_address_p): Likewise.
3754 (rs6000_eliminate_indexed_memrefs): Likewise.
3755 (rs6000_emit_le_vsx_store): Likewise.
3756 (rs6000_emit_move_si_sf_subreg): Likewise.
3757 (rs6000_emit_move): Likewise.
3758 (register_to_reg_type): Likewise.
3759 (rs6000_pre_atomic_barrier): Likewise.
3760 (rs6000_machopic_legitimize_pic_address): Likewise.
3761 (rs6000_allocate_stack_temp): Likewise.
3762 (rs6000_address_for_fpconvert): Likewise.
3763 (rs6000_address_for_altivec): Likewise.
3764 (rs6000_secondary_memory_needed_rtx): Delete function.
3765 (rs6000_check_sdmode): Likewise.
3766 (rs6000_alloc_sdmode_stack_slot): Likewise.
3767 (rs6000_instantiate_decls): Likewise.
3768 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
3769 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
3770 Delete reload_in_progress.
3771 (*vec_reload_and_plus_<mptrsize>): Likewise.
3772 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
3773 (vsx_div_v2di): Likewise.
3774 (vsx_udiv_v2di): Likewise.
3776 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3778 * config/rs6000/rs6000.opt (mlra): Replace with stub.
3779 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
3780 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
3781 (rs6000_debug_reg_global): Delete print of LRA status.
3782 (rs6000_option_override_internal): Delete dead LRA related code.
3783 (rs6000_lra_p): Delete function.
3784 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
3786 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
3788 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
3789 * config/riscv/rtems.h: New file.
3791 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3792 Sudakshina Das <sudi.das@arm.com>
3794 * config/aarch64/aarch64.md
3795 (define_split for and<mode>3nr_compare): Move
3796 non aarch64_logical_operand to a register.
3797 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
3798 register immediate operand to a register.
3799 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
3801 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3804 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
3806 2017-07-27 Richard Biener <rguenther@suse.de>
3808 PR tree-optimization/81573
3809 PR tree-optimization/81494
3810 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
3811 multi defuse cycle case.
3813 2017-07-27 Richard Biener <rguenther@suse.de>
3815 PR tree-optimization/81571
3816 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
3819 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
3821 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
3822 earlier and only if MASK_FPU is set. Adjust formatting.
3824 2017-07-27 Martin Liska <mliska@suse.cz>
3826 * opt-functions.awk: Add validation of value of Init.
3827 * optc-gen.awk: Pass new argument.
3829 2017-07-27 Martin Liska <mliska@suse.cz>
3831 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
3832 Fix wrong condition.
3834 2017-07-27 Martin Liska <mliska@suse.cz>
3836 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
3837 BBs and edges seen by autoFDO.
3839 2017-07-27 Richard Biener <rguenther@suse.de>
3841 PR tree-optimization/81502
3842 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
3843 with incompatible but same sized type.
3844 (execute_update_addresses_taken): Likewise.
3846 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
3848 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
3849 flag_tree_loop_vectorize rather than flag_tree_vectorize.
3851 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3854 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
3855 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
3856 Change s_operand to memory_operand.
3858 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
3860 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
3861 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
3862 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
3863 Emit instructions rather than returning an expression. Handle TFmode
3864 and KFmode by casting to TImode.
3865 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
3866 (rs6000_emit_le_vsx_store): Likewise.
3867 * config/rs6000/vsx.md (VSX_TI): New iterator.
3868 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
3869 (*vsx_le_undo_permute_<mode>): Likewise.
3870 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
3871 emit the split sequence.
3872 (*vsx_le_perm_store_<mode>): Likewise.
3874 2017-07-27 Jakub Jelinek <jakub@redhat.com>
3876 PR tree-optimization/81555
3877 PR tree-optimization/81556
3878 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
3879 if true, force CHANGED for the recursive invocation.
3880 (reassociate_bb): Remember original length of ops array, pass
3881 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
3883 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
3884 attributes for noipa attribute. For naked attribute use
3885 lookup_attribute first before lookup_attribute_spec.
3886 * final.c (rest_of_handle_final): Disable IPA RA for functions with
3888 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
3889 for functions with noipa attribute.
3890 (cgraph_externally_visible_p): Return true for functions with noipa
3892 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
3893 for functions with noipa attribute.
3894 * doc/extend.texi: Document noipa function attribute.
3895 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
3896 also for functions with noipa attribute.
3897 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
3899 2017-07-26 Andrew Pinski <apinski@cavium.com>
3901 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
3902 vec_unalign_load_cost and vec_unalign_store_cost.
3904 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3906 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
3907 -mvsx-small-integer option.
3908 (ISA_3_0_MASKS_IEEE): Likewise.
3909 (OTHER_VSX_VECTOR_MASKS): Likewise.
3910 (POWERPC_MASKS): Likewise.
3911 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
3912 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
3913 code, only testing for DImode being allowed in non-VSX floating
3915 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
3916 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
3918 (rs6000_option_override_internal): Delete -mvsx-small-integer.
3919 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
3920 TARGET_P8_VECTOR test.
3921 (rs6000_secondary_reload_simple_move): Likewise.
3922 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
3923 since TARGET_P9_VECTOR was already tested.
3924 (rs6000_opt_masks): Remove -mvsx-small-integer.
3925 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
3926 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
3928 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
3929 test for TARGET_VEXTRACTUB was used, and that uses
3931 (p9 extract splitter): Likewise.
3932 (vsx_extract_<mode>_di_p9): Likewise.
3933 (vsx_extract_<mode>_store_p9): Likewise.
3934 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
3935 for TARGET_P9_VECTOR was used. Delete code that is now dead with
3936 the elimination of TARGET_VSX_SMALL_INTEGER.
3937 (vsx_extract_<mode>_p8): Likewise.
3938 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
3939 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
3940 (vsx_set_<mode>_p9): Likewise.
3941 (vsx_set_v4sf_p9): Likewise.
3942 (vsx_set_v4sf_p9_zero): Likewise.
3943 (vsx_insert_extract_v4sf_p9): Likewise.
3944 (vsx_insert_extract_v4sf_p9_2): Likewise.
3945 * config/rs6000/rs6000.md (sign extend splitter): Change
3946 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
3947 (floatsi<mode>2_lfiwax_mem): Likewise.
3948 (floatunssi<mode>2_lfiwzx_mem): Likewise.
3949 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
3950 since a test for TARGET_P9_VECTOR was used.
3951 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
3952 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
3953 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
3954 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
3955 TARGET_P8_VECTOR test.
3956 (fix_trunc<mode>si2_stfiwx): Likewise.
3957 (fix_trunc<mode>si2_internal): Likewise.
3958 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
3959 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
3961 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
3962 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
3963 TARGET_P8_VECTOR test.
3964 (fixuns_trunc<mode>si2_stfiwx): Likewise.
3965 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
3966 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
3968 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
3969 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
3970 since a test for TARGET_P9_VECTOR was used.
3971 (splitter for loading small constants): Likewise.
3973 2017-07-26 Andrew Pinski <apinski@cavium.com>
3975 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
3978 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
3981 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
3982 (fp_valid_at): Likewise.
3984 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
3986 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
3987 (qdf24xx_addrcost_table): Likewise.
3988 (cortexa57_tunings): Update to use generic_branch_cost.
3989 (cortexa72_tunings): Likewise.
3990 (cortexa73_tunings): Likewise.
3991 (qdf24xx_tunings): Likewise.
3993 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
3995 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
3996 (thunderx2t99_branch_cost): Likewise.
3997 (cortexa35_tunings): Update to use generic_branch_cost.
3998 (cortexa53_tunings): Likewise.
3999 (cortexa57_tunings): Likewise.
4000 (cortexa72_tunings): Likewise.
4001 (cortexa73_tunings): Likewise.
4002 (thunderx2t99_tunings): Likewise.
4004 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4006 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4007 (sparc_option_override): Honour MASK_FSMULD.
4008 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4009 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4010 * config/sparc/sparc.opt (mfsmuld): New option.
4011 * doc/invoke.texi (mfsmuld): Document option.
4013 2017-07-26 Marek Polacek <polacek@redhat.com>
4016 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4018 2017-07-26 Richard Biener <rguenther@suse.de>
4020 * gimple-match-head.c (do_valueize): Return OP if valueize
4022 (get_def): New helper to get at the def stmt of a SSA name
4024 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4025 do_valueize to get at the def stmt.
4026 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4028 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4031 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4033 2017-07-26 Martin Liska <mliska@suse.cz>
4036 * function.c (expand_function_start): Make expansion of
4037 nonlocal_goto_save_area after parm_birth_insn.
4039 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4041 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4042 from all CPU target flags enable members.
4044 2017-07-26 Richard Biener <rguenther@suse.de>
4046 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4047 (decision_tree::gen): Make 'type' const.
4048 (write_predicate): Likewise.
4050 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4052 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4053 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4054 (rs6000_option_override_internal): Likewise.
4055 (rs6000_expand_vector_set): Likewise.
4056 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4057 (TARGET_UPPER_REGS_SF): Likewise.
4058 (TARGET_UPPER_REGS_DI): Likewise.
4059 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4060 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4061 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4062 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4063 (Splitters for DI constants in Altivec registers): Likewise.
4064 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4065 (vsx_set_v4sf_p9): Likewise.
4066 (vsx_set_v4sf_p9_zero): Likewise.
4067 (vsx_insert_extract_v4sf_p9): Likewise.
4068 (vsx_insert_extract_v4sf_p9_2): Likewise.
4070 2017-07-25 Carl Love <cel@us.ibm.com>
4072 * doc/extend.texi: Update the built-in documentation file for the
4073 existing built-in functions
4074 vector signed char vec_cnttz (vector signed char);
4075 vector unsigned char vec_cnttz (vector unsigned char);
4076 vector signed short vec_cnttz (vector signed short);
4077 vector unsigned short vec_cnttz (vector unsigned short);
4078 vector signed int vec_cnttz (vector signed int);
4079 vector unsigned int vec_cnttz (vector unsigned int);
4080 vector signed long long vec_cnttz (vector signed long long);
4081 vector unsigned long long vec_cnttz (vector unsigned long long);
4083 2017-07-25 Andrew Pinski <apinski@cavium.com>
4085 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4086 accesses where the use is for the first operand of a BIT_INSERT.
4088 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4091 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4092 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4094 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4096 * config/i386/gstabs.h: Delete.
4097 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4099 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4101 * config/i386/i386.c (ix86_decompose_address): Do not check for
4102 register RTX when looking at index_reg or base_reg.
4103 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4105 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4107 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4108 to update EH info here.
4110 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4112 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4114 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4116 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4118 2017-07-25 Torsten Duwe <duwe@suse.de>
4120 * common.opt: Introduce -fpatchable-function-entry
4121 command line option, and its variables function_entry_patch_area_size
4122 and function_entry_patch_area_start.
4123 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4124 including a two-value parser.
4125 * target.def (print_patchable_function_entry): New target hook.
4126 * targhooks.h (default_print_patchable_function_entry): New function.
4127 * targhooks.c (default_print_patchable_function_entry): Likewise.
4128 * toplev.c (process_options): Switch off IPA-RA if
4129 patchable function entries are being generated.
4130 * varasm.c (assemble_start_function): Look at the
4131 patchable-function-entry command line switch and current
4132 function attributes and maybe generate NOP instructions by
4133 calling the print_patchable_function_entry hook.
4134 * doc/extend.texi: Document patchable_function_entry attribute.
4135 * doc/invoke.texi: Document -fpatchable_function_entry
4136 command line option.
4137 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4139 * doc/tm.texi: Re-generate.
4141 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4144 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4145 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4147 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4149 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4151 2017-07-25 Richard Biener <rguenther@suse.de>
4153 PR tree-optimization/81455
4154 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4155 not walk in cycles when looking for guards.
4157 2017-07-25 Richard Biener <rguenther@suse.de>
4159 PR tree-optimization/81529
4160 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4161 when optimizing backedge uses.
4163 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4165 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4167 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4168 to dl_section_ref. On AIX, append an expression to subtract
4169 the size of the section length to dl_section_ref.
4171 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4173 * configure.ac: If any of the config.* scripts fail, exit 1.
4174 * configure: Regenerate.
4176 2017-07-25 Richard Biener <rguenther@suse.de>
4179 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4180 of immediate uses, be more verbose on errors.
4182 2017-07-25 Richard Biener <rguenther@suse.de>
4184 PR tree-optimization/81510
4185 * tree-vect-loop.c (vect_is_simple_reduction): When the
4186 reduction stmt is not inside the loop bail out.
4188 2017-07-25 Richard Biener <rguenther@suse.de>
4190 PR tree-optimization/81303
4191 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4192 profitability check against LOOP_VINFO_NITERSM1.
4194 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4196 * domwalk.c (cmp_bb_postorder): Simplify.
4197 (sort_bbs_postorder): New function. Use it...
4198 (dom_walker::walk): ...here to optimize common cases.
4200 2017-07-25 Martin Liska <mliska@suse.cz>
4203 * ipa-visibility.c (function_and_variable_visibility): Make the
4204 redirection just on target that supports aliasing.
4205 Fix GNU coding style.
4207 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4210 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4212 * config/arm/rtems.h: Likewise.
4213 * config/bfin/rtems.h: Likewise.
4214 * config/i386/rtemself.h: Likewise.
4215 * config/lm32/rtems.h: Likewise.
4216 * config/m32c/rtems.h: Likewise.
4217 * config/m68k/rtemself.h: Likewise.
4218 * config/microblaze/rtems.h: Likewise.
4219 * config/mips/rtems.h: Likewise.
4220 * config/moxie/rtems.h: Likewise.
4221 * config/nios2/rtems.h: Likewise.
4222 * config/powerpcspe/rtems.h: Likewise.
4223 * config/rs6000/rtems.h: Likewise.
4224 * config/rtems.h: Likewise.
4225 * config/sh/rtems.h: Likewise.
4226 * config/sh/rtemself.h: Likewise.
4227 * config/sparc/rtemself.h: Likewise.
4229 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4232 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4233 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4234 * tree-ssa-structalias.c (alias_get_name): Same.
4236 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4239 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4240 instructions if no du chain is found.
4242 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4244 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4246 2017-07-25 Richard Biener <rguenther@suse.de>
4249 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4252 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4254 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4256 (ISA_2_7_MASKS_SERVER): Likewise.
4257 (ISA_3_0_MASKS_IEEE): Likewise.
4258 (OTHER_P8_VECTOR_MASKS): Likewise.
4259 (OTHER_VSX_VECTOR_MASKS): Likewise.
4260 (POWERPC_MASKS): Likewise.
4261 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4262 duplicate list of options.
4263 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4264 explicit -mupper-regs options.
4265 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4266 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4267 alias for -mupper-regs-df.
4268 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4269 (rs6000_init_hard_regno_mode_ok): Likewise.
4270 (rs6000_option_override_internal): Likewise.
4271 (rs6000_opt_masks): Likewise.
4272 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4273 options in terms of whether -mvsx or -mpower8-vector was used.
4274 (TARGET_UPPER_REGS_DI): Likewise.
4275 (TARGET_UPPER_REGS_SF): Likewise.
4276 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4277 -mupper-regs-* options.
4279 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4281 * passes.c (emergency_dump_function): Print some empty lines and a
4282 header before the RTL dump.
4284 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4286 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4288 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4291 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4292 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4294 2017-07-24 Carl Love <cel@us.ibm.com>
4296 * config/rs6000/rs6000-c.c: Add support for built-in functions
4297 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4298 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4299 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4300 vec_extract_fp_from_shortl): Add defines for the two builtins.
4301 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4302 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4304 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4305 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4306 * doc/extend.texi: Update the built-in documentation file for the
4307 new built-in function.
4309 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4312 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4314 * doc/generic.texi: Likewise.
4315 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4316 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4318 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4320 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4321 (aarch64_mls_elt_merge<mode>): Likewise.
4323 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4325 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4326 having __cxa_atexit.
4328 2017-07-23 Michael Collison <michael.collison@arm.com>
4330 * config/arm/arm.c (arm_option_override): Deprecate
4331 use of -mstructure-size-boundary.
4332 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4333 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4335 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4338 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4339 Reduce cost estimate for direct moves.
4341 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4344 * config/i386/i386.c (ix86_option_override_internal): Disable
4345 BMI, BMI2 and TBM instructions for -m16.
4347 2017-07-21 Carl Love <cel@us.ibm.com>
4349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4350 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4351 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4352 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4353 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4354 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4355 VMULOSW): New enum "unspec" values.
4356 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4357 altivec_vmulosw): New patterns.
4358 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4359 VMULOSW): Add definitions.
4361 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4363 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4364 (qdf24xx): Likewise.
4365 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4366 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4367 (AARCH64_FL_V8_1): Renumber.
4368 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4369 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4370 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4371 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4372 rdma to feature modifiers list.
4374 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4377 * ipa-visibility (function_and_variable_visibility): Convert
4378 recursive PLT call to direct call if appropriate.
4380 2017-07-21 Andrew Pinski <apinski@cavium.com>
4382 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4383 operand 1 to see if the types precision matches.
4384 * fold-const.c (operand_equal_p): Likewise.
4386 2017-07-21 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/81303
4389 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4390 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4391 (vect_peeling_hash_get_lowest_cost): Adjust.
4392 (vect_enhance_data_refs_alignment): Likewise. Use
4393 vect_get_peeling_costs_all_drs to compute the penalty for no
4394 peeling to match up costs.
4396 2017-07-21 Richard Biener <rguenther@suse.de>
4398 PR tree-optimization/81500
4399 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4400 we didn't identify a reduction path.
4402 2017-07-21 Tom de Vries <tom@codesourcery.com>
4403 Cesar Philippidis <cesar@codesourcery.com>
4405 PR gcov-profile/81442
4406 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
4409 2017-07-21 Tom de Vries <tom@codesourcery.com>
4412 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
4414 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
4415 nvptx_override_options_after_change.
4417 2017-07-21 Ulrich Drepper <drepper@redhat.com>
4419 * dwarf2out.c (output_file_names): Avoid double testing for
4422 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
4424 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
4426 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4428 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
4430 (try_crossjump_to_edge): Do not punt on partitioned functions.
4432 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4434 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4435 Put all BBs reachable only via paths crossing cold region to cold
4437 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
4439 2016-07-21 Richard Biener <rguenther@suse.de>
4441 PR tree-optimization/81303
4442 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
4443 into account prologue and epilogue iterations when raising
4444 min_profitable_iters to sth at least covering one vector iteration.
4446 2017-07-21 Tamar Christina <tamar.christina@arm.com>
4448 * config/arm/arm.c (arm_test_cpu_arch_dat):
4451 2017-07-20 Nathan Sidwell <nathan@acm.org>
4453 Remove TYPE_METHODS.
4454 * tree.h (TYPE_METHODS): Delete.
4455 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
4456 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
4457 (dbxout_type_methods): Scan TYPE_FIELDS.
4458 (dbxout_type): Don't check TYPE_METHODS here.
4459 * function.c (use_register_for_decl): Always ignore register for
4460 class types when not optimizing.
4461 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
4462 * tree.c (free_lang_data_in_type): Stitch out member functions and
4463 templates from TYPE_FIELDS.
4464 (build_distinct_type_copy, verify_type_variant,
4465 verify_type): Member fns are on TYPE_FIELDS.
4466 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
4467 * tree-pretty-print.c (dump_generic_node): Likewise.
4469 2017-07-20 Jakub Jelinek <jakub@redhat.com>
4472 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
4473 V2TImode and V4TImode.
4474 (ix86_expand_vector_extract): Likewise.
4475 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
4476 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
4477 (ssescalarmode): Handle V4TImode and V2TImode.
4478 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
4479 (*vec_extractv2ti, *vec_extractv4ti): New insns.
4480 (VEXTRACTI128_MODE): New mode iterator.
4481 (splitter for *vec_extractv?ti first element): New.
4482 (VEC_INIT_MODE): New mode iterator.
4483 (vec_init<mode>): Consolidate 3 expanders into one using
4484 VEC_INIT_MODE mode iterator.
4486 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4488 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
4489 non_spilled_static_chain_regno_p.
4491 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4493 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
4495 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
4497 * bb-reorder.c (connect_traces): Allow copying of blocks within
4500 2017-07-20 Richard Biener <rguenther@suse.de>
4502 * gimple.h (gimple_phi_result): Add gphi * overload.
4503 (gimple_phi_result_ptr): Likewise.
4504 (gimple_phi_arg): Likewise. Adjust index assert to only
4505 allow actual argument accesses rather than all slots available
4507 (gimple_phi_arg_def): Add gphi * overload.
4508 * tree-phinodes.c (make_phi_node): Initialize only actual
4510 (resize_phi_node): Clear memory not covered by old node,
4511 do not initialize excess argument slots.
4512 (reserve_phi_args_for_new_edge): Initialize new argument slot
4515 2017-07-20 Bin Cheng <bin.cheng@arm.com>
4517 PR tree-optimization/81388
4519 2016-07-21 Bin Cheng <bin.cheng@arm.com>
4521 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
4522 by removing computation of may_be_zero.
4524 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4525 Tom de Vries <tom@codesourcery.com>
4528 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
4529 when gimple level profile disagrees with what RTL expander did.
4531 2017-07-20 Richard Biener <rguenther@suse.de>
4533 PR tree-optimization/61171
4534 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
4535 (vect_analyze_stmt): Add slp instance parameter.
4536 (vectorizable_reduction): Likewise.
4537 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
4538 (vect_is_simple_reduction): Deal with chains not detected
4539 as SLP reduction chain, specifically not properly associated
4540 chains containing a mix of plus/minus.
4541 (get_reduction_op): Remove.
4542 (get_initial_defs_for_reduction): Simplify, pass in whether
4543 this is a reduction chain, pass in the SLP node for the PHIs.
4544 (vect_create_epilog_for_reduction): Get the SLP instance as
4546 (vectorizable_reduction): Get the SLP instance as arg.
4547 During analysis remember the SLP node with the PHIs in the
4548 instance. Simplify getting at the vectorized reduction PHIs.
4549 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
4550 through SLP instance.
4551 (vect_slp_analyze_operations): Likewise.
4552 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
4553 (vect_transform_stmt): Likewise.
4555 2017-07-20 Tom de Vries <tom@codesourcery.com>
4557 PR tree-optimization/81489
4558 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
4559 read of phi arg location to before loop that modifies phi.
4561 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4563 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
4566 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4569 * except.c (execute): Fix ordering issue.
4571 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4573 PR rtl-optimization/81423
4574 * combine.c (make_compound_operation_int): Don't try to optimize
4575 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
4577 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4579 PR rtl-optimization/81423
4580 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
4581 with a constant that is -1 in the truncated to mode.
4583 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4585 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
4586 (determine_unlikely_bbs): ... here.
4587 * predict.h (propagate_unlikely_bbs_forward): Declare.
4588 * cfgexpand.c (pass_expand::execute): Use it.
4589 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
4591 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
4592 propagate_unlikely_bbs_forward.
4594 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4597 * except.c (maybe_add_nop_after_section_switch): New function.
4600 2017-07-19 Tom de Vries <tom@codesourcery.com>
4602 * gimple.h (gimple_phi_set_arg): Make assert more strict.
4604 2017-07-19 Tom de Vries <tom@codesourcery.com>
4606 * gimple.h (gimple_phi_arg): Make assert more strict.
4608 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
4610 * config.gcc (powerpc*-*-*): Add mmintrin.h.
4611 * config/rs6000/mmintrin.h: New file.
4612 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
4614 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4616 PR tree-optimization/81346
4617 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
4619 2017-07-19 Tom de Vries <tom@codesourcery.com>
4621 * config/nvptx/nvptx.md (VECIM): Add V2DI.
4623 2017-07-19 Tom de Vries <tom@codesourcery.com>
4625 * config/nvptx/nvptx-modes.def: Add V2DImode.
4626 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
4627 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
4628 (nvptx_output_mov_insn): Handle lack of mov.b128.
4629 (nvptx_print_operand): Handle 'H' and 'L' codes.
4630 (nvptx_vector_mode_supported): Allow V2DImode.
4631 (nvptx_preferred_simd_mode): New function.
4632 (nvptx_data_alignment): New function.
4633 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
4634 nvptx_preferred_simd_mode.
4635 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
4637 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
4639 2017-07-19 Tom de Vries <tom@codesourcery.com>
4641 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
4642 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
4643 (nvptx_vector_mode_supported): New function. Allow V2SImode.
4644 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
4645 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
4646 (mov<VECIM>_insn): New define_insn.
4647 (define_expand "mov<VECIM>): New define_expand.
4649 2017-07-19 Tom de Vries <tom@codesourcery.com>
4651 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
4653 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4655 PR tree-optimization/81346
4656 * fold-const.h (fold_div_compare, range_check_type): Declare.
4657 * fold-const.c (range_check_type): New function.
4658 (build_range_check): Use range_check_type.
4659 (fold_div_compare): No longer static, rewritten into
4660 a match.pd helper function.
4661 (fold_comparison): Don't call fold_div_compare here.
4662 * match.pd (X / C1 op C2): New optimization using fold_div_compare
4665 2017-07-19 Nathan Sidwell <nathan@acm.org>
4667 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
4668 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
4669 * tree.c (find_decls_types_r, verify_type): Use
4670 TYPE_{MIN,MAX}_VALUE_RAW.
4671 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
4672 (hash_tree): Likewise.
4673 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4675 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4678 2017-07-18 Tom de Vries <tom@codesourcery.com>
4681 * omp-expand.c (expand_omp_for_static_chunk): Handle
4682 equal-argument loop exit phi.
4684 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
4687 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
4688 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
4689 operand 2 predicate.
4690 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
4691 operand 2 predicate.
4692 (ror,rol -> rorx splitters): Use const_int_operand as
4693 operand 2 predicate.
4695 2017-06-18 Richard Biener <rguenther@suse.de>
4697 PR tree-optimization/81410
4698 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
4699 the gap in the ! slp_perm SLP case after each group.
4701 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4704 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
4707 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4710 * predict.c (set_even_probabilities): Cleanup; do not affect
4711 probabilities that are already known.
4712 (combine_predictions_for_bb): Call even when count is set.
4714 2017-07-18 Nathan Sidwell <nathan@acm.org>
4716 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
4719 2017-07-18 Bin Cheng <bin.cheng@arm.com>
4722 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
4723 optimization for loop niter analysis.
4725 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
4728 * config/avr/avr.c (avr_optimize_casesi): Don't use
4729 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
4731 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4733 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
4734 body_cost_vec from _vect_peel_extended_info.
4735 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
4736 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
4739 2017-07-18 Bin Cheng <bin.cheng@arm.com>
4741 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
4743 2017-07-18 Richard Biener <rguenther@suse.de>
4745 PR tree-optimization/80620
4746 PR tree-optimization/81403
4747 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
4748 info when re-using a VN table entry.
4750 2017-07-18 Richard Biener <rguenther@suse.de>
4752 PR tree-optimization/81418
4753 * tree-vect-loop.c (vectorizable_reduction): Properly compute
4754 vectype_in. Verify that with lane-reducing reduction operations
4755 we have a single def-use cycle.
4757 2017-07-17 Carl Love <cel@us.ibm.com>
4759 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
4761 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4762 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4763 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4764 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4765 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4766 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4767 VMULOSW): New enum "unspec" values.
4768 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4769 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4770 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4771 altivec_vmulosw): New patterns.
4772 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4773 VMULOSW): Add definitions.
4774 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
4776 * config/alpha/alpha.c: Include predict.h.
4778 2017-07-17 Yury Gribov <tetra2005@gmail.com>
4780 * tree-vrp.c (compare_assert_loc): Fix comparison function
4781 to return predictable results.
4783 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4785 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
4788 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
4789 * doc/invoke.texi (mexpand-adddi): Update text.
4791 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4793 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
4794 that also clobbers the CC register. The old expand code is moved
4796 (*arc_clzsi2): ... here.
4797 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
4798 the CC register. The old expand code is moved to ...
4799 (arc_ctzsi2): ... here.
4801 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4803 * config/arc/arc.opt (mindexed-loads): Use initial value
4804 TARGET_INDEXED_LOADS_DEFAULT.
4805 (mauto-modify-reg): Use initial value
4806 TARGET_AUTO_MODIFY_REG_DEFAULT.
4807 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
4808 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
4809 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
4810 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
4812 2017-07-17 Martin Liska <mliska@suse.cz>
4815 * opts.c (finish_options): Do not allow -fgnu-tm
4816 w/ -fsanitize={kernel-,}address. Say sorry.
4818 2017-07-17 Bin Cheng <bin.cheng@arm.com>
4821 * tree-loop-distribution.c (classify_partition): Only assert on
4822 numer of iterations.
4823 (merge_dep_scc_partitions): Delete prameter. Update function call.
4824 (distribute_loop): Remove code handling loop with unknown niters.
4825 (pass_loop_distribution::execute): Skip loop with unknown niters.
4827 2017-07-17 Bin Cheng <bin.cheng@arm.com>
4830 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
4831 function sort_partitions_by_post_order.
4833 2017-07-17 Bin Cheng <bin.cheng@arm.com>
4835 PR tree-optimization/81374
4836 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
4837 the max index of basic blocks, rather than number of basic blocks.
4839 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4841 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
4843 (arc_legitimate_pic_operand_p): Likewise.
4844 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
4846 (arc_needs_pcl_p): Likewise.
4847 (arc_legitimate_pc_offset_p): Likewise.
4848 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
4849 function is also used in constrains.md.
4850 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
4851 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
4852 PLUS. Only return true/false in known cases, otherwise assert.
4853 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
4854 is already called in arc_legitimate_constant_p.
4855 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
4857 (LEGITIMATE_PIC_OPERAND_P): Use
4858 arc_raw_symbolic_reference_mentioned_p function.
4859 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
4864 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4865 Andrew Burgess <andrew.burgess@embecosm.com>
4867 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
4868 (arc_return_address_register): New function.
4869 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
4870 (arc_handle_fndecl_attribute): Add naked attribute.
4871 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4872 (TARGET_WARN_FUNC_RETURN): Likewise.
4873 (arc_allocate_stack_slots_for_args): New function.
4874 (arc_warn_func_return): Likewise.
4875 (machine_function): Change type fn_type.
4876 (arc_compute_function_type): Consider new naked function type,
4877 change function return type.
4878 (arc_must_save_register): Adapt to handle new
4879 arc_compute_function_type's return type.
4880 (arc_expand_prologue): Likewise.
4881 (arc_expand_epilogue): Likewise.
4882 (arc_return_address_regs): Delete.
4883 (arc_return_address_register): New function.
4884 (arc_epilogue_uses): Use above function.
4885 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
4886 (arc_function_type): Change encoding, add naked type.
4887 (ARC_INTERRUPT_P): Change to handle the new encoding.
4888 (ARC_FAST_INTERRUPT_P): Likewise.
4889 (ARC_NORMAL_P): Define.
4890 (ARC_NAKED_P): Likewise.
4891 (arc_compute_function_type): Delete prototype.
4892 * config/arc/arc.md (in_ret_delay_slot): Use
4893 arc_return_address_register function.
4894 (simple_return): Likewise.
4895 (p_return_i): Likewise.
4897 2017-07-17 Jakub Jelinek <jakub@redhat.com>
4899 PR tree-optimization/81428
4900 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
4901 can't be built for those types.
4903 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
4905 Remove stuff dead since r239246.
4907 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
4908 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
4909 (avr_inform_devices): Remove dead stuff.
4911 2017-07-17 Tamar Christina <tamar.christina@arm.com>
4913 * config/arm/arm_neon.h: Fix softp typo.
4915 2017-07-17 Jakub Jelinek <jakub@redhat.com>
4917 PR tree-optimization/81365
4918 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
4919 aggregate moves onto bb predecessor edges, make sure there are no
4920 loads that could alias the lhs in between the start of bb and the
4923 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
4926 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
4927 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
4928 [LSHIFTRT, outer_code = TRUNCATE]: Same.
4930 2017-07-17 Jakub Jelinek <jakub@redhat.com>
4932 PR tree-optimization/81396
4933 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
4934 (init_symbolic_number): Initialize it to 1.
4935 (perform_symbolic_merge): Add n_ops from both operands into the new
4937 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
4938 without base_addr as useless if they need more than one operation.
4939 (bswap_replace): Handle !bswap case for NULL base_addr.
4941 2017-07-17 Tom de Vries <tom@codesourcery.com>
4944 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
4947 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
4949 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
4950 conditional builtin define __FIX_LEON3FT_B2BST.
4952 2017-07-17 Daniel Cederman <cederman@gaisler.com>
4954 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
4955 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
4958 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
4960 PR rtl-optimization/81424
4961 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
4962 to remove potential trapping from operands if -fnon-call-exceptions.
4964 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4966 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
4967 profile_proability for scalling.
4968 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
4970 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4972 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
4974 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4976 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
4977 fixpoint arithmetics.
4979 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4981 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
4982 fixpoint arithmetics.
4984 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4986 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
4987 fixpoint arithmetics.
4989 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4991 * profile-count.h (profile_probability::from_reg_br_prob_note,
4992 profile_probability::to_reg_br_prob_note): New functions.
4993 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
4994 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
4995 * predict.c (probability_reliable_p): Update.
4996 (edge_probability_reliable_p): Update.
4997 (br_prob_note_reliable_p): Update.
4998 (invert_br_probabilities): Update.
4999 (add_reg_br_prob_note): New function.
5000 (combine_predictions_for_insn): Update.
5001 * asan.c (asan_clear_shadow): Update.
5002 * cfgbuild.c (compute_outgoing_frequencies): Update.
5003 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5004 (update_br_prob_note): Update.
5005 (rtl_verify_edges): Update.
5006 (purge_dead_edges): Update.
5007 (fixup_reorder_chain): Update.
5008 * emit-rtl.c (try_split): Update.
5009 * ifcvt.c (cond_exec_process_insns): Update.
5010 (cond_exec_process_if_block): Update.
5011 (dead_or_predicable): Update.
5012 * internal-fn.c (expand_addsub_overflow): Update.
5013 (expand_neg_overflow): Update.
5014 (expand_mul_overflow): Update.
5015 * loop-doloop.c (doloop_modify): Update.
5016 * loop-unroll.c (compare_and_jump_seq): Update.
5017 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5018 * predict.h: Update.
5019 * reorg.c (mostly_true_jump): Update.
5021 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5022 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5023 * config/arc/arc.c: (emit_unlikely_jump): Update.
5024 * config/arm/arm.c: (emit_unlikely_jump): Update.
5025 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5026 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5027 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5028 (ix86_print_operand): Update.
5029 (ix86_split_fp_branch): Update.
5030 (predict_jump): Update.
5031 * config/ia64/ia64.c (ia64_print_operand): Update.
5032 * config/mmix/mmix.c (mmix_print_operand): Update.
5033 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5034 (rs6000_expand_split_stack_prologue): Update.
5035 * config/rs6000/rs6000.c: Update.
5036 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5037 (s390_expand_vec_movstr): Update.
5038 (s390_expand_cs_tdsi): Update.
5039 (s390_expand_split_stack_prologue): Update.
5040 * config/sh/sh.c (sh_print_operand): Update.
5041 (expand_cbranchsi4): Update.
5042 (expand_cbranchdi4): Update.
5043 * config/sparc/sparc.c (output_v9branch): Update.
5044 * config/spu/spu.c (get_branch_target): Update.
5045 (ea_load_store_inline): Update.
5046 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5047 * config/tilepro/tilepro.c: Update.
5049 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5051 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5052 (gimplify_save_expr): Likewise.
5054 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5056 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5058 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5060 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5062 * timevar.def (TV_IPA_FNSUMMARY): Define.
5064 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5066 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5067 to back store errata sensitive sequence from being generated.
5068 (sqrtdf2_fix): Likewise.
5070 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5072 * tree-ssa-threadupdate.c (compute_path_counts,
5073 update_joiner_offpath_counts): Use profile_probability.
5075 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5078 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5080 * config/arm/arm-c.c (arm_cpu_builtins): Define
5081 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5083 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5085 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5086 array entries to represent __ieee128 versions of the
5087 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5088 scalar_extract_sig, and scalar_insert_exp built-in functions.
5089 (altivec_resolve_overloaded_builtin): Add special case handling
5090 for the __builtin_scalar_insert_exp function, as represented by
5091 the P9V_BUILTIN_VEC_VSIEDP constant.
5092 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5093 exponent support for __ieee128 argument.
5094 (VSESQP): Add scalar extract signature support for __ieee128
5096 (VSTDCNQP): Add scalar test negative support for __ieee128
5098 (VSIEQP): Add scalar insert exponent support for __int128 argument
5099 with __ieee128 result.
5100 (VSIEQPF): Add scalar insert exponent support for __ieee128
5101 argument with __ieee128 result.
5102 (VSTDCQP): Add scalar test data class support for __ieee128
5104 (VSTDCNQP): Add overload support for scalar test negative with
5106 (VSTDCQP): Add overload support for scalar test data class
5108 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5110 (UNSPEC_VSX_SIEXPQP): New constant.
5111 (xsxexpqp): New insn for VSX scalar extract exponent quad
5113 (xsxsigqp): New insn for VSX scalar extract significand quad
5115 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5116 precision with floating point argument.
5117 (xststdcqp): New expand for VSX scalar test data class quad
5119 (xststdcnegqp): New expand for VSX scalar test negative quad
5121 (xststdcqp): New insn to match expansions for VSX scalar test data
5122 class quad precision and VSX scalar test negative quad precision.
5123 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5124 special case operand checking to enforce that second operand of
5125 VSX scalar test data class with quad precision argument is a 7-bit
5127 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5128 prototypes and descriptions of __ieee128 versions of
5129 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5130 scalar_test_data_class, and scalar_test_neg built-in functions.
5132 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5134 PR tree-optimization/81162
5135 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5136 replace a negate with an add.
5138 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5140 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5142 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5144 * config/arm/arm-c.c (arm_cpu_builtins): Define
5145 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5147 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5149 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5150 (armv8-r): Set ARM Cortex-R52 as default CPU.
5151 * config/arm/arm-tables.opt: Regenerate.
5152 * config/arm/arm-tune.md: Regenerate.
5153 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5155 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5156 extension for -mcpu=cortex-r52.
5158 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5160 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5161 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5162 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5163 (fp-armv8): Define it as FP_ARMv8 only.
5164 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5165 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5167 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5168 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5169 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5170 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5171 than TARGET_FPU_ARMV8.
5172 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5173 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5174 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5176 * config/arm/neon.md (neon_vrint): Likewise.
5177 (neon_vcvt): Likewise.
5178 (neon_<fmaxmin_op><mode>): Likewise.
5179 (<fmaxmin><mode>3): Likewise.
5180 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5181 * config/arm/predicates.md (arm_cond_move_operator): Check against
5182 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5184 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5186 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5189 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5191 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5192 loop in comment with memset.
5194 2017-07-14 Martin Liska <mliska@suse.cz>
5196 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5197 * dwarf2out.c (is_java): Remove the function.
5198 (output_pubname): Remove usage of the function.
5199 (lower_bound_default): Remove usage of DW_LANG_Java.
5200 (gen_compile_unit_die): Likewise.
5201 * gcc.c: Remove compiler defaults for .java and .zip files.
5202 * gimple-expr.c (remove_suffix): Change as there's no longer
5203 extension than 4-letter one.
5204 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5205 (gimplify_save_expr): Likewise.
5206 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5207 as it's possible even for other languages than Java.
5208 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5209 * lto-opts.c (lto_write_options): Remove reference to Java.
5210 * opts.c (strip_off_ending): Update file extension handling.
5211 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5212 * tree-eh.c (lower_resx): Likewise.
5213 * tree.c (free_lang_data_in_type): Remove dead code.
5214 (find_decls_types_r): Likewise.
5215 (build_common_builtin_nodes): Remove Java from a comment.
5216 (verify_type): Remove dead code.
5217 * varasm.c (assemble_external): Remove Java from a comment.
5219 2017-07-14 Martin Liska <mliska@suse.cz>
5221 * opts.c (finish_options): Add quotes.
5222 (common_handle_option): Likewise.
5224 2017-07-14 Martin Liska <mliska@suse.cz>
5226 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5227 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5229 * dwarf2out.c (lower_bound_default): Do not handle
5231 (gen_compile_unit_die): Likewise.
5232 * gcc.c: Remove default extension binding for GNU Pascal.
5233 * stmt.c: Remove Pascal language from a comment.
5234 * xcoffout.c: Likewise.
5236 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5239 * diagnostic-show-locus.c (fixit_cmp): New function.
5240 (layout::layout): Sort m_fixit_hints.
5241 (column_range::column_range): Assert that the values are valid.
5242 (struct char_span): New struct.
5243 (correction::overwrite): New method.
5244 (struct source_line): New struct.
5245 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5246 calls in terms of classes source_line and char_span, and
5247 correction::overwrite.
5248 (selftest::test_overlapped_fixit_printing_2): New function.
5249 (selftest::diagnostic_show_locus_c_tests): Call it.
5251 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5253 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5254 early if there is no lhs.
5256 2017-07-13 Martin Liska <mliska@suse.cz>
5258 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5259 (gen_reference_type_die): Likewise.
5260 * stor-layout.c: Remove Pascal-related comment.
5262 2017-07-13 Martin Liska <mliska@suse.cz>
5264 * opts.c (finish_options): Add quotes to error messages.
5265 (parse_sanitizer_options): Likewise.
5267 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5269 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5271 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5273 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5275 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5277 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5279 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5281 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5284 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5285 provides the hardware capability bits, define the macro
5286 __BUILTIN_CPU_SUPPORTS__.
5287 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5288 if GLIBC does not provide the hardware capability bits. Add a
5289 gcc_unreachable call if the built-in cpu function is neither
5290 __builtin_cpu_is nor __builtin_cpu_supports.
5291 (rs6000_get_function_versions_dispatcher): Change the warning
5292 that an old GLIBC is used which does not export the capability
5293 bits to be an error.
5294 * doc/extend.texi (target_clones attribute): Document the
5295 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5296 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5297 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5298 the macros defined by GCC if the newer GLIBC is available.
5300 2017-07-12 Jeff Law <law@redhat.com>
5302 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5303 remaining includes slightly.
5304 * config/riscv/riscv-builtins.c: Include profile-count.h.
5306 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5309 * config/avr/avr.c (avr_set_current_function): In diagnostic
5310 messages: Quote keywords and (parts of) identifiers.
5311 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5314 2017-07-12 Carl Love <cel@us.ibm.com>
5316 * config/rs6000/rs6000-c.c: Add support for built-in functions
5317 vector bool char vec_revb (vector bool char);
5318 vector bool short vec_revb (vector short char);
5319 vector bool int vec_revb (vector bool int);
5320 vector bool long long vec_revb (vector bool long long);
5321 * doc/extend.texi: Update the built-in documentation file for the
5322 new built-in functions.
5324 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5326 * config/s390/s390.md: Remove movcc splitter.
5328 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5330 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5331 load/store on condition.
5333 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5336 * config/avr/avr.c (avr_encode_section_info)
5337 [progmem && !TREE_READONLY]: Error if progmem object needs
5340 2017-07-11 Michael Collison <michael.collison@arm.com>
5342 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5345 2017-07-11 Carl Love <cel@us.ibm.com>
5347 * config/rs6000/rs6000-c.c: Add support for builtins
5348 vector unsigned int vec_parity_lsbb (vector signed int);
5349 vector unsigned int vec_parity_lsbb (vector unsigned int);
5350 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5351 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5352 vector unsigned long long vec_parity_lsbb (vector signed long long);
5353 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5354 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5355 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5356 * doc/extend.texi: Update the built-in documentation file for the
5357 new built-in functions.
5359 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5361 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5362 (layout::m_primary_loc): New field.
5363 (layout::layout): Initialize new field. Move location filtering
5364 logic from here to...
5365 (layout::maybe_add_location_range): ...this new method. Add
5366 support for filtering to just the lines already specified by other
5368 (layout::will_show_line_p): New method.
5369 (gcc_rich_location::add_location_if_nearby): New method.
5370 (selftest::test_add_location_if_nearby): New test function.
5371 (selftest::diagnostic_show_locus_c_tests): Call it.
5372 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5375 2017-07-11 Tom de Vries <tom@codesourcery.com>
5377 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5378 (bb_first_real_insn): New function.
5379 (nvptx_single): Add extra initialization of broadcasted condition
5382 2017-07-11 Nathan Sidwell <nathan@acm.org>
5384 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5386 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5388 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5389 Binutils doc as TEXI will mess them up.
5390 * doc/invoke.texi (AVR Options): Same here.
5392 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5394 * config/sparc/sparc.opt (mfix-ut700): New option.
5395 (mfix-gr712rc): Likewise.
5396 (sparc_fix_b2bst): New variable.
5397 * doc/invoke.texi (SPARC options): Document them.
5398 (ARM options): Fix warnings.
5399 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5400 instructions to prevent sequences that can trigger the store-store
5401 errata for certain LEON3FT processors.
5402 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
5403 (sparc_option_override): Set sparc_fix_b2bst appropriately.
5404 * config/sparc/sparc.md (fix_b2bst): New attribute.
5405 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
5407 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
5410 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
5412 (*rsqrtsf2_sse): Ditto.
5414 (div<mode>3): Macroize insn from divdf3 and divsf3
5415 using MODEF mode iterator.
5417 2017-07-10 Martin Sebor <msebor@redhat.com>
5419 PR tree-optimization/80397
5420 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
5421 instead of testing for equality to INTEGER_TYPE.
5423 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
5425 * config.gcc: Remove uclibc from arc target spec.
5427 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
5429 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
5431 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5434 * lto-wrapper.c (remove_option): New function.
5435 (merge_and_complain): Merge PIC/PIE options more realistically.
5437 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5439 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
5443 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
5444 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
5445 * config.in: Regenerate.
5446 * configure: Regenerate.
5447 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
5448 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
5449 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
5450 (TARGET_GASISR_PROLOGUES): ...target mask.
5451 * common/config/avr/avr-common.c
5452 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
5453 Set -mgas-isr-prologues.
5454 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
5455 INSERT_PASS_BEFORE for it.
5456 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
5457 * config/avr/avr.c (avr_option_override)
5458 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
5459 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
5460 (avr_attribute_table) <no_gccisr>: Add new function attribute.
5461 (avr_set_current_function) <is_no_gccisr>: Init machine field.
5462 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
5464 (make_avr_pass_pre_proep): New function.
5465 (emit_push_sfr) <treg>: Add argument to function and use it
5467 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
5468 and set machine->gasisr.yes.
5469 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
5470 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
5471 __gcc_isr.n_pushed to .L__stack_usage.
5472 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
5473 (avr_asm_final_postscan_insn): ...this new static function.
5474 * config/avr/avr.h (machine_function)
5475 <is_no_gccisr, use_L__stack_usage>: New fields.
5476 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
5477 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
5478 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
5479 (gasisr, *gasisr): New expander and insn.
5480 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5481 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
5482 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
5484 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
5486 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
5489 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5491 Move jump-tables out of .text again.
5494 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
5495 (ASM_OUTPUT_ADDR_VEC): New function.
5496 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
5497 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
5498 INSN_ADDRESSes as asm comment.
5499 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
5500 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
5501 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
5502 * config/avr/avr.md (*tablejump): Adjust comment.
5503 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
5504 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
5506 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
5507 (avr_output_addr_vec): New proto.
5508 (avr_log_t) <insn_addresses>: New field.
5510 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
5513 * config/i386/i386.c (ix86_function_arg_advance): Set
5514 outgoing_args_on_stack to true if there are outgoing arguments
5516 (ix86_function_arg): Likewise.
5517 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
5518 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
5519 * config/i386/i386.h (machine_function): Add
5520 outgoing_args_on_stack.
5522 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
5524 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5526 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
5528 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
5530 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
5531 (REAL_H): Remove $(MACHMODE_H).
5532 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
5534 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
5535 $(MACHMODE_H) and double-int.h.
5536 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
5538 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
5541 2017-07-07 Andrew Pinski <apinski@cavium.com>
5543 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
5544 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
5546 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5548 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
5549 Add warning if GCC was not configured to link against a GLIBC that
5550 exports the hardware capability bits.
5551 (make_resolver_func): Make resolver function private and not a
5552 COMDAT function. Create the name with clone_function_name instead
5553 of make_unique_name.
5556 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
5557 correct operand in doing the split.
5559 2017-07-07 Carl Love <cel@us.ibm.com>
5561 * config/rs6000/rs6000-c: Add support for built-in function
5562 vector unsigned short vec_pack_to_short_fp32 (vector float,
5564 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
5565 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
5566 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
5567 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
5568 (convert_4f32_8i16): Add define_expand.
5569 * doc/extend.texi: Update the built-in documentation file for the
5570 new built-in function.
5572 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5574 * config/sparc/m8.md: New file.
5575 * config/sparc/sparc.md: Include m8.md.
5577 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5579 * config/sparc/sparc.opt: New option -mvis4b.
5580 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
5581 (sparc_option_override): Handle VIS4B.
5582 (enum sparc_builtins): Define
5583 SPARC_BUILTIN_DICTUNPACK{8,16,32},
5584 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
5585 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
5586 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
5587 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
5588 (check_constant_argument): New function.
5589 (sparc_vis_init_builtins): Define builtins
5590 __builtin_vis_dictunpack{8,16,32},
5591 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
5592 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
5593 __builtin_vis_fpcmpde{8,16,32}shl and
5594 __builtin_vis_fpcmpur{8,16,32}shl.
5595 (sparc_expand_builtin): Check that the constant operands to
5596 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
5597 constant and in range.
5598 * config/sparc/sparc-c.c (sparc_target_macros): Handle
5600 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
5601 (SPARC_IMM5_P): Likewise.
5602 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
5603 (enabled): Handle vis4b.
5604 (UNSPEC_DICTUNPACK): New unspec.
5605 (UNSPEC_FPCMPSHL): Likewise.
5606 (UNSPEC_FPUCMPSHL): Likewise.
5607 (UNSPEC_FPCMPDESHL): Likewise.
5608 (UNSPEC_FPCMPURSHL): Likewise.
5609 (cpu_feature): New CPU feature `vis4b'.
5610 (dictunpack{8,16,32}): New insns.
5611 (FPCSMODE): New mode iterator.
5612 (fpcscond): New code iterator.
5613 (fpcsucond): Likewise.
5614 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
5615 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
5616 (fpcmpde{8,16,32}{si,di}shl): Likewise.
5617 (fpcmpur{8,16,32}{si,di}shl): Likewise.
5618 * config/sparc/constraints.md: Define constraints `q' for unsigned
5619 2-bit integer constants and `t' for unsigned 5-bit integer
5621 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
5623 (imm5_operand_dictunpack16): Likewise.
5624 (imm5_operand_dictunpack32): Likewise.
5625 (imm2_operand): Likewise.
5626 * doc/invoke.texi (SPARC Options): Document -mvis4b.
5627 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
5628 ditunpack* and fpcmp*shl builtins.
5630 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5632 * config.gcc: Handle m8 in --with-{cpu,tune} options.
5633 * config.in: Add HAVE_AS_SPARC6 define.
5634 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
5636 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
5638 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
5639 (CPP_CPU_SPEC): Handle m8.
5640 (ASM_CPU_SPEC): Likewise.
5641 * config/sparc/sparc-opts.h (enum processor_type): Add
5643 * config/sparc/sparc.c (m8_costs): New struct.
5644 (sparc_option_override): Handle TARGET_CPU_m8.
5645 (sparc32_initialize_trampoline): Likewise.
5646 (sparc64_initialize_trampoline): Likewise.
5647 (sparc_issue_rate): Likewise.
5648 (sparc_register_move_cost): Likewise.
5649 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
5650 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
5651 (ASM_CPU64_DEFAULT_SPEC): Likewise.
5652 (CPP_CPU_SPEC): Handle M8.
5653 (ASM_CPU_SPEC): Likewise.
5654 (AS_M8_FLAG): Define.
5655 * config/sparc/sparc.md: Add m8 to the cpu attribute.
5656 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
5657 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
5659 * configure: Regenerate.
5660 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
5663 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5665 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
5667 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
5668 ("*movdi_insn_sp32"): Do not set v3pipe.
5669 ("*movsi_insn"): Likewise.
5670 ("*movdi_insn_sp64"): Likewise.
5671 ("*movsf_insn"): Likewise.
5672 ("*movdf_insn_sp32"): Likewise.
5673 ("*movdf_insn_sp64"): Likewise.
5674 ("*zero_extendsidi2_insn_sp64"): Likewise.
5675 ("*sign_extendsidi2_insn"): Likewise.
5676 ("*mov<VM32:mode>_insn"): Likewise.
5677 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5678 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5679 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5680 ("<vlop:code><VL:mode>3"): Likewise.
5681 ("*not_<vlop:code><VL:mode>3"): Likewise.
5682 ("*nand<VL:mode>_vis"): Likewise.
5683 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
5684 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
5685 ("one_cmpl<VL:mode>2"): Likewise.
5686 ("faligndata<VM64:mode>_vis"): Likewise.
5687 ("alignaddrsi_vis"): Likewise.
5688 ("alignaddrdi_vis"): Likweise.
5689 ("alignaddrlsi_vis"): Likewise.
5690 ("alignaddrldi_vis"): Likewise.
5691 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5692 ("bmaskdi_vis"): Likewise.
5693 ("bmasksi_vis"): Likewise.
5694 ("bshuffle<VM64:mode>_vis"): Likewise.
5695 ("cmask8<P:mode>_vis"): Likewise.
5696 ("cmask16<P:mode>_vis"): Likewise.
5697 ("cmask32<P:mode>_vis"): Likewise.
5698 ("pdistn<P:mode>_vis"): Likewise.
5699 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5701 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5703 * config/sparc/sparc.md ("subtype"): New insn attribute.
5704 ("*wrgsr_sp64"): Set insn subtype.
5705 ("*rdgsr_sp64"): Likewise.
5706 ("alignaddrsi_vis"): Likewise.
5707 ("alignaddrdi_vis"): Likewise.
5708 ("alignaddrlsi_vis"): Likewise.
5709 ("alignaddrldi_vis"): Likewise.
5710 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5711 ("fexpand_vis"): Likewise.
5712 ("fpmerge_vis"): Likewise.
5713 ("faligndata<VM64:mode>_vis"): Likewise.
5714 ("bshuffle<VM64:mode>_vis"): Likewise.
5715 ("cmask8<P:mode>_vis"): Likewise.
5716 ("cmask16<P:mode>_vis"): Likewise.
5717 ("cmask32<P:mode>_vis"): Likewise.
5718 ("fchksm16_vis"): Likewise.
5719 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
5720 ("fmean16_vis"): Likewise.
5721 ("fp<plusminus_insn>64_vis"): Likewise.
5722 ("<plusminus_insn>v8qi3"): Likewise.
5723 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5724 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
5725 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
5726 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
5727 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
5728 ("*movqi_insn"): Likewise.
5729 ("*movhi_insn"): Likewise.
5730 ("*movsi_insn"): Likewise.
5731 ("movsi_pic_gotdata_op"): Likewise.
5732 ("*movdi_insn_sp32"): Likewise.
5733 ("*movdi_insn_sp64"): Likewise.
5734 ("movdi_pic_gotdata_op"): Likewise.
5735 ("*movsf_insn"): Likewise.
5736 ("*movdf_insn_sp32"): Likewise.
5737 ("*movdf_insn_sp64"): Likewise.
5738 ("*zero_extendhisi2_insn"): Likewise.
5739 ("*zero_extendqihi2_insn"): Likewise.
5740 ("*zero_extendqisi2_insn"): Likewise.
5741 ("*zero_extendqidi2_insn"): Likewise.
5742 ("*zero_extendhidi2_insn"): Likewise.
5743 ("*zero_extendsidi2_insn_sp64"): Likewise.
5744 ("ldfsr"): Likewise.
5745 ("prefetch_64"): Likewise.
5746 ("prefetch_32"): Likewise.
5747 ("tie_ld32"): Likewise.
5748 ("tie_ld64"): Likewise.
5749 ("*tldo_ldub_sp32"): Likewise.
5750 ("*tldo_ldub1_sp32"): Likewise.
5751 ("*tldo_ldub2_sp32"): Likewise.
5752 ("*tldo_ldub_sp64"): Likewise.
5753 ("*tldo_ldub1_sp64"): Likewise.
5754 ("*tldo_ldub2_sp64"): Likewise.
5755 ("*tldo_ldub3_sp64"): Likewise.
5756 ("*tldo_lduh_sp32"): Likewise.
5757 ("*tldo_lduh1_sp32"): Likewise.
5758 ("*tldo_lduh_sp64"): Likewise.
5759 ("*tldo_lduh1_sp64"): Likewise.
5760 ("*tldo_lduh2_sp64"): Likewise.
5761 ("*tldo_lduw_sp32"): Likewise.
5762 ("*tldo_lduw_sp64"): Likewise.
5763 ("*tldo_lduw1_sp64"): Likewise.
5764 ("*tldo_ldx_sp64"): Likewise.
5765 ("*mov<VM32:mode>_insn"): Likewise.
5766 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5767 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5769 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5771 * config/sparc/sparc.md ("type"): New insn type viscmp.
5772 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
5774 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
5775 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
5776 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5777 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
5779 ("n7_vis_logical_11cycle"): Likewise.
5780 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
5781 * config/sparc/niagara2.md ("niag3_vis": Likewise.
5782 * config/sparc/niagara.md ("niag_vis"): Likewise.
5783 * config/sparc/ultra3.md ("us3_fga"): Likewise.
5784 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
5786 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5788 * config/sparc/sparc.md: New instruction type `bmask'.
5789 (bmaskdi_vis): Use the `bmask' type.
5790 (bmasksi_vis): Likewise.
5791 * config/sparc/ultra3.md (us3_array): Likewise.
5792 * config/sparc/niagara7.md (n7_array): Likewise.
5793 * config/sparc/niagara4.md (n4_array): Likewise.
5794 * config/sparc/niagara2.md (niag2_vis): Likewise.
5795 (niag3_vis): Likewise.
5796 * config/sparc/niagara.md (niag_vis): Likewise.
5798 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5800 * ipa-comdats.c: Remove optimize check from gate.
5801 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
5802 for functions not optimized.
5803 (ipa_fn_summary_read): Skip optimize check.
5804 (ipa_fn_summary_write): Likewise.
5805 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
5807 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
5809 (can_inline_edge_p): Check flag_pcc_struct_return for match.
5810 (check_callers): Give up on caller which is not optimized.
5811 (inline_small_functions): Likewise.
5812 (ipa_inline): Do not give up when not optimizing.
5813 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
5814 away unoptimizes cdtors.
5815 (whole_program_function_and_variable_visibility): Do
5816 ipa_discover_readonly_nonaddressable_vars in LTO mode.
5817 * ipa.c (process_references): Do not check optimize.
5818 (symbol_table::remove_unreachable_nodes): Update optimize check.
5819 (set_writeonly_bit): Update optimize check.
5820 (pass_ipa_cdtor_merge::gate): Do not check optimize.
5821 (pass_ipa_single_use::gate): Remove.
5823 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5825 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
5826 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
5827 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
5828 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
5829 permute_load, permute_store, adjust_extract, adjust_splat,
5830 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
5831 replace_swap_with_copy, dump_swap_insn_table,
5832 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
5833 recombine_lvx_pattern, recombine_stvx_pattern,
5834 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
5835 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
5836 to file rs6000-p8swap.c.
5837 * config/rs6000/rs6000-p8swap.c: New file.
5838 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
5839 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
5840 and rs6000*-*-* targets.
5842 2017-07-06 David Malcolm <dmalcolm@redhat.com>
5844 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
5846 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5848 * lto-wrapper.c (merge_and_complain): Do not merge
5849 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
5850 fsigned_zeros, ftrapping_math, fwrapv.
5851 (append_compiler_options): Do not track these options.
5852 (append_linker_options): Likewie
5854 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5856 * cgraphunit.c (cgraph_node::finalize_function): When
5857 !flag_toplevel_reorde set no_reorder flag.
5858 (varpool_node::finalize_decl): Likewise.
5859 (symbol_table::compile): Drop no toplevel reorder path.
5861 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5863 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
5864 edges; zero probability is not better than uninitialized.
5866 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
5868 * asan.h (asan_sanitize_allocas_p): Declare.
5869 * asan.c (asan_sanitize_allocas_p): New function.
5870 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
5871 (handle_builtin_alloca): Likewise.
5872 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
5873 if !asan_sanitize_allocas_p.
5874 * params.def (asan-instrument-allocas): Add new option.
5875 * params.h (ASAN_PROTECT_ALLOCAS): Define.
5876 * opts.c (common_handle_option): Disable allocas sanitization for
5879 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
5881 * asan.c: Include gimple-fold.h.
5882 (get_last_alloca_addr): New function.
5883 (handle_builtin_stackrestore): Likewise.
5884 (handle_builtin_alloca): Likewise.
5885 (asan_emit_allocas_unpoison): Likewise.
5886 (get_mem_refs_of_builtin_call): Add new parameter, remove const
5887 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
5888 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
5889 (instrument_builtin_call): Pass gimple iterator to
5890 get_mem_refs_of_builtin_call.
5891 (last_alloca_addr): New global.
5892 * asan.h (asan_emit_allocas_unpoison): Declare.
5893 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
5894 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
5895 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
5896 if function calls alloca.
5897 * gimple-fold.c (replace_call_with_value): Remove static keyword.
5898 * gimple-fold.h (replace_call_with_value): Declare.
5899 * internal-fn.c: Include asan.h.
5900 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
5901 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
5903 2017-07-06 David Malcolm <dmalcolm@redhat.com>
5905 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
5906 (C_SELFTEST_FLAGS): New.
5907 (CPP_SELFTEST_FLAGS): New.
5908 (SELFTEST_DEPS): New, from deps of s-selftest.
5909 (C_SELFTEST_DEPS): New, from deps of s-selftest.
5910 (CPP_SELFTEST_DEPS): New.
5911 (selftest): Add dependency on s-selftest-c++.
5912 (s-selftest): Rename to...
5913 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
5914 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
5915 than SELFTEST_FLAGS.
5916 (selftest-gdb): Rename to...
5917 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
5919 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
5920 (selftest-valgrind): Rename to...
5921 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
5923 (selftest-valgrind): Reintroduce as an alias for
5924 selftest-c-valgrind.
5925 (s-selftest-c++): New.
5926 (selftest-c++-gdb): New.
5927 (selftest-c++-valgrind): New.
5929 2017-07-06 Olivier Hainque <hainque@adacore.com>
5931 * gcc.c (process_command): When deciding if undefined variables
5932 should be ignored when processing specs, accept "gcc -v" as well.
5934 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5936 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
5937 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
5939 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5941 * config/arm/arm-cpus.in (armv8-r): Add new entry.
5942 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
5943 * config/arm/arm-tables.opt: Regenerate.
5944 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
5946 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
5948 2017-07-06 Carl Love <cel@us.ibm.com>
5950 * ChangeLog: Clean up from mid air collision
5952 2017-07-06 Carl Love <cel@us.ibm.com>
5954 * config/rs6000/rs6000-c.c: Add support for built-in functions
5955 vector signed int vec_subc (vector signed int, vector signed int);
5956 vector signed __int128 vec_subc (vector signed __int128,
5957 vector signed __int128);
5958 vector unsigned __int128 vec_subc (vector unsigned __int128,
5959 vector unsigned __int128);
5960 vector signed int vec_sube (vector signed int, vector signed int,
5962 vector unsigned int vec_sube (vector unsigned int,
5963 vector unsigned int,
5964 vector unsigned int);
5965 vector signed __int128 vec_sube (vector signed __int128,
5966 vector signed __int128,
5967 vector signed__int128);
5968 vector unsigned __int128 vec_sube (vector unsigned __int128,
5969 vector unsigned __int128,
5970 vector unsigned __int128);
5971 vector signed int vec_subec (vector signed int, vector signed int,
5973 vector unsigned int vec_subec (vector unsigned int,
5974 vector unsigned int,
5975 vector unsigned int);
5976 vector signed __int128 vec_subec (vector signed __int128,
5977 vector signed __int128,
5978 vector signed__int128);
5979 vector unsigned __int128 vec_subec (vector unsigned __int128,
5980 vector unsigned __int128,
5981 vector unsigned __int128);
5982 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
5983 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
5984 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
5985 BU_ALTIVEC_OVERLOAD_X definitions.
5986 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
5987 * doc/extend.texi: Update the built-in documentation file for the new
5990 2017-07-06 David Malcolm <dmalcolm@redhat.com>
5993 * diagnostic-show-locus.c (layout::layout): Use start and finish
5994 spelling location for the start and finish of each range.
5995 * genmatch.c (linemap_client_expand_location_to_spelling_point):
5996 Add unused aspect param.
5997 * input.c (expand_location_1): Add "aspect" param, and use it
5998 to access the correct part of the location.
5999 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6001 (expand_location_to_spelling_point): Likewise.
6002 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6003 param, and pass it to expand_location_1.
6005 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6007 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6008 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6009 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6010 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6011 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6012 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6013 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6014 _mm_maskz_getmant_ss): New intrinsics.
6015 (__builtin_ia32_getexpss128_mask): Changed to ...
6016 __builtin_ia32_getexpss128_round ... this.
6017 (__builtin_ia32_getexpsd128_mask): Changed to ...
6018 __builtin_ia32_getexpsd128_round ... this.
6019 * config/i386/i386-builtin-types.def
6020 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6021 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6022 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6023 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6024 __builtin_ia32_getmantss_mask_round): New builtins.
6025 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6026 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6027 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6028 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6029 * config/i386/sse.md
6030 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6031 avx512f_sgetexp<mode><mask_scalar_name>
6032 <round_saeonly_scalar_name> ... this.
6033 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6034 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6035 vgetexp<ssescalarmodesuffix>
6036 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6037 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6038 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6039 avx512f_vgetmant<mode><mask_scalar_name>
6040 <round_saeonly_scalar_name> ... this.
6041 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6042 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6043 vgetmant<ssescalarmodesuffix>
6044 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6045 %0<mask_scalar_operand4>, %1, %2
6046 <round_saeonly_scalar_mask_op4>, %3} ... this.
6047 * config/i386/subst.md (mask_scalar_operand4,
6048 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6049 round_saeonly_scalar_nimm_predicate): New subst attributes.
6051 2017-07-06 Julia Koval <julia.koval@intel.com>
6053 * config/i386/i386.c (ix86_erase_embedded_rounding):
6054 Remove code for old rounding pattern.
6056 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6058 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6060 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6062 * doc/sourcebuild.texi (Test Directives, Variants of
6063 dg-require-support): Add documentation for dg-require-stack-check.
6065 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6067 * config/i386/subst.md (mask_scalar, round_scalar,
6068 round_saeonly_scalar): New meta-templates.
6069 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6070 round_scalar_mask_operand3, round_scalar_mask_op3,
6071 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6072 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6073 round_saeonly_scalar_constraint,
6074 round_saeonly_scalar_prefix): New subst attribute.
6075 * config/i386/sse.md
6076 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6077 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6078 <round_scalar_name> ... this.
6079 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6080 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6081 <round_scalar_name> ... this.
6082 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6083 <sse>_vm<code><mode>3<mask_scalar_name>
6084 <round_saeonly_scalar_name> ... this.
6085 (v<plusminus_mnemonic><ssescalarmodesuffix>
6086 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6087 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6088 v<plusminus_mnemonic><ssescalarmodesuffix>
6089 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6090 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6091 (v<multdiv_mnemonic><ssescalarmodesuffix>
6092 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6093 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6094 v<multdiv_mnemonic><ssescalarmodesuffix>
6095 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6096 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6097 (v<maxmin_float><ssescalarmodesuffix>
6098 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6099 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6100 v<maxmin_float><ssescalarmodesuffix>
6101 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6102 %0<mask_scalar_operand3>, %1, %<iptr>2
6103 <round_saeonly_scalar_mask_op3>} ... this.
6105 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6107 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6108 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6110 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6111 Alan Hayward <alan.hayward@arm.com>
6112 David Sherwood <david.sherwood@arm.com>
6114 * combine.c (simplify_if_then_else): Remove "enum" before
6116 * compare-elim.c (can_eliminate_compare): Likewise.
6117 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6119 (aarch64_lookup_simd_builtin_type): Likewise.
6120 (aarch64_simd_builtin_type): Likewise.
6121 (aarch64_init_simd_builtin_types): Likewise.
6122 (aarch64_simd_expand_args): Likewise.
6123 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6125 (aarch64_reverse_mask): Likewise.
6126 (aarch64_simd_emit_reg_reg_move): Likewise.
6127 (aarch64_gen_adjusted_ldpstp): Likewise.
6128 (aarch64_ccmp_mode_to_code): Likewise.
6129 (aarch64_operands_ok_for_ldpstp): Likewise.
6130 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6131 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6133 (aarch64_min_divisions_for_recip_mul): Likewise.
6134 (aarch64_reassociation_width): Likewise.
6135 (aarch64_get_condition_code_1): Likewise.
6136 (aarch64_simd_emit_reg_reg_move): Likewise.
6137 (aarch64_simd_attr_length_rglist): Likewise.
6138 (aarch64_reverse_mask): Likewise.
6139 (aarch64_operands_ok_for_ldpstp): Likewise.
6140 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6141 (aarch64_gen_adjusted_ldpstp): Likewise.
6142 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6144 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6145 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6146 (arm_lookup_simd_builtin_type): Likewise.
6147 (arm_simd_builtin_type): Likewise.
6148 (arm_init_simd_builtin_types): Likewise.
6149 (arm_expand_builtin_args): Likewise.
6150 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6151 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6152 (ft32_setup_incoming_varargs): Likewise.
6153 (ft32_function_arg): Likewise.
6154 (ft32_function_arg_advance): Likewise.
6155 (ft32_pass_by_reference): Likewise.
6156 (ft32_arg_partial_bytes): Likewise.
6157 (ft32_valid_pointer_mode): Likewise.
6158 (ft32_addr_space_pointer_mode): Likewise.
6159 (ft32_addr_space_legitimate_address_p): Likewise.
6160 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6162 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6163 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6164 (iamcu_alignment): Likewise.
6165 (canonicalize_vector_int_perm): Likewise.
6166 (ix86_noce_conversion_profitable_p): Likewise.
6167 (ix86_mpx_bound_mode): Likewise.
6168 (ix86_operands_ok_for_move_multiple): Likewise.
6169 * config/microblaze/microblaze-protos.h
6170 (microblaze_expand_conditional_branch_reg): Likewise.
6171 * config/microblaze/microblaze.c
6172 (microblaze_expand_conditional_branch_reg): Likewise.
6173 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6175 (rs6000_reassociation_width): Likewise.
6176 (rs6000_invalid_binary_op): Likewise.
6177 (fusion_p9_p): Likewise.
6178 (emit_fusion_p9_load): Likewise.
6179 (emit_fusion_p9_store): Likewise.
6180 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6182 (riscv_hard_regno_mode_ok_p): Likewise.
6183 (riscv_address_insns): Likewise.
6184 (riscv_split_symbol): Likewise.
6185 (riscv_legitimize_move): Likewise.
6186 (riscv_function_value): Likewise.
6187 (riscv_hard_regno_nregs): Likewise.
6188 (riscv_expand_builtin): Likewise.
6189 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6190 (riscv_build_integer): Likewise.
6191 (riscv_split_integer): Likewise.
6192 (riscv_legitimate_constant_p): Likewise.
6193 (riscv_cannot_force_const_mem): Likewise.
6194 (riscv_regno_mode_ok_for_base_p): Likewise.
6195 (riscv_valid_base_register_p): Likewise.
6196 (riscv_valid_offset_p): Likewise.
6197 (riscv_valid_lo_sum_p): Likewise.
6198 (riscv_classify_address): Likewise.
6199 (riscv_legitimate_address_p): Likewise.
6200 (riscv_address_insns): Likewise.
6201 (riscv_load_store_insns): Likewise.
6202 (riscv_force_binary): Likewise.
6203 (riscv_split_symbol): Likewise.
6204 (riscv_force_address): Likewise.
6205 (riscv_legitimize_address): Likewise.
6206 (riscv_move_integer): Likewise.
6207 (riscv_legitimize_const_move): Likewise.
6208 (riscv_legitimize_move): Likewise.
6209 (riscv_address_cost): Likewise.
6210 (riscv_subword): Likewise.
6211 (riscv_output_move): Likewise.
6212 (riscv_canonicalize_int_order_test): Likewise.
6213 (riscv_emit_int_order_test): Likewise.
6214 (riscv_function_arg_boundary): Likewise.
6215 (riscv_pass_mode_in_fpr_p): Likewise.
6216 (riscv_pass_fpr_single): Likewise.
6217 (riscv_pass_fpr_pair): Likewise.
6218 (riscv_get_arg_info): Likewise.
6219 (riscv_function_arg): Likewise.
6220 (riscv_function_arg_advance): Likewise.
6221 (riscv_arg_partial_bytes): Likewise.
6222 (riscv_function_value): Likewise.
6223 (riscv_pass_by_reference): Likewise.
6224 (riscv_setup_incoming_varargs): Likewise.
6225 (riscv_print_operand): Likewise.
6226 (riscv_elf_select_rtx_section): Likewise.
6227 (riscv_save_restore_reg): Likewise.
6228 (riscv_for_each_saved_reg): Likewise.
6229 (riscv_register_move_cost): Likewise.
6230 (riscv_hard_regno_mode_ok_p): Likewise.
6231 (riscv_hard_regno_nregs): Likewise.
6232 (riscv_class_max_nregs): Likewise.
6233 (riscv_memory_move_cost): Likewise.
6234 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6235 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6236 (rl78_addr_space_address_mode): Likewise.
6237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6239 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6240 (rs6000_reassociation_width): Likewise.
6241 (rs6000_invalid_binary_op): Likewise.
6242 (fusion_p9_p): Likewise.
6243 (emit_fusion_p9_load): Likewise.
6244 (emit_fusion_p9_store): Likewise.
6245 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6246 (ok_for_simple_move_operands): Likewise.
6247 (ok_for_simple_move_strict_operands): Likewise.
6248 (ok_for_simple_arith_logic_operands): Likewise.
6249 (visium_legitimize_reload_address): Likewise.
6250 (visium_select_cc_mode): Likewise.
6251 (output_cbranch): Likewise.
6252 (visium_split_double_move): Likewise.
6253 (visium_expand_copysign): Likewise.
6254 (visium_expand_int_cstore): Likewise.
6255 (visium_expand_fp_cstore): Likewise.
6256 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6257 (visium_function_arg): Likewise.
6258 (visium_function_arg_advance): Likewise.
6259 (visium_libcall_value): Likewise.
6260 (visium_setup_incoming_varargs): Likewise.
6261 (visium_legitimate_constant_p): Likewise.
6262 (visium_legitimate_address_p): Likewise.
6263 (visium_legitimize_address): Likewise.
6264 (visium_secondary_reload): Likewise.
6265 (visium_register_move_cost): Likewise.
6266 (visium_memory_move_cost): Likewise.
6267 (prepare_move_operands): Likewise.
6268 (ok_for_simple_move_operands): Likewise.
6269 (ok_for_simple_move_strict_operands): Likewise.
6270 (ok_for_simple_arith_logic_operands): Likewise.
6271 (visium_function_value_1): Likewise.
6272 (rtx_ok_for_offset_p): Likewise.
6273 (visium_legitimize_reload_address): Likewise.
6274 (visium_split_double_move): Likewise.
6275 (visium_expand_copysign): Likewise.
6276 (visium_expand_int_cstore): Likewise.
6277 (visium_expand_fp_cstore): Likewise.
6278 (visium_split_cstore): Likewise.
6279 (visium_select_cc_mode): Likewise.
6280 (visium_split_cbranch): Likewise.
6281 (output_cbranch): Likewise.
6282 (visium_print_operand_address): Likewise.
6283 * expmed.c (flip_storage_order): Likewise.
6284 * expmed.h (emit_cstore): Likewise.
6285 (flip_storage_order): Likewise.
6286 * genrecog.c (validate_pattern): Likewise.
6287 * hsa-gen.c (gen_hsa_addr): Likewise.
6288 * internal-fn.c (expand_arith_overflow): Likewise.
6289 * ira-color.c (allocno_copy_cost_saving): Likewise.
6290 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6291 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6292 (process_invariant_for_inheritance): Likewise.
6293 * lra-eliminations.c (move_plus_up): Likewise.
6294 * omp-low.c (lower_oacc_reductions): Likewise.
6295 * simplify-rtx.c (simplify_subreg): Likewise.
6296 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6297 (TARGET_CHKP_BOUND_MODE): Likewise..
6298 * targhooks.c (default_chkp_bound_mode): Likewise.
6299 (default_setup_incoming_vararg_bounds): Likewise.
6300 * targhooks.h (default_chkp_bound_mode): Likewise.
6301 (default_setup_incoming_vararg_bounds): Likewise.
6302 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6303 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6304 (have_whole_vector_shift): Likewise.
6305 * tree-vect-stmts.c (vectorizable_load): Likewise.
6306 * doc/tm.texi: Regenerate.
6308 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6310 Graceful degrade if Binutils PR21472 is not available.
6313 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6314 .rodata in flash test fails.
6315 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6316 * confgure: Regenerate.
6317 * config.in: Regenerate.
6318 * config/avr/avr.c (avr_asm_named_section)
6319 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6320 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6321 (avr_asm_init_sections): Same.
6323 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6325 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6326 (fma<VH:mode>4_intrinsic): Likewise.
6327 (*fmsub<VCVTF:mode>4): Likewise.
6328 (*fmsub<VH:mode>4_intrinsic): Likewise.
6330 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6333 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6334 Don't depend on "optimize > 0".
6335 (out_movhi_r_mr, out_movqi_mr_r): Same.
6336 (out_movhi_mr_r, out_movqi_r_mr): Same.
6337 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6338 io_address_operand on "optimize > 0".
6340 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6342 * tree-loop-distribution.c: Add general explanantion on the pass.
6343 (generate_loops_for_partition): Mark distributed loop.
6344 (pg_add_dependence_edges): New parameter. Handle alias data
6345 dependence specially and record it in the parameter if asked.
6346 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6347 (init_partition_graph_vertices, add_partition_graph_edge): New.
6348 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6349 (free_partition_graph_vdata, build_partition_graph): New.
6350 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6351 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6352 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6353 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6354 (version_for_distribution_p, finalize_partitions): New.
6355 (distribute_loop): Handle alias data dependence specially. Factor
6356 out loop fusion code as functions and call these functions.
6358 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6360 * tree-loop-distribution.c (classify_partition): New parameter and
6361 better handle reduction statement.
6362 (rdg_build_partitions): Revise comment.
6363 (distribute_loop): Compute statements in all partitions and pass it
6364 to classify_partition.
6366 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6368 * tree-loop-distribution.c (enum partition_type): New.
6369 (struct partition): New field type.
6370 (partition_merge_into): Add parameter. Update partition type.
6371 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6372 (build_rdg_partition_for_vertex): Compute partition type.
6373 (rdg_build_partitions): Dump partition type.
6374 (distribute_loop): Update calls to partition_merge_into.
6376 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6378 * tree-loop-distribution.c (struct ddr_hasher): New.
6379 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6381 (classify_partition): Call get_data_dependence.
6382 (pg_add_dependence_edges): Ditto.
6383 (distribute_loop): Release data dependence hash table.
6385 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6387 * tree-loop-distribution.c (ref_base_address): Delete.
6388 (similar_memory_accesses): Rename ...
6389 (share_memory_accesses): ... to this. Check if partitions access
6390 the same memory reference.
6391 (distribute_loop): Call share_memory_accesses.
6393 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6395 * tree-loop-distribution.c (struct partition): New field recording
6397 (partition_alloc, partition_free): Init and release data refs.
6398 (partition_merge_into): Merge data refs.
6399 (build_rdg_partition_for_vertex): Collect data refs for partition.
6400 (pg_add_dependence_edges): Change parameters from vector to bitmap.
6402 (distribute_loop): Remve data refs from vertice data of partition
6405 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6407 * tree-loop-distribution.c (params.h): Include header file.
6408 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
6409 (datarefs_vec): New global var.
6410 (create_rdg_vertices): Use datarefs_vec directly.
6411 (free_rdg): Don't free data references.
6412 (build_rdg): Update use. Don't free data references.
6413 (distribute_loop): Compute global variable for data references.
6414 Bail out if there are too many data references.
6416 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6418 * tree-loop-distribution.c (loop_nest): New global var.
6419 (build_rdg): Use loop directly, rather than loop nest.
6420 (pg_add_dependence_edges): Remove loop nest parameter. Use global
6422 (distribute_loop): Compute global variable loop nest. Update use.
6424 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6426 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
6427 (partition_merge_into): New parameter. Dump reason for fusion.
6428 (distribute_loop): Update use of partition_merge_into.
6430 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6432 * tree-loop-distribution.c (bb_top_order_index): New.
6433 (bb_top_order_index_size, bb_top_order_cmp): New.
6434 (stmts_from_loop): Use topological order.
6435 (pass_loop_distribution::execute): Compute and release topological
6436 order for basic blocks.
6438 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6440 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
6443 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6445 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
6446 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
6447 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
6448 * internal-fn.def (LOOP_DIST_ALIAS): New.
6449 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
6450 (fold_loop_internal_call): ... this.
6451 (vect_loop_dist_alias_call): New function.
6452 (set_uid_loop_bbs): Call fold_loop_internal_call.
6453 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
6456 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6459 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
6460 Require dead FLAGS_REG at the beginning of a peephole.
6462 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6465 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
6466 arguments in the call to __builtin_ia32_sbb_u32.
6467 (_subborrow_u64): Swap _X and _Y arguments in the call to
6468 __builtin_ia32_sbb_u64.
6470 2017-07-04 Jakub Jelinek <jakub@redhat.com>
6473 * tree-vrp.c (compare_assert_loc): Turn into a function template
6474 with stable template parameter. Only test if a->e is NULL,
6475 !a->e == !b->e has been verified already. Use e == NULL or
6476 e != NULL instead of e or ! e tests. If stable is true, don't use
6477 iterative_hash_expr, on the other side allow a or b or both NULL
6478 and sort the NULLs last.
6479 (process_assert_insertions): Sort using compare_assert_loc<false>
6480 instead of compare_assert_loc, later sort using
6481 compare_assert_loc<true> before calling process_assert_insertions_for
6482 in a loop. Use break instead of continue once seen NULL pointer.
6484 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6486 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6487 Cortex-R7 and Cortex-R8 processors.
6489 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6491 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
6492 uninitialized while src is not.
6494 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
6496 * common/config/arm/arm-common.c: Adjust include path for
6498 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
6499 (arm-cpu.h): Create in build directory. Adjust dependency rules.
6500 (arm-cpu-data.h): Likewise.
6501 (arm-cpu-cdata.h): Likewise.
6502 * config/arm/arm-cpu.h: Delete.
6503 * config/arm/arm-cpu-cdata.h: Delete.
6504 * config/arm/arm-cpu-data.h: Delete.
6506 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
6508 * config/arm/arm-cpus.in (cortex-a55): New.
6509 (cortex-a75): Likewise.
6510 (cortex-a75.cortex-a55): Likewise.
6511 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
6513 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
6514 * config/arm/arm-cpu-cdata.h: Regenerate.
6515 * config/arm/arm-cpu-data.h: Regenerate.
6516 * config/arm/arm-cpu.h: Regenerate.
6517 * config/arm/arm-tables.opt: Regenerate.
6518 * config/arm/arm-tune.md: Regenerate.
6520 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6522 * haifa-sched.c (sched_create_recovery_edges): Update profile.
6524 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6526 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
6529 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
6531 PR tree-optimization/81292
6532 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
6533 full_string_p, also call adjust_related_strinfos if the adjustment
6534 is simple, otherwise invalidate related strinfos.
6536 2017-07-04 Martin Liska <mliska@suse.cz>
6539 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
6540 newly created variable as DECL_IGNORED_P.
6542 2017-07-04 Martin Liska <mliska@suse.cz>
6545 * ipa-inline.c (inline_small_functions):
6546 Use xstrdup_for_dump.
6548 2017-07-04 Tom de Vries <tom@codesourcery.com>
6550 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
6552 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6555 * config/darwin.c (darwin_function_switched_text_sections):
6558 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
6560 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
6562 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6564 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
6566 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6568 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
6569 min_profitable_iters, and th as inclusive lower bounds.
6570 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
6571 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
6572 for min_profitable_iters and min_profitable_estimate.
6573 (vect_transform_loop): Treat th as an inclusive lower bound.
6574 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6576 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6579 * config/darwin.c (darwin_function_switched_text_sections):
6580 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
6581 in two pieces, and suppress the use of buf.
6583 2017-07-03 Nathan Sidwell <nathan@acm.org>
6585 * hash-table.h (hash_table_mod1): Fix indentation.
6587 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6590 * predict.c (force_edge_cold): Be more careful about propagation
6592 * profile-count.h (profile_probability::guessed,
6593 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
6595 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
6597 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
6599 * doc/invoke.texi (rcpc architecture extension): Document it.
6601 2017-07-03 Richard Biener <rguenther@suse.de>
6603 PR tree-optimization/60510
6604 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
6605 the scalar reduction PHI and use it.
6606 (vectorizable_reduction): Properly guard the single_defuse_cycle
6607 path for non-SLP reduction chains where we cannot use it.
6608 Rework reduc_def/index and vector type deduction. Rework
6609 vector operand gathering during reduction op code-gen.
6610 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
6611 chains dissolve the chain and leave it to non-SLP reduction
6614 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6616 * tree-data-ref.h (dr_alignment): Declare.
6617 * tree-data-ref.c (dr_alignment): New function.
6618 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
6619 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
6621 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
6623 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6625 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
6626 and base_misalignment fields.
6627 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
6628 * tree-data-ref.c: Include builtins.h.
6629 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
6630 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
6631 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
6632 * tree-vect-data-refs.c: Include tree-cfg.h.
6633 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
6634 fields instead of calculating an alignment here.
6635 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
6636 innermost_loop_behavior fields.
6638 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6640 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
6642 (DR_STEP_ALIGNMENT): New macro.
6643 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
6644 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
6645 (create_data_ref): Print it.
6646 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
6647 to tell whether the step preserves vector (mis)alignment.
6648 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6649 Move the check for an integer step and generalise to all INTEGER_CST.
6650 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
6651 Print the outer step alignment.
6653 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6655 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
6656 with offset_alignment.
6657 (DR_ALIGNED_TO): Delete.
6658 (DR_OFFSET_ALIGNMENT): New macro.
6659 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
6660 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
6661 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
6662 (create_data_ref): Likewise.
6663 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6664 (vect_analyze_data_refs): Likewise.
6665 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
6666 creating dummy innermost behavior.
6668 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6670 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
6671 with a "innermost_loop_behavior *" and refeence tree.
6672 * tree-data-ref.c (dr_analyze_innermost): Likewise.
6673 (create_data_ref): Update call accordingly.
6674 * tree-predcom.c (find_looparound_phi): Likewise.
6676 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6678 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
6679 fields with dr_wrt_vec_loop.
6680 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
6681 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
6682 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
6683 (vect_dr_behavior): New function.
6684 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6685 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
6686 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
6687 track whether the step preserves the misalignment.
6688 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6689 Use vect_dr_behavior.
6690 (vect_setup_realignment): Update call accordingly.
6691 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
6692 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
6693 call to vect_create_addr_base_for_vector_ref.
6694 (vect_create_cond_for_align_checks): Likewise.
6695 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
6696 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
6697 (vect_recog_mask_conversion_pattern): Likewise.
6698 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
6699 (new_stmt_vec_info): Remove redundant zeroing.
6701 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6703 * common/config/arm/arm-common.c (arm_be8_option): New function.
6704 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
6705 (ISA_ARMv6): Add isa_bit_be8.
6706 * config/arm/arm.h (arm_be8_option): Add prototype.
6707 (BE8_SPEC_FUNCTION): New define.
6708 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
6709 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
6710 (mlittle-endian): Similarly.
6711 (mbe8, mbe32): New options.
6712 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
6713 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
6715 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6717 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
6719 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6721 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
6722 (cleanup_tree_cfg_bb): Use it.
6723 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
6725 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
6727 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6730 * loop-doloop.c (add_test): Update profile.
6732 2017-07-03 Martin Liska <mliska@suse.cz>
6735 * sanopt.c (rewrite_usage_of_param): New function.
6736 (sanitize_rewrite_addressable_params): Likewise.
6737 (pass_sanopt::execute): Call rewrite_usage_of_param.
6739 2017-07-03 Richard Biener <rguenther@suse.de>
6741 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
6742 back to using VIEW_CONVERT_EXPR.
6744 2017-07-03 Martin Liska <mliska@suse.cz>
6747 * doc/extend.texi: Document when a resolver function is
6748 generated for target_clones.
6750 2017-07-03 Martin Liska <mliska@suse.cz>
6752 * asan.c (asan_emit_stack_protection): Unpoison just red zones
6753 and shadow memory of auto variables which are subject of
6754 use-after-scope sanitization.
6755 (asan_expand_mark_ifn): Add do set only when is_poison.
6757 2016-07-03 Richard Biener <rguenther@suse.de>
6759 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
6761 (vect_force_simple_reduction): Record reduction def -> phi mapping.
6762 (vectorizable_reduction): Perform reduction PHI creation when
6763 visiting a reduction PHI and adjust and simplify code generation
6764 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
6765 (vect_transform_loop): Visit reduction PHIs.
6766 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
6767 defs into the SLP tree.
6768 (vect_build_slp_tree): Reduction defs terminate the recursion.
6769 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
6771 (vect_get_vec_defs_for_stmt_copy): Export.
6772 (vect_get_vec_defs): Likewise.
6773 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
6775 (vect_get_vec_defs_for_stmt_copy): Declare.
6776 (vect_get_vec_defs): Likewise.
6778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6780 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
6781 parameter with a "loop" parameter and use it instead of the
6782 loop containing DR_STMT. Don't check simple_iv when doing
6783 BB analysis. Describe the two analysis modes in the comment.
6785 2017-07-03 Tom de Vries <tom@codesourcery.com>
6787 PR tree-optimization/69468
6788 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
6789 (find_same_succ_bb): Handle ignore_edge_flags.
6791 2017-07-03 Tom de Vries <tom@codesourcery.com>
6793 PR tree-optimization/81192
6794 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
6796 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
6798 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
6800 2017-07-03 Tom de Vries <tom@codesourcery.com>
6802 PR tree-optimization/81192
6803 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
6804 BB_SAME_SUCC (bb) == NULL.
6806 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6808 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
6811 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6813 * dumpfile.c: Include profile-count.h
6814 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
6816 (insert_cond_bb): Update profile.
6817 * tree-cfg.h (insert_cond_bb): Update prototype.
6818 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
6819 * tree-dump.c: Do not include tree-cfg.
6821 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6823 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
6825 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6827 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
6830 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6832 * tree-complex.c (expand_complex_div_wide): update profile.
6834 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6835 Alan Hayward <alan.hayward@arm.com>
6836 David Sherwood <david.sherwood@arm.com>
6838 * Makefile.in (MACHMODE_H): Remove insn-modes.h
6839 (CORETYPES_H): New define.
6840 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
6841 (insn-modes-inline.h, s-modes-inline-h): New rules.
6842 (generated_files): Add insn-modes-inline.h.
6843 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
6844 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
6845 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
6846 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
6847 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
6848 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
6849 (build/gencodes.o, build/genconditions.o): Likewise.
6850 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
6851 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
6852 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
6853 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
6854 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
6855 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
6856 * coretypes.h: Include everything up to real.h for generators.
6857 Include insn-modes.h first. Include wide-int-print.h after
6858 wide-int.h. Include insn-modes-inline.h and then machmode.h.
6859 * machmode.h: Don't include insn-modes.h here.
6860 * function-tests.c: Remove includes of signop.h, machmode.h,
6861 double-int.h and wide-int.h.
6863 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
6865 * optc-save-gen.awk: Likewise.
6866 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
6867 * godump.c: Remove include of wide-int-print.h.
6868 * pretty-print.h: Likewise.
6869 * wide-int-print.cc: Likewise.
6870 * wide-int.cc: Likewise.
6871 * hash-map-tests.c: Remove include of signop.h.
6872 * hash-set-tests.c: Likewise.
6873 * rtl-tests.c: Likewise.
6874 * mkconfig.sh: Remove include of machmode.h.
6875 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
6877 (emit_insn_modes_inline_h): ...this new function. Emit the code
6878 into an insn-modes-inline.h header file, adding appropriate
6879 include guards and end comments.
6880 (emit_insn_modes_c_header): Remove include of machmode.h.
6881 (emit_min_insn_modes_c_header): Include coretypes.h rather than
6883 (main): Handle -i flag and call emit_insn_modes_inline_h when
6886 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6888 * tree-ssa-strlen.c (strinfo): Rename the length field to
6889 nonzero_chars. Add a full_string_p field.
6890 (compare_nonzero_chars, zero_length_string_p): New functions.
6891 (get_addr_stridx): Add an offset_out parameter.
6892 Use compare_nonzero_chars.
6893 (get_stridx): Update accordingly. Use compare_nonzero_chars.
6894 (new_strinfo): Update after above changes to strinfo.
6895 (set_endptr_and_length): Set full_string_p.
6896 (get_string_length): Update after above changes to strinfo.
6897 (unshare_strinfo): Update call to new_strinfo.
6898 (maybe_invalidate): Likewise.
6899 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
6900 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
6901 as a uhwi instead of an shwi. Update after above changes to
6902 strinfo and new_strinfo.
6903 (zero_length_string): Assert that chainsi contains full strings.
6904 Use zero_length_string_p. Update call to new_strinfo.
6905 (adjust_related_strinfos): Update after above changes to strinfo.
6906 Copy full_string_p from origsi.
6907 (adjust_last_stmt): Use zero_length_string_p.
6908 (handle_builtin_strlen): Update after above changes to strinfo and
6909 new_strinfo. Install the lhs as the string length if the previous
6910 entry didn't describe a full string.
6911 (handle_builtin_strchr): Update after above changes to strinfo
6913 (handle_builtin_strcpy): Likewise.
6914 (handle_builtin_strcat): Likewise.
6915 (handle_builtin_malloc): Likewise.
6916 (handle_pointer_plus): Likewise.
6917 (handle_builtin_memcpy): Likewise. Track nonzero characters
6918 that aren't necessarily followed by a nul terminator.
6919 (handle_char_store): Likewise.
6921 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6923 PR tree-optimization/80769
6924 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
6925 for malloc and calloc. Document the new invariant that all related
6926 strinfos have delayed lengths or none do.
6927 (verify_related_strinfos): Move earlier in file.
6928 (set_endptr_and_length): New function, split out from...
6929 (get_string_length): ...here. Also set the lengths of related
6931 (zero_length_string): Assert that chainsi has known (rather than
6933 (adjust_related_strinfos): Likewise.
6935 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6937 PR tree-optimization/81136
6938 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
6939 assert that two references with the same misalignment have the same
6940 compile-time misalignment if those compile-time misalignments
6943 2017-07-01 Andi Kleen <ak@linux.intel.com>
6945 * print-tree.c (print_node): Print all attributes.
6947 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6949 * cfg.c (scale_bbs_frequencies): New function.
6950 * cfg.h (scale_bbs_frequencies): Declare it.
6951 * cfgloopanal.c (single_likely_exit): Cleanup.
6952 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
6954 (scale_loop_profile): Likewise.
6955 (loop_version): Likewise.
6956 (create_empty_loop_on_edge): Update.
6957 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
6958 scale_loop_frequencies, scale_loop_profile, loopify,
6959 loop_version): Update prototypes.
6960 * modulo-sched.c (sms_schedule): Update.
6961 * predict.c (unlikely_executed_edge_p): Also check probability.
6962 (probably_never_executed_edge_p): Fix typo.
6963 * tree-if-conv.c (version_loop_for_if_conversion): Update.
6964 * tree-parloops.c (gen_parallel_loop): Update.
6965 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
6966 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6967 * tree-ssa-loop-split.c (split_loop): Update.
6968 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
6969 * tree-vect-loop-manip.c (vect_do_peeling): Update.
6970 (vect_loop_versioning): Update.
6971 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6973 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6975 * trans-mem.c (split_bb_make_tm_edge): Update profile.
6977 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6979 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
6980 to keep profile consistent.
6982 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6984 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
6985 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
6986 * profile-count.h (max_safe_multiplier): Make unsigned.
6987 (profile_count::guessed_zero): New.
6989 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6991 * bb-reorder.c (fix_up_crossing_landing_pad,
6992 fix_crossing_conditional_branches): Use make_single_succ_edge
6993 to keep profile consistent.
6995 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6997 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7000 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7003 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7004 the right scopes, make sure cond_jump isn't preserved between multiple
7005 iterations. Search for fallthru edge whenever there are 3+ edges and
7006 use find_fallthru_edge for it.
7008 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7010 Patch by Alexander Monakov <amonakov@ispras.ru>
7011 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7012 probabilities consistently.
7014 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7016 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7017 emit_cmp_and_jump_insns.
7019 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7022 * tree-inline.c (expand_call_inline): Combine profile statuses.
7024 2017-06-30 Andrew Pinski <apinski@cavium.com>
7026 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7027 fold_stmt returned true.
7029 2017-06-30 Nathan Sidwell <nathan@acm.org>
7031 * ggc.h (empty_string): Delete.
7032 * cfgexpand.c (expand_asm_stmt): Use plain "".
7033 * optabs.c (expand_asm_memory_barrier): Likewise.
7034 * stringpool.c (empty_string): Delete.
7035 (digit_vector, digit_string): Delete.
7036 (ggc_alloc_string): Use plain "", don't optimize single digit
7037 strings. Use ggc_alloc_atomic.
7039 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7041 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7042 comparison set and one other set, use the cost of the non-comparison
7045 2017-06-30 Nathan Sidwell <nathan@acm.org>
7047 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7050 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7052 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7053 loops. Remove now unneeded calls to gimple_switch_set_label() that
7054 just set removed labels to NULL_TREE.
7056 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7058 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7059 (set_range_info): ...here. Only call set_range_info_raw if domain
7061 (set_nonzero_bits): Call set_range_info_raw.
7062 * tree-ssanames.h (set_range_info_raw): New.
7064 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7067 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7068 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7069 of nonimmediate_operand and <store_mask_constraint> instead of m
7070 for the input operand. For V8FI iterator, always split if input
7071 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7072 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7073 <store_mask_predicate> instead of register_operand and
7074 <store_mask_constraint> instead of v for the input operand. Make
7075 sure both operands aren't MEMs for if not <mask_applied>.
7077 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7079 * lto-wrapper.c (copy_file) Close both file descriptors before
7082 2017-06-30 Martin Liska <mliska@suse.cz>
7085 * multiple_target.c (create_dispatcher_calls): Make ifunc
7086 also for function that don't have calls or are not referenced.
7088 2017-06-30 Richard Biener <rguenther@suse.de>
7090 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7091 analyze the first scalar stmt. Move vector type computation
7092 for the BB case here from ...
7093 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7094 live operation processing in the SLP case properly.
7096 2017-06-30 Richard Biener <rguenther@suse.de>
7098 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7100 2017-06-30 Martin Liska <mliska@suse.cz>
7103 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7104 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7106 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7108 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7111 2017-06-30 Richard Biener <rguenther@suse.de>
7113 PR tree-optimization/81249
7114 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7115 condition reduction result to original scalar type.
7117 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7119 * profile-count.h (enum profile_quality): Fix typos and whitespace
7122 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7124 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7125 type for branch probabilities.
7127 2017-06-29 Julian Brown <julian@codesourcery.com>
7128 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7130 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7131 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7132 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7133 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7135 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7137 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7138 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7139 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7140 CC usage from generic code to here.
7141 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7142 CC usage into the target macros.
7144 2017-06-29 Maya Rashish <coypu@sdf.org>
7146 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7149 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7151 * arm/arm-builtins.c: Include profile-count.h
7152 * except.c (sjlj_emit_function_enter): Use
7153 profile_probability::unlikely.
7155 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7157 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7158 and tocrel_offset be pointer args rather than implicitly using
7160 (legitimate_constant_pool_address_p, rs6000_emit_move,
7161 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7162 tocrel_offset and use in toc_relative_expr_p call.
7163 (print_operand, print_operand_address): Use static tocrel_base_oac
7164 and tocrel_offset_oac.
7165 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7168 2017-06-29 Maya Rashish <coypu@sdf.org>
7170 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7172 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7174 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7175 objects, take into account only the alignment of 'op0' and 'mode1' if
7178 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7180 * ccmp.c (ccmp_tree_comparison_p): New function.
7181 (ccmp_candidate_p): Update to use above function.
7182 (get_compare_parts): New function.
7183 (expand_ccmp_next): Update to use new functions.
7184 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7186 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7187 take mode as argument.
7188 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7189 * expr.c (expand_expr_real_1): Pass mode as argument.
7191 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7193 * combine.c (combine_instructions): Print insns to dump_file, together
7196 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7198 * asan.c (asan_emit_stack_protection): Update.
7199 (create_cond_insert_point): Update.
7200 * auto-profile.c (afdo_propagate_circuit): Update.
7201 * basic-block.h (struct edge_def): Turn probability to
7202 profile_probability.
7203 (EDGE_FREQUENCY): Update.
7204 * bb-reorder.c (find_traces_1_round): Update.
7205 (better_edge_p): Update.
7206 (sanitize_hot_paths): Update.
7207 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7208 (make_single_succ_edge): Update.
7209 (check_bb_profile): Update.
7210 (dump_edge_info): Update.
7211 (update_bb_profile_for_threading): Update.
7212 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7213 probabilitycount to 0.
7214 * cfgbuild.c (compute_outgoing_frequencies): Update.
7215 * cfgcleanup.c (try_forward_edges): Update.
7216 (outgoing_edges_match): Update.
7217 (try_crossjump_to_edge): Update.
7218 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7219 (expand_gimple_tailcall): Update.
7220 (construct_init_block): Use make_single_succ_edge.
7221 (construct_exit_block): Use make_single_succ_edge.
7222 * cfghooks.c (verify_flow_info): Update.
7223 (redirect_edge_succ_nodup): Update.
7224 (split_edge): Update.
7225 (account_profile_record): Update.
7226 * cfgloopanal.c (single_likely_exit): Update.
7227 * cfgloopmanip.c (scale_loop_profile): Update.
7228 (set_zero_probability): Remove.
7229 (duplicate_loop_to_header_edge): Update.
7230 * cfgloopmanip.h (loop_version): Update prototype.
7231 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7232 (force_nonfallthru_and_redirect): Update.
7233 (update_br_prob_note): Update.
7234 (rtl_verify_edges): Update.
7235 (purge_dead_edges): Update.
7236 (rtl_lv_add_condition_to_bb): Update.
7237 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7238 * cgraphunit.c (init_lowered_empty_function): Update.
7239 (cgraph_node::expand_thunk): Update.
7240 * cilk-common.c: Include profile-count.h
7241 * dojump.c (inv): Remove.
7242 (jumpifnot): Update.
7243 (jumpifnot_1): Update.
7244 (do_jump_1): Update.
7246 (do_jump_by_parts_greater_rtx): Update.
7247 (do_compare_rtx_and_jump): Update.
7248 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7249 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7250 * dwarf2cfi.c: Include profile-count.h
7251 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7252 (sjlj_emit_dispatch_table): Likewise.
7253 * explow.c: Include profile-count.h
7254 * expmed.c (emit_store_flag_force): Update.
7255 (do_cmp_and_jump): Update.
7256 * expr.c (compare_by_pieces_d::generate): Update.
7257 (compare_by_pieces_d::finish_mode): Update.
7258 (emit_block_move_via_loop): Update.
7259 (store_expr_with_bounds): Update.
7260 (store_constructor): Update.
7261 (expand_expr_real_2): Update.
7262 (expand_expr_real_1): Update.
7263 * expr.h (try_casesi, try_tablejump): Update prototypes.
7264 * gimple-pretty-print.c (dump_probability): Update.
7265 (dump_profile): New.
7266 (dump_gimple_label): Update.
7267 (dump_gimple_bb_header): Update.
7268 * graph.c (draw_cfg_node_succ_edges): Update.
7269 * hsa-gen.c (convert_switch_statements): Update.
7270 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7271 (find_if_case_1): Update.
7272 (find_if_case_2): Update.
7273 * internal-fn.c (expand_arith_overflow_result_store): Update.
7274 (expand_addsub_overflow): Update.
7275 (expand_neg_overflow): Update.
7276 (expand_mul_overflow): Update.
7277 (expand_vector_ubsan_overflow): Update.
7278 * ipa-cp.c (good_cloning_opportunity_p): Update.
7279 * ipa-split.c (split_function): Use make_single_succ_edge.
7280 * ipa-utils.c (ipa_merge_profiles): Update.
7281 * loop-doloop.c (add_test): Update.
7282 (doloop_modify): Update.
7283 * loop-unroll.c (compare_and_jump_seq): Update.
7284 (unroll_loop_runtime_iterations): Update.
7285 * lra-constraints.c (lra_inheritance): Update.
7286 * lto-streamer-in.c (input_cfg): Update.
7287 * lto-streamer-out.c (output_cfg): Update.
7288 * mcf.c (adjust_cfg_counts): Update.
7289 * modulo-sched.c (sms_schedule): Update.
7290 * omp-expand.c (expand_omp_for_init_counts): Update.
7291 (extract_omp_for_update_vars): Update.
7292 (expand_omp_ordered_sink): Update.
7293 (expand_omp_for_ordered_loops): Update.
7294 (expand_omp_for_generic): Update.
7295 (expand_omp_for_static_nochunk): Update.
7296 (expand_omp_for_static_chunk): Update.
7297 (expand_cilk_for): Update.
7298 (expand_omp_simd): Update.
7299 (expand_omp_taskloop_for_outer): Update.
7300 (expand_omp_taskloop_for_inner): Update.
7301 * omp-simd-clone.c (simd_clone_adjust): Update.
7302 * optabs.c (expand_doubleword_shift): Update.
7303 (expand_abs): Update.
7304 (emit_cmp_and_jump_insn_1): Update.
7305 (expand_compare_and_swap_loop): Update.
7306 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7307 * predict.c (predictable_edge_p): Update.
7308 (edge_probability_reliable_p): Update.
7309 (set_even_probabilities): Update.
7310 (combine_predictions_for_insn): Update.
7311 (combine_predictions_for_bb): Update.
7312 (propagate_freq): Update.
7313 (estimate_bb_frequencies): Update.
7314 (force_edge_cold): Update.
7315 * profile-count.c (profile_count::dump): Add missing space into dump.
7316 (profile_count::debug): Add newline.
7317 (profile_count::differs_from_p): Explicitly convert to unsigned.
7318 (profile_count::stream_in): Update.
7319 (profile_probability::dump): New member function.
7320 (profile_probability::debug): New member function.
7321 (profile_probability::differs_from_p): New member function.
7322 (profile_probability::differs_lot_from_p): New member function.
7323 (profile_probability::stream_in): New member function.
7324 (profile_probability::stream_out): New member function.
7325 * profile-count.h (profile_count_quality): Rename to ...
7326 (profile_quality): ... this one.
7327 (profile_probability): New.
7328 (profile_count): Update.
7329 * profile.c (compute_branch_probabilities): Update.
7330 * recog.c (peep2_attempt): Update.
7331 * sched-ebb.c (schedule_ebbs): Update.
7332 * sched-rgn.c (find_single_block_region): Update.
7333 (compute_dom_prob_ps): Update.
7334 (schedule_region): Update.
7335 * sel-sched-ir.c (compute_succs_info): Update.
7336 * stmt.c (struct case_node): Update.
7337 (do_jump_if_equal): Update.
7338 (get_outgoing_edge_probs): Update.
7339 (conditional_probability): Update.
7340 (emit_case_dispatch_table): Update.
7341 (expand_case): Update.
7342 (expand_sjlj_dispatch_table): Update.
7343 (emit_case_nodes): Update.
7344 * targhooks.c: Update.
7345 * tracer.c (better_p): Update.
7346 (find_best_successor): Update.
7347 * trans-mem.c (expand_transaction): Update.
7348 * tree-call-cdce.c: Update.
7349 * tree-cfg.c (gimple_split_edge): Upate.
7350 (move_sese_region_to_fn): Upate.
7351 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7352 * tree-eh.c (lower_resx): Upate.
7353 (cleanup_empty_eh_move_lp): Upate.
7354 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7355 * tree-inline.c (copy_edges_for_bb): Update.
7356 (copy_cfg_body): Update.
7357 * tree-parloops.c (gen_parallel_loop): Update.
7358 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7359 (gimple_gen_time_profiler): Update.
7360 * tree-ssa-dce.c (remove_dead_stmt): Update.
7361 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7362 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7363 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7364 (unloop_loops): Update.
7365 (try_peel_loop): Update.
7366 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7367 * tree-ssa-loop-split.c (connect_loops): Update.
7368 (split_loop): Update.
7369 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7370 (hoist_guard): Update.
7371 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7372 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7373 (value_replacement): Update.
7374 * tree-ssa-reassoc.c (branch_fixup): Update.
7375 * tree-ssa-tail-merge.c (replace_block_by): Update.
7376 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7377 (create_edge_and_update_destination_phis): Update.
7378 (compute_path_counts): Update.
7379 (recompute_probabilities): Update.
7380 (update_joiner_offpath_counts): Update.
7381 (freqs_to_counts_path): Update.
7382 (duplicate_thread_path): Update.
7383 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7384 (struct switch_conv_info): Update.
7385 (gen_inbound_check): Update.
7386 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7387 (vect_do_peeling): Update.
7388 (vect_loop_versioning): Update.
7389 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7390 (optimize_mask_stores): Update.
7391 * ubsan.c (ubsan_expand_null_ifn): Update.
7392 * value-prof.c (gimple_divmod_fixed_value): Update.
7393 (gimple_divmod_fixed_value_transform): Update.
7394 (gimple_mod_pow2): Update.
7395 (gimple_mod_pow2_value_transform): Update.
7396 (gimple_mod_subtract): Update.
7397 (gimple_mod_subtract_transform): Update.
7398 (gimple_ic): Update.
7399 (gimple_stringop_fixed_value): Update.
7400 (gimple_stringops_transform): Update.
7401 * value-prof.h: Update.
7403 2017-06-29 Carl Love <cel@us.ibm.com>
7405 * config/rs6000/rs6000-c.c: Add support for built-in functions
7406 vector signed int vec_signed (vector float);
7407 vector signed long long vec_signed (vector double);
7408 vector signed int vec_signed2 (vector double, vector double);
7409 vector signed int vec_signede (vector double);
7410 vector signed int vec_signedo (vector double);
7411 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
7412 instruction generator.
7413 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7414 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
7415 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
7417 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
7418 vunsignede_v2df): Add define_expands.
7419 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
7420 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
7421 VEC_UNSIGNEDO): Add definitions.
7422 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7423 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
7424 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
7425 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
7426 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
7427 * config/rs6000/altivec.h (vec_signed, vec_signed2,
7428 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
7429 vec_unsignede, vec_unsignedo): Add builtin defines.
7430 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
7432 * doc/extend.texi: Update the built-in documentation file for the
7433 new built-in functions.
7435 2017-06-29 Richard Biener <rguenther@suse.de>
7437 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
7438 reduction chains to LOOP_VINFO_REDUCTIONS.
7439 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
7440 SLP reductions after processing reduction chains.
7442 2017-06-29 Nathan Sidwell <nathan@acm.org>
7444 * builtins.c (fold_builtin_FUNCTION): Use
7445 lang_hooks.decl_printable_name.
7447 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
7450 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
7451 with only one label.
7452 * stmt.c (expand_case): Assert NCASES is greater than one.
7454 2017-06-29 Richard Biener <rguenther@suse.de>
7456 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
7458 (group_case_labels): Likewise.
7459 (find_taken_edge): Push sanity checking on val to workers...
7460 (find_taken_edge_cond_expr): ... here
7461 (find_taken_edge_switch_expr): ... and here, handle cases
7462 with just a default label.
7463 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
7464 (group_case_labels): Likewise.
7465 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
7466 group_case_labels does anything cleanup the CFG again.
7468 2017-06-29 Bin Cheng <bin.cheng@arm.com>
7470 PR tree-optimization/81196
7471 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
7472 exit condition comparing two IVs.
7474 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
7476 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
7477 profile to the dummy entry at the end of the list of architectures.
7478 * config/arm/arm-cpu-cdata.h: Regenerated.
7480 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7481 Michael Collison <michael.collison@arm.com>
7484 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
7486 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
7487 (*aarch64_reg_<mode>3_minus_mask): New pattern.
7488 (*aarch64_<optab>_reg_di3_mask2): New pattern.
7489 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
7490 of shift when the shift amount is masked with constant equal to
7491 the size of the mode.
7492 * config/aarch64/predicates.md (subreg_lowpart_operator): New
7495 2017-06-29 Martin Liska <mliska@suse.cz>
7497 * config/i386/i386.opt: Change range from [1,5] to [0,5].
7499 2017-06-29 Yury Gribov <tetra2005@gmail.com>
7502 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
7505 (edge_growth_cache_entry::edge_growth_cache_entry): New
7507 (reset_edge_growth_cache): Update to use constructor.
7509 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7511 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
7512 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7513 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7515 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
7517 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
7518 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
7520 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
7522 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
7523 (*-linux-uclibc*): Add t-uclibc tmake_file.
7524 * config/t-musl: New.
7525 * config/t-uclibc: New.
7527 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
7529 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
7531 (gen_comm_data): Emit architectural setting of arch_prof.
7532 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
7534 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
7535 (armv8-m.base, armv8-m.main): Likewise.
7536 * arm-protos.h (arm_build_target): Add profile field.
7537 (arch_option): Likewise.
7538 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
7540 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
7541 arm_active_target.profile.
7543 2017-06-28 Richard Biener <rguenther@suse.de>
7546 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
7547 TYPE_OVERFLOW_WRAPS.
7548 * match.pd (negate_expr_p): Likewise.
7549 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
7550 fold_build2, not fold_binary.
7552 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7554 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
7555 Convert memory address to Pmode.
7556 (aarch64_print_operand): Assert MEM operands are always Pmode.
7558 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7561 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
7562 Remove redundant if.
7563 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
7564 * config/arm/aarch-common-protos.h
7565 (aarch_forward_to_shift_is_not_shifted_re): Remove.
7566 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
7568 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
7571 * multiple_target.c (create_dispatcher_calls): Set the default
7572 clone to be static, not public.
7574 2017-06-28 Richard Biener <rguenther@suse.de>
7576 * tree-vect-loop.c (vectorizable_reduction): Move special
7577 cond reduction IV var creation ...
7578 (vect_create_epilog_for_reduction): ... here. Remove induction_index
7579 parameter. Use STMT_VINFO_VECTYPE.
7580 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
7583 2017-06-28 Martin Liska <mliska@suse.cz>
7586 * ipa-visibility.c (non_local_p): Handle visibility.
7588 2017-06-28 Martin Liska <mliska@suse.cz>
7591 * common.opt: Add IntegerRange to various options.
7592 * opt-functions.awk (integer_range_info): New function.
7593 * optc-gen.awk: Add integer_range_info to cl_options struct.
7594 * opts-common.c (decode_cmdline_option): Handle
7595 CL_ERR_INT_RANGE_ARG.
7596 (cmdline_handle_error): Likewise.
7597 * opts.c (print_filtered_help): Show valid interval in
7598 when --help is provided.
7599 * opts.h (struct cl_option): Add range_min and range_max fields.
7600 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
7602 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
7604 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
7605 (x * C EQ/NE y * C): New transformation.
7607 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
7609 * genmultilib (combination_space): Accept '+' in option names.
7611 2017-06-28 Martin Liska <mliska@suse.cz>
7614 * asan.c (instrument_derefs): Bail out inner references
7615 that are hard register variables.
7617 2017-06-28 Jakub Jelinek <jakub@redhat.com>
7620 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
7621 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
7623 2017-06-28 Richard Biener <rguenther@suse.de>
7625 * tree-vectorizer.h (vect_get_vec_defs): Remove.
7626 (vect_get_slp_defs): Adjust.
7627 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
7629 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
7631 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7632 get_initial_defs_for_reduction instead of vect_get_vec_defs.
7633 (vectorizable_reduction): Adjust.
7634 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
7636 (vect_get_slp_defs): Likewise.
7637 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
7638 (vectorizable_bswap): Adjust.
7639 (vectorizable_call): Likewise.
7640 (vectorizable_conversion): Likewise.
7641 (vectorizable_assignment): Likewise.
7642 (vectorizable_shift): Likewise.
7643 (vectorizable_operation): Likewise.
7644 (vectorizable_store): Likewise.
7645 (vectorizable_condition): Likewise.
7646 (vectorizable_comparison): Likewise.
7648 2017-06-28 Michael Collison <michael.collison@arm.com>
7651 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
7653 (arm_gen_movmemqi): Removed unused variable 'i'.
7654 Convert 'for' loop into 'while' loop.
7655 (arm_expand_prologue): Remove last unnecessary set of insn.
7656 (thumb_pop): Remove unused variable 'pushed_words'.
7657 (thumb_exit): Remove last unnecessary set of regs_to_pop.
7659 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7661 * config/s390/predicates.md: Use s390_rel_address_ok_p.
7662 * config/s390/s390-protos.h: Add prototype of
7663 s390_rel_address_ok_p.
7664 * config/s390/s390.c (s390_got_symbol): New function.
7665 (s390_rel_address_ok_p): New function.
7666 (legitimize_pic_address): Use s390_rel_address_ok_p.
7667 (s390_load_got): Use s390_got_symbol.
7668 (s390_option_override): Issue error if
7669 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
7670 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
7672 * config/s390/s390.opt: New option mpic-data-is-text-relative.
7674 2017-06-27 Andrew Pinski <apinski@cavium.com>
7676 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
7677 (X * copysign (1.0, X)): New pattern.
7678 (X * copysign (1.0, -X)): New pattern.
7679 (copysign (-1.0, CST)): New pattern.
7681 2017-06-27 Joseph Myers <joseph@codesourcery.com>
7683 * genmultilib (combination_space): Remove variable.
7684 Validate reuse rules against regular expression for any sequence
7685 of multilib options in any order.
7687 2017-06-27 Michael Collison <michael.collison@arm.com>
7689 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
7690 call aarch64_split_simd_combine.
7691 * (aarch64_combine_internal<mode>): Delete pattern.
7692 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
7693 Allow register and subreg operands.
7695 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7697 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
7698 specific need, just fallback on defaults.
7699 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
7701 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7702 Olivier Hainque <hainque@adacore.com>
7704 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
7706 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
7707 targets. Pick a default if no particular attempt applied.
7708 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
7711 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7713 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
7714 (x86_64-wrs-vxworks7): Likewise.
7716 2017-06-27 Marek Polacek <polacek@redhat.com>
7719 * ubsan.c (instrument_null): Check get_base_address's result for null.
7721 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
7723 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
7725 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
7727 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
7728 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
7729 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
7730 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
7732 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
7733 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
7734 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
7735 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
7736 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
7737 BUILT_IN_FEUPDATEENV): New builtins.
7738 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
7739 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
7740 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
7741 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
7743 (builtin_structptr_types): Adjust size.
7744 * tree.c (builtin_structptr_types): Add four entries.
7746 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7747 Olivier Hainque <hainque@adacore.com>
7749 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
7750 (TLS_SYM): New local macro, forcing reference to __tls__ on
7751 link command lines for VxWorks 7 RTPs, triggering initialization
7753 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
7754 OS features TLS support, true for RTPs on VxWorks 7.
7755 * config/vxworks.c (vxworks_override_options): Setup emutls
7758 2017-06-27 Jakub Jelinek <jakub@redhat.com>
7760 * predict.c (test_prediction_value_range): Use -1U instead of -1
7761 to avoid narrowing conversion warning.
7762 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
7763 to avoid narrowing conversion warning.
7764 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
7766 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
7768 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7770 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
7771 64bit configurations.
7772 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
7773 (SIZE_TYPE): Likewise.
7774 * config/vxworks.c (vxworks_emutls_var_fields): Use
7775 long_unsigned_type_node instead of unsigned_type_node as the offset
7776 field type, which is "pointer" mode in emutls.c.
7778 2017-06-27 Jakub Jelinek <jakub@redhat.com>
7781 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
7784 * gimple-fold.c (replace_call_with_call_and_fold): Handle
7785 gimple_vuse copying separately from gimple_vdef copying.
7787 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7789 * value-prof.c (free_hist): Remove call to memset and the enclosing if
7792 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
7793 Olivier Hainque <hainque@adacore.com>
7795 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
7796 for all vxworks7 targets.
7797 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
7798 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
7799 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
7800 variations for VX6/VX7 and 32/64bits later on in ...
7801 (VXWORKS_LIB_SPEC): Leverage new macros.
7802 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
7803 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
7805 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
7807 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
7808 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
7810 2017-06-26 Carl Love <cel@us.ibm.com>
7812 * config/rs6000/rs6000-c.c: Add support for built-in functions
7813 vector bool char vec_reve (vector bool char);
7814 vector signed char vec_reve (vector signed char);
7815 vector unsigned char vec_reve (vector unsigned char);
7816 vector bool int vec_reve (vector bool int);
7817 vector signed int vec_reve (vector signed int);
7818 vector unsigned int vec_reve (vector unsigned int);
7819 vector bool long long vec_reve (vector bool long long);
7820 vector signed long long vec_reve (vector signed long long);
7821 vector unsigned long long vec_reve (vector unsigned long long);
7822 vector bool short vec_reve (vector bool short);
7823 vector signed short vec_reve (vector signed short);
7824 vector double vec_reve (vector double);
7825 vector float vec_reve (vector float);
7826 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
7827 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
7828 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
7829 (altivec_vreve): New pattern.
7830 * config/rs6000/altivec.h (vec_reve): New define.
7831 * doc/extend.texi (vec_rev): Update the built-in documentation file
7832 for the new built-in functions.
7834 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7836 PR tree-optimization/71815
7837 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
7839 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
7841 (slsr_process_phi): Likewise.
7842 (replace_uncond_cands_and_profitable_phis): Don't replace a
7843 multiply candidate with a stride of 1 (copy or cast).
7844 (phi_incr_cost): Call uses_consumed_by_stmt rather than
7846 (lowest_cost_path): Likewise.
7847 (total_savings): Likewise.
7849 2017-06-26 Richard Biener <rguenther@suse.de>
7852 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
7853 Use def_builtin_pure for all gather builtins.
7855 2017-06-26 Richard Biener <rguenther@suse.de>
7857 PR tree-optimization/81203
7858 * tree-tailcall.c (find_tail_calls): Do not move stmts into
7861 2017-06-26 Marek Polacek <polacek@redhat.com>
7864 * doc/invoke.texi: Document -Wmultistatement-macros.
7866 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
7868 * doc/sourcebuild.texi (ARM-specific attributes): Document new
7869 arm_neon_ok_no_float_abi effective target.
7871 2017-06-26 Richard Biener <rguenther@suse.de>
7873 PR tree-optimization/80928
7874 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
7875 (copy_bbs): Set BB_DUPLICATED flag early.
7876 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
7878 (execute_on_shrinking_pred): Likewise.
7879 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
7880 BB_DUPLICATED blocks.
7881 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
7882 iterate over all PHIs considering removal of *gsi.
7884 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
7886 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
7889 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7891 * config/rs6000/rs6000-string.c: (expand_block_clear,
7892 do_load_for_compare, select_block_compare_mode,
7893 compute_current_alignment, expand_block_compare,
7894 expand_strncmp_align_check, expand_strn_compare,
7895 expand_block_move, rs6000_output_load_multiple)
7896 Move functions related to string/block move/compare
7898 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
7899 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
7900 for this function which is now used in two files.
7901 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
7902 * config.gcc: Add rs6000-string.o to extra_objs for
7903 targets powerpc*-*-* and rs6000*-*-*.
7905 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
7908 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
7909 32-bit, since indexed is not valid for DImode.
7910 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
7911 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
7912 (define_peephole2 for Altivec d-form load): Add 32-bit support.
7913 (define_peephole2 for Altivec d-form store): Likewise.
7916 * multiple_target.c (create_dispatcher_calls): Only create the
7917 dispatcher call if the function is the default clone of a
7920 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
7923 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
7924 a call, force the call to not be a tail call.
7926 2017-06-23 Jeff Law <law@redhat.com>
7928 * doc/contrib.texi: Add entry for Steven Pemberton's work on
7931 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
7933 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
7934 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
7935 handling for early expansion of vector shifts (sl,sr,sra,rl).
7936 (builtin_function_type): Add vector shift right instructions
7937 to the unsigned argument list.
7939 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7941 rtl-optimizatoin/79286
7942 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
7943 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
7944 trap. PIC register plus a const unspec without offset can never trap.
7946 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
7948 * tree.h (builtin_structptr_type): New type.
7949 (builtin_structptr_types): Declare new array.
7950 * tree.c (builtin_structptr_types): New array.
7951 (free_lang_data, build_common_tree_nodes): Use it.
7953 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
7956 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
7959 2017-06-22 Matt Turner <mattst88@gmail.com>
7961 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
7962 Lake models to skylake case. Assume skylake for unknown
7963 models with clflushopt.
7965 2017-06-22 Jeff Law <law@redhat.com>
7967 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
7968 frame sizes that do not satisfy aarch64_uimm12_shift.
7970 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
7972 * profile-count.h (apply_probability,
7973 apply_scale, probability_in): Fix checks for zero.
7975 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7977 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
7978 * doc/cppdiropts.texi (-I @var{dir}): Document it.
7980 2016-06-22 Richard Biener <rguenther@suse.de>
7982 * tree-vect-loop.c (vect_model_reduction_cost): Handle
7983 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
7984 REDUC_MAX_EXPR support.
7985 (vectorizable_reduction): Likewise.
7986 (vect_create_epilog_for_reduction): Likewise.
7988 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
7990 * match.pd (A / (1 << B) -> A >> B): New.
7991 * generic-match-head.c: Include optabs-tree.h.
7992 * gimple-match-head.c: Likewise.
7993 * optabs-tree.h (target_supports_op_p): New.
7994 * optabs-tree.c (target_supports_op_p): New.
7996 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7998 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
7999 $gcc_cv_ld --help output.
8000 (gcc_cv_ld_demangle): Likewise.
8001 (gcc_cv_ld_eh_frame_hdr): Likewise.
8002 (gcc_cv_ld_pie): Likewise.
8003 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8004 (gcc_cv_ld_buildid): Likewise.
8005 (gcc_cv_ld_sysroot): Likewise.
8006 (ld_bndplt_support): Likewise.
8007 (ld_pushpopstate_support): Likewise.
8008 * configure: Regenerate.
8009 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8011 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8014 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8015 operands indexes to avoid gap between operands and match_dups.
8017 2017-06-21 Andrew Pinski <apinski@cavium.com>
8019 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8020 Increment Arith_shift and Arith_shift_reg by 1.
8021 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8023 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8024 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8025 (aarch64_strip_extend): Add new argument and test for it.
8026 (aarch64_cheap_mult_shift_p): New function.
8027 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8028 add a cost if it is true.
8029 Update calls to aarch64_strip_extend.
8030 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8032 2017-06-21 Andrew Pinski <apinski@cavium.com>
8034 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8036 (thunderxt88): Likewise.
8037 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8038 (thunderx_prefetch_tune): New variable.
8039 (thunderx2t99_prefetch_tune): Update for the correct values.
8040 (thunderxt88_tunings): New variable.
8041 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8042 generic_prefetch_tune.
8043 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8045 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8047 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8048 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8049 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8050 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8051 (aarch64_atomic_cas<mode>, GPI): Likewise.
8053 2017-06-21 Martin Liska <mliska@suse.cz>
8055 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8056 statements on cold and hot labels.
8057 * predict.c (tree_estimate_probability_bb): Remove the
8058 prediction from this place.
8060 2017-06-21 Martin Liska <mliska@suse.cz>
8062 PR tree-optimization/79489
8063 * gimplify.c (maybe_add_early_return_predict_stmt): New
8065 (gimplify_return_expr): Call the function.
8066 * predict.c (tree_estimate_probability_bb): Remove handling
8068 * predict.def: Update comment about early return predictor.
8069 * gimple-predict.h (is_gimple_predict): New function.
8070 * predict.def: Change default value of early return to 66.
8071 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8073 * passes.def: Put pass_strip_predict_hints to the beginning of
8076 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8078 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8079 FUNCTION_DECL declarations.
8080 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8082 (dwaf2out_decl): Likewise.
8083 * godump.c (go_early_global_decl): Skip call to the real debug hook
8084 for FUNCTION_DECL declarations.
8085 * passes.c (rest_of_decl_compilation): Skip call to the
8086 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8087 -fdump-go-spec is passed.
8089 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8091 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8092 Reorder for compactness.
8093 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8094 (def_builtin_pure, def_builtin_pure2): New functions.
8095 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8097 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8099 * match.pd (nop_convert): New predicate.
8100 ((A +- CST1) +- CST2): Allow some NOP conversions.
8102 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8105 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8106 with ctors/dtors if GOVD_SHARED is set.
8108 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8110 * config/aarch64/aarch64.md (movti_aarch64):
8111 Emit mov rather than orr.
8112 (movtf_aarch64): Likewise.
8113 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8114 Emit mov rather than orr.
8116 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8118 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8119 Swap alternatives, make integer dup more expensive.
8121 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8123 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8124 Return true for non-tls symbols.
8126 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8128 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8129 (cortex-a75): Likewise.
8130 (cortex-a75.cortex-a55): Likewise.
8131 * config/aarch64/aarch64-tune.md: Regenerate.
8132 * doc/invoke.texi (-mtune): Document new values for -mtune.
8134 2017-06-21 Tom de Vries <tom@codesourcery.com>
8136 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8138 (Effective-Target Keywords, Other attributes): Suggest using
8139 dg-add-options stack_size feature to get stack limit in stack_size
8140 effective target documentation.
8142 2017-06-21 Julian Brown <julian@codesourcery.com>
8143 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8145 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8146 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8147 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8149 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8150 attribute type list for neon_multiply.
8151 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8152 attribute type list for neon_multiply.
8153 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8154 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8155 attribute type list for neon_multiply.
8156 * config/arm/types.md (crypto_pmull): Add.
8157 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8158 attribute type list.
8160 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8162 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8165 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8167 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8168 to make sure not to dereference a NULL cost_classes_ptr pointer.
8170 2017-06-20 Carl Love <cel@us.ibm.com>
8172 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8173 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8174 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8175 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8176 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8177 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8178 VMULOSW): New enum "unspec" values.
8179 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8180 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8181 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8182 altivec_vmulosw): New patterns.
8183 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8184 VMULOSW): Add definitions.
8186 2017-06-20 Julia Koval <julia.koval@intel.com>
8188 * config/i386/i386.c: Fix rounding expand for new pattern.
8189 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8191 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8193 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8194 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8196 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8198 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8201 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8203 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8204 architecture, then by implementer ID.
8205 * config/aarch64/aarch64-tune.md: Regenerate.
8207 2017-06-20 Richard Biener <rguenther@suse.de>
8210 * fold-const.c (split_tree): Fold to type before negating.
8212 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8214 * diagnostic-show-locus.c
8215 (selftest::test_fixit_deletion_affecting_newline): New function.
8216 (selftest::diagnostic_show_locus_c_tests): Call it.
8218 2017-06-20 Andreas Schwab <schwab@suse.de>
8221 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8224 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8226 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8227 __ARM_FEATURE_COPROC according to support.
8229 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8231 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8232 Rewritten to avoid overflow for > 32-bit pointers.
8235 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8236 by removing enum keyword.
8237 (ubsan_type_descriptor): Likewise. Formatting fix.
8240 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8241 splitter): Require TARGET_SSE2 in the condition.
8243 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8246 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8247 for doing vector set of SFmode on ISA 3.0.
8248 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8249 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8251 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8252 SFmode value into a V4SF variable that was extracted from another
8253 V4SF variable without converting the element to double precision
8254 and back to single precision vector format.
8255 (vsx_insert_extract_v4sf_p9_2): Likewise.
8257 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8259 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8260 in UWHI to avoid undefined overflow.
8263 * ubsan.h (enum ubsan_encode_value_phase): New.
8264 (ubsan_encode_value): Change second argument to
8265 enum ubsan_encode_value_phase with default value of
8266 UBSAN_ENCODE_VALUE_GENERIC.
8267 * ubsan.c (ubsan_encode_value): Change second argument to
8268 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8269 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8270 create_tmp_var_raw instead of create_tmp_var and use a
8272 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8273 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8274 ubsan_encode_value callers.
8277 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8278 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8279 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8281 2017-06-19 Richard Biener <rguenther@suse.de>
8284 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8285 estimates if we changed anything.
8287 2017-06-19 Richard Biener <rguenther@suse.de>
8289 PR tree-optimization/80887
8290 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8291 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8292 simplified lookups, then reset mprts_hook.
8293 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8295 (try_to_simplify): Likewise.
8297 2017-06-19 Martin Liska <mliska@suse.cz>
8300 * gimplify.c (gimplify_switch_expr):
8301 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8303 2017-06-19 Martin Liska <mliska@suse.cz>
8305 * doc/install.texi: Document that PGO runs in 4 stages.
8307 2017-06-19 Martin Liska <mliska@suse.cz>
8310 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8311 to dispatcher function name.
8312 * multiple_target.c (replace_function_decl): New function.
8313 (create_dispatcher_calls): Redirect both edges and references.
8315 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8317 * profile-count.c (profile_count::dump): Dump quality.
8318 (profile_count::differs_from_p): Update for unsigned val.
8319 * profile-count.h (profile_count_quality): New enum.
8320 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8322 2017-06-19 Richard Biener <rguenther@suse.de>
8324 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8325 struct function as arg.
8326 (estimate_numbers_of_iterations): Export overload with loop arg.
8327 (free_numbers_of_iterations_estimates_loop): Use an overload of
8328 free_numbers_of_iterations_estimates instead.
8329 * tree-cfg.c (remove_bb): Adjust.
8330 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8331 * tree-parloops.c (gen_parallel_loop): Likewise.
8332 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8334 (tree_unroll_loops_completely): Likewise.
8335 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8336 Use an overload instead and export.
8337 (estimated_loop_iterations): Adjust.
8338 (max_loop_iterations): Likewise.
8339 (likely_max_loop_iterations): Likewise.
8340 (estimate_numbers_of_iterations): Take struct function as arg
8342 (loop_exits_before_overflow): Adjust.
8343 (free_numbers_of_iterations_estimates_loop): Use an overload.
8344 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8345 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8347 2017-06-19 Richard Biener <rguenther@suse.de>
8350 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8351 RANGE_EXPR conservatively.
8353 2017-06-16 Carl Love <cel@us.ibm.com>
8355 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8356 definitions for vec_float, vec_float2, vec_floato,
8357 vec_floate built-ins.
8358 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8359 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8361 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8362 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8363 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8364 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8365 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8366 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8367 vec_floato): Add builtin defines.
8368 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8369 Update the built-in documentation file for the new built-in
8372 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8374 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8375 (mthumb): Mark as the negative of -marm.
8377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8379 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8381 (ARM Options, -mtune): Document that this accepts the same
8382 extension options as -mcpu.
8383 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8387 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8388 permitted extensions.
8390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8392 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8393 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8394 (armv8-m.main): Add option +nodsp.
8395 * config/arm/arm-cpu-cdata.h: Regenerated.
8397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8399 * config/arm/t-fuchsia: New file.
8400 * config.gcc (arm*-*-fuchsia*): Use it.
8402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8404 * config/arm/t-symbian: Rewrite for new option infrastructure.
8406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8408 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
8409 (MULTILIB_REQUIRED): Likewise.
8411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8413 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
8414 (MULTILIB_RESUE): Likewise.
8415 (MULTILIB_MATCHES): Likewise.
8416 (MULTLIB_REQUIRED): Likewise.
8418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8420 * config/arm/t-rtems: Rewrite for new option framework.
8422 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8424 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
8425 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
8426 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
8427 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
8428 * config/arm/t-multilib: ... here.
8429 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
8430 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
8431 armv7-a and armv8*-a when A-profile libraries have not been built.
8432 * config/arm/t-rmprofile: Rewrite.
8434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8436 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
8437 with a backslash. Remove the backslash after substituting unescaped
8439 * doc/fragments.texi (MULTILIB_REUSE): Document it.
8441 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8443 * config.gcc: (arm*-*-*): When building a-profile libraries, force
8444 the driver to pass through the default setting of -mfloat-abi.
8445 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
8447 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
8448 (all_feat_combs): New rule.
8449 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
8451 * config/arm/t-aprofile: Rewrite.
8453 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8455 * config/arm/arm.h (FPUTYPE_AUTO): Define.
8456 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
8457 fpu is not specified by the user/command-line.
8458 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
8459 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
8460 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
8461 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
8462 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
8463 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
8465 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8467 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
8468 * config/arm/t-arm-elf: Rewritten.
8470 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8472 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
8473 have some floating-point instructions.
8474 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
8475 (TARGET_MAYBE_HARD_FLOAT): New macro.
8476 * config/arm/arm-builtins.c (arm_init_builtins): Use
8477 TARGET_MAYBE_HARD_FLOAT.
8478 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
8480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8482 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
8483 (configargs.h): Include it.
8484 (arm_print_hint_for_fpu_option): New function.
8485 (arm_parse_fpu_option): New function.
8486 (candidate_extension): New class.
8487 (arm_canon_for_multilib): New function.
8488 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
8489 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
8490 (ARCH_CANONICAL_SPECS): New macro.
8491 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
8493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8495 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
8496 are set after handling multilib fragments. Set target_cpu_default2
8499 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8501 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
8503 (arm*-*-*): Set target_cpu_default2 to a quoted string.
8504 * config/arm/parsecpu.awk (check_cpu): Validate any extension
8506 (check_arch): Likewise.
8507 * config/arm/arm.c (arm_configure_build_target): Handle
8508 TARGET_CPU_DEFAULT being a string constant. Scan any feature
8509 options in the default.
8511 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8513 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
8514 when an option is an alias of another.
8515 * config/arm/parsecpu.awk (optalias): New parser token.
8516 (gen_comm_data): Mark non-alias options as such. Emit entries
8517 for extension aliases.
8518 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
8519 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
8520 (armv6kz, armv6zk, armv6t2): Likewise.
8521 (armv7): Make vfpv3-d16 an alias.
8522 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
8524 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
8525 Sort in canonical order.
8526 (armv8-a): Sort in canonical order.
8527 (armv8.1-a, armv8.2-a): Likewise.
8528 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
8530 (cortex-a9): Sort in canonical order.
8531 * config/arm/arm.c (selftests.h): Include it.
8532 (arm_test_cpu_arch_data): New function.
8533 (arm_run_self_tests): New function.
8534 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
8535 (targetm): Move declaration to the end of the file.
8536 * arm-cpu-cdata.h: Regenerated.
8538 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8540 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
8541 call to target_mode_check describing the type of option passed.
8542 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
8543 (arm_target_thumb_only): Use arm_parse_arch_option_name or
8544 arm_parse_cpu_option_name to match parameters against list of
8546 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
8547 arm_arch_core_flags data structure.
8548 * config/arm/arm-cpu_cdata.h: Regenerated.
8550 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8552 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
8554 (arm_print_hint_for_cpu_option): Likewise.
8555 (arm_print_hint_for_arch_option): Likewise.
8556 (arm_parse_cpu_option_name): Likewise.
8557 (arm_parse_arch_option_name): Likewise.
8558 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
8559 of entries in the all_fpus list.
8560 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
8561 (arm_parse_cpu_option_name): Declare.
8562 (arm_parse_arch_option_name): Declare.
8563 (arm_parse_option_features): Declare.
8564 (arm_intialize_isa): Declare.
8565 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
8567 (gen_comm_data): ... here. Make definitions non-static.
8568 * config/arm/arm-cpu-data.h: Regenerated.
8569 * config/arm/arm-cpu-cdata.h: Regenerated.
8571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8573 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
8574 (cpu_arch_extension): New structure.
8575 (cpu_arch_option, arch_option, cpu_option): New structures.
8576 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
8578 (gen_data): Generate new format data tables.
8579 * config/arm/arm.c (cpu_tune): New structure.
8580 (cpu_option, processors): Delete.
8581 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
8582 (arm_print_hint_for_cpu_option): ... this and ...
8583 (arm_print_hint_for_arch_option): ... this.
8584 (arm_parse_arch_cpu_name): Delete. Replace with ...
8585 (arm_parse_cpu_option_name): ... this and ...
8586 (arm_parse_arch_option_name): ... this.
8587 (arm_unrecognized_feature): Change type of target parameter to
8589 (arm_parse_arch_cpu_features): Delete. Replace with ...
8590 (arm_parse_option_features): ... this.
8591 (arm_configure_build_target): Rework to use new configuration data
8593 (arm_print_tune_info): Rework for new configuration data tables.
8594 * config/arm/arm-cpu-data.h: Regenerated.
8595 * config/arm/arm-cpu.h: Regenerated.
8597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8599 * Makefile.in (OBJS): Move sbitmap.o from here ...
8600 (OBJS-libcommon): ... to here.
8602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8604 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
8605 (ISA_ALL_CRYPTO): New macro.
8606 (ISA_ALL_SIMD): New macro
8607 (ISA_ALL_FP): New macro.
8608 * config/arm/arm.c (fpu_bitlist): Update initializer.
8609 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
8611 (arm9e): Add fpu. Add option for nofp
8612 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
8613 (arm926ej-s, arm1026ej-s): Likewise.
8614 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
8615 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
8616 neon-fp16, neon-vfpv4, nofp and nosimd.
8617 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
8618 (cortex-a8): Add fpu. Add option for nofp.
8619 (cortex-a9): Add fpu. Add options for nosimd and nofp.
8620 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
8621 (cortex-r4f): Add fpu.
8622 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
8623 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
8625 (cortex-r8): Likewise.
8626 (cortex-m4): Add fpu. Add option for nofp.
8627 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
8628 (cortex-a17.cortex-a7): Likewise.
8629 (cortex-a32): Add fpu. Add options for crypto and nofp.
8630 (cortex-a35, cortex-a53): Likewise.
8631 (cortex-a57): Add fpu. Add option for crypto.
8632 (cortex-a72, cortex-a73): Likewise.
8633 (exynos-m1): Likewise.
8634 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
8635 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
8636 (cortex-m33): Add fpu. Add option for nofp.
8637 * config/arm/arm-cpu-cdata.h: Regenerated
8638 * config/arm/arm-cpu-data.h: Regenerated.
8640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8642 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
8643 (armv5te, armv5tej): Likewise.
8644 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
8645 (armv7): Add options fp and vfpv3-d16.
8646 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
8647 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
8649 (armv7ve): Likewise.
8650 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
8651 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
8652 (armv8-a): Add nocrypto option.
8653 (armv8.1-a, armv8.2-a): Likewise.
8654 (armv8-m.main): add options fp, fp.dp and nofp.
8656 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8658 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
8660 (armv8-a+crc): Delete.
8661 (armv8.1-a): Add options simd, crypto and nofp.
8662 (armv8.2-a): Add options fp16, simd, crypto and nofp.
8663 (armv8.2-a+fp16): Delete.
8664 (armv8-m.main): Add option dsp.
8665 (armv8-m.main+dsp): Delete.
8666 (cortex-a8): Add fpu. Add nofp option.
8667 (cortex-a9): Add fpu. Add nofp and nosimd options.
8668 * config/arm/parsecpu.awk (gen_data): Generate option tables and
8669 link to main cpu and architecture data structures.
8670 (gen_comm_data): Only put isa attributes from the main architecture
8672 (option): New statement for architecture and CPU entries.
8673 * arm.c (struct cpu_option): New structure.
8674 (struct processors): Add entry for options.
8675 (arm_unrecognized_feature): New function.
8676 (arm_parse_arch_cpu_name): Ignore any characters after the first
8678 (arm_parse_arch_cpu_feature): New function.
8679 (arm_configure_build_target): Separate out any CPU and architecture
8680 features and parse separately. Don't error out if -mfpu=auto is
8681 used with only an architecture string.
8682 (arm_print_asm_arch_directives): New function.
8683 (arm_file_start): Call it.
8684 * config/arm/arm-cpu-cdata.h: Regenerated.
8685 * config/arm/arm-cpu-data.h: Likewise.
8686 * config/arm/arm-tables.opt: Likewise.
8688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8690 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
8691 assembler when it is not -mfpu=auto.
8693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8695 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
8696 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
8697 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
8698 (ASM_CPU_SPEC): Rewrite.
8699 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
8700 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
8701 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
8702 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
8703 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
8704 copied string is NUL-terminated. Also strip any characters prefixed
8706 (arm_rewrite_selected_arch): New function.
8707 (arm_rewrite_march): New function.
8709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8711 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
8712 (x_arm_cpu_string, x_arm_tune_string): Likewise.
8713 (march, mcpu, mtune): Convert to string-based options.
8714 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
8715 (arm_parse_arch_cpu_name): New function.
8716 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
8717 identify selected architecture or CPU.
8718 (arm_option_save): New function.
8719 (TARGET_OPTION_SAVE): Redefine.
8720 (arm_option_restore): Restore string options.
8721 (arm_option_print): Print string options.
8723 2017-06-16 Martin Sebor <msebor@redhat.com>
8725 PR tree-optimization/80933
8726 PR tree-optimization/80934
8727 * builtins.c (fold_builtin_3): Do not handle bcmp here.
8728 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
8729 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
8730 (gimple_fold_builtin): Call them.
8732 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8734 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
8735 as unlikely; update profile.
8737 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8739 * predict.c (force_edge_cold): Handle declaring edges impossible
8742 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8744 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
8746 (try_unroll_loop_completely): Fix reporting.
8748 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8750 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
8752 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
8755 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
8756 if given a non-constant argument for an intrinsic which requires a
8759 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8761 * profile.c (compare_freqs): New function.
8762 (branch_prob): Sort edge list.
8763 (find_spanning_tree): Assume that the list is priority sorted.
8765 2017-06-16 Richard Biener <rguenther@suse.de>
8767 PR tree-optimization/81090
8768 * passes.def (pass_record_bounds): Remove.
8769 * tree-pass.h (make_pass_record_bounds): Likewise.
8770 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
8771 make_pass_record_bounds): Likewise.
8772 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
8773 not free niter estimates at the beginning but at the end.
8774 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
8776 2017-06-16 Richard Biener <rguenther@suse.de>
8778 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
8779 initializer to workaround ICE in host GCC 4.8.
8781 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8783 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
8785 (clone_inlined_nodes): Update.
8787 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8789 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
8790 prefetch settings, and enable prefetching by default at -O3.
8792 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8794 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8795 Set flag_prefetch_loop_arrays according to tuning data.
8797 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8799 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
8801 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
8802 [Unrelated to main purpose of the patch] Place the pointer field last
8803 to enable type checking errors when tune structure are wrongly merged.
8804 * config/aarch64/aarch64.c (generic_prefetch_tune,)
8805 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
8806 (thunderx2t99_prefetch_tune): New tune constants.
8807 (tune_params *_tunings): Update all tunings (no functional change).
8808 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
8809 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
8810 from tunings structures.
8812 2017-06-16 Jakub Jelinek <jakub@redhat.com>
8815 * ubsan.c (instrument_null): Add T argument, use it instead
8816 of computing it based on IS_LHS.
8817 (instrument_object_size): Likewise.
8818 (pass_ubsan::execute): Adjust instrument_null and
8819 instrument_object_size callers to pass gimple_get_lhs or
8820 gimple_assign_rhs1 result to it. Use instrument_null instead of
8821 calling get_base_address and instrument_mem_ref. Handle
8822 aggregate call arguments for object-size sanitization.
8824 2017-06-16 Yury Gribov <tetra2005@gmail.com>
8826 PR tree-optimization/81089
8827 * tree-vrp.c (is_masked_range_test): Validate operands of
8830 2017-06-15 Martin Sebor <msebor@redhat.com>
8833 * dumpfile.c (dump_register): Avoid calling memset to initialize
8834 a class with a default ctor.
8835 * gcc.c (struct compiler): Remove const qualification.
8836 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
8837 * hash-table.h: Ditto.
8838 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
8840 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
8841 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
8843 * params.h (struct param_info): Make struct members non-const.
8844 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
8845 with default initialization.
8846 * vec.h (vec_copy_construct, vec_default_construct): New helper
8848 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
8849 with vec_copy_construct.
8850 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
8851 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
8852 * doc/invoke.texi (-Wclass-memaccess): Document.
8854 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8856 * emit-rtl.h (is_leaf): Update comment about local
8859 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8862 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
8863 for a variable to have a section before checking if the section has a
8865 Set section to.persistent if persistent attribute is set.
8866 Warn if .persistent attribute is used on an automatic variable.
8868 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
8870 PR rtl-optimization/80474
8871 * reorg.c (update_block): Do not ignore instructions in a delay slot.
8873 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
8875 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
8878 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
8880 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
8881 (casesi): Emit bounds checking as RTL.
8882 (casesi_internal_mips16_<mode>): Remove bounds checking.
8884 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
8886 * config/xtensa/xtensa.c (xtensa_option_override): Append
8887 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
8888 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
8889 xtensa_doloop_hooks): Define unconditionally.
8890 (xtensa_reorg_loops): Only call reorg_loops in the presence of
8892 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
8893 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
8894 for it in xtensa_option_override.
8895 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
8896 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
8898 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
8900 * doc/cppopts.texi: Document '-' special value to -MF.
8902 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
8904 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
8905 (cortex_a53_fconst): Likewise.
8906 (cortex_a53_fpmul): Likewise.
8907 (cortex_a53_f_load_64): Likewise.
8908 (cortex_a53_f_load_many): Likewise.
8909 (cortex_a53_advsimd_alu): Likewise.
8910 (cortex_a53_advsimd_alu_q): Likewise.
8911 (cortex_a53_advsimd_mul): Likewise.
8912 (cortex_a53_advsimd_mul_q): Likewise.
8913 (fpmac bypass): Add new bypass for fpmac-fpmac case.
8914 Add missing fmul, r2f_cvt and fconst cases.
8916 2017-06-14 Richard Biener <rguenther@suse.de>
8919 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
8921 (fold_binary_loc): When associating do not treat pre-existing
8922 TREE_OVERFLOW on literal constants as a reason to allow
8923 TREE_OVERFLOW on associated literal constants.
8925 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
8927 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
8928 (MASK_FEATURES): New macro.
8929 * config/sparc/sparc.c (sparc_option_override): Remove the special
8930 handling of -mfpu and generalize it to all MASK_FEATURES switches.
8932 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
8934 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
8935 a division of 0 if non-call exceptions are enabled.
8937 2017-06-14 Andrew Pinski <apinski@cavium.com>
8938 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8941 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
8942 Improve vector initialization code gen for only variable case.
8944 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
8946 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
8948 2017-06-14 Richard Biener <rguenther@suse.de>
8950 PR tree-optimization/81083
8951 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
8954 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8956 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
8957 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
8958 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
8959 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
8960 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
8961 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
8963 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8965 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
8966 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
8968 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8970 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
8972 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8974 * config/rs6000/t-rtems: Don't handle SPE.
8976 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8978 * config/rs6000/t-linux: Don't handle SPE.
8980 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8982 * config/rs6000/eabispe.h: Delete file.
8984 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8986 * config/rs6000/t-spe: Delete file.
8988 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8990 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
8991 (rs6000_legitimate_offset_address_p): Return false for anything in
8992 V2SImode or V2SFmode.
8994 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8996 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
8997 except V2SF and V2SI. Rearrange the vector modes, and add comments.
8998 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9000 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9001 (rs6000_legitimate_offset_address_p): Ditto.
9002 (rs6000_emit_move): Ditto.
9003 (rs6000_init_builtins): Remove V4HI_type_node.
9005 2017-06-13 Martin Liska <mliska@suse.cz>
9008 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9009 (gate_asan): Likewise.
9010 * asan.h (asan_no_sanitize_address_p): Remove the function.
9011 (sanitize_flags_p): New function.
9012 * builtins.def: Fix coding style.
9013 * common.opt: Use renamed enum value.
9014 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9015 * doc/extend.texi: Document no_sanitize attribute.
9016 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9017 to SANITIZE_UNDEFINED_NONDEFAULT.
9018 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9019 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9020 Use sanitize_flags_p.
9021 * gimplify.c (gimplify_function_tree): Likewise.
9022 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9023 * opts.c (parse_no_sanitize_attribute): New function.
9024 (common_handle_option): Use renamed enum value.
9025 * opts.h (parse_no_sanitize_attribute): Declare.
9026 * tree.c (sanitize_flags_p): New function.
9027 * tree.h: Declared here.
9028 * tsan.c: Use sanitize_flags_p.
9029 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9030 (instrument_mem_ref): Likewise.
9031 (instrument_bool_enum_load): Likewise.
9032 (do_ubsan_in_current_function): Remove the function.
9033 (pass_ubsan::execute): Use sanitize_flags_p.
9034 * ubsan.h: Remove do_ubsan_in_current_function
9035 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9036 (dump_function_to_file): Use it here.
9038 2017-06-13 Martin Jambor <mjambor@suse.cz>
9040 PR tree-optimization/80803
9041 PR tree-optimization/81063
9042 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9043 (propagate_subaccesses_across_link): Enqueue subtree whenever
9044 necessary instead of relying on the caller.
9046 2017-06-13 Martin Jambor <mjambor@suse.cz>
9048 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9049 that have a first_link.
9050 (sort_and_splice_var_accesses): Do not check first_link before
9052 (subtree_mark_written_and_enqueue): Likewise.
9053 (propagate_all_subaccesses): Likewise and do not stop at first
9054 parent with a first_link.
9056 2017-06-13 Martin Jambor <mjambor@suse.cz>
9058 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9061 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9063 * match.pd: New pattern.
9065 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9067 * tree-vrp.c (is_masked_range_test): New function.
9068 (register_edge_assert_for): Determine ranges for
9071 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9073 PR tree-optimization/67328
9074 * fold-const.c (maskable_range_p): New function.
9075 (build_range_check): Generate bittests if possible.
9077 2017-06-13 Martin Liska <mliska@suse.cz>
9079 * gimple-pretty-print.c (dump_probability): Add new argument.
9080 (dump_edge_probability): Dump both probability and count.
9081 (dump_gimple_label): Likewise.
9082 (dump_gimple_bb_header): Likewise.
9084 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9087 * config/avr/avr-devices.c: Fix indentation.
9088 * config/avr/gen-avr-mmcu-specs.c: Dito.
9090 2017-06-13 Richard Biener <rguenther@suse.de>
9092 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9093 instead get vector type from stmt_info.
9094 (vectorizable_reduction): Adjust. Remove dead code.
9096 2017-06-13 Richard Biener <rguenther@suse.de>
9099 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9100 case of C * (x * C2 + C3).
9101 (fold_addr_of_array_ref_difference): Properly fold index difference.
9103 2017-06-12 David S. Miller <davem@davemloft.net>
9106 * config/sparc/sparc.md (return expander): Emit frame blockage if
9107 function uses alloca.
9109 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9111 * combine.c (make_field_assignment): Check len rather than the mode
9112 precision when calling force_to_mode.
9114 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9116 Support multilibs and devices that see flash in RAM address range.
9119 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9120 (avr_mcu_t) <flash_pm_offset>: New field.
9121 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9122 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9123 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9124 (AVR_TINY_PM_OFFSET): Remove macro.
9125 * config/avr/avr.opt (-mshort-calls): New option.
9126 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9127 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9128 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9129 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9130 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9131 instead of avr_arch->have_jmp_call.
9132 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9133 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9134 avr_arch->flash_pm_offset to define.
9135 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9136 new field flash_pm_offset. Add entry for avrxmega3.
9137 (avr_texinfo): Add entry for avrxmega3.
9138 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9139 attiny212, attiny214,
9140 attiny412, attiny414, attiny416, attiny417,
9141 attiny814, attiny816, attiny817,
9142 attiny1614, attiny1616, attiny1617,
9143 attiny3214, attiny3216, attiny3217.
9144 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9145 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9146 (avr_print_operand_address) [AVR_TINY]: Same.
9147 (avr_asm_init_sections) <readonly_data_section>: Only patch
9148 callback if avr_arch->flash_pm_offset = 0.
9149 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9150 for rodata if avr_arch->flash_pm_offset != 0.
9151 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9152 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9153 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9154 (m_options): Append opt_rcall.
9155 (m_dirnames): Append dir_rcall.
9156 * config/avr/t-multilib: Regenerate.
9158 * configure.ac [target=avr]: Check whether avrxmega3 default
9159 linker description file works as needed.
9160 * configure: Regenerate.
9161 * doc/avr-mmcu.texi: Regenerate.
9162 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9163 <__AVR_ARCH__>: Document avrxmega3 and 103.
9164 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9165 <__AVR_SHORT_CALLS__>: Document it.
9166 <__AVR_PM_BASE_ADDRESS__>: Document it.
9167 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9168 (AVR Variable Attributes) <progmem>: Document this is
9169 not needed for avrxmega3.
9170 (AVR Named Address Spaces) <__flash>: Dito.
9172 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9174 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9176 2017-06-12 Doug Rupp <rupp@adacore.com>
9178 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9179 Append vxworks-stdint.h to the tm_file list.
9180 * config/vxworks-stdint.h: New file.
9182 2017-06-12 Martin Liska <mliska@suse.cz>
9184 PR tree-optimization/81041
9185 * tree-profile.c (gimple_gen_ic_func_profiler):
9186 Create an extra BB in profile-generate
9187 (gimple_gen_time_profiler): Likewise.
9189 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9191 PR tree-optimization/81003
9192 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9193 (update_range_test): Use it instead of force_gimple_operand_gsi.
9195 2017-06-12 Richard Biener <rguenther@suse.de>
9197 PR tree-optimization/81053
9198 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9199 with backedge value not defined in loop. Simplify def stmt
9202 2017-06-11 Tom de Vries <tom@codesourcery.com>
9205 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9207 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9208 nvptx_cannot_force_const_mem.
9210 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9212 * opts.c (finish_options): Move test for flag_split_stack after
9213 it has been initialized.
9215 2017-06-11 Jason Merrill <jason@redhat.com>
9217 * tree.h (id_equal): New.
9218 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9219 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9220 instead of strcmp of IDENTIFIER_POINTER.
9222 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9224 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9225 (mark_all_inlined_calls_cdtor): Fix formating.
9226 (inline_transform): Rescale profile before inlining.
9228 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9230 * cgraph.h (cgraph_edge::clone): Update prototype.
9231 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9232 (cgraph_node::create_clone): Update.
9233 (cgraph_node::create_version_clone): Update.
9234 * tree-inline.c (copy_bb): Update.
9235 (expand_call_inline): Update.
9237 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9239 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9240 factored out from ...
9241 (rs6000_emit_prologue): ... here.
9243 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9245 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9246 factored out from ...
9247 (rs6000_emit_prologue): ... here.
9249 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9251 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9253 (handle_missing_profiles): Fix computation of tp_first_run.
9254 (counts_to_freqs): Do not touch freqs when count is 0.
9256 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9258 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9261 2017-06-10 Tom de Vries <tom@codesourcery.com>
9263 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9264 attributes): Document signal effective target.
9266 2017-06-10 Tom de Vries <tom@codesourcery.com>
9268 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9269 Document effective target stack_size.
9271 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9273 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9274 to the edit_context if they can be auto-applied.
9276 2017-06-9 Ian Lance Taylor <iant@golang.org>
9278 * opts.c (finish_options): If -fsplit-stack, disable implicit
9279 -forder-blocks-and-partition.
9280 * doc/invoke.texi (Optimize Options): Document that when using
9281 -fsplit-stack -forder-blocks-and-partition is not implicitly
9284 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9286 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9287 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9288 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9289 * builtins.def (abort, trap, unreachable): Declare cold.
9290 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9291 * tree-core.h (ECF_COLD): New.
9292 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9293 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9295 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9297 * predict.c (unlikely_executed_stmt_p): Cleanup.
9299 2017-06-09 Richard Biener <rguenther@suse.de>
9301 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9302 model if the ref is always written to.
9304 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9306 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9308 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9310 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9313 2017-06-09 Tom de Vries <tom@codesourcery.com>
9316 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9317 "target cannot support label values" when encountering LABEL_REF.
9319 2017-06-09 Martin Liska <mliska@suse.cz>
9321 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9322 (gimple_gen_ic_func_profiler): Emit direct comparison
9323 of __gcov_indirect_call_callee with NULL.
9324 (gimple_gen_time_profiler): Change probability from
9325 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9327 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9329 * profile.c (edge_gcov_counts): Turn to pointer.
9330 (compute_branch_probabilities, compute_branch_probabilities): Update.
9331 (branch_prob): Do not clear edge_gcov_count.
9332 * profile.h (edge_gcov_counts): Turn to pointer.
9333 (edge_gcov_count): Update.
9335 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9337 * gimple.h (gimple_check_failed): Mark cold.
9339 2017-06-09 Richard Biener <rguenther@suse.de>
9341 PR tree-optimization/66623
9342 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9343 refactor check_reduction into two parts, properly computing
9344 whether we have to check reduction validity for outer loop
9347 2017-06-09 Richard Biener <rguenther@suse.de>
9349 PR tree-optimization/79483
9350 * graphite-scop-detection.c (order): New global.
9351 (get_order): Compute bb to order mapping that satisfies code
9352 generation constraints.
9353 (cmp_pbbs): New helper.
9354 (build_scops): Start domwalk at entry block, sort generated
9357 2017-06-09 Richard Biener <rguenther@suse.de>
9360 * ipa-polymorphic-call.c
9361 (ipa_polymorphic_call_context::restrict_to_inner_class):
9362 Skip FIELD_DECLs with error_mark_node type.
9363 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9366 2017-06-09 Martin Liska <mliska@suse.cz>
9368 * predict.c (struct branch_predictor): New struct.
9369 (test_prediction_value_range): New test.
9370 (predict_c_tests): New function.
9371 * selftest-run-tests.c (selftest::run_tests): Run the function.
9372 * selftest.h: Declare new tests.
9374 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9377 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9378 gen_add3_insn did not fail.
9379 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9380 r0, construct that number in a temporary reg and add that reg to r0.
9381 If asked to put the result in r0 as well, fail.
9383 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9385 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9386 for early expansion of vec_eqv.
9388 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9391 * ubsan.c (instrument_null): Avoid pointless code temporary.
9392 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9394 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9398 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
9401 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
9404 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9406 * predict.c (unlikely_executed_edge_p): Move ahead.
9407 (probably_never_executed_edge_p): Use it.
9409 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9412 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
9413 gimple_call_builtin_p call.
9415 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9417 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
9418 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
9419 rtl_check_failed_type2, rtl_check_failed_code1,
9420 rtl_check_failed_code2, rtl_check_failed_code_mode,
9421 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
9422 rtvec_check_failed_bounds, rtl_check_failed_flag,
9423 _fatal_insn_not_found, _fatal_insn): Likewise.
9424 * tree.h (tree_contains_struct_check_failed,
9425 tree_check_failed, tree_not_check_failed,
9426 tree_class_check_failed, tree_range_check_failed,
9427 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
9428 tree_vec_elt_check_failed, phi_node_elt_check_failed,
9429 tree_operand_check_failed, omp_clause_check_failed,
9430 omp_clause_operand_check_failed, omp_clause_range_check_failed):
9433 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9435 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
9436 flag_branch_probabilities.
9437 * ipa-inline.c (edge_badness): Likewise.
9438 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9439 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
9440 * predict.c (maybe_hot_frequency_p): Likewise.
9441 (probably_never_executed): Likewise.
9442 * sched-ebb.c (schedule_ebbs): Likewise.
9443 * sched-rgn.c (find_single_block_region): Likewise.
9444 * tracer.c (tail_duplicate): Likewise.
9446 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9448 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
9449 longer requires x_flag_profile_use.
9451 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9453 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
9454 instead of flag_reorder_blocks_and_partition.
9455 * dbxout.c (dbxout_function_end): Likewise.
9456 * dwarf2out.c (gen_subprogram_die): Likewise.
9457 * haifa-sched.c (sched_create_recovery_edges): Likewise.
9458 * hw-doloop.c (reorg_loops): Likewise.
9459 * varasm.c (assemble_start_function,
9460 assemble_end_function): Likewise.
9461 (decide_function_section): Do not check for
9462 flag_reorder_blocks_and_partition.
9464 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9466 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9468 (chkp_get_hard_register_fake_addr_expr): Ditto.
9469 (chkp_build_addr_expr): Add check for hard reg case.
9470 (chkp_parse_array_and_component_ref): Ditto.
9471 (chkp_find_bounds_1): Ditto.
9472 (chkp_process_stmt): Don't generate bounds store for
9475 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9477 * predict.c (maybe_hot_bb_p): Do not check profile status.
9478 (maybe_hot_edge_p): Likewise.
9479 (probably_never_executed): Check for zero counts even if profile
9481 (unlikely_executed_edge_p): New function.
9482 (unlikely_executed_stmt_p): New function.
9483 (unlikely_executed_bb_p): New function.
9484 (set_even_probabilities): Use unlikely predicates.
9485 (combine_predictions_for_bb): Likewise.
9486 (predict_paths_for_bb): Likewise.
9487 (predict_paths_leading_to_edge): Likewise.
9488 (determine_unlikely_bbs): New function.
9489 (estimate_bb_frequencies): Use it.
9490 (compute_function_frequency): Use zero counts even if profile is
9492 * profile-count.h: Fix typo.
9494 2017-08-08 Julia Koval <julia.koval@intel.com>
9496 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
9497 _mm512_mask_cvtsepi16_storeu_epi8,
9498 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
9499 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
9500 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
9501 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
9502 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
9503 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
9504 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
9505 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
9506 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
9507 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
9508 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
9509 __builtin_ia32_pmovuswb256mem_mask,
9510 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
9511 __builtin_ia32_pmovwb512mem_mask): New builtins.
9513 2017-08-08 Julia Koval <julia.koval@intel.com>
9515 PR target/73350,80862
9516 * config/i386/subst.md (round): Fix round pattern.
9517 * config/i386/i386.c (ix86_erase_embedded_rounding):
9518 Fix erasing rounding for the fixed pattern.
9520 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9522 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
9524 2017-06-08 Martin Liska <mliska@suse.cz>
9526 PR gcov-profile/80911
9527 * gcov.c (block_info::block_info): New constructor.
9529 2017-06-07 Carl Love <cel@us.ibm.com>
9531 * config/rs6000/rs6000-c: The return type of the following
9532 built-in functions was implemented as int not long long. Fix sign
9533 of return value for the unsigned version of vec_mulo and vec_mule.
9534 vector unsigned long long vec_bperm (vector unsigned long long,
9535 vector unsigned char)
9536 vector signed long long vec_mule (vector signed int,
9538 vector unsigned long long vec_mule (vector unsigned int,
9539 vector unsigned int)
9540 vector signed long long vec_mulo (vector signed int,
9542 vector unsigned long long vec_mulo (vector unsigned int,
9543 vector unsigned int)
9544 * doc/extend.texi: Fix the documentation for the built-in
9547 2017-06-07 Carl Love <cel@us.ibm.com>
9550 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
9553 2017-06-07 Carl Love <cel@us.ibm.com>
9555 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
9556 support, Generate doublehv for signed int/float for BE case only.
9558 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
9560 * doc/invoke.texi (mcx16): Rewrite.
9562 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9564 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
9565 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
9566 *mov<mode>_softfloat, and an anonymous splitter): Use
9567 nonimmediate_operand instead of rs6000_nonimmediate_operand.
9569 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9571 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
9573 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9574 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
9575 (rs6000_debug_reg_global): Adjust.
9576 (rs6000_init_hard_regno_mode_ok): Adjust.
9577 (rs6000_dbx_register_number): Adjust.
9578 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
9579 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9580 Remove SPE_ACC and SPEFSCR.
9581 (REG_ALLOC_ORDER): Ditto.
9582 (FRAME_POINTER_REGNUM): Change to 111.
9583 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
9584 (REG_CLASS_NAMES): Ditto.
9585 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
9586 (REGISTER_NAMES): Ditto.
9587 (ADDITIONAL_REG_NAMES): Ditto.
9588 (rs6000_reg_names): Ditto.
9589 * config/rs6000/rs6000.md: Renumber some register number
9592 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9594 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
9596 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9597 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
9599 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
9600 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9601 Delete the SPE high registers.
9602 (REG_ALLOC_ORDER): Ditto.
9603 (enum reg_class): Remove SPE_HIGH_REGS.
9604 (REG_CLASS_NAMES): Ditto.
9605 (REG_CLASS_CONTENTS): Delete the SPE high registers.
9606 (REGISTER_NAMES): Ditto.
9607 (rs6000_reg_names): Ditto.
9608 * doc/tm.texi.in: Remove SPE as example.
9609 * doc/tm.texi: Regenerate.
9611 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9613 * config/rs6000/8540.md (ppc8540_brinc): Delete.
9614 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
9615 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
9616 * config/rs6000/rs6000.md (type): Remove "brinc".
9618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9620 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
9621 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
9622 * config/rs6000/linuxspe.h: Delete file.
9623 * config/rs6000/rs6000.md: Don't include spe.md.
9624 * config/rs6000/spe.h: Delete file.
9625 * config/rs6000/spe.md: Delete file.
9626 * config/rs6000/t-rs6000: Remove spe.md.
9628 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9630 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
9631 (reg_or_none500mem_operand): Delete.
9632 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
9633 instead of reg_or_none500mem_operand.
9635 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9637 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
9638 handling of SPE flags.
9639 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
9641 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9643 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
9645 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
9646 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
9647 paired_divv2sf3): Similar.
9648 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
9649 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
9650 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
9652 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
9653 Rename the paired_* instruction patterns.
9654 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
9656 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
9657 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
9658 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
9659 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
9660 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
9662 (struct machine_function): Delete field spe_insn_chain_scanned_p.
9663 (spe_func_has_64bit_regs_p): Delete.
9664 (spe_expand_predicate_builtin): Delete.
9665 (spe_expand_evsel_builtin): Delete.
9666 (TARGET_DWARF_REGISTER_SPAN): Do not define.
9667 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
9668 (invalid_e500_subreg): Delete.
9669 (rs6000_legitimize_address): Always force_reg op2 as well, for
9670 paired single memory accesses.
9671 (rs6000_member_type_forces_blk): Delete.
9672 (rs6000_spe_function_arg): Delete.
9673 (rs6000_expand_unop_builtin): Delete SPE handling.
9674 (rs6000_expand_binop_builtin): Ditto.
9675 (spe_expand_stv_builtin): Delete.
9676 (bdesc_2arg_spe): Delete.
9677 (spe_expand_builtin): Delete.
9678 (spe_expand_predicate_builtin): Delete.
9679 (spe_expand_evsel_builtin): Delete.
9680 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
9681 (spe_init_builtins): Delete.
9682 (spe_func_has_64bit_regs_p): Delete.
9683 (savres_routine_name): Delete "info" parameter. Adjust callers.
9684 (rs6000_emit_stack_reset): Ditto.
9685 (rs6000_dwarf_register_span): Delete.
9686 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
9687 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
9688 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
9690 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
9692 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
9693 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
9694 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
9695 mulv2sf3, divv2sf3): Delete expanders.
9697 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9699 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
9701 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9703 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
9704 * config/rs6000/rs6000.c: Ditto.
9706 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9708 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
9709 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
9710 comparison_operator.
9712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9714 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
9715 * config/rs6000/rs6000.opt: Ditto.
9716 * config/rs6000/t-rtems: Ditto.
9718 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9720 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
9721 TARGET_E500_SINGLE by 0, simplify.
9722 * config/rs6000/rs6000.c: Ditto.
9723 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
9724 (spe_build_register_parallel): Delete.
9725 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
9726 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
9727 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
9728 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
9729 (E500_CONVERT): Delete.
9730 * config/rs6000/spe.md: Remove many patterns and all define_constants.
9732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9734 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
9735 * config/rs6000/dfp.md: Ditto.
9736 (negdd2, *negdd2_fpr): Merge.
9737 (absdd2, *absdd2_fpr): Merge.
9738 (negtd2, *negtd2_fpr): Merge.
9739 (abstd2, *abstd2_fpr): Merge.
9740 * config/rs6000/e500.h: Delete file.
9741 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
9742 TARGET_FPRS by 1 and simplify.
9743 * config/rs6000/rs6000-c.c: Ditto.
9744 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
9746 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
9748 * config/rs6000/rs6000.md: Ditto.
9749 (floatdidf2, *floatdidf2_fpr): Merge.
9750 (move_from_CR_gt_bit): Delete.
9751 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
9752 (E500_CR_IOR_COMPARE): Delete.
9753 (All patterns that require !TARGET_FPRS): Delete.
9754 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
9756 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9758 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
9760 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9762 * graphds.c (add_edge): Intitialize edge's attached data.
9763 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
9764 pointer parameter. Call pointed function on each edge during
9765 graph traversing. Skip traversing the edge when the function
9767 (graphds_dfs, graphds_scc): Ditto.
9768 (for_each_edge): New parameter. Pass the new parameter to callback
9770 * graphds.h (skip_edge_callback): New function pointer type.
9771 (graphds_dfs, graphds_scc): New function pointer parameter.
9772 (graphds_edge_callback, for_each_edge): New parameter.
9774 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9776 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
9777 out code checking if runtime alias check is possible to below ...
9778 Call the new function.
9779 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
9780 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
9782 2017-06-07 Marek Polacek <polacek@redhat.com>
9785 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
9786 TYPE_OVERFLOW_WRAPS check.
9788 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9790 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
9791 if versioning is required.
9792 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
9793 peeling with the check for versioning.
9795 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9797 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
9798 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
9799 Set true to new parameter if new ssa variable is defined.
9800 (vect_gen_vector_loop_niters): Refactor. Set range information
9801 for the new vector loop bound variable.
9802 (vect_do_peeling): Ditto.
9804 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9806 * tree-affine.c (ssa.h): Include header file.
9807 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
9808 has wrapping overflow behavior.
9810 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9812 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
9814 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9816 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
9817 (tree_to_aff_combination): ... here.
9819 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9821 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
9822 reg_pressure model function.
9823 (ivopts_global_cost_for_size): Delete.
9824 (determine_set_costs, iv_ca_recount_cost): Call new model function
9825 ivopts_estimate_reg_pressure.
9827 2017-06-07 Tamar Christina <tamar.christina@arm.com>
9829 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
9830 expensive than udiv. Remove floating point cases from mod.
9832 2017-06-07 Tamar Christina <tamar.christina@arm.com>
9834 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
9837 2017-06-07 Tamar Christina <tamar.christina@arm.com>
9839 * config/aarch64/aarch64.md
9840 (copysignsf3): Fix mask generation.
9842 2017-06-07 Jakub Jelinek <jakub@redhat.com>
9844 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
9846 (class dump_manager): Add register_dumps method.
9847 * dumpfile.c: Include langhooks.h.
9848 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
9849 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
9850 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
9851 (dump_manager::dump_register): Start with 512 entries instead of 32.
9852 (dump_manager::register_dumps): New method.
9853 * toplev.c (general_init): Instead of invoking register_dumps
9854 langhook, invoke register_dumps method on the dump manager.
9855 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
9858 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
9860 * doc/md.texi: Clarify the restrictions on a define_insn condition.
9861 Say that # requires an associated define_split to exist, and that
9862 the define_split must be suitable for use after register allocation.
9864 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9866 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
9867 (compute_outgoing_frequencies): Also initialize zero counts.
9868 (find_many_sub_basic_blocks): Do not produce uninitialized profile
9869 around loops; preserve more of profile when nothing changes.
9871 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
9873 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
9875 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
9876 * config/arm/arm-cpu-cdata.h: Regenerate.
9877 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
9878 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
9879 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
9880 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
9881 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
9883 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
9885 * config/arm/t-rmprofile: Likewise.
9886 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
9888 2017-06-06 David S. Miller <davem@davemloft.net>
9891 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
9892 blockage if function uses alloca.
9894 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9896 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
9897 New "uid" fields to hold pretty-print IDs of group and ref.
9898 Memory references are now identified as <group_id>:<ref_id>
9899 instead of using [random] addresses.
9900 (dump_mem_details): Simplify, no functional change.
9901 (dump_mem_ref): Simplify and make output more concise.
9902 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
9903 (find_or_create_group): Initialize group uid.
9904 (record_ref): Initialize ref uid. Improve debug output.
9905 (prune_group_by_reuse, should_issue_prefetch_p,)
9906 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
9907 (mark_nontemporal_store, determine_loop_nest_reuse):
9908 Improve debug output.
9910 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9912 * dbgcnt.def (prefetch): New debug counter.
9913 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
9914 (schedule_prefetches): Stop issueing prefetches if debug counter
9917 2017-06-06 Tom de Vries <tom@codesourcery.com>
9919 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
9920 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
9922 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9924 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
9925 Use aarch64_reg_or_zero predicate for operand 4.
9926 (aarch64_compare_and_swap<mode> define_insn_and_split):
9927 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
9928 (aarch64_store_exclusive<mode>): Likewise for operand 2.
9930 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9932 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
9933 (arm_compute_save_core_reg_mask): This.
9934 (thumb1_compute_save_reg_mask): Rename into ...
9935 (thumb1_compute_save_core_reg_mask): This.
9936 (arm_compute_save_reg0_reg12_mask): Adapt comment.
9937 (arm_compute_frame_layout): Likewise.
9939 2017-06-06 Richard Biener <rguenther@suse.de>
9941 PR tree-optimization/80974
9942 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
9943 keep or clear leaders SSA info.
9945 2017-06-06 Tom de Vries <tom@codesourcery.com>
9947 * config/nvptx/nvptx.c (split_mode_p): New function.
9948 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
9950 2017-06-06 Tom de Vries <tom@codesourcery.com>
9952 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
9954 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9957 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
9960 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9962 * shrink-wrap.c (handle_simple_exit): Update profile.
9963 (try_shrink_wrapping): Upate profile.
9965 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9967 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
9968 (tree_guess_outgoing_edge_probabilities): New.
9969 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
9970 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
9972 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9974 * ipa-split.c (split_function): Initialize return bb profile.
9976 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9978 * profile.c (compute_branch_probabilities): Also initialize
9981 2017-06-06 Richard Biener <rguenther@suse.de>
9983 PR tree-optimization/80928
9984 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
9985 (vect_analyze_loop_operations): Properly guard analysis for
9987 (vect_transform_loop): Likewise.
9988 (vect_analyze_loop_2): Also reset SLP type on PHIs.
9989 (vect_model_induction_cost): Do not cost for pure SLP.
9990 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
9991 of induction in inner loop vectorization.
9992 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
9993 (vect_get_and_check_slp_defs): Handle vect_induction_def.
9994 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
9996 (vect_analyze_slp_cost_1): Cost induction.
9997 (vect_detect_hybrid_slp_stmts): Handle PHIs.
9998 (vect_get_slp_vect_defs): Likewise.
9999 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10000 (vect_transform_stmt): Handle SLP reductions.
10001 * tree-vectorizer.h (vectorizable_induction): Adjust.
10003 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10005 * config/rs6000/rs6000.c (make_resolver_func): Update
10006 init_lowered_empty_function call.
10008 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10010 * doc/invoke.texi: Document the -fprofile-abs-path option.
10011 * common.opt (fprofile-abs-path): New option.
10012 * gcov-io.h (gcov_write_filename): Declare.
10013 * gcov-io.c (gcov_write_filename): New function.
10014 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10015 * profile.c (output_location): Likewise.
10017 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10019 * shring-wrap.c: Revert accidental commit.
10021 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10023 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10025 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10027 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10029 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10030 profile in callgraph edge.
10031 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10032 (apply_scale): Likewise.
10033 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10034 Also scale profile when inlining function with zero profile.
10035 (initialize_cfun): Update exit block profile even when it is zero.
10036 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10037 when profile is read.
10039 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10041 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10042 (CLONE_*): New constants to define the processors we can generate
10043 code for with the target_clone attribute.
10044 (rs6000_clone_map): New array to identify which clone processors
10045 the current program is running on.
10046 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10047 target_clone attribute.
10048 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10049 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10050 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10051 (cpu_expand_builtin): Add support for target_clone attribute.
10052 (rs6000_valid_attribute_p): Allow "default" attribute.
10053 (get_decl_name): New debug function to simplify printing the
10054 current function name in debugging statements.
10055 (rs6000_clone_priority): New functions to support the target_clone
10056 attribute, and be able to generate code to switch between ISA 2.05
10057 through ISA 3.0 (power6 through power9).
10058 (rs6000_compare_version_priority): Likewise.
10059 (rs6000_get_function_versions_dispatcher): Likewise.
10060 (make_resolver_func): Likewise.
10061 (add_condition_to_bb): Likewise.
10062 (dispatch_function_versions): Likewise.
10063 (rs6000_generate_version_dispatcher_body): Likewise.
10064 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10065 (fusion_gpr_load_p): Fix a spacing issue.
10066 * doc/extend.texi (Common Function Attributes): Document that the
10067 PowerPC supports the target_clone attribute.
10069 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10071 * config/arm/arm.h: explain F symbol found in description of ARM
10072 register allocation in its legend.
10074 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10076 * config/mips/frame-header-opt.c: Include profile-count.h.
10077 * config/riscv/riscv.c: Include profile-count.h
10079 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10081 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10083 (sm_set_flag_if_changed): Add bbs field.
10084 (execute_sm_if_changed_flag_set): Pass BBS.
10085 (execute_sm): Update.
10087 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10089 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10092 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10094 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10095 (peephole2): New peephole2 to emit the above.
10096 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10098 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10100 * config/aarch64/aarch64.c (define_peephole2 above
10101 *sub_<shift>_<mode>): New peephole.
10103 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10105 * config/i386/i386.c (make_resolver_func): Update.
10106 * Makefile.in: Add profile-count.h and profile-count.o
10107 * auto-profile.c (afdo_indirect_call): Update to new API.
10108 (afdo_set_bb_count): Update.
10109 (afdo_propagate_edge): Update.
10110 (afdo_propagate_circuit): Update.
10111 (afdo_calculate_branch_prob): Update.
10112 (afdo_annotate_cfg): Update.
10113 * basic-block.h: Include profile-count.h
10114 (struct edge_def): Turn count to profile_count.
10115 (struct basic_block_def): Likewie.
10116 (REG_BR_PROB_BASE): Move to profile-count.h
10117 (RDIV): Move to profile-count.h
10118 * bb-reorder.c (max_entry_count): Turn to profile_count.
10119 (find_traces): Update.
10120 (rotate_loop):Update.
10121 (connect_traces):Update.
10122 (sanitize_hot_paths):Update.
10123 * bt-load.c (migrate_btr_defs): Update.
10124 * cfg.c (RDIV): Remove.
10125 (init_flow): Use alloc_block.
10126 (alloc_block): Uninitialize count.
10127 (unchecked_make_edge): Uninitialize count.
10128 (check_bb_profile): Update.
10129 (dump_edge_info): Update.
10130 (dump_bb_info): Update.
10131 (update_bb_profile_for_threading): Update.
10132 (scale_bbs_frequencies_int): Update.
10133 (scale_bbs_frequencies_gcov_type): Update.
10134 (scale_bbs_frequencies_profile_count): New.
10135 * cfg.h (update_bb_profile_for_threading): Update.
10136 (scale_bbs_frequencies_profile_count): Declare.
10137 * cfgbuild.c (compute_outgoing_frequencies): Update.
10138 (find_many_sub_basic_blocks): Update.
10139 * cfgcleanup.c (try_forward_edges): Update.
10140 (try_crossjump_to_edge): Update.
10141 * cfgexpand.c (expand_gimple_tailcall): Update.
10142 (construct_exit_block): Update.
10143 * cfghooks.c (verify_flow_info): Update.
10144 (dump_bb_for_graph): Update.
10145 (split_edge): Update.
10146 (make_forwarder_block): Update.
10147 (duplicate_block): Update.
10148 (account_profile_record): Update.
10149 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10150 (get_estimated_loop_iterations): Update.
10151 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10152 (single_likely_exit): Update.
10153 * cfgloopmanip.c (scale_loop_profile): Update.
10155 (set_zero_probability): Update.
10156 (lv_adjust_loop_entry_edge): Update.
10157 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10158 (purge_dead_edges): Update.
10159 (rtl_account_profile_record): Update.
10160 * cgraph.c (cgraph_node::create): Uninitialize count.
10161 (symbol_table::create_edge): Uninitialize count.
10162 (cgraph_update_edges_for_call_stmt_node): Update.
10163 (cgraph_edge::dump_edge_flags): Update.
10164 (cgraph_node::dump): Update.
10165 (cgraph_edge::maybe_hot_p): Update.
10166 * cgraph.h: Include profile-count.h
10167 (create_clone), create_edge, create_indirect_edge): Update.
10168 (cgraph_node): Turn count to profile_count.
10169 (cgraph_edge0: Likewise.
10170 (make_speculative, clone): Update.
10171 (create_edge): Update.
10172 (init_lowered_empty_function): Update.
10173 * cgraphclones.c (cgraph_edge::clone): Update.
10174 (duplicate_thunk_for_node): Update.
10175 (cgraph_node::create_clone): Update.
10176 * cgraphunit.c (cgraph_node::analyze): Update.
10177 (cgraph_node::expand_thunk): Update.
10178 * final.c (dump_basic_block_info): Update.
10179 * gimple-streamer-in.c (input_bb): Update.
10180 * gimple-streamer-out.c (output_bb): Update.
10181 * graphite.c (print_global_statistics): Update.
10182 (print_graphite_scop_statistics): Update.
10183 * hsa-brig.c: Include basic-block.h.
10184 * hsa-dump.c: Include basic-block.h.
10185 * hsa-gen.c (T sum_slice): Update.
10186 (convert_switch_statements):Update.
10187 * hsa-regalloc.c: Include basic-block.h.
10188 * ipa-chkp.c (chkp_produce_thunks): Update.
10189 * ipa-cp.c (struct caller_statistics): Update.
10190 (init_caller_stats): Update.
10191 (gather_caller_stats): Update.
10192 (ipcp_cloning_candidate_p): Update.
10193 (good_cloning_opportunity_p): Update.
10194 (get_info_about_necessary_edges): Update.
10195 (dump_profile_updates): Update.
10196 (update_profiling_info): Update.
10197 (update_specialized_profile): Update.
10198 (perhaps_add_new_callers): Update.
10199 (decide_about_value): Update.
10200 (ipa_cp_c_finalize): Update.
10201 * ipa-devirt.c (struct odr_type_warn_count): Update.
10202 (struct decl_warn_count): Update.
10203 (struct final_warning_record): Update.
10204 (possible_polymorphic_call_targets): Update.
10205 (ipa_devirt): Update.
10206 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10207 * ipa-icf.c (sem_function::merge): Update.
10208 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10209 * ipa-inline.c (compute_uninlined_call_time): Update.
10210 (compute_inlined_call_time): Update.
10211 (want_inline_small_function_p): Update.
10212 (want_inline_self_recursive_call_p): Update.
10213 (edge_badness): Update.
10214 (lookup_recursive_calls): Update.
10215 (recursive_inlining): Update.
10216 (inline_small_functions): Update.
10217 (dump_overall_stats): Update.
10218 (dump_inline_stats): Update.
10219 * ipa-profile.c (ipa_profile_generate_summary): Update.
10220 (ipa_propagate_frequency): Update.
10221 (ipa_profile): Update.
10222 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10223 * ipa-utils.c (ipa_merge_profiles): Update.
10224 * loop-doloop.c (doloop_modify): Update.
10225 * loop-unroll.c (report_unroll): Update.
10226 (unroll_loop_runtime_iterations): Update.
10227 * lto-cgraph.c (lto_output_edge): Update.
10228 (lto_output_node): Update.
10229 (input_node): Update.
10230 (input_edge): Update.
10231 (merge_profile_summaries): Update.
10232 * lto-streamer-in.c (input_cfg): Update.
10233 * lto-streamer-out.c (output_cfg): Update.
10234 * mcf.c (create_fixup_graph): Update.
10235 (adjust_cfg_counts): Update.
10236 (sum_edge_counts): Update.
10237 * modulo-sched.c (sms_schedule): Update.
10238 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10239 * predict.c (maybe_hot_count_p): Update.
10240 (probably_never_executed): Update.
10241 (dump_prediction): Update.
10242 (combine_predictions_for_bb): Update.
10243 (propagate_freq): Update.
10244 (handle_missing_profiles): Update.
10245 (counts_to_freqs): Update.
10246 (rebuild_frequencies): Update.
10247 (force_edge_cold): Update.
10248 * predict.h: Include profile-count.h
10249 (maybe_hot_count_p, counts_to_freqs): UPdate.
10250 * print-rtl-function.c: Do not include cfg.h
10251 * print-rtl.c: Include basic-block.h
10252 * profile-count.c: New file.
10253 * profile-count.h: New file.
10254 * profile.c (is_edge_inconsistent): Update.
10255 (correct_negative_edge_counts): Update.
10256 (is_inconsistent): Update.
10257 (set_bb_counts): Update.
10258 (read_profile_edge_counts): Update.
10259 (compute_frequency_overlap): Update.
10260 (compute_branch_probabilities): Update; Initialize and deinitialize
10262 (branch_prob): Update.
10263 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10264 (edge_gcov_count): New.
10265 (bb_gcov_count): New.
10266 * shrink-wrap.c (try_shrink_wrapping): Update.
10267 * tracer.c (better_p): Update.
10268 * trans-mem.c (expand_transaction): Update.
10269 (ipa_tm_insert_irr_call): Update.
10270 (ipa_tm_insert_gettmclone_call): Update.
10271 * tree-call-cdce.c: Update.
10272 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10273 (gimple_duplicate_sese_tail): Update.
10274 (gimple_account_profile_record): Update.
10275 (execute_fixup_cfg): Update.
10276 * tree-inline.c (copy_bb): Update.
10277 (copy_edges_for_bb): Update.
10278 (initialize_cfun): Update.
10279 (freqs_to_counts): Update.
10280 (copy_cfg_body): Update.
10281 (expand_call_inline): Update.
10282 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10283 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10284 (try_unroll_loop_completely): Update.
10285 (try_peel_loop): Update.
10286 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10287 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10288 * tree-ssa-loop-split.c (connect_loops): Update.
10289 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10290 * tree-ssa-reassoc.c (branch_fixup): Update.
10291 * tree-ssa-tail-merge.c (replace_block_by): Update.
10292 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10293 (compute_path_counts): Update.
10294 (update_profile): Update.
10295 (recompute_probabilities): Update.
10296 (update_joiner_offpath_counts): Update.
10297 (estimated_freqs_path): Update.
10298 (freqs_to_counts_path): Update.
10299 (clear_counts_path): Update.
10300 (ssa_fix_duplicate_block_edges): Update.
10301 (duplicate_thread_path): Update.
10302 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10303 (struct switch_conv_info): Update.
10304 * tree-tailcall.c (decrease_profile): Update.
10305 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10306 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10307 * value-prof.c (check_counter): Update.
10308 (gimple_divmod_fixed_value): Update.
10309 (gimple_mod_pow2): Update.
10310 (gimple_mod_subtract): Update.
10311 (gimple_ic_transform): Update.
10312 (gimple_stringop_fixed_value): Update.
10313 * value-prof.h (gimple_ic): Update.
10315 2017-06-02 Carl Love <cel@us.ibm.com>
10317 * config/rs6000/rs6000-c: Add support for built-in functions
10318 vector double vec_doublee (vector signed int);
10319 vector double vec_doublee (vector unsigned int);
10320 vector double vec_doublee (vector float);
10321 vector double vec_doubleh (vector signed int);
10322 vector double vec_doubleh (vector unsigned int);
10323 vector double vec_doubleh (vector float);
10324 vector double vec_doublel (vector signed int);
10325 vector double vec_doublel (vector unsigned int);
10326 vector double vec_doublel (vector float);
10327 vector double vec_doubleo (vector signed int);
10328 vector double vec_doubleo (vector unsigned int);
10329 vector double vec_doubleo (vector float);.
10330 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10331 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10333 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10334 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10335 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10337 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10338 vec_doublel, vec_doubleh.
10339 * doc/extend.texi: Update the built-in documentation file for the
10340 new built-in functions.
10342 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10345 * ipa-inline-analysis.c (free_growth_caches): Set
10346 edge_removal_hook_holder to NULL after removing it.
10348 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10350 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10351 comparision with zero.
10353 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10354 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10355 for early expansion of vec_min and vec_max builtins.
10356 (builtin_function_type): Add min/max unsigned variants to those
10357 identified as having unsigned arguments.
10359 2017-06-02 Olivier Hainque <hainque@adacore.com>
10361 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10363 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10365 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10366 Use VALL_F16 iterator rather than VALL.
10368 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10370 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10371 Emit CBNZ inside loop when doing a strong exchange and comparing
10372 against zero. Generate the CC flags after the loop.
10374 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10376 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10377 (dl_section_ref): New.
10378 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10379 On AIX, append an expression to subtract the size of the
10380 section length to dl_section_ref.
10382 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10384 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10385 for early expansion of vector absolute builtins.
10387 2017-06-02 Richard Biener <rguenther@suse.de>
10389 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10390 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10392 2017-06-02 Richard Biener <rguenther@suse.de>
10394 PR tree-optimization/80948
10395 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10396 stmt order as well.
10398 2017-06-02 Richard Biener <rguenther@suse.de>
10400 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
10402 * tree-vect-stmts.c (process_use): Do not mark backedge defs
10403 for inductions as relevant.
10405 2017-06-02 Richard Biener <rguenther@suse.de>
10407 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
10408 (vectorizable_induction): ... this. Remove dead code.
10410 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
10412 * builtins. (expand_builtin_alloca): Remove second parameter and
10413 infer its value from the first parameter instead.
10414 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
10416 2017-06-02 Jakub Jelinek <jakub@redhat.com>
10418 PR rtl-optimization/80903
10419 * loop-doloop.c (add_test): Unshare sequence.
10421 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
10423 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
10425 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
10427 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
10429 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
10430 xlogue_layout::get_instance, logue_layout::xlogue_layout,
10431 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
10432 (xlogue_layout::get_stub_rtx): Make static.
10433 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
10434 (xlogue_layout::compute_stub_managed_regs): Rename to...
10435 (xlogue_layout::count_stub_managed_regs): ...this.
10436 (xlogue_layout::is_stub_managed_reg): New function.
10437 (xlogue_layout::m_stub_names): Rename to...
10438 (xlogue_layout::s_stub_names): ...this, make static.
10439 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
10440 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
10441 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
10442 xlogue_layout::s_stub_names): Instantiate statics.
10443 (stub_managed_regs): Remove.
10444 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
10445 (disable_call_ms2sysv_xlogues): Rename to...
10446 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
10447 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
10449 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
10450 change after reload_completed.
10451 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
10453 (ix86_expand_prologue): Likewise.
10454 (ix86_expand_epilogue): Likewise.
10455 (ix86_expand_split_stack_prologue): Likewise.
10456 (ix86_compute_frame_layout): Remove frame parameter ...
10457 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
10458 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
10460 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
10461 (ix86_frame): Move from here ...
10462 * config/i386/i386.h (ix86_frame): ... to here.
10463 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
10464 complete ix86_frame data structure instead. Remove some_ld_name.
10466 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10468 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
10469 symbols that hold a DECL_VALUE_EXPR.
10471 2017-06-01 Martin Jambor <mjambor@suse.cz>
10473 PR tree-optimization/80898
10474 * tree-sra.c (process_subtree_disqualification): Removed.
10475 (disqualify_candidate): Do not acll
10476 process_subtree_disqualification.
10477 (subtree_mark_written_and_enqueue): New function.
10478 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
10479 RHS has been disqualified and re-queue LHS if necessary. Apart
10480 from that, ignore disqualified RHS.
10482 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10484 * config/s390/s390.c (s390_emit_epilogue): Disable early return
10485 address fetch for z10 or later.
10487 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10489 * config/arc/arc.md (tst_movb): Add guard when splitting.
10491 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10493 * config/arc/arc.c (arc_can_eliminate): Test against
10494 arc_frame_pointer_needed.
10496 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10498 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
10499 to prevent store reordering.
10500 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
10501 (type): Add block type.
10502 (stack_tie): Define special instruction to be used in
10505 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10507 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
10508 constraint. It is not valid for the pattern.
10509 (noncommutative_binary_comparison): Likewise.
10511 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10513 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
10516 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10518 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
10519 be used by the reg-alloc.
10521 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10523 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
10524 reg-alloc when having mul64 or mul32x16 instructions.
10525 (mulsidi3): Likewise.
10526 (umulsidi3): Likewise.
10527 (mulsi32x16): New pattern.
10528 (mulsi64): Likewise.
10529 (mulsidi64): Likewise.
10530 (umulsidi64): Likewise.
10531 (MUL32x16_REG): Define.
10532 (mul64_600): Use MUL32x16_REG.
10533 (mac64_600): Likewise.
10534 (umul64_600): Likewise.
10535 (umac64_600): Likewise.
10537 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10539 * config/arc/arc.md (mulsi3_700): Make it commutative.
10541 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
10543 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
10545 (*sign_extendsidi2_insn): Likewise for movstosw.
10547 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10549 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
10550 the type of the input discriminant value. Convert the
10551 discriminant value of signedness vary.
10553 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
10555 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
10556 Add to -Wall section.
10558 2017-06-01 Richard Biener <rguenther@suse.de>
10560 PR middle-end/66313
10561 * fold-const.c (fold_plusminus_mult_expr): If the factored
10562 factor may be zero use a wrapping type for the inner operation.
10563 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
10564 and handle moved defs.
10565 (process_assignment): Properly guard the unary op case. Return a
10566 tri-state indicating that moving the stmt before the call may allow
10567 to continue. Pass through to_move.
10568 (find_tail_calls): Handle moving unrelated defs before
10571 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10574 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
10575 splitter result in the canonical way.
10577 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
10579 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
10580 also for 32bit target. Update insn attributes.
10581 (zero-extendsidi2 splitter): Allow all registers for operand 1.
10583 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
10585 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
10586 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
10587 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
10588 (_mm_maskz_min_ss): New intrinsics.
10590 2017-05-31 Martin Liska <mliska@suse.cz>
10592 * tree-vect-loop.c (vect_create_epilog_for_reduction):
10593 Change comment style to one we normally use.
10594 (vectorizable_reduction): Likewise.
10595 (vectorizable_induction): Likewise.
10596 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10597 (vectorizable_call): Likewise.
10598 (vectorizable_simd_clone_call): Likewise.
10599 (vectorizable_conversion): Likewise.
10600 (vectorizable_assignment): Likewise.
10601 (vectorizable_shift): Likewise.
10602 (vectorizable_operation): Likewise.
10603 (vectorizable_store): Likewise.
10604 (vectorizable_load): Likewise.
10605 * tree-vectorizer.h: Likewise.
10607 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
10609 * passes.c (emergency_dump_function): New.
10610 * tree-pass.h (emergency_dump_function): Declare.
10611 * plugin.c (plugins_internal_error_function): Remove.
10612 * plugin.h (plugins_internal_error_function): Remove declaration.
10613 * toplev.c (internal_error_function): New static function. Use it...
10614 (general_init): ...here.
10616 2017-05-31 Graham Markall <graham.markall@embecosm.com>
10618 * config/arc/arc.c (arc_print_operand): Handle constant operands.
10619 (arc_rtx_costs): Add costs for new patterns.
10620 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
10621 * config/arc/predicates.md: Add _1_2_3_operand predicate.
10623 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10625 * tree-ssa-strlen.c (get_next_strinfo): New function.
10626 (get_stridx_plus_constant): Use it.
10627 (zero_length_string): Likewise.
10628 (adjust_related_strinfos): Likewise.
10629 (adjust_last_stmt): Likewise.
10631 2017-05-31 Richard Biener <rguenther@suse.de>
10634 * config/i386/i386.c (ix86_expand_builtin): Remove assert
10635 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
10637 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10639 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
10640 loop_vinfo argument and use of dependence distance vectors.
10641 Check instead whether the two references differ only in their
10642 initial value and assume that they have the same alignment if the
10643 difference is a multiple of the vector alignment.
10644 (vect_analyze_data_refs_alignment): Update call accordingly.
10646 2017-05-31 Martin Liska <mliska@suse.cz>
10649 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
10651 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10653 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
10654 (create_intersect_range_checks): Move from ...
10655 * tree-data-ref.c (create_intersect_range_checks_index)
10656 (create_intersect_range_checks): ... to here.
10657 (create_runtime_alias_checks): New function factored from ...
10658 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
10659 here. Call above function.
10660 * tree-data-ref.h (create_runtime_alias_checks): New function.
10662 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10664 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
10665 segment length for dr_b and compute it in wide_int.
10667 2017-05-31 Richard Biener <rguenther@suse.de>
10669 PR tree-optimization/80906
10670 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
10671 and pass through iv_map.
10672 (copy_bb_and_scalar_dependences): Adjust.
10673 (translate_pending_phi_nodes): Likewise.
10674 (copy_loop_close_phi_args): Handle code-generating IVs instead
10677 2017-05-30 David Malcolm <dmalcolm@redhat.com>
10679 * diagnostic-color.c (color_dict): Add "type-diff".
10680 (parse_gcc_colors): Update comment.
10681 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10682 -fdiagnostics-show-template-tree and -fno-elide-type.
10683 (GCC_COLORS): Add type-diff to example.
10685 (-fdiagnostics-show-template-tree): New.
10686 (-fno-elide-type): New.
10687 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
10688 the pp_format_decoder callback. Call any m_format_postprocessor's
10690 (pretty_printer::pretty_printer): Initialize
10691 m_format_postprocessor.
10692 (pretty_printer::~pretty_printer): Delete any
10693 m_format_postprocessor.
10694 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
10695 (class format_postprocessor): New class.
10696 (struct pretty_printer::format_decoder): Document the new parameters.
10697 (struct pretty_printer::m_format_postprocessor): New field.
10698 * tree-diagnostic.c (default_tree_printer): Update for new
10699 bool and const char ** params.
10700 * tree-diagnostic.h (default_tree_printer): Likewise.
10702 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
10704 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
10705 (lwa_operand): Delete rs6000_gen_cell_microcode test.
10706 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10707 rs6000_gen_cell_microcode code.
10708 (rs6000_final_prescan_insn): Delete.
10709 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
10710 "warn-cell-microcode" entries.
10711 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
10712 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
10713 throughout. Change cc_reg_not_micro_cr0_operand to
10714 cc_reg_not_cr0_operand throughout.
10715 (*extendhi<mode>2_noload): Delete.
10716 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
10717 (mwarn-cell-microcode): Delete.
10718 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
10719 -mgen-cell-microcode and -mwarn-cell-microcode.
10721 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
10724 * config/i386/constraints.md (Yd): New constraint.
10726 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
10727 and (?Yd, r) alternatives. Update insn attributes.
10728 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
10729 and (?*Yd, r) alternatives. Update insn attributes.
10730 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
10732 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
10734 * gimplify.c (gimplify_modify_expr): Don't create a
10735 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
10738 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10740 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
10742 2017-05-30 Richard Biener <rguenther@suse.de>
10744 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
10745 and reduc_def fields.
10746 (STMT_VINFO_REDUC_TYPE): New define.
10747 (STMT_VINFO_REDUC_DEF): Likewise.
10748 (vect_force_simple_reduction): Adjust prototype.
10749 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
10750 (vect_is_simple_reduction): Remove check_reduction argument.
10751 (vect_force_simple_reduction): Adjust and set
10752 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
10753 (vectorizable_reduction): Do not re-do reduction analysis
10754 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
10755 * tree-parloops.c (gather_scalar_reductions): Adjust.
10757 2017-05-30 Richard Biener <rguenther@suse.de>
10759 PR middle-end/80901
10760 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
10763 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
10765 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
10766 Introduce unknown_misalignment parameter and remove vf.
10767 (vect_peeling_hash_get_lowest_cost):
10768 Pass unknown_misalignment parameter.
10769 (vect_enhance_data_refs_alignment):
10770 Fix unsupportable data ref treatment.
10772 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10774 * tree-vect-data-refs.c (vect_get_data_access_cost):
10775 Workaround for SLP handling.
10776 (vect_enhance_data_refs_alignment):
10777 Compute costs for doing no peeling at all, compare to the best
10778 peeling costs so far and avoid peeling if cheaper.
10780 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10782 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
10783 Return peeling info and set costs to zero for unlimited cost
10785 (vect_enhance_data_refs_alignment): Also inspect all datarefs
10786 with unknown misalignment. Compute and costs for unknown
10787 misalignment, compare them to the costs for known misalignment
10788 and choose the cheapest for peeling.
10790 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10792 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
10793 (vect_get_peeling_costs_all_drs): Create function.
10794 (vect_peeling_hash_get_lowest_cost):
10795 Use vect_get_peeling_costs_all_drs.
10796 (vect_peeling_supportable): Create function.
10797 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
10799 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10801 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
10802 DR_HAS_NEGATIVE_STEP.
10803 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
10804 (vect_enhance_data_refs_alignment): Use.
10805 (vect_duplicate_ssa_name_ptr_info): Use.
10806 * tree-vectorizer.h (dr_misalignment): Use.
10807 (known_alignment_for_access_p): Use.
10809 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10812 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
10814 (has_section_name): New function.
10816 2017-05-30 Martin Liska <mliska@suse.cz>
10819 * auto-profile.c (get_function_decl_from_block): Fix
10822 2017-05-30 Richard Biener <rguenther@suse.de>
10824 PR middle-end/80876
10825 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
10827 2017-05-30 Martin Liska <mliska@suse.cz>
10829 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
10830 * dumpfile.h (struct dump_file_info): Remove ctors.
10832 2017-05-30 Martin Liska <mliska@suse.cz>
10834 * predict.def: Fix GNU coding style.
10836 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
10838 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
10839 Mark 'to' argument with ATTRIBUTE_UNUSED.
10841 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
10843 * config/xtensa/xtensa.c (xtensa_emit_call): Use
10844 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
10845 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
10848 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
10850 * doc/install.texi (Options specification): Restore entry of
10851 --enable-sjlj-exceptions.
10853 2017-05-27 Michael Eager <eager@eagercon.com>
10856 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
10858 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
10860 * config/microblaze/microblaze.h
10861 (FIXED_REGISTERS): Update in macro.
10862 (CALL_USED_REGISTERS): Update in macro.
10864 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
10866 * doc/install.texi: Add links to macOS binary distributions.
10868 2017-05-27 Jakub Jelinek <jakub@redhat.com>
10872 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10874 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
10876 2017-05-26 Martin Liska <mliska@suse.cz>
10878 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
10880 2017-05-26 Martin Liska <mliska@suse.cz>
10882 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
10883 always leading ';; '.
10884 (dump_bb_info): Likewise.
10885 (brief_dump_cfg): Likewise.
10886 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
10887 * dumpfile.c: Remove usage of TDF_VERBOSE.
10888 * dumpfile.h (enum dump_kind): Likewise.
10889 (dump_gimple_bb_header): Do not use TDF_COMMENT.
10890 * print-tree.c (debug_verbose): Remove.
10891 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
10892 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
10893 * tree-diagnostic.c (default_tree_printer): Replace
10894 TDF_DIAGNOSTIC with TDF_SLIM.
10896 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10898 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
10899 in parameter loop, rather than loop_vinfo.
10900 (create_intersect_range_checks): Ditto.
10901 (vect_create_cond_for_alias_checks): Update call to above functions.
10903 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10905 PR tree-optimization/80815
10906 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
10907 for merging runtime alias checks. Handle negative DR_STEPs.
10909 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10911 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
10913 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
10914 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
10915 out code pruning runtime alias checks.
10916 * tree-data-ref.c (prune_runtime_alias_test_list): New function
10917 factored out from above.
10918 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
10920 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
10922 (prune_runtime_alias_test_list): New decalaration.
10924 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10926 * tree-vect-data-refs.c (compare_tree): Rename and move ...
10927 * tree-data-ref.c (data_ref_compare_tree): ... to here.
10928 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
10929 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
10930 (operator==, comp_dr_with_seg_len_pair): Ditto.
10931 (vect_prune_runtime_alias_test_list): Ditto.
10933 2017-05-26 Martin Liska <mliska@suse.cz>
10936 * params.def: Bound partial-inlining-entry-probability param.
10938 2017-05-26 Marek Polacek <polacek@redhat.com>
10941 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
10944 2017-05-26 Richard Biener <rguenther@suse.de>
10946 PR tree-optimization/80842
10947 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
10950 2017-05-26 Richard Biener <rguenther@suse.de>
10952 PR tree-optimization/80844
10953 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
10955 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
10957 * doc/md.texi (Machine Constraints): Update x86 family
10958 machine constraints section to match 'config/i386/constraints.md'.
10960 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
10962 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
10964 2017-05-25 Nathan Sidwell <nathan@acm.org>
10966 * doc/invoke.texi (--enable-languages): Update documentation.
10968 2017-05-25 Martin Liska <mliska@suse.cz>
10970 * dumpfile.c: Add TDF_FOLDING.
10971 * dumpfile.h (enum dump_kind): Likewise.
10972 * genmatch.c (dt_simplify::gen_1): Use it.
10974 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10976 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
10978 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10980 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
10981 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
10983 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10985 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
10986 * match.pd (X == C): Rewrite it here.
10987 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
10988 with_certain_nonzero_bits2): New predicates.
10989 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
10991 2017-05-24 Nathan Sidwell <nathan@acm.org>
10993 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
10996 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
10999 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11001 * config/powerpcspe: New port. Files are copied from the rs6000
11002 port, with "rs6000" in filenames replaced by "powerpcspe".
11004 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11006 PR rtl-optimization/80754
11007 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11009 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11011 * config/sparc/sparc.md (length): Return the correct value for -mflat
11012 sibcalls to match output_sibcall.
11014 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11018 * config/rs6000/rs6000.c (struct machine_function): Add new field
11020 (rs6000_get_separate_components): Init that field, use it.
11021 (rs6000_components_for_bb): Use the field.
11023 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11025 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11027 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11029 PR middle-end/80823
11030 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11032 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11035 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11036 address against address_operand predicate.
11037 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11039 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11041 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11042 parameters passed indirectly.
11044 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11046 * config/i386/i386.md (*movdi_internal): Remove SSE4
11047 alternative 18 (?r, *v). Update insn attributes.
11048 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11049 Update insn attributes.
11050 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11051 Update insn attributes.
11052 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11053 alternative 1 (r, v). Remove isa attribute.
11054 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11055 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11056 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11058 2017-05-23 Tom de Vries <tom@codesourcery.com>
11060 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11063 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11065 * cgraphunit.c (symbol_table::process_new_functions): Update.
11066 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11067 (inline_generate_summary): Rename to ...
11068 (ipa_fn_summary_generate): ... this one.
11069 (inline_read_summary): Rename to ...
11070 (ipa_fn_summary_read): ... this one.
11071 (inline_write_summary): Rename to ...
11072 (ipa_fn_summary_write): ... this one.
11073 (inline_free_summary): Rename to ...
11074 (ipa_free_fn_summary): ... this one.
11075 (pass_data_local_fn_summary, pass_local_fn_summary,
11076 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11077 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11078 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11079 make_pass_ipa_fn_summary): New.
11080 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11081 inline_write_summary, inline_free_summary): Remove.
11082 (ipa_free_fn_summary) : New.
11083 * ipa-inline.c (ipa_inline): Update.
11084 (pass_ipa_inline): Do not generate summaries.
11085 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11087 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11088 and add pass_ipa_fn_summary.
11089 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11091 (make_pass_inline_parameters): Remove.
11093 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11095 * omp-low.c (struct omp_context): Remove "default_kind" member.
11098 * omp-offload.c (execute_oacc_device_lower): Remove the
11099 parallelism dimensions function attributes for unparallelized
11100 OpenACC kernels constructs.
11102 2017-05-23 Martin Liska <mliska@suse.cz>
11104 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11106 (cgraph_edge::make_speculative): Likewise.
11107 (cgraph_edge::resolve_speculation): Likewise.
11108 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11109 (cgraph_node::dump): Likewise.
11110 * cgraph.h: Likewise.
11111 * cgraphunit.c (analyze_functions): Likewise.
11112 (symbol_table::compile): Likewise.
11113 * ipa-cp.c (print_all_lattices): Likewise.
11114 (determine_versionability): Likewise.
11115 (initialize_node_lattices): Likewise.
11116 (ipcp_verify_propagated_values): Likewise.
11117 (estimate_local_effects): Likewise.
11118 (update_profiling_info): Likewise.
11119 (create_specialized_node): Likewise.
11120 (perhaps_add_new_callers): Likewise.
11121 (decide_about_value): Likewise.
11122 (decide_whether_version_node): Likewise.
11123 (identify_dead_nodes): Likewise.
11124 (ipcp_store_bits_results): Likewise.
11125 * ipa-devirt.c (dump_targets): Likewise.
11126 (ipa_devirt): Likewise.
11127 * ipa-icf.c (sem_item::dump): Likewise.
11128 (sem_function::equals): Likewise.
11129 (sem_variable::equals): Likewise.
11130 (sem_item_optimizer::read_section): Likewise.
11131 (sem_item_optimizer::execute): Likewise.
11132 (congruence_class::dump): Likewise.
11133 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11134 (dump_inline_summary): Likewise.
11135 (estimate_node_size_and_time): Likewise.
11136 (inline_analyze_function): Likewise.
11137 * ipa-inline-transform.c (inline_call): Likewise.
11138 * ipa-inline.c (report_inline_failed_reason): Likewise.
11139 (want_early_inline_function_p): Likewise.
11140 (edge_badness): Likewise.
11141 (update_edge_key): Likewise.
11142 (inline_small_functions): Likewise.
11143 * ipa-profile.c (ipa_profile): Likewise.
11144 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11145 (ipa_make_edge_direct_to_target): Likewise.
11146 (remove_described_reference): Likewise.
11147 (ipa_impossible_devirt_target): Likewise.
11148 (propagate_controlled_uses): Likewise.
11149 (ipa_print_node_params): Likewise.
11150 (ipcp_transform_function): Likewise.
11151 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11152 (propagate_pure_const): Likewise.
11153 * ipa-reference.c (generate_summary): Likewise.
11154 (read_write_all_from_decl): Likewise.
11155 (propagate): Likewise.
11156 (ipa_reference_read_optimization_summary): Likewise.
11157 * ipa-utils.c (ipa_merge_profiles): Likewise.
11158 * ipa.c (walk_polymorphic_call_targets): Likewise.
11159 (symbol_table::remove_unreachable_nodes): Likewise.
11160 (ipa_single_use): Likewise.
11161 * passes.c (execute_todo): Likewise.
11162 * predict.c (drop_profile): Likewise.
11163 * symtab.c (symtab_node::get_dump_name): New function.
11164 (symtab_node::dump_name): Likewise.
11165 (symtab_node::dump_asm_name): Likewise.
11166 (symtab_node::dump_references): Likewise.
11167 (symtab_node::dump_referring): Likewise.
11168 (symtab_node::dump_base): Likewise.
11169 (symtab_node::debug_symtab): Likewise.
11170 * tree-sra.c (convert_callers_for_node): Likewise.
11171 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11172 * value-prof.c (init_node_map): Likewise.
11174 2017-05-23 Martin Liska <mliska@suse.cz>
11176 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11177 and symtab_node::debug_symtab to symbol_table::debug.
11178 * cgraphunit.c (analyze_functions): Use the renamed function.
11179 (symbol_table::compile): Likewise.
11180 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11181 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11182 * passes.c (execute_todo): Likewise.
11183 * symtab.c (symbol_table::dump): New function.
11184 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11186 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11188 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11189 that nonconst implies exec.
11191 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11193 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11194 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11195 (inline_edge_summary_vec): Turn into ...
11196 (ipa_call_summaries): ... this one.
11197 (redirect_to_unreachable, edge_set_predicate,
11198 evaluate_properties_for_edge, inline_summary_alloc,
11199 reset_ipa_call_summary, reset_inline_summary,
11200 inline_summary_t::duplicate): Update.
11201 (inline_edge_duplication_hook): Turn to ...
11202 (ipa_call_summary_t::duplicate): ... this one.
11203 (inline_edge_removal_hook): Turn to ...
11204 (ipa_call_summary_t::remove): ... this one.
11205 (dump_inline_edge_summary): Turn to ...
11206 (dump_ipa_call_summary): ... this one.
11207 (estimate_function_body_sizes): Update.
11208 (inline_update_callee_summaries): Update.
11209 (remap_edge_change_prob): Update.
11210 (remap_edge_summaries): Update.
11211 (inline_merge_summary): Update.
11212 (do_estimate_edge_time): Update.
11213 (inline_generate_summary): Update.
11214 (inline_read_section): Update.
11215 (inline_read_summary): Update.
11216 (inline_free_summary): Update.
11217 * ipa-inline.c (can_inline_edge_p): Update.
11218 (compute_inlined_call_time): Update.
11219 (want_inline_small_function_p): Update.
11220 (edge_badness): Update.
11221 (early_inliner): Update.
11222 * ipa-inline.h (inline_edge_summary): Turn to ...
11223 (ipa_call_summary): ... this one.
11224 (ipa_call_summary_t): New class.
11225 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11226 (ipa_call_summaries): New.
11227 (inline_edge_summary): Remove.
11228 (estimate_edge_growth): Update.
11229 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11230 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11231 * ipa-split.c (execute_split_functions): Update.
11232 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11234 2017-05-23 Tom de Vries <tom@codesourcery.com>
11236 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11237 attributes): Document rdrand effective target.
11239 2017-05-23 Tom de Vries <tom@codesourcery.com>
11241 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11242 attributes): Sort alphabetically.
11244 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11246 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11248 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11251 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11252 V2DF/V2DI splat into two separate patterns, one that handles
11253 registers, and the other that only handles memory. Drop support
11254 for splatting from a GPR on ISA 2.07 and then splitting the
11255 splat into direct move and splat.
11256 (vsx_splat_<mode>_reg): Likewise.
11257 (vsx_splat_<mode>_mem): Likewise.
11259 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11261 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11263 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11265 PR middle-end/80809
11266 * omp-low.c (finish_taskreg_remap): New function.
11267 (finish_taskreg_scan): If unit size of ctx->record_type
11268 is non-constant, unshare the size expression and replace
11269 decls in it with possible outer var refs.
11271 PR middle-end/80809
11272 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11273 GOVD_SHARED rather than GOVD_PRIVATE with it.
11274 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11275 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11277 PR middle-end/80853
11278 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11279 as last argument to build_outer_var_ref for pointer bases of array
11280 section reductions.
11282 2017-05-19 Martin Sebor <msebor@redhat.com>
11284 * print-tree.c (print_node): Print DECL_READ_P flag.
11286 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11288 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11289 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11290 * cgraph.c: Likewise.
11291 * cgraphunit.c: Likewise.
11292 * gengtype.c: Likewise.
11293 * ipa-cp.c: Likewise.
11294 * ipa-devirt.c: Likewise.
11295 * ipa-icf.c: Likewise.
11296 * ipa-predicate.c: Likewise.
11297 * ipa-profile.c: Likewise.
11298 * ipa-prop.c: Likewise.
11299 * ipa-split.c: Likewise.
11301 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11302 edge_predicate_pool, dump_inline_hints,
11303 inline_summary::account_size_time, redirect_to_unreachable,
11304 edge_set_predicate, set_hint_predicate,
11305 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11306 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11307 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11308 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11309 ipa_call_summary_t::remove, initialize_growth_caches,
11310 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11311 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11312 mark_modified, unmodified_parm_1, unmodified_parm,
11313 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11314 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11315 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11316 will_be_nonconstant_predicate, record_modified_bb_info,
11317 get_minimal_bb, record_modified, param_change_prob,
11318 phi_result_unknown_predicate, predicate_for_phi_result,
11319 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11320 estimate_function_body_sizes, compute_inline_parameters,
11321 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11322 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11323 inline_update_callee_summaries, remap_edge_change_prob,
11324 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11325 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11326 inline_analyze_function, inline_summary_t::insert,
11327 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11328 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11329 inline_free_summary): Move to ipa-fnsummary.h
11330 (predicate_t): Remove.
11331 * ipa-fnsummary.c: New file.
11332 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11333 (enum inline_hints_vals, inline_hints, agg_position_info,
11334 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11335 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11336 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11337 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11338 inline_read_summary, inline_write_summary, inline_free_summary,
11339 inline_analyze_function, initialize_inline_failed,
11340 inline_merge_summary, inline_update_overall_summary,
11341 compute_inline_parameters): Move to ipa-fnsummary.h
11342 * ipa-fnsummary.h: New file.
11343 * ipa-inline-transform.h: Include ipa-inline.h.
11344 * ipa-inline.c: LIkewise.
11346 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11348 * ipa-inline.c (edge_badness): Use inlined_time instead of
11349 inline_summaries->get.
11351 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11353 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11355 2017-05-22 Nathan Sidwell <nathan@acm.org>
11357 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11358 (fdump-lang): Document 'raw' option.
11359 * dumpfile.h (TDI_tu): Delete.
11360 * dumpfile.c (dump_files): Remove translation-unit.
11361 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11363 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11365 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11366 command option from $(AWK) call.
11367 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11368 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11369 [FORMAT]: Remove handling of variable.
11370 * config/avr/t-multilib: Regenerate.
11372 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11374 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11376 (dump_inline_summary): Do not print self_time.
11377 (estimate_function_body_sizes): Do not set self_time.
11378 (compute_inline_parameters): Likewise.
11379 (inline_read_section, inline_write_summary): Do not stream self_time.
11380 * ipa-inline.h (inline_summary): Drop self_time.
11382 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11384 * ipa-inline-analysis.c (account_size_time): Rename to ...
11385 (inline_summary::account_size_time): ... this one.
11386 (reset_ipa_call_summary): Turn to ...
11387 (ipa_call_summary::reset): ... this one.
11388 (reset_inline_summary): Turn to ...
11389 (inline_summary::reset): ... this one.
11390 (inline_summary_t::remove): Update.
11391 (inline_summary_t::duplicate): Update.
11392 (ipa_call_summary_t::remove): Update.
11393 (dump_inline_summary): Update.
11394 (estimate_function_body_sizes): Update.
11395 (compute_inline_parameters): Update.
11396 (estimate_node_size_and_time): Update.
11397 (inline_merge_summary): Update.
11398 (inline_update_overall_summary): Update.
11399 (inline_read_section): Update.
11400 (inline_write_summary): Update.
11401 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
11402 add account_size_time and reset member functions.
11403 (ipa_call_summary): Add reset function.
11404 * ipa-predicate.h (predicate::operator &): Constify.
11406 2017-05-22 Richard Biener <rguenther@suse.de>
11408 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
11410 2017-05-19 Jason Merrill <jason@redhat.com>
11412 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
11414 2017-05-19 Marek Polacek <polacek@redhat.com>
11417 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
11418 TYPE_OVERFLOW_WRAPS checks.
11420 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
11422 * tree-core.h (enum omp_clause_default_kind): Add
11423 "OMP_CLAUSE_DEFAULT_PRESENT".
11424 * tree-pretty-print.c (dump_omp_clause): Handle it.
11425 * gimplify.c (enum gimplify_omp_var_data): Add
11426 "GOVD_MAP_FORCE_PRESENT".
11427 (gimplify_adjust_omp_clauses_1): Map it to
11428 "GOMP_MAP_FORCE_PRESENT".
11429 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
11431 * gimplify.c (oacc_default_clause): Clarify.
11433 2017-05-19 Nathan Sidwell <nathan@acm.org>
11435 LANG_HOOK_REGISTER_DUMPS
11436 * toplev.c (general_init): Call register dump lang hook.
11437 * doc/invoke.texi: Document -fdump-lang option family.
11438 * dumpfile.c (dump_files): Remove class dump here.
11439 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
11440 * dumpfile.h (tree_dump_index): Remove TDI_class.
11441 * langhooks-def.h (lhd_register_dumps): Declare.
11442 (LANG_HOOKS_REGISTER_DUMPS): Define.
11443 (LANG_HOOKS_INITIALIZER): Add it.
11444 * langhooks.c (lhd_register_dumps): Define.
11445 * langhooks.h (struct lang_hooks): Add register_dumps.
11447 2017-05-19 Nathan Sidwell <nathan@acm.org>
11449 * context.h (context::set_passes): New.
11450 * context.c (context::context): Do not create pass manager.
11451 * toplev.c (general_init): Create pass manager here.
11453 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
11455 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
11456 use this splitter if two add or or instructions would also work for
11457 the constant we want to generate.
11459 2017-05-19 Richard Biener <rguenther@suse.de>
11462 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
11463 predicate evaluation.
11465 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11467 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
11469 * ipa-inline.c (want_inline_small_function_p): Do not cast to
11472 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11474 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11475 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11476 (inline_edge_summary_vec): Turn into ...
11477 (ipa_call_summaries): ... this one.
11478 (redirect_to_unreachable, edge_set_predicate,
11479 evaluate_properties_for_edge, inline_summary_alloc,
11480 reset_ipa_call_summary, reset_inline_summary,
11481 inline_summary_t::duplicate): Update.
11482 (inline_edge_duplication_hook): Turn to ...
11483 (ipa_call_summary_t::duplicate): ... this one.
11484 (inline_edge_removal_hook): Turn to ...
11485 (ipa_call_summary_t::remove): ... this one.
11486 (dump_inline_edge_summary): Turn to ...
11487 (dump_ipa_call_summary): ... this one.
11488 (estimate_function_body_sizes): Update.
11489 (inline_update_callee_summaries): Update.
11490 (remap_edge_change_prob): Update.
11491 (remap_edge_summaries): Update.
11492 (inline_merge_summary): Update.
11493 (do_estimate_edge_time): Update.
11494 (inline_generate_summary): Update.
11495 (inline_read_section): Update.
11496 (inline_read_summary): Update.
11497 (inline_free_summary): Update.
11498 * ipa-inline.c (can_inline_edge_p): Update.
11499 (compute_inlined_call_time): Update.
11500 (want_inline_small_function_p): Update.
11501 (edge_badness): Update.
11502 (early_inliner): Update.
11503 * ipa-inline.h (inline_edge_summary): Turn to ...
11504 (ipa_call_summary): ... this one.
11505 (ipa_call_summary_t): New class.
11506 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11507 (ipa_call_summaries): New.
11508 (inline_edge_summary): Remove.
11509 (estimate_edge_growth): Update.
11510 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11511 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11512 * ipa-split.c (execute_split_functions): Update.
11513 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11515 2017-05-19 Richard Biener <rguenther@suse.de>
11517 PR middle-end/80764
11518 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
11520 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
11522 * config/rs6000/rs6000.c (struct machine_function): Add field
11523 fpr_is_wrapped_separately.
11524 (rs6000_get_separate_components): Use 64 components. Handle the
11525 new FPR components.
11526 (rs6000_components_for_bb): Handle the FPR components.
11527 (rs6000_emit_prologue_components): Handle the FPR components.
11528 (rs6000_emit_epilogue_components): Handle the FPR components.
11529 (rs6000_set_handled_components): Handle the FPR components.
11530 (rs6000_emit_prologue): Don't output prologue code for those FPRs
11531 that are already separately shrink-wrapped.
11532 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
11533 that are already separately shrink-wrapped.
11535 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11538 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
11541 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
11542 (define_peephole2 for Altivec d-form load): Add peepholes to catch
11543 cases where the register allocator uses a move and an offsettable
11544 memory operation to/from a FPR register on ISA 2.06/2.07.
11545 (define_peephole2 for Altivec d-form store): Likewise.
11547 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
11550 * config/i386/mmx.md (*mov<mode>_internal): Enable
11551 alternatives 11, 12, 13 and 14 also for 32bit targets.
11552 Remove alternatives 15, 16, 17 and 18.
11553 * config/i386/sse.md (vec_concatv2di): Change
11554 alternative (!x, *y) to (x, ?!*Yn).
11556 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
11558 * dumpfile.h (enum dump_kind): Remove stray comma.
11560 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11562 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
11563 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
11564 predicate::num_conditions
11565 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
11566 (CHANGED): turn into predicate::changed.
11567 (agg_position_info): Move to ipa-predicate.h
11568 (add_condition, predicate::add_clause, predicate::operator &=,
11569 predicate::or_with, predicate::evaluate, predicate::probability,
11570 dump_condition, dump_clause, predicate::dump,
11571 predicate::remap_after_duplication, predicate::remap_after_inlining,
11572 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
11573 (evaluate_conditions_for_known_args): Update.
11574 (set_cond_stmt_execution_predicate): Update.
11575 * ipa-inline.h: Include ipa-predicate.h
11576 (condition, inline_param_summary, conditions, agg_position_info,
11577 predicate): Move to ipa-predicate.h
11578 * ipa-predicate.c: New file.
11579 * ipa-predicate.h: New file.
11581 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
11583 * final.c (leaf_function_p): Check we are not in a sequence.
11585 2017-05-18 Martin Liska <mliska@suse.cz>
11587 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
11588 * dumpfile.c (dump_register): Use new enum dump_kind.
11589 (get_dump_file_name): Likewise.
11590 (dump_enable_all): Likewise.
11591 (dump_switch_p_1): Likewise.
11592 (enable_rtl_dump_file): Remove usage of TDF_RTL.
11593 * dumpfile.h (enum dump_kind): New enum type.
11594 (struct dump_file_info): Create constructor and
11595 format fields and comments.
11596 * passes.c (pass_manager::register_one_dump_file):
11598 * statistics.c (statistics_early_init): Likewise.
11599 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
11600 TDF_TREE with TDF_SLIM.
11601 (gather_memory_references_ref): Likewise.
11603 2017-05-18 Martin Liska <mliska@suse.cz>
11605 * vec.h (struct vnull): Use it.
11607 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11609 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
11610 (true_predicate, false_predicate, true_predicate_p,
11611 false_predicate_p): Remove.
11612 (single_cond_predicate, not_inlined_predicate): Turn to member function
11614 (add_condition): Update.
11615 (add_clause): Turn to...
11616 (predicate::add_clause): ... this one; update; allow passing NULL
11618 (and_predicates): Turn to ...
11619 (predicate::operator &=): ... this one.
11620 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
11621 (or_predicates): Turn to ...
11622 (predicate::or_with): ... this one.
11623 (evaluate_predicate): Turn to ...
11624 (predicate::evaluate): ... this one.
11625 (predicate_probability): Turn to ...
11626 (predicate::probability): ... this one.
11627 (dump_condition): Update.
11628 (dump_predicate): Turn to ...
11629 (predicate::dump): ... this one.
11630 (account_size_time): Update.
11631 (edge_set_predicate): Update.
11632 (set_hint_predicate): UPdate.
11633 (evaluate_conditions_for_known_args): Update.
11634 (evaluate_properties_for_edge): Update.
11635 (remap_predicate_after_duplication): Turn to...
11636 (predicate::remap_after_duplication): ... this one.
11637 (remap_hint_predicate_after_duplication): Update.
11638 (inline_summary_t::duplicate): UPdate.
11639 (dump_inline_edge_summary): Update.
11640 (dump_inline_summary): Update.
11641 (set_cond_stmt_execution_predicate): Update.
11642 (set_switch_stmt_execution_predicate): Update.
11643 (compute_bb_predicates): Update.
11644 (will_be_nonconstant_expr_predicate): Update.
11645 (will_be_nonconstant_predicate): Update.
11646 (phi_result_unknown_predicate): Update.
11647 (predicate_for_phi_result): Update.
11648 (array_index_predicate): Update.
11649 (estimate_function_body_sizes): Update.
11650 (estimate_node_size_and_time): Update.
11651 (estimate_ipcp_clone_size_and_time): Update.
11652 (remap_predicate): Rename to ...
11653 (predicate::remap_after_inlining): ... this one.
11654 (remap_hint_predicate): Update.
11655 (inline_merge_summary): Update.
11656 (inline_update_overall_summary): Update.
11657 (estimate_size_after_inlining): Update.
11658 (read_predicate): Rename to ...
11659 (predicate::stream_in): ... this one.
11660 (read_inline_edge_summary): Update.
11661 (write_predicate): Rename to ...
11662 (predicate::stream_out): ... this one.
11663 (write_inline_edge_summary): Update.
11664 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
11665 (clause_t): Turn to uint32_t
11666 (predicate): Turn to class; implement constructor and operators
11668 (size_time_entry): Update.
11669 (inline_summary): Update.
11670 (inline_edge_summary): Update.
11672 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
11674 * fold-const.c (fold_binary_loc): Move transformation...
11675 * match.pd (C - X CMP X): ... here.
11677 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
11679 * config/sparc/sparc.c (sparc_option_override): Set function
11680 alignment for -mcpu=niagara7 to 64 to match the I$ line.
11681 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
11683 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
11685 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
11687 2017-05-18 Marek Polacek <polacek@redhat.com>
11690 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
11691 (pass_ubsan::execute): Call gimple_assign_single_p instead of
11692 gimple_assign_load_p.
11694 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
11696 PR middle-end/80692
11697 * real.c (do_compare): Give decimal_do_compare preference over
11698 comparing just the signs.
11700 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
11702 * doc/md.texi (Canonicalization of Instructions): Describe the
11703 canonical form of instructions that inherently set a condition
11706 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
11708 PR middle-end/80775
11709 * tree-cfg.c: Move deletion of unreachable case statements to after
11710 the merging of consecutive case labels.
11712 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
11714 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
11715 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
11716 restoring of callee-saved registers.
11718 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
11720 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
11721 * config/visium/visium.c (single_set_and_flags): Likewise.
11722 * config/visium/visium.md (Substitutions): Likewise.
11724 2017-05-17 Martin Liska <mliska@suse.cz>
11726 * cfg.c: Introduce dump_flags_t type and
11727 use it instead of int type.
11729 * cfghooks.c: Likewise.
11730 * cfghooks.h (struct cfg_hooks): Likewise.
11731 * cfgrtl.c: Likewise.
11732 * cfgrtl.h: Likewise.
11733 * cgraph.c (cgraph_node::get_body): Likewise.
11734 * coretypes.h: Likewise.
11735 * domwalk.c: Likewise.
11736 * domwalk.h: Likewise.
11737 * dumpfile.c (struct dump_option_value_info): Likewise.
11738 (dump_enable_all): Likewise.
11739 (dump_switch_p_1): Likewise.
11740 (opt_info_switch_p): Likewise.
11741 * dumpfile.h (enum tree_dump_index): Likewise.
11742 (struct dump_file_info): Likewise.
11743 * genemit.c: Likewise.
11744 * generic-match-head.c: Likewise.
11745 * gengtype.c (open_base_files): Likewise.
11746 * gimple-pretty-print.c: Likewise.
11747 * gimple-pretty-print.h: Likewise.
11748 * graph.c (print_graph_cfg): Likewise.
11749 * graphite-scop-detection.c (dot_all_sese): Likewise.
11750 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
11751 * loop-unroll.c (report_unroll): Likewise.
11752 * passes.c (pass_manager::register_one_dump_file): Likewise.
11753 * print-tree.c: Likewise.
11754 * statistics.c: Likewise.
11755 * tree-cfg.c: Likewise.
11756 * tree-cfg.h: Likewise.
11757 * tree-dfa.c: Likewise.
11758 * tree-dfa.h: Likewise.
11759 * tree-dump.c (dump_function): Likewise.
11760 * tree-dump.h (struct dump_info): Likewise.
11761 * tree-pretty-print.c: Likewise.
11762 * tree-pretty-print.h: Likewise.
11763 * tree-ssa-live.c: Likewise.
11764 * tree-ssa-live.h: Likewise.
11765 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
11766 * tree-vect-loop.c: Likewise.
11767 * tree-vect-slp.c: Likewise.
11769 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
11770 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11772 PR tree-optimization/80457
11773 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
11774 of all arguments to a statement as scalar_to_vec operations.
11775 (vectorizable_call): Adjust call to vect_model_simple_cost for
11777 (vectorizable_conversion): Likewise.
11778 (vectorizable_assignment): Likewise.
11779 (vectorizable_shift): Likewise.
11780 (vectorizable_operation): Likewise.
11781 (vectorizable_comparison): Likewise.
11782 (vect_is_simple_cond): Record the def types for operands.
11783 (vectorizable_condition): Likewise, call vect_model_simple_cost.
11784 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
11785 for statement argument count.
11787 2017-05-16 Carl Love <cel@us.ibm.com>
11789 * config/rs6000/rs6000-c: Add support for built-in functions
11790 vector unsigned long long vec_bperm (vector unsigned long long,
11791 vector unsigned char)
11792 vector signed long long vec_mule (vector signed int,
11794 vector unsigned long long vec_mule (vector unsigned int,
11795 vector unsigned int)
11796 vector signed long long vec_mulo (vector signed int,
11798 vector unsigned long long vec_mulo (vector unsigned int,
11799 vector unsigned int)
11800 vector signed char vec_sldw (vector signed char,
11801 vector signed char,
11803 vector unsigned char vec_sldw (vector unsigned char,
11804 vector unsigned char,
11806 vector signed short vec_sldw (vector signed short,
11807 vector signed short,
11809 vector unsigned short vec_sldw (vector unsigned short,
11810 vector unsigned short,
11812 vector signed int vec_sldw (vector signed int,
11815 vector unsigned int vec_sldw (vector unsigned int,
11816 vector unsigned int,
11818 vector signed long long vec_sldw (vector signed long long,
11819 vector signed long long,
11821 vector unsigned long long vec_sldw (vector unsigned long long,
11822 vector unsigned long long,
11824 * config/rs6000/rs6000-c: Add support for built-in functions
11825 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
11826 * config/rs6000/altivec.h: Add defintion for vec_sldw.
11827 * doc/extend.texi: Update the built-in documentation for the
11828 new built-in functions.
11830 2017-05-16 Marek Polacek <polacek@redhat.com>
11834 * tree.c (save_expr): Don't fold the expression.
11836 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
11838 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
11839 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
11840 and (?*y,m). Update insn attributes.
11842 2017-05-16 Martin Liska <mliska@suse.cz>
11844 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
11845 flags argument of print_gimple_stmt, print_gimple_expr,
11846 print_generic_stmt and print_generic_expr.
11847 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
11848 * coretypes.h: Likewise.
11849 * except.c (dump_eh_tree): Likewise.
11850 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11851 * gimple-pretty-print.h: Likewise.
11852 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
11853 (backprop::push_to_worklist): Likewise.
11854 (backprop::pop_from_worklist): Likewise.
11855 (backprop::process_use): Likewise.
11856 (backprop::intersect_uses): Likewise.
11857 (note_replacement): Likewise.
11858 * gimple-ssa-store-merging.c
11859 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
11860 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
11861 (pass_store_merging::execute): Likewise.
11862 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
11863 (ssa_base_cand_dump_callback): Likewise.
11864 (dump_incr_vec): Likewise.
11865 (replace_refs): Likewise.
11866 (replace_mult_candidate): Likewise.
11867 (create_add_on_incoming_edge): Likewise.
11868 (create_phi_basis): Likewise.
11869 (insert_initializers): Likewise.
11870 (all_phi_incrs_profitable): Likewise.
11871 (introduce_cast_before_cand): Likewise.
11872 (replace_one_candidate): Likewise.
11873 * gimplify.c (gimplify_expr): Likewise.
11874 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
11875 (set_rename): Likewise.
11876 (rename_uses): Likewise.
11877 (copy_loop_phi_nodes): Likewise.
11878 (add_close_phis_to_merge_points): Likewise.
11879 (copy_loop_close_phi_args): Likewise.
11880 (copy_cond_phi_args): Likewise.
11881 (graphite_copy_stmts_from_block): Likewise.
11882 (translate_pending_phi_nodes): Likewise.
11883 * graphite-poly.c (print_pdr): Likewise.
11884 (dump_gbb_cases): Likewise.
11885 (dump_gbb_conditions): Likewise.
11886 (print_scop_params): Likewise.
11887 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
11888 (build_cross_bb_scalars_use): Likewise.
11889 (gather_bbs::before_dom_children): Likewise.
11890 * hsa-dump.c (dump_hsa_immed): Likewise.
11891 * ipa-cp.c (print_ipcp_constant_value): Likewise.
11892 (get_replacement_map): Likewise.
11893 * ipa-inline-analysis.c (dump_condition): Likewise.
11894 (estimate_function_body_sizes): Likewise.
11895 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
11896 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
11897 * ipa-prop.c (ipa_dump_param): Likewise.
11898 (ipa_print_node_jump_functions_for_edge): Likewise.
11899 (ipa_modify_call_arguments): Likewise.
11900 (ipa_modify_expr): Likewise.
11901 (ipa_dump_param_adjustments): Likewise.
11902 (ipa_dump_agg_replacement_values): Likewise.
11903 (ipcp_modif_dom_walker::before_dom_children): Likewise.
11904 * ipa-pure-const.c (check_stmt): Likewise.
11905 (pass_nothrow::execute): Likewise.
11906 * ipa-split.c (execute_split_functions): Likewise.
11907 * omp-offload.c (dump_oacc_loop_part): Likewise.
11908 (dump_oacc_loop): Likewise.
11909 * trans-mem.c (tm_log_emit): Likewise.
11910 (tm_memopt_accumulate_memops): Likewise.
11911 (dump_tm_memopt_set): Likewise.
11912 (dump_tm_memopt_transform): Likewise.
11913 * tree-cfg.c (gimple_verify_flow_info): Likewise.
11914 (print_loop): Likewise.
11915 * tree-chkp-opt.c (chkp_print_addr): Likewise.
11916 (chkp_gather_checks_info): Likewise.
11917 (chkp_get_check_result): Likewise.
11918 (chkp_remove_check_if_pass): Likewise.
11919 (chkp_use_outer_bounds_if_possible): Likewise.
11920 (chkp_reduce_bounds_lifetime): Likewise.
11921 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
11922 (chkp_mark_completed_bounds): Likewise.
11923 (chkp_register_incomplete_bounds): Likewise.
11924 (chkp_mark_invalid_bounds): Likewise.
11925 (chkp_maybe_copy_and_register_bounds): Likewise.
11926 (chkp_build_returned_bound): Likewise.
11927 (chkp_get_bound_for_parm): Likewise.
11928 (chkp_build_bndldx): Likewise.
11929 (chkp_get_bounds_by_definition): Likewise.
11930 (chkp_generate_extern_var_bounds): Likewise.
11931 (chkp_get_bounds_for_decl_addr): Likewise.
11932 * tree-chrec.c (chrec_apply): Likewise.
11933 * tree-data-ref.c (dump_data_reference): Likewise.
11934 (dump_subscript): Likewise.
11935 (dump_data_dependence_relation): Likewise.
11936 (analyze_overlapping_iterations): Likewise.
11937 * tree-inline.c (expand_call_inline): Likewise.
11938 (tree_function_versioning): Likewise.
11939 * tree-into-ssa.c (dump_defs_stack): Likewise.
11940 (dump_currdefs): Likewise.
11941 (dump_names_replaced_by): Likewise.
11942 (dump_update_ssa): Likewise.
11943 (update_ssa): Likewise.
11944 * tree-object-size.c (pass_object_sizes::execute): Likewise.
11945 * tree-parloops.c (build_new_reduction): Likewise.
11946 (try_create_reduction_list): Likewise.
11947 (ref_conflicts_with_region): Likewise.
11948 (oacc_entry_exit_ok_1): Likewise.
11949 (oacc_entry_exit_single_gang): Likewise.
11950 * tree-pretty-print.h: Likewise.
11951 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
11952 (get_scalar_evolution): Likewise.
11953 (add_to_evolution): Likewise.
11954 (get_loop_exit_condition): Likewise.
11955 (analyze_evolution_in_loop): Likewise.
11956 (analyze_initial_condition): Likewise.
11957 (analyze_scalar_evolution): Likewise.
11958 (instantiate_scev): Likewise.
11959 (number_of_latch_executions): Likewise.
11960 (gather_chrec_stats): Likewise.
11961 (final_value_replacement_loop): Likewise.
11962 (scev_const_prop): Likewise.
11963 * tree-sra.c (dump_access): Likewise.
11964 (disqualify_candidate): Likewise.
11965 (create_access): Likewise.
11966 (reject): Likewise.
11967 (maybe_add_sra_candidate): Likewise.
11968 (create_access_replacement): Likewise.
11969 (analyze_access_subtree): Likewise.
11970 (analyze_all_variable_accesses): Likewise.
11971 (sra_modify_assign): Likewise.
11972 (initialize_constant_pool_replacements): Likewise.
11973 (find_param_candidates): Likewise.
11974 (decide_one_param_reduction): Likewise.
11975 (replace_removed_params_ssa_names): Likewise.
11976 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
11977 * tree-ssa-copy.c (dump_copy_of): Likewise.
11978 (copy_prop_visit_cond_stmt): Likewise.
11979 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
11980 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
11981 (record_equivalences_from_stmt): Likewise.
11982 * tree-ssa-dse.c (compute_trims): Likewise.
11983 (delete_dead_call): Likewise.
11984 (delete_dead_assignment): Likewise.
11985 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
11986 (forward_propagate_into_cond): Likewise.
11987 (pass_forwprop::execute): Likewise.
11988 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11989 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
11991 (move_computations_worker): Likewise.
11992 (execute_sm): Likewise.
11993 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
11994 (remove_exits_and_undefined_stmts): Likewise.
11995 (remove_redundant_iv_tests): Likewise.
11996 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
11997 (adjust_iv_update_pos): Likewise.
11998 * tree-ssa-math-opts.c (bswap_replace): Likewise.
11999 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12000 (value_replacement): Likewise.
12001 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12002 * tree-ssa-pre.c (print_pre_expr): Likewise.
12003 (get_representative_for): Likewise.
12004 (create_expression_by_pieces): Likewise.
12005 (insert_into_preds_of_block): Likewise.
12006 (eliminate_insert): Likewise.
12007 (eliminate_dom_walker::before_dom_children): Likewise.
12008 (eliminate): Likewise.
12009 (remove_dead_inserted_code): Likewise.
12010 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12011 * tree-ssa-reassoc.c (get_rank): Likewise.
12012 (eliminate_duplicate_pair): Likewise.
12013 (eliminate_plus_minus_pair): Likewise.
12014 (eliminate_not_pairs): Likewise.
12015 (undistribute_ops_list): Likewise.
12016 (eliminate_redundant_comparison): Likewise.
12017 (update_range_test): Likewise.
12018 (optimize_range_tests_var_bound): Likewise.
12019 (optimize_vec_cond_expr): Likewise.
12020 (rewrite_expr_tree): Likewise.
12021 (rewrite_expr_tree_parallel): Likewise.
12022 (linearize_expr): Likewise.
12023 (break_up_subtract): Likewise.
12024 (linearize_expr_tree): Likewise.
12025 (attempt_builtin_powi): Likewise.
12026 (attempt_builtin_copysign): Likewise.
12027 (transform_stmt_to_copy): Likewise.
12028 (transform_stmt_to_multiply): Likewise.
12029 (dump_ops_vector): Likewise.
12030 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12031 (print_scc): Likewise.
12032 (set_ssa_val_to): Likewise.
12033 (visit_reference_op_store): Likewise.
12034 (visit_use): Likewise.
12035 (sccvn_dom_walker::before_dom_children): Likewise.
12036 (run_scc_vn): Likewise.
12037 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12039 (expr_hash_elt::print): Likewise.
12040 (const_and_copies::pop_to_marker): Likewise.
12041 (const_and_copies::record_const_or_copy_raw): Likewise.
12042 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12043 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12044 (dump_predicates): Likewise.
12045 (find_uninit_use): Likewise.
12046 (warn_uninitialized_phi): Likewise.
12047 (pass_late_warn_uninitialized::execute): Likewise.
12048 * tree-ssa.c (verify_vssa): Likewise.
12049 (verify_ssa): Likewise.
12050 (maybe_optimize_var): Likewise.
12051 * tree-vrp.c (dump_value_range): Likewise.
12052 (dump_all_value_ranges): Likewise.
12053 (dump_asserts_for): Likewise.
12054 (register_edge_assert_for_2): Likewise.
12055 (vrp_visit_cond_stmt): Likewise.
12056 (vrp_visit_switch_stmt): Likewise.
12057 (vrp_visit_stmt): Likewise.
12058 (vrp_visit_phi_node): Likewise.
12059 (simplify_cond_using_ranges_1): Likewise.
12060 (fold_predicate_in): Likewise.
12061 (evrp_dom_walker::before_dom_children): Likewise.
12062 (evrp_dom_walker::push_value_range): Likewise.
12063 (evrp_dom_walker::pop_value_range): Likewise.
12064 (execute_early_vrp): Likewise.
12066 2017-05-16 Richard Biener <rguenther@suse.de>
12068 * dwarf2out.c (loc_list_from_tree_1): Do not create
12069 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12071 2017-05-16 Richard Biener <rguenther@suse.de>
12073 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12075 (note_variable_value_in_expr): If we resolved the decl ref
12076 do not push to the stack.
12078 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12080 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12081 operations in fast-math mode.
12082 (vaddq_f16): Likewise.
12083 (vmul_f16): Likewise.
12084 (vmulq_f16): Likewise.
12085 (vsub_f16): Likewise.
12086 (vsubq_f16): Likewise.
12087 * config/arm/neon.md (add<mode>3): New.
12089 (fma:<VH:mode>3): New. Also remove outdated comment.
12092 2017-05-16 Martin Liska <mliska@suse.cz>
12096 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12097 (odr_types_equivalent_p): Likewise.
12099 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12101 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12103 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12106 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12107 non-interunit SSE move alternatives with '?'.
12108 (zero-extendsidi peephole2): New peephole to skip intermediate
12109 general register in SSE zero-extend sequence.
12111 2017-05-15 Jeff Law <law@redhat.com>
12113 * reorg.c (relax_delay_slots): Create a new variable to hold
12114 the temporary target rather than clobbering TARGET_LABEL.
12116 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12117 missing argument to extract_bit_field call.
12118 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12120 2017-05-15 Martin Liska <mliska@suse.cz>
12123 * gcc.c (process_command): Do not allow empty argument of -o option.
12125 2017-05-15 Renlin Li <renlin.li@arm.com>
12127 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12128 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12129 * config/aarch64/constraints.md (Usf): Add long call check.
12130 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12131 (call_value): Likewise.
12132 (sibcall): Likewise.
12133 (sibcall_value): Likewise.
12135 (call_value_insn): New.
12136 (sibcall_insn): Update rtx pattern.
12137 (sibcall_value_insn): Likewise.
12138 (call_internal): Remove.
12139 (call_value_internal): Likewise.
12140 (sibcall_internal): Likewise.
12141 (sibcall_value_internal): Likewise.
12142 (call_reg): Likewise.
12143 (call_symbol): Likewise.
12144 (call_value_reg): Likewise.
12145 (call_value_symbol): Likewise.
12147 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12150 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12152 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12154 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12155 compatible with CCGOCmode and with CCZmode.
12157 2017-05-14 Martin Sebor <msebor@redhat.com>
12159 PR middle-end/77671
12160 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12161 (gimple_fold_builtin_snprintf): Same.
12162 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12163 (gimple_fold_builtin_snprintf): Same.
12164 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12165 of character types.
12166 (is_call_safe): New function.
12167 (try_substitute_return_value): Call it.
12168 (try_simplify_call): New function.
12169 (pass_sprintf_length::handle_gimple_call): Call it.
12171 2017-05-14 Martin Sebor <msebor@redhat.com>
12173 PR middle-end/80669
12174 * builtins.c (expand_builtin_stpncpy): Simplify.
12176 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12178 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12179 * config/i386/i386.h
12180 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12181 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12182 (struct machine_function): Add new members call_ms2sysv,
12183 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12184 (struct machine_frame_state): New fields sp_realigned and
12185 sp_realigned_offset.
12186 * config/i386/i386.c
12187 (enum xlogue_stub): New enum.
12188 (enum xlogue_stub_sets): New enum.
12189 (class xlogue_layout): New class.
12190 (struct ix86_frame): New fields stack_realign_allocate_offset,
12191 stack_realign_offset and outlined_save_offset. Modify comments to
12192 detail stack layout when using out-of-line stubs.
12193 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12194 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12195 -mcall-ms2sysv-xlogues.
12196 (stub_managed_regs): New static variable.
12197 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12198 registers managed by out-of-line stub.
12199 (disable_call_ms2sysv_xlogues): New function.
12200 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12201 m->call_ms2sysv when appropriate and compute frame layout for
12203 (sp_valid_at, fp_valid_at): New inline functions.
12204 (choose_basereg): New function.
12205 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12207 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12208 Use align parameter of choose_baseaddr to generated aligned SSE movs
12210 (pro_epilogue_adjust_stack): Modify to track
12211 machine_frame_state::sp_realigned.
12212 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12213 (ix86_nsaved_sseregs): Likewise.
12214 (ix86_emit_save_regs): Likewise.
12215 (ix86_emit_save_regs_using_mov): Likewise.
12216 (ix86_emit_save_sse_regs_using_mov): Likewise.
12217 (get_scratch_register_on_entry): Likewise.
12218 (gen_frame_set): New function.
12219 (gen_frame_load): Likewise.
12220 (gen_frame_store): Likewise.
12221 (emit_outlined_ms2sysv_save): Likewise.
12222 (emit_outlined_ms2sysv_restore): Likewise.
12223 (ix86_expand_prologue): Modify stack re-alignment code and call
12224 emit_outlined_ms2sysv_save when appropriate.
12225 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12226 parameter rtx_insn *insn, which allows the function to be used to only
12227 generate the notes.
12228 (ix86_expand_epilogue): Modify validity checks of frame and stack
12229 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12230 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12231 * config/i386/predicates.md
12232 (save_multiple): New predicate.
12233 (restore_multiple): Likewise.
12234 * config/i386/sse.md
12235 (save_multiple<mode>): New pattern.
12236 (save_multiple_realign<mode>): Likewise.
12237 (restore_multiple<mode>): Likewise.
12238 (restore_multiple_and_return<mode>): Likewise.
12239 (restore_multiple_leave_return<mode>): Likewise.
12240 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12242 2017-05-14 Julia Koval <julia.koval@intel.com>
12244 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12245 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12246 (__builtin_ia32_xsetbv): New builtins.
12247 * config/i386/i386.c (ix86_expand_special_args_builtin):
12249 (ix86_expand_builtin): Special expand for new intrinsics.
12250 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12251 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12252 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12256 * cfganal.c (inverted_post_order_compute): Change argument type
12258 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12259 * df-core.c (rest_of_handle_df_initialize): Adjust.
12260 (rest_of_handle_df_finish): Likewise.
12261 (df_analyze_1): Likewise.
12262 (df_analyze): Likewise.
12263 (loop_inverted_post_order_compute): Change argument to be a vec *.
12264 (df_analyze_loop): Adjust.
12265 (df_get_n_blocks): Likewise.
12266 (df_get_postorder): Likewise.
12267 * df.h (struct df_d): Change field to be a vec.
12268 * lcm.c (compute_laterin): Adjust.
12269 (compute_available): Likewise.
12270 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12271 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12272 * tree-ssa-pre.c (compute_antic): Likewise.
12274 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12276 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12277 (depth_first_search::depth_first_search): Change structure init
12278 function to this constructor.
12279 (depth_first_search::add_bb): Rename function to this member.
12280 (depth_first_search::execute): Likewise.
12281 (flow_dfs_compute_reverse_finish): Adjust.
12283 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12285 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12286 (longest_simple_path): Likewise.
12287 * shrink-wrap.c (spread_components): Likewise.
12288 (disqualify_problematic_components): Likewise.
12289 (emit_common_heads_for_components): Likewise.
12290 (emit_common_tails_for_components): Likewise.
12291 (insert_prologue_epilogue_for_components): Likewise.
12293 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12295 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12298 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12300 * df-core.c (df_set_blocks): Start using auto_bitmap.
12301 (df_compact_blocks): Likewise.
12302 * df-problems.c (df_rd_confluence_n): Likewise.
12303 * df-scan.c (df_insn_rescan_all): Likewise.
12304 (df_process_deferred_rescans): Likewise.
12305 (df_update_entry_block_defs): Likewise.
12306 (df_update_exit_block_uses): Likewise.
12307 (df_entry_block_bitmap_verify): Likewise.
12308 (df_exit_block_bitmap_verify): Likewise.
12309 (df_scan_verify): Likewise.
12310 * lra-constraints.c (lra_constraints): Likewise.
12311 (undo_optional_reloads): Likewise.
12312 (lra_undo_inheritance): Likewise.
12313 * lra-remat.c (calculate_gen_cands): Likewise.
12314 (do_remat): Likewise.
12315 * lra-spills.c (assign_spill_hard_regs): Likewise.
12316 (spill_pseudos): Likewise.
12317 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12318 (bitmap_set_subtract_values): Likewise.
12320 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12322 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12323 management with auto_bitmap.
12324 (fix_inter_tick): Likewise.
12325 (fix_recovery_deps): Likewise.
12326 * ira.c (add_store_equivs): Likewise.
12327 (find_moveable_pseudos): Likewise.
12328 (split_live_ranges_for_shrink_wrap): Likewise.
12329 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12330 (rtx_reuse_manager::seen_def_p): Likewise.
12331 (rtx_reuse_manager::set_seen_def): Likewise.
12332 * print-rtl.h (class rtx_reuse_manager): Likewise.
12334 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12336 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12338 (migrate_btr_def): Likewise.
12339 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12340 * df-core.c (loop_post_order_compute): Likewise.
12341 (loop_inverted_post_order_compute): Likewise.
12342 * hsa-common.h: Likewise.
12343 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12344 * init-regs.c (initialize_uninitialized_regs): Likewise.
12345 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12346 (inline_small_functions): Likewise.
12347 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12348 * ira.c (combine_and_move_insns): Likewise.
12349 (build_insn_chain): Likewise.
12350 * loop-invariant.c (find_invariants): Likewise.
12351 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12352 * predict.c (tree_predict_by_opcode): Likewise.
12353 (predict_paths_leading_to): Likewise.
12354 (predict_paths_leading_to_edge): Likewise.
12355 (estimate_loops_at_level): Likewise.
12356 (estimate_loops): Likewise.
12357 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12358 (spread_components): Likewise.
12359 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12360 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12361 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12362 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12363 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12364 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12365 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12366 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12367 (mark_threaded_blocks): Likewise.
12368 (thread_through_all_blocks): Likewise.
12369 * tree-ssa.c (verify_ssa): Likewise.
12370 (execute_update_addresses_taken): Likewise.
12371 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12373 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12375 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12377 (post_order_compute): Likewise.
12378 (inverted_post_order_compute): Likewise.
12379 (pre_and_rev_post_order_compute_fn): Likewise.
12381 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12383 * genrecog.c (int_set::int_set): Explicitly construct our
12384 auto_vec base class.
12385 * vec.h (auto_vec::auto_vec): New constructor.
12387 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12389 * bitmap.h (class auto_bitmap): New constructor taking
12390 bitmap_obstack * argument.
12392 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12394 * bitmap.h (class auto_bitmap): Change type of m_bits to
12395 bitmap_head, and adjust ctor / dtor and member operators.
12397 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12399 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12400 when returned register mode doesn't match original mode.
12402 2017-05-12 Jeff Law <law@redhat.com>
12403 Jakub Jelinek <jakub@redhat.com>
12405 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
12406 we look for cc setter after the compare-elim changes.
12407 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
12408 within the vector to match what compare-elim now expects.
12409 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
12410 (xorsi3_flags, one_cmplsi2_flags): Likewise.
12412 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
12413 after the compare-elim changes.
12414 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
12415 the vector to match what compare-elim now expects.
12416 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
12417 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
12418 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
12419 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
12420 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
12422 * config/visium/visium.c (single_set_and_flags): Fix where
12423 we look for cc setter after the compare-elim changes.
12424 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
12425 with the vector to match what compare-elim now expects.
12426 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
12427 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
12428 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
12429 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
12430 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
12431 (neg<mode>2_insn_set_overflow): Likewise.
12433 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
12435 PR middle-end/79794
12436 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
12437 maybe_expand_insn call, set ops[0].target. If still set after call,
12438 set alt_rtl. Add extra arg to recursive calls.
12439 (extract_bit_field): Add alt_rtl argument. Pass to
12441 * expmed.h (extract_bit_field): Fix prototype.
12442 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
12443 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
12444 to extract_bit_field_calls.
12445 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
12446 Pass alt_rtl to extract_bit_field calls.
12447 * calls.c (store_unaligned_arguments_into_psuedos)
12448 load_register_parameters): Pass extra NULL to extract_bit_field calls.
12449 * optabs.c (maybe_legitimize_operand): Clear op->target when call
12451 * optabs.h (struct expand_operand): Add target bitfield.
12453 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12455 * compare-elim.c (try_eliminate_compare): Canonicalize
12456 operation with embedded compare to
12457 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
12458 (set (reg) (operation)].
12460 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
12462 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12465 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
12466 cost of adding a carry flag for ADC instruction.
12467 [case MINUS]: Ignore the cost of subtracting a carry flag
12468 for SBB instruction.
12470 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
12472 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
12474 * config/rs6000/bmiintrin.h: New file.
12475 * config/rs6000/bmi2intrin.h: New file.
12476 * config/rs6000/x86intrin.h: New file.
12478 2017-05-12 Jeff Law <law@redhat.com>
12480 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
12483 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
12485 PR middle-end/80707
12486 * tree-cfg.c: Remove cfg edges of unreachable case statements.
12488 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12490 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12491 early expansion of vector divide builtins.
12492 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
12493 builtins identified as having unsigned arguments.
12495 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12497 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
12498 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
12499 expansion of vector logical operations (and, andc, or, xor,
12502 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12504 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
12505 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
12507 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12510 early GIMPLE expansion of vector multiplies.
12512 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12514 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
12515 TARGET_HAVE_MOVT conditional.
12516 (movt splitter): Likewise.
12518 2017-05-12 Richard Biener <rguenther@suse.de>
12520 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
12521 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12522 Fold all stmts not inplace.
12524 2017-05-12 Richard Biener <rguenther@suse.de>
12526 PR tree-optimization/80713
12527 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
12528 inserted_exprs bit for not removed stmts.
12530 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
12532 PR middle-end/69921
12533 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
12534 parallelized" attribute for parallelized OpenACC kernels.
12535 * omp-offload.c (execute_oacc_device_lower): Use it.
12537 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
12538 Set "oacc kernels" attribute.
12539 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
12540 parameter. Adjust all users.
12541 (oacc_fn_attrib_kernels_p): Remove function.
12542 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
12543 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
12544 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
12545 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
12546 assert "oacc kernels" attribute is set.
12548 2017-05-11 Carl Love <cel@us.ibm.com>
12550 * config/rs6000/rs6000-c: Add support for built-in functions
12551 vector unsigned char vec_popcnt (vector signed char)
12552 vector unsigned char vec_popcnt (vector unsigned char)
12553 vector unsigned short vec_popcnt (vector signed short)
12554 vector unsigned short vec_popcnt (vector unsigned short)
12555 vector unsigned int vec_popcnt (vector signed int)
12556 vector unsigned int vec_popcnt (vector unsigned int)
12557 vector unsigned long long vec_popcnt (vector signed long long)
12558 vector unsigned long long vec_popcnt (vector unsigned long long)
12559 vector signed long long vec_slo (vector signed long long,
12560 vector signed char)
12561 vector signed long long vec_slo (vector signed long long,
12562 vector unsigned char)
12563 vector unsigned long long vec_slo (vector unsigned long long,
12564 vector signed char)
12565 vector unsigned long long vec_slo (vector unsigned long long,
12566 vector unsigned char)
12567 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
12568 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
12569 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
12570 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
12571 * doc/extend.texi: Update the built-in documentation file for the
12572 new built-in functions.
12574 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12576 * attribs.h (sorted_attr_string): Move machine independent
12577 functions for target clone support from the i386 port to common
12578 code. Rename ix86_function_versions to common_function_versions.
12579 Rename make_name to make_unique_name.
12580 (common_function_versions): Likewise.
12581 (make_unique_name): Likewise.
12582 (make_dispatcher_decl): Likewise.
12583 (is_function_default_version): Likewise.
12584 * attribs.c (attr_strcmp): Likewise.
12585 (sorted_attr_string): Likewise.
12586 (common_function_versions): Likewise.
12587 (make_unique_name): Likewise.
12588 (make_dispatcher_decl): Likewise.
12589 (is_function_default_version): Likewise.
12590 * config/i386/i386.c (attr_strcmp): Likewise.
12591 (sorted_attr_string): Likewise.
12592 (ix86_function_versions): Likewise.
12593 (make_name): Likewise.
12594 (make_dispatcher_decl): Likewise.
12595 (is_function_default_version): Likewise.
12596 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
12598 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12601 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12602 Account for direct move costs for vec_construct of integer
12605 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
12608 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
12609 (UNSPEC_STX_ATOMIC): Ditto.
12610 (loaddi_via_sse): New insn.
12611 (storedi_via_sse): Ditto.
12612 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
12613 Update corresponding peephole2 patterns.
12614 (atomic_storedi_fpu): Ditto.
12616 2017-05-11 Julia Koval <julia.koval@intel.com>
12618 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
12619 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
12621 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
12622 (__builtin_ia32_rsqrt14ss_mask): New builtins.
12623 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
12625 2017-05-11 Nathan Sidwell <nathan@acm.org>
12627 * graphite-poly.c: Include dumpfile.h.
12629 * dumpfle.h (dump_function): Declare here ...
12630 * tree-dump.h (dump_function): ... not here.
12631 * dumpfile.c: #include tree-cfg.h.
12632 (dump_function): Move here from ...
12633 * tree-dump.c (dump_function): ... here.
12634 * gimplify.c: #include splay-tree.h, not tree-dump.h.
12635 * graphite-poly.c: Don't include tree-dump.h.
12636 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
12637 * print-tree.c: Likewise.
12638 * stor-layout.c: Likewise.
12639 * tree-nested.c: Likewise.
12641 * dumpfile.c (dump_start): Use TDF_FLAGS.
12642 (dump_enable_all): Fix TDF_KIND check thinko.
12644 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12646 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12647 array entries to represent two legal parameterizations of the
12648 overloaded __builtin_cmpb function, as represented by the
12649 P6_OV_BUILTIN_CMPB constant.
12650 (altivec_resolve_overloaded_builtin): Add special case handling
12651 for the __builtin_cmpb function, as represented by the
12652 P6_OV_BUILTIN_CMPB constant.
12653 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
12654 (BU_P6_64BIT_2): New macro.
12655 (BU_P6_OVERLOAD_2): New macro
12656 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
12657 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
12658 (CMPB): Add overload support to represent both 32-bit and 64-bit
12659 compare-bytes function.
12660 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12661 support for TARGET_CMPB.
12662 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
12663 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12664 documentation of the __builtin_cmpb overloaded built-in function.
12666 2017-05-11 Richard Biener <rguenther@suse.de>
12668 PR tree-optimization/80705
12669 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
12670 bases are not vectorizable.
12672 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12674 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
12675 when counting register pressure.
12677 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12679 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
12680 (iv_ca_more_deps): Renamed to ...
12681 (iv_ca_compare_deps): ... this.
12682 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
12684 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12686 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
12688 (determine_group_iv_costs): ... here.
12689 (find_inv_vars_cb): Record inv var if it's not recorded before.
12691 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12693 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
12694 (get_shiftadd_cost): Ditto.
12696 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12698 * tree-ssa-address.c: Include header file.
12699 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
12701 (add_to_parts): Refactor.
12702 (addr_to_parts): New parameter. Update use of move_hint_to_base.
12703 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
12706 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12708 PR tree-optimization/53090
12709 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
12711 (extract_cond_operands): Detect condition with IV on both sides
12712 and return COMP_IV_EXPR_2.
12713 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
12714 (rewrite_use_compare): Simplify by removing call to function
12715 extract_cond_operands.
12717 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12719 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
12720 (extract_cond_operands): Detect condition comparing against non-
12721 invariant bound and return appropriate enum value.
12722 (find_interesting_uses_cond): Update use of extract_cond_operands.
12723 Handle its return value accordingly.
12724 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
12726 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12728 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
12729 nonlinear iv_use computation in loop invariant sensitive way.
12731 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12733 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
12734 (find_iv_candidates): Call relate_compare_use_with_all_cands.
12736 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12738 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
12739 (dump_cand): Support iv_cand.inv_exprs.
12740 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
12742 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
12745 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12747 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
12749 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
12750 as local function. Include necessary header files.
12751 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
12753 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12755 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
12757 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12759 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
12760 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
12761 RSHIFT_EXPR and BIT_NOT_EXPR.
12763 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12765 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
12766 (adjust_setup_cost): New parameter supporting round up adjustment.
12767 (struct address_cost_data): Delete.
12768 (force_expr_to_var_cost): Don't bound cost with spill_cost.
12769 (split_address_cost, ptr_difference_cost): Delete.
12770 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
12771 (struct ainc_cost_data): New struct.
12772 (get_address_cost_ainc): New function.
12773 (get_address_cost, get_computation_cost): Reimplement.
12774 (determine_group_iv_cost_address): Record inv_expr for all uses of
12776 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
12777 (iv_ca_has_deps): Reimplemented to ...
12778 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
12780 (iv_ca_extend): Call iv_ca_more_deps.
12782 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12784 * tree-ssa-address.c (struct mem_address): Move to header file.
12785 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
12786 * tree-ssa-address.h (struct mem_address): Move from C file.
12787 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
12789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12791 * tree-affine.h (aff_combination_type): New interface.
12792 (aff_combination_zero_p): Remove static.
12793 (aff_combination_const_p): New interface.
12794 (aff_combination_singleton_var_p): New interfaces.
12796 2017-05-11 Richard Biener <rguenther@suse.de>
12798 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12799 Skip unreachable blocks and destinations.
12800 (eliminate): Move stmt removal and fixup ...
12801 (fini_eliminate): ... here. Skip inserted exprs.
12802 (pass_pre::execute): Move fini_pre after fini_eliminate.
12803 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
12804 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
12805 PRE to get rid of dead code that has invalid SSA form and
12806 split critical edges again.
12808 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12810 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
12812 2017-05-11 Richard Biener <rguenther@suse.de>
12814 * passes.c (execute_function_todo): Verify loops if they are
12815 said to be up-to-date.
12816 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
12817 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
12819 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
12822 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
12823 handle calling assemble_external ourself.
12826 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
12827 modes with zero size. Enhance comment.
12829 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12831 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
12832 built-ins for vec_xl and vec_xst with short and char pointer
12835 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
12837 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
12838 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
12839 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
12840 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
12841 (_mm_maskz_min_round_ss): New intrinsics.
12842 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
12843 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
12844 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
12845 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
12846 (__builtin_ia32_minss_mask_round): New builtins.
12847 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
12848 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
12849 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
12851 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
12852 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
12854 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
12857 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
12859 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
12860 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
12861 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
12862 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
12863 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
12864 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
12865 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
12866 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
12867 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
12868 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
12869 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
12870 (__builtin_ia32_mulss_mask_round): New builtins.
12871 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
12872 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
12873 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
12875 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
12876 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
12878 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
12881 2017-05-10 Julia Koval <julia.koval@intel.com>
12883 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
12884 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
12885 (_mm256_setr_m128i): New intrinsics.
12887 2017-05-10 Julia Koval <julia.koval@intel.com>
12889 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
12890 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
12891 (_mm_maskz_rcp14_ss): New intrinsics.
12892 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
12893 (__builtin_ia32_rcp14ss_mask): New builtins.
12894 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
12896 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
12898 PR tree-optimization/51513
12899 * tree-cfg.c (gimple_seq_unreachable_p): New function.
12900 (assert_unreachable_fallthru_edge_p): Use it.
12901 (group_case_labels_stmt): Likewise.
12902 * tree-cfg.h: Prototype it.
12903 * stmt.c: Include cfghooks.h and tree-cfg.h.
12904 (emit_case_dispatch_table) <gap_label>: New local variable.
12905 Use it to fill dispatch table gaps.
12906 Test for default_label before updating probabilities.
12907 (expand_case) <default_label>: Remove unneeded initialization.
12908 Test for unreachable default case statement and remove its edge.
12909 Set default_label accordingly.
12910 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
12912 2017-05-10 Carl Love <cel@us.ibm.com>
12914 * config/rs6000/rs6000-c: Add support for built-in functions
12915 vector signed char vec_neg (vector signed char)
12916 vector signed short int vec_neg (vector short int)
12917 vector signed int vec_neg (vector signed int)
12918 vector signed long long vec_neg (vector signed long long)
12919 vector float vec_neg (vector float)
12920 vector double vec_neg (vector double)
12921 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
12923 * config/rs6000/altivec.h: Add define for vec_neg
12924 * doc/extend.texi: Update the built-in documentation for the
12925 new built-in functions.
12927 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12929 PR tree-optimization/77644
12930 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
12932 2017-05-10 Nathan Sidwell <nathan@acm.org>
12934 * dumpfile.h (TDI_lang_all): New.
12935 (TDF_KIND): New. Renumber others
12936 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
12938 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
12940 (get_dump_file_name): Adjust suffix generation.
12941 (dump_enable_all): Use TDF_KIND.
12942 * doc/invoke.texi (-fdump-lang-all): Document.
12944 * dumpfile.h: Tabify.
12946 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
12949 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
12950 Move member access before delete.
12952 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
12954 * tree-inline.c (expand_call_inline): Split block at stmt
12957 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12960 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
12961 are now unused after splitting mov{sf,sd}_hardfloat.
12962 (f32_lr2): Likewise.
12963 (f32_lm): Likewise.
12964 (f32_lm2): Likewise.
12965 (f32_li): Likewise.
12966 (f32_li2): Likewise.
12967 (f32_lv): Likewise.
12968 (f32_sr): Likewise.
12969 (f32_sr2): Likewise.
12970 (f32_sm): Likewise.
12971 (f32_sm2): Likewise.
12972 (f32_si): Likewise.
12973 (f32_si2): Likewise.
12974 (f32_sv): Likewise.
12975 (f32_dm): Likewise.
12976 (f32_vsx): Likewise.
12977 (f32_av): Likewise.
12978 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
12979 For movsf, order stores so the VSX stores occur before the GPR
12980 store which encourages the register allocator to use a traditional
12981 FPR instead of a GPR. For movsd, order the stores so that the GPR
12982 store comes before the VSX stores to allow the power6 to work.
12983 This is due to the power6 not having a 32-bit integer store
12984 instruction from a FPR.
12985 (movsf_hardfloat): Likewise.
12986 (movsd_hardfloat): Likewise.
12988 2017-05-09 Martin Sebor <msebor@redhat.com>
12990 PR translation/80280
12991 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
12994 PR translation/80280
12995 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
12996 data member added in r247778.
12997 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
12999 2017-05-09 Nathan Sidwell <nathan@acm.org>
13001 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13003 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13006 2017-05-09 Marek Polacek <polacek@redhat.com>
13008 * doc/invoke.texi: Fix typo.
13010 2017-05-09 Richard Biener <rguenther@suse.de>
13012 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13013 (vrp_val_is_min): Likewise.
13014 (set_value_range_to_value): Likewise.
13015 (set_value_range_to_nonnegative): Likewise.
13016 (gimple_assign_nonzero_p): Likewise.
13017 (gimple_stmt_nonzero_p): Likewise.
13018 (vrp_int_const_binop): Likewise. Remove unreachable case.
13019 (adjust_range_with_scev): Adjust comments.
13020 (compare_range_with_value): Likewise.
13021 (extract_range_from_phi_node): Likewise.
13022 (test_for_singularity): Likewise.
13024 2017-05-09 Richard Biener <rguenther@suse.de>
13026 * tree-vrp.c (get_single_symbol): Add assert that we don't
13027 get overflowed constants as invariant part.
13028 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13029 checks. Use wi::cmp instead of recursing for integer constants.
13030 (compare_values): Just ignore whether we assumed undefined
13031 overflow instead of failing the compare.
13032 (extract_range_for_var_from_comparison_expr): Add comment before the
13033 TREE_NO_WARNING sets.
13034 (test_for_singularity): Likewise.
13035 (extract_range_from_comparison): Do not disable optimization
13036 when we assumed undefined overflow.
13037 (extract_range_basic): Remove init of unused var.
13039 2017-05-09 Richard Biener <rguenther@suse.de>
13041 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13042 (extract_range_from_multiplicative_op_1): Adjust.
13043 (extract_range_from_binary_expr_1): Use int_const_binop.
13045 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13048 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13049 rs6000_store_data_bypass_p in seven define_bypass directives and
13050 in several comments.
13051 * config/rs6000/rs6000-protos.h: Add prototype for
13052 rs6000_store_data_bypass_p function.
13053 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13054 function implements slightly different (rs6000-specific) semantics
13055 than store_data_bypass_p, returning false rather than aborting
13056 with assertion error when arguments do not satisfy the
13057 requirements of store data bypass.
13058 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13059 rs6000_store_data_bypass_p.
13061 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13063 * config/xtensa/xtensa-protos.h
13064 (xtensa_initial_elimination_offset): New declaration.
13065 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13066 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13067 macro definition, add case for FRAME_POINTER_REGNUM when
13068 FRAME_GROWS_DOWNWARD.
13069 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13070 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13071 xtensa_initial_elimination_offset.
13073 2017-05-08 Nathan Sidwell <nathan@acm.org>
13075 * doc/invoke.texi: Alphabetize -fdump options.
13077 2017-05-08 Martin Sebor <msebor@redhat.com>
13079 PR translation/80280
13080 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13082 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13084 * target.def (compute_frame_layout): New optional target hook.
13085 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13086 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13087 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13089 * reload1.c (verify_initial_elim_offsets): Likewise.
13090 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13091 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13092 (arm_get_frame_offsets): Split up into this ...
13093 (arm_compute_frame_layout): ... and this function.
13095 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13097 * config/aarch64/constraints.md (Usa): New constraint.
13098 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13100 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13102 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13103 with_multilib_list after it has been checked.
13105 2017-05-08 Richard Biener <rguenther@suse.de>
13107 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13108 (bitmap_set_subtract_values): Likewise.
13110 2017-05-08 Richard Biener <rguenther@suse.de>
13112 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13113 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13115 (gimple_stmt_nonzero_warnv_p): Rename to ...
13116 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13118 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13119 (extract_range_basic): Adjust, do not disable propagation on
13120 strict overflow sensitive simplification.
13121 (vrp_visit_cond_stmt): Likewise.
13123 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13125 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13126 body size unconditionally.
13128 2017-05-07 Jeff Law <law@redhat.com>
13131 2017-05-06 Jeff Law <law@redhat.com>
13132 PR tree-optimization/78496
13133 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13136 PR tree-optimization/78496
13137 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13138 (simplify_stmt_using_ranges): Call it.
13139 (vrp_dom_walker::before_dom_children): Extract equivalences
13140 from an ASSERT_EXPR with an equality comparison against a
13143 2017-05-06 Jeff Law <law@redhat.com>
13145 PR tree-optimization/78496
13146 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13149 PR tree-optimization/78496
13150 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13151 (simplify_stmt_using_ranges): Call it.
13152 (vrp_dom_walker::before_dom_children): Extract equivalences
13153 from an ASSERT_EXPR with an equality comparison against a
13156 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13158 * lra-constraints.c (lra_copy_reg_equiv): New function.
13159 (split_reg): Use it to copy equivalence information from the
13160 original register to the spill register.
13162 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13164 PR rtl-optimization/75964
13165 * simplify-rtx.c (simplify_const_relational_operation): Remove
13166 invalid handling of comparisons of integer ABS.
13168 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13170 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13171 initialize to zero.
13172 (init_regs): Remove declaration.
13173 (function_arg_advance_32): Initialize error_p as boolean variable.
13175 2017-05-05 Nathan Sidwell <nathan@acm.org>
13177 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13178 lines. Use for (;;).
13180 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13182 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13183 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13184 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13185 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13186 VF=2 that require versioning.
13188 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13190 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13193 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13195 * diagnostic.h (diagnostic_override_option_index): Convert from
13196 macro to inline function.
13198 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13200 * diagnostic.c (last_module_changed_p): New function.
13201 (set_last_module): New function.
13202 (diagnostic_report_current_module): Convert macro usage to
13203 the above functions.
13204 * diagnostic.h (diagnostic_context::last_module): Strengthen
13205 from const line_map * to const line_map_ordinary *.
13206 (diagnostic_last_module_changed): Delete macro.
13207 (diagnostic_set_last_module): Delete macro.
13209 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13211 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13212 with diagnostic_report_diagnostic.
13213 (diagnostic_n_impl_richloc): Likewise.
13214 * diagnostic.h (report_diagnostic): Delete macro.
13215 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13216 with diagnostic_report_diagnostic.
13217 * substring-locations.c (format_warning_va): Likewise.
13219 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13221 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13222 save/restor of format_spec. Move option-printing code to...
13223 (print_option_information): ...this new function, and
13224 reimplement by simply printing to the pretty_printer,
13225 rather than appending to the format string.
13227 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13229 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13230 handling logic into...
13231 (update_effective_level_from_pragmas): ...this new function.
13233 2017-05-04 Andrew Waterman <andrew@sifive.com>
13235 * config/riscv/riscv.opt (mstrict-align): New option.
13236 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13237 (SLOW_UNALIGNED_ACCESS): Define.
13238 (riscv_slow_unaligned_access): Declare.
13239 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13241 (riscv_slow_unaligned_access): New variable.
13242 (rocket_tune_info): Set slow_unaligned_access to true.
13243 (optimize_size_tune_info): Set slow_unaligned_access to false.
13244 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13245 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13246 (riscv_option_override): Set riscv_slow_unaligned_access.
13247 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13249 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13251 * config/riscv/riscv.md: Unify indentation.
13253 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13258 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13260 (extendsi<mode>2): Add support for doing sign extension via
13261 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13262 don't have ISA 3.0 instructions.
13263 (extendsi<mode>2 splitter): Likewise.
13264 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13265 generate the normal insns since SImode can now go in vector
13266 registers. Disallow the special UNSPECs needed for previous
13267 machines to hide SImode being used. Add new insns
13268 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13269 (fix_trunc<mode>si2_stfiwx): Likewise.
13270 (fix_trunc<mode>si2_internal): Likewise.
13271 (fixuns_trunc<mode>si2): Likewise.
13272 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13273 (fctiw<u>z_<mode>_smallint): Likewise.
13274 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13275 of floating point to 32-bit integer from doing a direct move to
13276 the GPR registers to do a store.
13277 (fctiwz_<mode>): Break long line.
13279 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13281 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13282 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13283 (addr_list, addr_offset_valid_p): New.
13284 (split_address_groups): Check offset validity with above function.
13285 (gt-tree-ssa-loop-ivopts.h): Include header file.
13287 2017-05-05 Nathan Sidwell <nathan@acm.org>
13289 * config.gcc (arm*-*-*): Add missing 'fi'.
13291 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13293 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13294 included in -fopt-info does not matter.
13295 * doc/optinfo.texi (-fopt-info): Fix description of default
13296 behavour. Explicitly say order of options included in -fopt-info
13299 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13301 * config.gcc: Allow combinations of aprofile and rmprofile values for
13302 --with-multilib-list.
13303 * config/arm/t-multilib: New file.
13304 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13305 variables. Remove setting of ISA and floating-point ABI in
13306 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13307 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13308 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13309 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13311 * config/arm/t-rmprofile: Likewise except for the matches changes.
13312 * doc/install.texi (--with-multilib-list): Document the combination of
13313 aprofile and rmprofile values and warn about pitfalls in doing that.
13315 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13317 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13318 (movdi_aarch64): Likewise.
13320 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13322 PR tree-optimization/80632
13323 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13325 (build_arrays): Initialize it for virtual phis.
13326 (fix_phi_nodes): Use it for virtual phis.
13328 PR tree-optimization/80558
13329 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13330 [x, y] op z into [x op, y op z] for op & or | if conditions
13333 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13334 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13337 * config/arm/arm.md (use_literal_pool): Remove.
13338 (64-bit immediate split): No longer takes cost into consideration
13339 if arm_disable_literal_pool is enabled.
13340 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13341 used when arm_disable_literal_pool is enabled.
13342 (arm_max_const_double_inline_cost): Remove use of
13343 arm_disable_literal_pool.
13344 (push_minipool_fix): Add assert.
13345 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13346 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13347 (no_literal_pool_sf_immediate): New.
13349 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13351 PR tree-optimization/80613
13352 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13353 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13355 2017-05-05 Richard Biener <rguenther@suse.de>
13357 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13359 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13361 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13362 of this flag from insn conditions due to removal from r247495.
13364 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13366 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13368 (arm_early_store_addr_dep_ptr): Likewise.
13369 * config/arm/aarch-common-protos.h
13370 (arm_early_load_addr_dep_ptr): Add prototype.
13371 (arm_early_store_addr_dep_ptr): Likewise.
13372 * config/arm/cortex-a53.md: Add new bypasses.
13374 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13376 * tree.c (next_type_uid): Change type to unsigned.
13377 (type_hash_canon): Decrement back next_type_uid if
13378 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13379 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13382 2017-05-04 Martin Sebor <msebor@redhat.com>
13384 * builtins.c: Fix a trivial typo in a comment.
13386 PR middle-end/79234
13387 * builtins.c (check_sizes): Adjust to handle reading past the end.
13388 Avoid printing excessive upper bound of ranges. Use %E to print
13389 tree nodes instead of converting them to %wu.
13390 (expand_builtin_memchr): New function.
13391 (compute_dest_size): Rename...
13392 (compute_objsize): ...to this.
13393 (expand_builtin_memcpy): Adjust.
13394 (expand_builtin_mempcpy): Adjust.
13395 (expand_builtin_strcat): Adjust.
13396 (expand_builtin_strcpy): Adjust.
13397 (check_strncat_sizes): Adjust.
13398 (expand_builtin_strncat): Adjust.
13399 (expand_builtin_strncpy): Adjust and simplify.
13400 (expand_builtin_memset): Adjust.
13401 (expand_builtin_bzero): Adjust.
13402 (expand_builtin_memcmp): Adjust.
13403 (expand_builtin): Handle memcmp.
13404 (maybe_emit_chk_warning): Check strncat just once.
13406 2017-05-04 Martin Sebor <msebor@redhat.com>
13408 PR preprocessor/79214
13409 PR middle-end/79222
13410 PR middle-end/79223
13411 * builtins.c (check_sizes): Add inlining context and issue
13412 warnings even when -Wno-system-headers is set.
13413 (check_strncat_sizes): Same.
13414 (expand_builtin_strncat): Same.
13415 (expand_builtin_memmove): New function.
13416 (expand_builtin_stpncpy): Same.
13417 (expand_builtin): Handle memmove and stpncpy.
13419 2017-05-04 Bin Cheng <bin.cheng@arm.com>
13421 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
13422 which is not used any more.
13424 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13426 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
13428 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13430 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
13431 (cortexa53_tunings): Likewise.
13432 (cortexa57_tunings): Likewise.
13433 (cortexa72_tunings): Likewise.
13434 (cortexa73_tunings): Likewise.
13436 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13438 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
13439 Set loop alignment to 8.
13441 2017-05-04 Martin Sebor <msebor@redhat.com>
13443 PR translation/80280
13444 * builtins.c (expand_builtin_object_size): Add missing quoting to
13445 %D and like directives.
13446 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
13447 (hsa_type_for_tree_type): Same.
13448 (verify_function_arguments): Same.
13449 * symtab.c (symbol_table::change_decl_assembler_name): Same.
13450 * varasm.c (get_section): Same.
13453 2017-05-04 Martin Sebor <msebor@redhat.com>
13455 PR translation/80280
13456 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
13458 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13460 * config/aarch64/aarch64.c (generic_addrcost_table):
13461 Change HI/TI mode setting.
13463 2017-05-04 Martin Jambor <mjambor@suse.cz>
13465 PR tree-optimization/80622
13466 * tree-sra.c (comes_initialized_p): New function.
13467 (build_accesses_from_assign): Only set write lazily when
13468 comes_initialized_p is false.
13469 (analyze_access_subtree): Use comes_initialized_p.
13470 (propagate_subaccesses_across_link): Assert !comes_initialized_p
13471 instead of testing for PARM_DECL.
13473 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13475 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
13476 constraint on operand 0 to allow more general addressing modes.
13477 Adjust output template.
13478 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
13480 * config/aarch64/aarch64-protos.h
13481 (aarch64_address_valid_for_prefetch_p): Declare prototype.
13482 * config/aarch64/constraints.md (Dp): New address constraint.
13483 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
13486 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
13488 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
13489 update use of estimate_ipcp_clone_size_and_time.
13490 (estimate_local_effects): Update use of
13491 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
13492 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
13493 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
13494 Return nonspecialized time.
13496 2017-05-04 Richard Biener <rguenther@suse.de>
13498 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
13499 for the last VUSE which def dominates the PHI. Directly call
13501 (get_continuation_for_phi_1): Remove.
13503 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
13505 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
13506 to explain the use of truncating division. Cap the number of
13507 iterations to the maximum given by nb_iterations_upper_bound,
13510 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13512 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
13513 * configure: Regenerate.
13514 * config.in: Regenerate.
13515 * config/i386/driver-mingw32.c: new file.
13516 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
13517 * config.host: Link driver-mingw32.o on MinGW host.
13518 * doc/install.texi: Document new --enable-mingw-wildcard configure
13521 2017-05-04 Marek Polacek <polacek@redhat.com>
13523 PR tree-optimization/80612
13524 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
13526 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13527 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
13529 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
13530 (movt splitter): Likewise.
13531 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
13532 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
13533 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
13534 block for Thumb-1 with MOVT.
13535 (thumb2_legitimate_address_p): Move code block ...
13536 (can_avoid_literal_pool_for_label_p): ... into this new function.
13537 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
13539 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
13540 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
13541 "M-profile targets with the MOVT instruction".
13543 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13545 * config/arm/arm-builtins.c (arm_init_builtins): Rename
13546 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
13547 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
13549 2017-05-04 Martin Liska <mliska@suse.cz>
13551 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
13552 variable cond_code.
13554 2017-05-04 Richard Biener <rguenther@suse.de>
13556 * tree.c (array_at_struct_end_p): Handle arrays at struct
13557 end with flexarrays more conservatively. Refactor and treat
13558 arrays of arrays or aggregates more strict. Fix
13559 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
13560 * tree.c (array_at_struct_end_p): Adjust prototype.
13561 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13562 * gimple-fold.c (get_range_strlen): Likewise.
13563 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
13565 2017-05-04 Richard Biener <rguenther@suse.de>
13567 PR tree-optimization/31130
13568 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
13570 (supports_overflow_infinity): Likewise.
13571 (is_negative_overflow_infinity): Likewise.
13572 (is_positive_overflow_infinity): Likewise.
13573 (is_overflow_infinity): Likewise.
13574 (stmt_overflow_infinity): Likewise.
13575 (overflow_infinity_range_p): Likewise.
13576 (usable_range_p): Remove as always returning true.
13577 (make_overflow_infinity): Remove.
13578 (negative_overflow_infinity): Likewise.
13579 (positive_overflow_infinity): Likewise.
13580 (avoid_overflow_infinity): Likewise.
13581 (set_value_range): Adjust accordingly.
13582 (set_value_range_to_nonnegative): Likewise, remove now unused
13583 overflow_infinity arg.
13584 (vrp_operand_equal_p): Adjust.
13585 (update_value_range): Likewise.
13586 (range_int_cst_singleton_p): Likewise.
13587 (operand_less_p): Likewise.
13588 (compare_values_warnv): Likewise.
13589 (extract_range_for_var_from_comparison_expr): Likewise.
13590 (vrp_int_const_binop): Likewise.
13591 (zero_nonzero_bits_from_vr): Likewise.
13592 (extract_range_from_multiplicative_op_1): Likewise.
13593 (extract_range_from_binary_expr_1): Likewise.
13594 (extract_range_from_unary_expr): Likewise.
13595 (extract_range_from_comparison): Likewise.
13596 (extract_range_basic): Likewise.
13597 (adjust_range_with_scev): Likewise.
13598 (compare_ranges): Likewise.
13599 (compare_range_with_value): Likewise.
13600 (dump_value_range): Likewise.
13601 (test_for_singularity): Likewise, remove strict_overflow_p parameter
13603 (simplify_cond_using_ranges): Adjust.
13605 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13607 * brig-builtins.def: Added a builtin for class_f64.
13608 * builtin-types.def: Added a builtin type needed by class_f64.
13610 2017-05-03 Jason Merrill <jason@redhat.com>
13612 * timevar.def: Add TV_CONSTEXPR.
13614 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13616 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
13618 2017-05-03 Martin Jambor <mjambor@suse.cz>
13620 * ipa-prop.c (ipa_update_after_lto_read): Removed.
13621 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
13622 * ipa-cp.c (ipcp_propagate_stage): Do not call
13623 ipa_update_after_lto_read.
13624 * ipa-inline.c (ipa_inline): Likewise.
13626 2017-05-03 Martin Jambor <mjambor@suse.cz>
13628 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
13629 tag. Added a default constructor and a destructor.
13630 (ipa_edge_args_sum_t): New class;
13631 (ipa_edge_args_sum): Declare.
13632 (ipa_edge_args_vector): Remove declaration.
13633 (IPA_EDGE_REF): Use ipa_edge_args_sum.
13634 (ipa_free_edge_args_substructures): Remove declaration.
13635 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
13636 (ipa_edge_args_info_available_for_edge_p): Likewise.
13637 * ipa-prop.c (ipa_edge_args_vector): Removed.
13638 (edge_removal_hook_holder): Likewise.
13639 (edge_duplication_hook_holder): Likewise.
13640 (ipa_edge_args_sum): New variable.
13641 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
13642 ipa_edge_args_vector.
13643 (ipa_free_edge_args_substructures): Likewise.
13644 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
13645 ipa_edge_args_vector.
13646 (ipa_edge_removal_hook): Turned into method
13647 ipa_edge_args_sum_t::remove.
13648 (ipa_edge_duplication_hook): Turned into method
13649 ipa_edge_args_sum_t::duplicate.
13650 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
13651 registering edge hooks.
13652 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
13653 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
13654 ipa_edge_args_sum instead of ipa_edge_args_vector.
13655 * ipa-profile.c (ipa_profile): Likewise.
13657 2017-05-03 Martin Jambor <mjambor@suse.cz>
13659 * symbol-summary.h (function_summary): New method exists.
13660 (function_summary::symtab_removal): Deallocate through release.
13661 (call_summary): New class.
13662 (gt_ggc_mx): New overload.
13663 (gt_pch_nx): Likewise.
13664 (gt_pch_nx): Likewise.
13666 2017-05-03 Jeff Law <law@redhat.com>
13668 PR tree-optimization/78496
13669 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
13670 from simplify_cond_using_ranges. Split off code to walk
13671 backwards through casts into ...
13672 (simplify_cond_using_ranges_2): New function.
13673 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
13674 (execute_vrp): After identifying jump threads, call
13675 simplify_cond_using_ranges_2.
13677 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
13680 * ipa-inline.h (inline_summary): Add ctor.
13681 (create_ggc): Do not use ggc_cleared_alloc.
13683 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
13684 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13686 * gcc.c (handle_braces): Support escaping in switch matching
13688 * doc/invoke.texi (Spec Files): Document it.
13689 Remove superfluous @code markup in items.
13691 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13693 * diagnostic-show-locus.c (struct column_range): New struct.
13694 (get_affected_columns): New function.
13695 (get_printed_columns): New function.
13696 (struct correction): New struct.
13697 (correction::ensure_capacity): New function.
13698 (correction::ensure_terminated): New function.
13699 (struct line_corrections): New struct.
13700 (line_corrections::~line_corrections): New dtor.
13701 (line_corrections::add_hint): New function.
13702 (layout::print_trailing_fixits): Reimplement in terms of the new
13704 (selftest::test_overlapped_fixit_printing): New function.
13705 (selftest::diagnostic_show_locus_c_tests): Call it.
13707 2017-05-03 Nathan Sidwell <nathan@acm.org>
13709 Canonicalize canonical type hashing
13710 * tree.h (type_hash_canon_hash): Declare.
13711 * tree.c (type_hash_list, attribute_hash_list): Move into
13712 type_hash_canon_hash.
13713 (build_type_attribute_qual_variant): Break out hash code calc into
13714 type_hash_canon_hash.
13715 (type_hash_canon_hash): New. Generic type hash computation.
13716 (build_range_type_1, build_array_type_1, build_function_type,
13717 build_method_type_directly, build_offset_type, build_complex_type,
13718 make_vector_type): Call it.
13720 2017-05-03 Richard Biener <rguenther@suse.de>
13722 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
13723 When all DRs have unknown misaligned do not always peel
13724 when there is a store but apply the same costing model as if
13725 there were only loads.
13727 2017-05-03 Richard Biener <rguenther@suse.de>
13730 PR tree-optimization/80492
13731 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
13732 compare_base_decls returning dont-know properly.
13734 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
13736 * config/arm/iterators.md (CCSI): New mode iterator.
13737 (arch): New mode attribute.
13738 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
13739 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
13740 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
13741 code iterator for success result mode.
13742 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
13743 the corresponding new insn generators.
13745 2017-05-03 Bin Cheng <bin.cheng@arm.com>
13748 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13749 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13751 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
13753 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
13754 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
13755 (DDR_A): Wrap DDR argument in brackets.
13756 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
13757 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
13758 (DDR_REVERSED_P): Likewise.
13760 2017-05-03 Jakub Jelinek <jakub@redhat.com>
13762 PR tree-optimization/79472
13763 * tree-switch-conversion.c (struct switch_conv_info): Add
13764 contiguous_range and default_case_nonstandard fields.
13765 (collect_switch_conv_info): Compute contiguous_range and
13766 default_case_nonstandard fields, don't clear final_bb if
13767 contiguous_range and only the default case doesn't have the required
13769 (check_all_empty_except_final): Set default_case_nonstandard instead
13770 of failing if contiguous_range and the default case doesn't have empty
13772 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
13773 and only the default case doesn't have the required constants. Skip
13775 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
13776 if default_case_nonstandard.
13777 (build_constructors): Build constant 1 just once. Assert that default
13778 values aren't inserted in between cases if contiguous_range. Skip
13780 (build_arrays): Skip virtual phis.
13781 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
13782 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
13783 Handle virtual phis.
13784 (gen_inbound_check): Handle default_case_nonstandard case.
13785 (process_switch): Adjust check_final_bb caller. Call
13786 gather_default_values with the first non-default case instead of
13787 default case if default_case_nonstandard.
13789 2017-05-02 Nathan Sidwell <nathan@acm.org>
13791 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
13792 check. Fix formatting.
13794 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
13796 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
13797 errors when comparing specialized and unspecialized times.
13799 2017-05-02 David Malcolm <dmalcolm@redhat.com>
13801 * diagnostic-show-locus.c
13802 (layout::should_print_annotation_line_p): Make private.
13803 (layout::print_annotation_line): Make private.
13804 (layout::annotation_line_showed_range_p): Make private.
13805 (layout::show_ruler): Make private.
13806 (layout::print_source_line): Make private. Pass in line and
13807 line_width, rather than calling location_get_source_line. Drop
13809 (layout::print_leading_fixits): New method.
13810 (layout::print_any_fixits): Rename to...
13811 (layout::print_trailing_fixits): ...this, and make private.
13812 Don't print newline fixits.
13813 (diagnostic_show_locus): Move logic for printing one row into...
13814 (layout::print_line): ...this new function. Move the
13815 location_get_source_line call and error-handling from
13816 print_source_line to here. Call print_leading_fixits, and rename
13817 print_any_fixits to print_trailing_fixits.
13818 (selftest::test_fixit_insert_containing_newline): Update now that
13819 newlines are partially supported.
13820 (selftest::test_fixit_insert_containing_newline_2): New test.
13821 (selftest::test_fixit_replace_containing_newline): Update comments.
13822 (selftest::diagnostic_show_locus_c_tests): Call the new test.
13823 * edit-context.c (class added_line): New class.
13824 (class edited_line): Describe newline handling in comment.
13825 (edited_line::actually_edited_p): New method.
13826 (edited_line::print_content): Delete redundant decl.
13827 (edited_line::m_predecessors): New field.
13828 (edited_file::print_content): Call edited_line::print_content.
13829 (edited_file::print_diff): Update to support newlines.
13830 (edited_file::print_diff_hunk): Likewise.
13831 (edited_file::print_run_of_changed_lines): New function.
13832 (edited_file::print_diff_line): Convert to...
13833 (print_diff_line): ...this.
13834 (edited_file::get_effective_line_count): New function.
13835 (edited_line::edited_line): Initialize new field m_predecessors.
13836 (edited_line::~edited_line): Clean up m_predecessors.
13837 (edited_line::apply_fixit): Handle newlines.
13838 (edited_line::get_effective_line_count): New function.
13839 (edited_line::print_content): New function.
13840 (edited_line::print_diff_lines): New function.
13841 (selftest::test_applying_fixits_insert_containing_newline): New
13843 (selftest::test_applying_fixits_replace_containing_newline): New
13845 (selftest::insert_line): New function.
13846 (selftest::test_applying_fixits_multiple_lines): Add example of
13848 (selftest::edit_context_c_tests): Call the new tests.
13850 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13852 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
13853 parameter cand. Update dump information.
13854 (get_computation_cost): Update uses.
13856 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13858 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
13859 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
13860 (get_computation_at, rewrite_use_address): Update use of
13861 get_computation_aff.
13863 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13865 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
13866 (get_computation): Delete.
13867 (get_computation_cost): Implement like get_computation_cost_at.
13868 Use get_computation_at.
13869 (get_computation_cost_at): Delete.
13870 (rewrite_use_nonlinear_expr): Use get_computation_at.
13871 (rewrite_use_compare, remove_unused_ivs): Ditto.
13873 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13875 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
13877 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13879 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
13880 (ivopts_global_cost_for_size): Rename parameter and update uses.
13881 (iv_ca_recount_cost): Update uses.
13882 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
13883 candidates seperately in n_invs and n_cands.
13884 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
13886 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13888 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
13889 (find_inv_vars_cb): New.
13890 (find_depends): Renamed to ...
13891 (find_inv_vars): ... this.
13892 (add_candidate_1, force_var_cost): Call find_inv_vars.
13893 (split_address_cost, determine_group_iv_cost_cond): Ditto.
13895 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13897 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
13898 inv_vars. Add inv_exprs.
13899 (struct iv_cand): Rename depends_on to inv_vars.
13900 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
13901 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
13902 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
13903 (dump_cand): Dump inv_vars.
13904 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
13905 (record_invariant, find_depends, add_candidate_1): Ditto.
13906 (set_group_iv_cost, force_var_cost): Ditto.
13907 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
13908 (get_computation_cost_at, get_computation_cost): Ditto.
13909 (determine_group_iv_cost_generic): Ditto.
13910 (determine_group_iv_cost_address): Ditto.
13911 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
13912 (determine_group_iv_costs): Ditto.
13913 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
13914 (iv_ca_set_remove_invariants): Renamed to ...
13915 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
13916 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
13917 (iv_ca_set_add_invariants): Renamed to ...
13918 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
13919 (iv_ca_set_cp): Use iv_ca_set_add_invs.
13920 (iv_ca_has_deps): Support inv_vars and inv_exprs.
13921 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
13922 (create_new_ivs): Remove useless dump.
13924 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13926 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
13928 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
13929 (iv_ca_set_no_cp, create_new_iv): Ditto.
13931 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13933 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13935 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
13937 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
13938 function tree_check2.
13940 2017-05-02 Martin Liska <mliska@suse.cz>
13942 * doc/gcov.texi: Add missing preposition.
13943 * gcov.c (function_info::function_info): Properly fill up
13944 all member variables.
13946 2017-05-02 Tamar Christina <tamar.christina@arm.com>
13948 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
13950 2017-05-02 Tamar Christina <tamar.christina@arm.com>
13952 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
13954 2017-05-02 Martin Liska <mliska@suse.cz>
13957 * lto-streamer-in.c (lto_read_tree_1): Remove
13958 LTO_STREAMER_DEBUG.
13959 * lto-streamer.c (struct tree_hash_entry): Likewise.
13960 (struct tree_entry_hasher): Likewise.
13961 (tree_entry_hasher::hash): Likewise.
13962 (tree_entry_hasher::equal): Likewise.
13963 (lto_streamer_init): Likewise.
13964 (lto_orig_address_map): Likewise.
13965 (lto_orig_address_get): Likewise.
13966 (lto_orig_address_remove): Likewise.
13967 * lto-streamer.h: Likewise.
13968 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
13969 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
13971 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
13973 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
13974 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
13975 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
13976 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
13977 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
13978 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
13979 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
13980 (mm_maskz_sub_ss): New intrinsics.
13981 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13982 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13983 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
13984 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
13985 (__builtin_ia32_subss_mask_round): New builtins.
13986 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13987 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13988 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
13990 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
13991 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13993 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13996 2017-05-02 Martin Jambor <mjambor@suse.cz>
13998 PR tree-optimization/78687
13999 * tree-sra.c (access): New field parent.
14000 (process_subtree_disqualification): New function.
14001 (disqualify_candidate): Call it.
14002 (build_accesses_from_assign): Reset write flag if creating an
14004 (build_access_subtree): Fill in parent field and also prpagate
14005 down grp_write flag.
14006 (create_artificial_child_access): New parameter set_grp_write, set
14007 grp_write to its value.
14008 (propagate_subaccesses_across_link): Also propagate grp_write flag
14010 (propagate_all_subaccesses): Push the closest parent back to work
14011 queue if add_access_to_work_queue returned true.
14013 2017-05-02 Richard Biener <rguenther@suse.de>
14015 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14016 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14017 -fstrict-overflow documentation.
14018 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14019 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14020 flag_strict_overflow.
14021 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14022 * lto-opts.c (lto_write_options): Do not stream it.
14023 * lto-wrapper.c (merge_and_complain): Do not handle it.
14024 * opts.c (default_options_table): Do not set -fstrict-overflow.
14025 (finish_options): Likewise do not clear it when sanitizing.
14026 * simplify-rtx.c (simplify_const_relational_operation): Do not
14027 test flag_strict_overflow.
14029 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14031 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14032 using enabled attribute.
14033 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14034 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14035 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14036 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14037 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14038 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14039 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14040 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14041 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14042 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14044 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14046 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14048 2017-05-02 Richard Biener <rguenther@suse.de>
14050 PR tree-optimization/80591
14052 2017-04-10 Richard Biener <rguenther@suse.de>
14054 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14057 2017-05-02 Richard Biener <rguenther@suse.de>
14059 PR tree-optimization/80549
14060 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14061 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14062 headers if they do not have a preheader.
14064 2017-05-02 Martin Liska <mliska@suse.cz>
14067 * common.opt: Fix typo.
14068 * doc/invoke.texi: Likewise.
14070 2017-05-01 Jan Beulich <jbeulich@suse.com>
14072 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14073 swapping, add (x,x,m,x,n) alternative.
14075 2017-05-01 Nathan Sidwell <nathan@acm.org>
14077 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14078 unnecessary unadjusted_alignment check.
14080 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14083 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14085 * gimplify.c (gimplify_cilk_detach): New function.
14086 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14087 * tree-core.h: Document EXPR_CILK_SPAWN.
14088 * tree.h (EXPR_CILK_SPAWN): Define.
14090 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14092 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14093 to use new fixit_hint representation, using the "replace" logic.
14094 (get_line_span_for_fixit_hint): Likewise.
14095 (layout::print_any_fixits): Likewise.
14096 (selftest::test_one_liner_many_fixits): Rename to...
14097 (selftest::test_one_liner_many_fixits_1): ...this, and update
14098 comment and expected output to reflect that the multiple fix-it
14099 hints are now consolidated into one insertion.
14100 (selftest::test_one_liner_many_fixits_2): New test.
14101 (selftest::test_diagnostic_show_locus_one_liner): Update for
14103 (selftest::test_fixit_consolidation): Update for fix-it API
14105 * diagnostic.c (print_parseable_fixits): Likewise.
14106 * edit-context.c (edited_line::m_line_events): Convert from
14107 auto_vec <line_event *> to auto_vec <line_event>.
14108 (class line_event): Convert from abstract base class to a concrete
14109 class, taking over the role of replace_event.
14110 (class insert_event): Delete.
14111 (class replace_event): Rename to class line_event. Convert to
14113 (edit_context::add_fixits): Reimplement.
14114 (edit_context::apply_insert): Delete.
14115 (edit_context::apply_replace): Rename to...
14116 (edit_context::apply_fixit): ...this. Convert to half-open range.
14117 (edited_file::apply_insert): Delete.
14118 (edited_file::apply_replace): Rename to...
14119 (edited_file::apply_fixit): ...this.
14120 (edited_line::~edited_line): Drop deletion of events.
14121 (edited_line::apply_insert): Delete.
14122 (edited_line::apply_replace): Rename to...
14123 (edited_line::apply_fixit): ...this. Convert to half-open range.
14124 Update for change to type of m_line_events.
14125 * edit-context.h (edit_context::apply_insert): Delete.
14126 (edit_context::apply_replace): Rename to...
14127 (edit_context::apply_fixit): ...this.
14129 2017-05-01 Martin Sebor <msebor@redhat.com>
14131 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14134 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14137 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14138 __get_cpuid_max returns 0.
14139 (__get_cpuid_count): Ditto.
14141 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14143 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14144 replacement expression is another instance of one of its arguments.
14146 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14149 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14150 check for stack push/pop autoinc.
14151 * config/i386/i386.c (ix86_agi_dependent): Return false
14152 if the only reason why modified_in_p returned true is that
14153 addr is SP based and set_insn is a push or pop.
14155 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14157 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14160 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14163 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14164 (account_size_time): Use two predicates - exec_pred and
14166 (evaluate_conditions_for_known_args): Compute both clause and
14168 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14169 (inline_summary_t::duplicate): Update.
14170 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14172 (compute_inline_parameters): Likewise.
14173 (estimate_edge_size_and_time): Update caluclation of time.
14174 (estimate_node_size_and_time): Compute both time and nonspecialized
14176 (estimate_ipcp_clone_size_and_time): Update.
14177 (inline_merge_summary): Update.
14178 (do_estimate_edge_time): Update.
14179 (do_estimate_edge_size): Update.
14180 (do_estimate_edge_hints): Update.
14181 (inline_read_section, inline_write_summary): Stream both new predicates.
14182 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14184 (compute_inlined_call_time): Cleanup.
14185 (big_speedup_p): Update.
14186 (edge_badness): Update.
14187 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14188 (size_time_entry): Replace predicate by exec_predicate and
14189 nonconst_predicate.
14190 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14191 (estimate_edge_time): Return also nonspec_time.
14192 (reset_edge_growth_cache): Update.
14194 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14196 PR rtl-optimization/80491
14197 * ifcvt.c (noce_process_if_block): When looking for x setter
14198 with missing else_bb, don't check only the insn right before
14199 cond_earliest, but look for the last insn that x is modified in
14200 within the same bb.
14202 PR rtl-optimization/80491
14203 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14205 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14207 PR tree-optimization/80487
14208 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14210 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14212 PR tree-optimization/79697
14213 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14214 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14215 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14217 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14218 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14220 2017-04-28 Martin Sebor <msebor@redhat.com>
14222 PR tree-optimization/80523
14223 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14224 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14226 (maybe_warn, format_directive, parse_directive): Use new functions.
14227 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14229 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14231 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14233 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14235 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14236 target_header_dir): Set correctly.
14237 * configure: Regenerated.
14238 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14239 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14240 instead of SYSTEM_HEADER_DIR.
14242 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14244 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14245 (estimate_local_effects): Likewise.
14246 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14247 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14248 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14249 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14250 do_estimate_edge_time, estimate_edge_time): Likewise.
14251 * ipa-inline-analysis.c (estimate_node_size_and_time,
14252 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14253 (estimate_time_after_inlining): Remove.
14255 2017-04-28 Martin Liska <mliska@suse.cz>
14257 * doc/gcov.texi: Enhance documentation of gcov.
14259 2017-04-28 Martin Liska <mliska@suse.cz>
14261 * doc/gcov.texi: Sort options in alphabetic order.
14262 * doc/gcov-dump.texi: Likewise.
14263 * doc/gcov-tool.texi: Likewise.
14264 * gcov.c (print_usage): Likewise.
14265 * gcov-dump.c (print_usage): Likewise.
14266 * gcov-tool.c (print_merge_usage_message): Likewise.
14267 (print_rewrite_usage_message): Likewise.
14268 (print_overlap_usage_message): Likewise.
14270 2017-04-28 Martin Liska <mliska@suse.cz>
14272 PR gcov-profile/53915
14273 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14275 2017-04-28 Martin Liska <mliska@suse.cz>
14277 PR gcov-profile/79891
14278 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14279 is marked by compiler as living on a line.
14280 (get_cycles_count): Remove usage of the union.
14281 (output_intermediate_file): Likewise.
14282 (find_source): Fix GNU coding style.
14283 (accumulate_line_counts): Remove old non-all block mode.
14284 (output_lines): Remove usage of the union.
14285 * profile.c (output_location): Include all BBs, even if
14286 belonging to a same line (and file) as a previous BB.
14288 2017-04-28 Martin Liska <mliska@suse.cz>
14290 * gcov.c (process_args): Handle new argument 'w'.
14291 (read_graph_file): Assign ID to BBs.
14292 (output_branch_count): Display BB # if verbose flag is set.
14293 (output_lines): Likewise for arcs.
14294 (print_usage): Add '--verbose' option help.
14295 * doc/gcov.texi: Document --verbose (-w) option.
14297 2017-04-28 Martin Liska <mliska@suse.cz>
14299 * gcov.c (struct block_location_info): New struct.
14300 (process_file): Fill up the new structure.
14301 (read_graph_file): Replace usage of encoding by the newly added
14303 (add_line_counts): Likewise.
14304 (accumulate_line_counts): Remove usage of the union.
14305 (function_info::function_info): New function.
14306 (function_info::~function_info): Likewise.
14307 (process_file): Call delete instead of release_function.
14308 (release_function): Release the function.
14309 (release_structures): Call delete instead of release_function.
14310 (solve_flow_graph): Replace usage of num_blocks.
14311 (find_exception_blocks): Likewise.
14312 (output_lines): Fix GNU coding style.
14314 2017-04-28 Martin Liska <mliska@suse.cz>
14317 * coverage.c (coverage_remove_note_file): New function.
14318 * coverage.h: Declare the function.
14319 * toplev.c (finalize): Clean if an error has been seen.
14321 2017-04-28 Martin Liska <mliska@suse.cz>
14323 PR gcov-profile/80031
14324 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14325 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14326 * gcov.c (read_graph_file): Read just number of blocks.
14327 * profile.c (branch_prob): Do not stream 0 flags per a basic
14330 2017-04-28 Martin Liska <mliska@suse.cz>
14332 * gcov-dump.c (tag_*): Add new argument to declarations.
14333 (dump_gcov_file): Likewise.
14334 (tag_blocks): Add and use new argument depth.
14335 (tag_arcs): Likewise.
14336 (tag_lines): Likewise.
14337 (tag_counters): Likewise.
14338 (tag_summary): Likewise.
14339 (dump_working_sets): Use depth to do a proper indentation.
14341 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14344 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14345 * symtab.c (symtab_node::debug_symtab): Move definition here.
14347 2017-04-28 Richard Biener <rguenther@suse.de>
14349 * lto-streamer.h (LTO_major_version): Bump to 7.
14351 2017-04-28 Richard Biener <rguenther@suse.de>
14353 * tree-vrp.c (assert_info): New struct.
14354 (add_assert_info): New helper.
14355 (register_edge_assert_for_2): Refactor to add asserts to a vector
14357 (register_edge_assert_for_1): Likewise.
14358 (register_edge_assert_for): Likewise.
14359 (finish_register_edge_assert_for): New helper actually registering
14360 asserts where live on edge.
14361 (find_conditional_asserts): Adjust.
14362 (find_switch_asserts): Likewise.
14363 (evrp_dom_walker::try_find_new_range): Generalize.
14364 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14366 2017-04-27 Marek Polacek <polacek@redhat.com>
14369 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14370 arg10 and arg11 to itype.
14372 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14374 * doc/extend.texi (Object Size Checking): Improve grammar.
14376 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14379 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14380 that the logic for permitting reciprocal estimates matches that
14383 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14386 * tree.c (type_cache_hasher::equal): Only compare
14387 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14388 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14389 non-aggregate element types.
14390 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14391 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14393 2017-04-27 Richard Biener <rguenther@suse.de>
14395 PR middle-end/80533
14396 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14397 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14398 keeping a reference to a trailing array.
14400 2017-04-27 Richard Biener <rguenther@suse.de>
14402 PR middle-end/80539
14403 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
14404 being in loop-closed SSA form conservatively.
14405 (chrec_fold_multiply_poly_poly): Likewise.
14407 2017-04-27 Tamar Christina <tamar.christina@arm.com>
14409 PR middle-end/79665
14410 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
14411 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
14413 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14416 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
14417 (aarch64_function_arg_alignment): Return unsigned int again, but still
14418 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
14419 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14420 Don't emit -Wpsabi note.
14421 (aarch64_function_arg_boundary): Likewise.
14422 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14425 2017-04-26 Nathan Sidwell <nathan@acm.org>
14427 * tree.h (crc32_unsigned_n): Declare.
14428 (crc32_unsigned, crc32_unsigned): Make inline.
14429 * tree.c (crc32_unsigned_bits): Replace with ...
14430 (crc32_unsigned_n): ... this.
14431 (crc32_unsigned, crc32_byte): Remove.
14432 (crc32_string): Remove unnecessary braces.
14434 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14436 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
14437 * ipa-inline-analysis.c (MAX_TIME): Remove.
14438 (account_size_time): Use sreal for time.
14439 (dump_inline_summary): Update.
14440 (estimate_function_body_sizes): Update.
14441 (estimate_edge_size_and_time): Update.
14442 (estimate_calls_size_and_time): Update.
14443 (estimate_node_size_and_time): Update.
14444 (inline_merge_summary): Update.
14445 (inline_update_overall_summary): Update.
14446 (estimate_time_after_inlining): Update.
14447 (inline_read_section): Update.
14448 (inline_write_summary): Update.
14449 * ipa-inline.c (compute_uninlined_call_time): Update.
14450 (compute_inlined_call_time): Update.
14451 (recursive_inlining): Update.
14452 (inline_small_functions): Update.
14453 (dump_overall_stats): Update.
14454 * ipa-inline.h: Include sreal.h.
14455 (size_time_entry): Turn time to sreal.
14456 (inline_summary): Turn self_time nad time to sreal.
14458 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14460 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
14462 (sreal::stream_out, sreal::stream_in): New.
14463 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
14465 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14467 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
14470 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
14473 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
14475 (dimode_scalar_chain::compute_convert_gain): Ditto.
14476 (dimode_scalar_chain::make_vector_copies): Ditto.
14477 (dimode_scalar_chain::convert_reg): Ditto.
14478 (dimode_scalar_chain::convert_insn): Ditto.
14479 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
14480 (VI248_AVX512BW_1): New mode iterator.
14481 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
14482 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
14485 2017-04-25 Martin Sebor <msebor@redhat.com>
14487 PR tree-optimization/80497
14488 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
14489 constants are representable in HOST_WIDE_INT.
14490 (parse_directive): Ditto.
14492 2017-04-25 Martin Sebor <msebor@redhat.com>
14495 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
14496 (new_zero_array): Adjust signature.
14497 (dom_info::dom_init): Used unsigned rather that size_t.
14498 (dom_info::dom_info): Same.
14500 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14501 Jakub Jelinek <jakub@redhat.com>
14504 * config/arm/arm.c: Include gimple.h.
14505 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14506 returns negative, increment ncrn only if it returned positive.
14507 (arm_needs_doubleword_align): Return int instead of bool,
14508 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
14509 members, but if there is any such non-FIELD_DECL
14510 > PARM_BOUNDARY aligned decl, return -1 instead of false.
14511 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14512 returns negative, increment nregs only if it returned positive.
14513 (arm_setup_incoming_varargs): Likewise.
14514 (arm_function_arg_boundary): Emit -Wpsabi note if
14515 arm_needs_doubleword_align returns negative, return
14516 DOUBLEWORD_ALIGNMENT only if it returned positive.
14518 2017-04-25 Marek Polacek <polacek@redhat.com>
14521 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
14522 first argument to type.
14524 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
14527 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
14528 type checks to test for compatibility instead of equality.
14530 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14531 Jakub Jelinek <jakub@redhat.com>
14534 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
14536 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
14537 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
14538 the alignment computation, but return their maximum in warn_alignment.
14539 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14540 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
14542 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
14543 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14546 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14548 * config/arc/simdext.md (dmpyh): Fix typo.
14550 2017-04-25 Richard Biener <rguenther@suse.de>
14552 PR tree-optimization/80492
14553 * alias.c (compare_base_decls): Handle registers with asm
14554 specification conservatively.
14555 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14556 compare_base_decls returning dont-know properly.
14558 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14560 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
14561 (legitimate_offset_address_p): New function.
14562 (arc_legitimate_address_p): Use above function.
14564 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14566 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
14568 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14570 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
14571 ACCH registers whenever they are available.
14573 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14575 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
14576 double regs fix when not used.
14578 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14580 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
14582 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
14583 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
14585 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14587 * config/arc/arc.c (arc_output_addsi): Check for h-register class
14588 when emitting short ADD instructions.
14590 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14592 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
14594 (cmpsi_cc_c_insn): Likewise.
14595 (cbranchsi4_scratch): Compute proper instruction length using
14596 compact_hreg_operand.
14597 * config/arc/predicates.md (compact_hreg_operand): New predicate.
14599 2017-04-25 Richard Biener <rguenther@suse.de>
14601 PR middle-end/80509
14602 * passes.c (pass_manager::pass_manager): Initialize
14603 m_name_to_pass_map.
14605 2017-04-25 Richard Biener <rguenther@suse.de>
14607 PR tree-optimization/79201
14608 * tree-ssa-sink.c (statement_sink_location): Handle calls.
14610 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14613 * config/s390/vector.md: Split MEM->GPR vector moves for
14614 non-s_operand addresses.
14616 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14619 * config/s390/predicates.md (reload_const_wide_int_operand): New
14621 * config/s390/s390.md ("movti"): Remove d/P alternative.
14622 ("movti_bigconst"): New pattern definition.
14624 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14627 * s390-protos.h (s390_expand_cs_hqi): Removed.
14628 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
14629 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
14630 modes as well as CCZ1mode and CCZmode.
14631 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
14632 signature of s390_emit_compare_and_swap.
14633 (s390_expand_cs_hqi): Likewise, make static.
14634 (s390_expand_cs_tdsi): Generate an explicit compare before trying
14635 compare-and-swap, in some cases.
14636 (s390_expand_cs): Wrapper function.
14637 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
14639 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
14640 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
14641 patterns for small and large integers. Forbid symref memory operands.
14642 Move expander to s390.c. Require cc register.
14643 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
14644 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
14645 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
14646 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
14647 symref memory operands. Remove CC mode and call s390_match_ccmode
14649 ("atomic_exchange<mode>"): Allow and implement all integer modes.
14651 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14653 * config/s390/s390.md (define_peephole2): New peephole to help
14654 combining the load-and-test pattern with volatile memory.
14656 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14658 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
14659 with CCZmode for TARGET_Z196.
14661 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14663 PR rtl-optimization/80501
14664 * combine.c (make_compound_operation_int): Set subreg_code to SET
14665 even for AND with mask of the sign bit of mode.
14667 PR rtl-optimization/80500
14668 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
14669 sum's initial value.
14671 2017-04-25 Julian Brown <julian@codesourcery.com>
14672 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14674 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
14676 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
14678 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
14680 2017-04-25 Julian Brown <julian@codesourcery.com>
14681 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14683 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
14684 (thunderx2t99_sha): New Reservation.
14686 2017-04-25 Julian Brown <julian@codesourcery.com>
14687 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14689 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
14690 type for 1-element load.
14692 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
14694 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
14696 2017-04-24 Martin Jambor <mjambor@suse.cz>
14698 PR tree-optimization/80293
14699 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
14700 char arrays not totally scalarizable if it is false.
14701 (analyze_all_variable_accesses): Pass correct value in the new
14702 parameter. Add a statistics counter.
14704 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
14706 PR middle-end/79931
14707 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
14709 2017-04-24 Richard Biener <rguenther@suse.de>
14711 PR tree-optimization/80494
14712 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
14713 out for complex types.
14715 2017-04-24 Richard Biener <rguenther@suse.de>
14717 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
14718 * tree-ssa-sccvn.c (print_scc): Print SCC size.
14719 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
14720 (DFS): Adjust and never fail.
14721 (sccvn_dom_walker::fail): Remove.
14722 (sccvn_dom_walker::before_dom_children): Adjust.
14723 (run_scc_vn): Likewise and never fail.
14724 * tree-ssa-pre.c (pass_pre::execute): Adjust.
14725 (pass_fre::execute): Likewise.
14727 2017-04-24 Richard Biener <rguenther@suse.de>
14729 PR tree-optimization/79725
14730 * tree-ssa-sink.c (statement_sink_location): Return whether
14731 failure reason was zero uses. Move that check later.
14732 (sink_code_in_bb): Deal with zero uses by removing the stmt
14735 2017-04-24 Richard Biener <rguenther@suse.de>
14738 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
14739 pointer-based references.
14741 2017-04-24 Richard Biener <rguenther@suse.de>
14744 * pass_manager.h (pass_manager::operator new): Remove.
14745 (pass_manager::operator delete): Likewise.
14746 * passes.c (pass_manager::operator new): Remove.
14747 (pass_manager::operator delete): Likewise.
14748 (pass_manager::pass_manager): Zero individual pass members.
14750 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
14753 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
14754 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
14755 Check "XEXP (src, 1)" operand here.
14756 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
14757 Check "XEXP (src, 1)" operand here.
14758 (dimode_scalar_chain::make_vector_copies): Detect count register
14759 of a shift instruction. Zero extend count register from QImode
14760 to DImode to satisfy vector shift pattern count operand predicate.
14761 Substitute vector shift count operand with a DImode copy.
14762 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
14765 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
14767 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
14768 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
14769 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
14770 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
14771 (UNSPEC_NOREX_MEM): Remove definition.
14773 2017-04-21 Richard Biener <rguenther@suse.de>
14775 PR tree-optimization/79547
14776 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14777 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
14778 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
14779 without any constraints.
14781 2017-04-21 Richard Biener <rguenther@suse.de>
14783 PR tree-optimization/78847
14784 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
14786 2017-04-21 Richard Biener <rguenther@suse.de>
14788 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
14789 (build_distinct_type_copy): Likewise.
14790 (build_variant_type_copy): Likewise.
14791 * tree.c (build_qualified_type): Pass down mem-stat info.
14792 (build_distinct_type_copy): Likewise.
14793 (build_variant_type_copy): Likewise.
14795 2017-04-21 Richard Biener <rguenther@suse.de>
14797 PR tree-optimization/80237
14798 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
14800 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
14801 for a simplified result.
14803 2016-04-21 Richard Biener <rguenther@suse.de>
14805 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
14806 sth as strict as a simple_iv but a chrec without symbols and an
14807 operand defined in the loop we are peeling (and not some subloop).
14808 (propagate_constants_for_unrolling): Propagate all constants.
14810 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
14813 * config/i386/i386.c (print_reg): Remove assert for disalowed
14814 regno values, call output_operand_lossage instead.
14816 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
14819 * config/i386/constraints.md (Yc): New register constraint.
14820 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
14821 Use Yc constraint for alternative 2 of operand 0. Remove
14822 preferred_for_speed attribute.
14824 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
14826 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
14827 lastprivate clauses in SIMT case.
14829 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
14831 * doc/invoke.texi (-Wextra-semi): Document new warning option.
14833 2017-04-20 Richard Biener <rguenther@suse.de>
14835 PR tree-optimization/57796
14836 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
14837 as N scalar stores.
14838 (vect_model_load_cost): Cost gathers as N scalar loads.
14840 2017-04-20 Richard Biener <rguenther@suse.de>
14842 * ggc-page.c (ggc_allocated_p): Rename to ...
14843 (safe_lookup_page_table_entry): ... this and return the lookup
14845 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
14847 2017-04-20 Richard Biener <rguenther@suse.de>
14849 PR tree-optimization/80453
14850 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
14851 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
14852 from the conditions.
14853 (vn_phi_eq): Pass them down.
14854 (vn_phi_lookup): Record them.
14855 (vn_phi_insert): Likewise.
14857 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
14859 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
14860 uninitialized variable warning to avoid buffer overrun.
14862 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
14865 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
14866 is suppressed for '{ 0 }' in C.
14868 2017-04-20 Jakub Jelinek <jakub@redhat.com>
14870 * BASE-VER: Set to 8.0.0.
14872 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
14874 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
14875 priority .init_array and .fini_array section with SECTION_NOTYPE
14878 2017-04-20 Jakub Jelinek <jakub@redhat.com>
14880 PR middle-end/80423
14881 * tree.h (build_array_type): Add typeless_storage default argument.
14882 * tree.c (type_cache_hasher::equal): Also compare
14883 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
14884 (build_array_type): Add typeless_storage argument, set
14885 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
14887 (build_nonshared_array_type): Adjust build_array_type_1 caller.
14888 (build_array_type): Likewise. Add typeless_storage argument.
14890 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
14891 Jakub Jelinek <jakub@redhat.com>
14893 PR tree-optimization/80426
14894 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
14895 operation on symbolic operands, also compute the overflow for the
14896 invariant part when the operation degenerates into a negation.
14898 2017-04-19 Jakub Jelinek <jakub@redhat.com>
14901 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
14902 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
14905 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
14907 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
14910 * config/avr/avr.c (tree.h): Include it.
14911 (cgraph.h): Include it.
14912 (avr_encode_section_info): Don't warn for uninitialized progmem
14913 variable if it's just an alias.
14915 2017-04-19 Richard Biener <rguenther@suse.de>
14918 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
14919 when needed by AutoPGO.
14921 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
14924 * doc/lto.texi: Remove an extra 'that'.
14926 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
14928 PR rtl-optimization/80429
14929 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
14930 are only used in debug insns.
14932 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
14933 Vladimir Makarov <vmakarov@redhat.com>
14935 * config/sparc/predicates.md (input_operand): Add comment. Return
14936 true for any memory operand when LRA is in progress.
14937 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
14939 2017-04-18 Jeff Law <law@redhat.com>
14942 * mips.md ({return,simple_return}_internal): Do not overwrite
14945 2017-04-18 Jakub Jelinek <jakub@redhat.com>
14947 PR tree-optimization/80443
14948 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
14949 instead of adding 1, subtract -1 and similarly instead of subtracting
14952 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
14954 PR rtl-optimization/80357
14955 * haifa-sched.c (tmp_bitmap): New variable.
14956 (model_recompute): Handle duplicate use records.
14957 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
14958 (free_global_sched_pressure_data): Free it.
14960 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
14963 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14964 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14965 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14966 instead of SYSTEM_HEADER_DIR.
14968 2017-04-18 Jeff Law <law@redhat.com>
14970 PR middle-end/80422
14971 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
14972 predecessors after walking up the insn chain.
14974 2017-04-18 Jakub Jelinek <jakub@redhat.com>
14977 * dwarf2out.c (modified_type_die): Try harder not to emit internal
14978 sizetype type into debug info.
14980 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14983 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
14984 unneeded test for TARGET_UPPER_REGS_SF.
14985 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
14987 2017-04-18 Jakub Jelinek <jakub@redhat.com>
14990 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
14991 instead of gsi_after_labels.
14993 2017-04-18 Jeff Law <law@redhat.com>
14995 * regcprop.c (maybe_mode_change): Avoid creating copies of the
14999 2017-04-13 Jeff Law <law@redhat.com>
15000 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15001 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15003 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15006 * config/avr/avr.c (intl.h): Include it.
15007 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15009 2017-04-18 Martin Liska <mliska@suse.cz>
15011 PR gcov-profile/78783
15012 * gcov-tool.c (gcov_output_files): Validate that destination
15013 file is either removed by the tool or by a user.
15015 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15016 Guy Benyei <guybe@mellanox.com>
15018 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15019 block, and do not negate it, the stored id is already negative.
15021 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15023 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15025 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15028 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15029 masks of options that should be turned off if the VSX vector
15030 options are turned off.
15031 (OTHER_P8_VECTOR_MASKS): Likewise.
15032 (OTHER_VSX_VECTOR_MASKS): Likewise.
15033 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15034 rs6000_disable_incompatible_switches to validate no type switches
15036 (rs6000_incompatible_switch): New function to disallow turning on
15037 other vector options if -mno-vsx, -mno-power8-vector, or
15038 -mno-power9-vector are specified.
15040 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15042 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15044 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15046 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15047 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15048 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15049 (ARG_POINTER_CFA_OFFSET): Likewise.
15051 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15053 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15054 conditions to take advantage of various optimizations.
15056 2017-04-13 Jeff Law <law@redhat.com>
15058 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15059 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15060 (zero_extendsidi2_dext): Likewise.
15062 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15065 * fold-const.c (fold_ternary_loc): Revert
15066 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15069 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15071 PR rtl-optimization/80343
15072 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15073 new scratch pseudo.
15075 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15078 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15079 to ubsan_encode_value.
15081 2017-04-13 Jeff Law <law@redhat.com>
15083 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15084 appearing in DEBUG_INSNs.
15086 2017-04-13 Martin Liska <mliska@suse.cz>
15088 PR gcov-profile/80413
15089 * gcov-io.c (gcov_write_string): Copy to buffer just when
15090 allocated size is greater than zero.
15092 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15095 * dwarf2out.c (decls_for_scope): Ignore declarations of
15096 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15098 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15101 * ipa-visibility.c (non_local_p): Fix typos.
15102 (localize_node): When localizing symbol in same comdat group,
15103 dissolve the group only when we know external symbols are going
15105 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15107 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15109 PR tree-optimization/79390
15110 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15111 order does not result in usable sequence, retry with reversed operand
15117 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15118 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15119 op0 instead of fold_convert_loc (loc, type, arg0).
15121 2017-04-12 Jeff Law <law@redhat.com>
15123 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15124 has a delay slot in the generated code.
15126 * config/cris/cris.md (cris_preferred_reload_class): Return
15127 GENNONACR_REGS rather than GENERAL_REGS.
15129 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15132 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15133 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15134 signedness of the result type.
15136 2017-04-12 Richard Biener <rguenther@suse.de>
15137 Jeff Law <law@redhat.com>
15139 PR tree-optimization/80359
15140 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15141 trim stores to TARGET_MEM_REFs.
15143 2017-04-12 Richard Biener <rguenther@suse.de>
15145 PR tree-optimization/79390
15146 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15147 threading case even more.
15149 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15152 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15153 for quad_address_p for TImode, instead of just not indexed_address.
15155 2017-04-12 Richard Biener <rguenther@suse.de>
15156 Bernd Edlinger <bernd.edlinger@hotmail.de>
15158 PR middle-end/79671
15159 * alias.c (component_uses_parent_alias_set_from): Handle
15160 TYPE_TYPELESS_STORAGE.
15161 (get_alias_set): Likewise.
15162 * tree-core.h (tree_type_common): Add typeless_storage flag.
15163 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15164 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15165 for types containing members with TYPE_TYPELESS_STORAGE.
15166 (place_field): Likewise.
15167 (layout_type): Likewise for ARRAY_TYPE.
15168 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15169 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15170 TYPE_TYPELESS_STORAGE.
15171 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15173 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15176 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15177 first argument to type.
15179 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15183 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15184 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15185 (rs6000_expand_binop_builtin): Likewise.
15186 (rs6000_expand_ternop_builtin): Likewise; also add missing
15187 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15188 vshasigma built-ins.
15189 * doc/extend.texi: Document that vec_xxpermdi's third argument
15190 must be a constant.
15192 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15194 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15195 Use shift_const cost parameter when calculating gain of STV shifts.
15197 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15199 PR rtl-optimization/70478
15200 * lra-constraints.c (process_alt_operands): Check memory for
15201 disfavoring memory insn operand.
15203 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15205 PR middle-end/80100
15206 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15207 left shift in unsigned HOST_WIDE_INT type.
15209 PR rtl-optimization/80385
15210 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15211 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15214 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15215 if they have any depend clauses.
15217 2017-04-11 Martin Liska <mliska@suse.cz>
15220 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15221 * ipa-split.c (split_function): Create a local comdat symbol
15222 if caller is in a comdat group.
15224 2017-04-11 Martin Liska <mliska@suse.cz>
15227 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15230 2017-04-11 Martin Sebor <msebor@redhat.com>
15232 PR middle-end/80364
15233 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15234 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15236 (directive::set_width, directive::set_precision, format_character):
15238 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15241 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15244 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15245 conflict, set target->arch_name instead of target->cpu_name.
15247 2017-04-11 Richard Biener <rguenther@suse.de>
15249 PR tree-optimization/80374
15250 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15251 build_zero_cst, remove fold_convertible_p check again.
15253 2017-04-11 Martin Liska <mliska@suse.cz>
15256 * ubsan.c (instrument_object_size): Do not instrument register
15259 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15262 * config/i386/i386-builtin-types.def
15263 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15264 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15265 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15266 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15267 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15268 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15269 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15270 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15271 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15272 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15273 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15274 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15275 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15276 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15277 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15278 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15279 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15280 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15281 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15282 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15283 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15284 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15285 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15286 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15287 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15288 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15289 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15290 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15291 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15292 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15293 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15294 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15295 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15296 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15297 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15298 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15299 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15300 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15301 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15302 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15303 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15304 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15305 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15306 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15307 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15308 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15310 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15311 flag to second_arg_count, handle 4 argument function type _COUNT
15312 aliases, handle second_arg_count on second argument rather than last.
15314 2017-04-10 Jeff Law <law@redhat.com>
15316 PR tree-optimization/80374
15317 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15318 record anything if we can not convert integer_zero_node to the
15321 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15324 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15325 Enhance special handling given to the TARGET_P9_MINMAX option in
15326 relation to certain other options.
15328 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15330 PR tree-optimization/80153
15331 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15332 remove POINTER_PLUS_EXPR's base part directly, rather than through
15335 2017-04-10 Richard Biener <rguenther@suse.de>
15336 Bin Cheng <bin.cheng@arm.com>
15338 PR tree-optimization/80153
15339 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15340 the first element of pointer type aff_tree. Build result expr in
15342 (add_elt_to_tree): Convert to type unconditionally. Remove other
15343 fold_convert calls.
15344 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15345 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15347 2017-04-10 Richard Biener <rguenther@suse.de>
15349 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15352 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15354 PR rtl-optimization/70478
15355 * lra-constraints.c (curr_small_class_check): New.
15356 (update_and_check_small_class_inputs): New.
15357 (process_alt_operands): Update curr_small_class_check. Disfavor
15358 alternative insn memory operands. Check available regs for small
15361 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15364 * config/mips/mips.opt (-mvirt): Update description.
15365 * doc/invoke.texi (-mvirt): Likewise.
15367 2017-04-10 Richard Biener <rguenther@suse.de>
15369 PR middle-end/80362
15370 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15371 looking for NEGATE_EXPR in -A / -B to A / B folding.
15373 2017-04-10 Martin Liska <mliska@suse.cz>
15375 PR gcov-profile/80224
15376 * gcov.c (print_usage): Fix usage string.
15377 (get_gcov_intermediate_filename): Remove.
15378 (output_gcov_file): Use both for normal and intermediate format.
15379 (generate_results): Do not initialize special file for
15380 intermediate format.
15382 2017-04-10 Richard Biener <rguenther@suse.de>
15384 PR tree-optimization/80304
15385 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15388 2017-04-10 Nathan Sidwell <nathan@acm.org>
15391 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15392 (rs6000_init_builtins): Use it.
15394 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15396 * config/arm/arm.md (<mrc>): Add mode to SET source.
15397 (<mrrc>): Likewise.
15399 2017-04-10 Richard Biener <rguenther@suse.de>
15401 PR middle-end/80344
15402 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
15404 2017-04-10 Jakub Jelinek <jakub@redhat.com>
15407 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
15408 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
15409 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
15410 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
15411 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
15412 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
15413 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
15414 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
15415 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
15416 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
15417 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
15418 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
15419 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
15420 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
15421 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
15422 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
15423 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
15424 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
15425 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
15426 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
15427 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
15428 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
15429 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
15431 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
15433 PR rtl-optimization/70478
15434 * lra-constraints.c: Reverse the last patch.
15436 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
15438 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
15439 Add comment for WCHAR_T.
15441 2017-04-08 Martin Liska <mliska@suse.cz>
15444 2017-04-07 Martin Liska <mliska@suse.cz>
15447 * ipa-split.c (split_function): Add function part to a same comdat
15450 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15453 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
15455 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
15457 * rs6000/rs6000.c (vec_load_pendulum): Rename...
15458 (vec_pairing): ...to this.
15459 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
15460 (rs6000_sched_init): Adjust for name change.
15461 (struct rs6000_sched_context): Likewise.
15462 (rs6000_init_sched_context): Likewise.
15463 (rs6000_set_sched_context): Likewise.
15465 2017-04-07 Jakub Jelinek <jakub@redhat.com>
15471 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
15473 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
15474 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
15475 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
15477 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
15479 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
15481 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15483 PR rtl-optimization/70703
15484 * ira-color.c (update_conflict_hard_regno_costs): Use
15485 int64_t instead of HOST_WIDE_INT.
15487 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15489 PR rtl-optimization/70478
15490 * lra-constraints.c (process_alt_operands): Disfavor alternative
15491 insn memory operands.
15493 2017-04-07 Jeff Law <law@redhat.com>
15495 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
15496 CALL and NOTE_INSN_CALL_ARG_LOCATION.
15498 2017-04-07 Martin Liska <mliska@suse.cz>
15501 * config/aarch64/aarch64.c (aarch64_process_target_attr):
15502 Show error message instead of an ICE.
15504 2017-04-07 Martin Liska <mliska@suse.cz>
15507 * ipa-split.c (split_function): Add function part to a same comdat
15510 2017-04-07 Richard Biener <rguenther@suse.de>
15512 PR middle-end/80341
15513 * tree.c (get_unwidened): Also handle ! for_type case for
15515 * convert.c (do_narrow): Split out from ...
15516 (convert_to_integer_1): ... here. Do not pass final truncation
15517 type to get_unwidened for TRUNC_DIV_EXPR.
15519 2017-04-07 Richard Biener <rguenther@suse.de>
15521 * tree-affine.c (wide_int_ext_for_comb): Take type rather
15523 (aff_combination_const): Adjust.
15524 (aff_combination_scale): Likewise.
15525 (aff_combination_add_elt): Likewise.
15526 (aff_combination_add_cst): Likewise.
15527 (aff_combination_convert): Likewise.
15528 (add_elt_to_tree): Likewise. Remove unused argument.
15529 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
15531 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
15533 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
15535 * config/arm/arm.c (arm_default_short_enums): Use
15536 ARM_DEFAULT_SHORT_ENUMS.
15537 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
15539 2017-04-06 Jakub Jelinek <jakub@redhat.com>
15542 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
15543 members with redundant out-of-class redeclaration.
15545 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15548 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
15549 * config/i386/i386.md (*zero_extendsidi2):
15550 Add (?*x,*x) and (?*v,*v) alternatives.
15552 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15555 * config/i386/i386.c (ix86_expand_builtin)
15556 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
15557 mode from insn data. Convert operands to insn operand mode.
15558 Copy operands that don't satisfy insn predicate to a register.
15560 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
15562 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
15565 2017-04-06 Richard Biener <rguenther@suse.de>
15567 PR tree-optimization/80334
15568 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
15569 preserve alignment of accesses.
15571 2017-04-06 Richard Biener <rguenther@suse.de>
15573 PR tree-optimization/80262
15574 * tree-sra.c (build_ref_for_offset): Preserve address-space
15576 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
15577 Drop useless address-space information on MEM_REF offsets.
15579 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
15581 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
15583 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15585 PR rtl-optimization/70703
15586 * ira-color.c (update_conflict_hard_regno_costs): Use
15587 HOST_WIDE_INT instead of long.
15589 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
15592 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
15593 not defined for x86_64 target. Add -mmmx target option when __SSE2__
15595 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
15596 for x86_64 target. Handle -m3dnowa option.
15598 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15600 PR rtl-optimization/70703
15601 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
15602 (update_conflict_hard_regno_costs): Use long instead of unsigned
15603 arithmetic for cost calculation.
15605 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15606 Bernd Edlinger <bernd.edlinger@hotmail.de>
15609 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
15612 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
15615 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
15616 ptr_mode with Pmode throughout.
15617 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
15618 into probe_stack_range and use DImode.
15620 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
15623 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
15624 call_eh_return is true.
15626 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15628 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
15629 Initialize last_match_fntype_index.
15631 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15634 * tree-nvr.c: Include internal-fn.h.
15635 (pass_return_slot::execute): Ignore internal calls without
15638 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15639 Richard Biener <rguenther@suse.de>
15642 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
15643 captures used multiple times, except for the last use.
15644 * generic-match-head.c: Include gimplify.h.
15646 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15648 PR tree-optimization/79390
15649 * target.h (struct noce_if_info): Declare.
15650 * targhooks.h (default_noce_conversion_profitable_p): Declare.
15651 * target.def (noce_conversion_profitable_p): New target hook.
15652 * ifcvt.h (struct noce_if_info): New type, moved from ...
15653 * ifcvt.c (struct noce_if_info): ... here.
15654 (noce_conversion_profitable_p): Renamed to ...
15655 (default_noce_conversion_profitable_p): ... this. No longer
15657 (noce_try_store_flag_constants, noce_try_addcc,
15658 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
15659 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
15660 instead of noce_conversion_profitable_p.
15661 * config/i386/i386.c: Include ifcvt.h.
15662 (ix86_option_override_internal): Don't override
15663 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
15664 (ix86_noce_conversion_profitable_p): New function.
15665 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
15666 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
15667 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
15668 * doc/tm.texi: Regenerated.
15670 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15672 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
15675 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15678 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
15679 instructions for small multiply cores.
15681 2017-04-04 Jeff Law <law@redhat.com>
15683 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
15685 (mips_expand_vec_perm_const): Initialize elements in orig_perm
15686 that are not set by the loop over the elements.
15688 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15691 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
15692 int mode, convert_modes it to mode as unsigned, otherwise use
15693 lowpart_subreg to mode rather than SImode.
15694 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
15695 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
15696 Use DImode instead of SImode for the shift count operand.
15697 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
15700 2017-04-04 Richard Biener <rguenther@suse.de>
15702 PR middle-end/80281
15703 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
15704 arithmetic done for the negate or the plus. Simplify.
15705 (A - (-B) -> A + B): Likewise.
15706 * fold-const.c (split_tree): Make sure to not negate pointers.
15708 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
15710 PR rtl-optimization/60818
15711 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
15712 a compare of comparisons with the thing compared if this results
15713 in a different machine mode.
15715 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
15717 * alias.c (base_alias_check): Fix typo in comment.
15718 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
15719 * cgraphunit.c (symbol_table::compile): Likewise.
15720 * collect2.c (maybe_run_lto_and_relink): Likewise.
15721 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
15722 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
15723 * config/avr/avr.c (avr_map_op_t): Likewise.
15724 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
15725 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
15726 * config/epiphany/epiphany.md (movcc): Likewise.
15727 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
15728 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
15730 * config/mips/mips.c (mips_save_restore_reg): Likewise.
15731 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
15732 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
15733 * config/sh/sh.c (sh_rtx_costs): Likewise.
15734 * fold-const.c (fold_truth_andor): Likewise.
15735 * genautomata.c (collapse_flag): Likewise.
15736 * gengtype.h (struct type::u::s): Likewise.
15737 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
15738 * input.c (FORMAT_AMOUNT): Likewise.
15739 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
15740 (known_aggs_to_agg_replacement_list): Likewise.
15741 * ipa-inline-analysis.c: Likewise.
15742 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
15743 * ipa-polymorphic-call.c
15744 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
15745 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
15746 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
15748 * modulo-sched.c (apply_reg_moves): Likewise.
15749 * omp-expand.c (build_omp_regions_1): Likewise.
15750 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
15751 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
15752 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
15753 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
15754 * value-prof.c: Likewise.
15755 * var-tracking.c (val_reset): Likewise.
15757 2017-04-03 Richard Biener <rguenther@suse.de>
15759 PR tree-optimization/80275
15760 * fold-const.c (split_address_to_core_and_offset): Handle
15763 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
15765 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
15766 descriptors is at least equal to that of functions.
15768 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
15770 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
15772 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
15775 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
15776 (mov<IMOD4:mode>): New expander.
15777 (*mov<IMOD4:mode>_internal): New insn and split pattern.
15779 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15781 PR rtl-optimization/79405
15782 * fwprop.c (propagations_left): New variable.
15783 (forward_propagate_into): Decrement it.
15784 (fwprop_init): Initialize it.
15785 (fw_prop): If the variable has reached zero, stop propagating.
15786 (fwprop_addr): Ditto.
15788 2017-03-31 Jakub Jelinek <jakub@redhat.com>
15791 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
15792 a FUNCTION_DECL, pass it as decl instead of origin to
15795 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
15797 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
15800 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
15803 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
15804 TARGET_VSX_SMALL_INTEGER.
15806 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15808 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15809 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
15811 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15813 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
15814 extraction from odd-numbered MSA register.
15816 2017-03-31 Jakub Jelinek <jakub@redhat.com>
15818 PR middle-end/80173
15819 * expmed.c (store_bit_field_1): Don't attempt to create
15820 a word subreg out of hard registers wider than word if they
15821 have HARD_REGNO_NREGS of 1 for their mode.
15823 PR middle-end/80163
15824 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
15825 conversions to integer types wider than word and pointer.
15828 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
15829 (rtx_equal_for_cselib_p): Pass 0 to it.
15830 * cselib.c (cselib_hasher::equal): Likewise.
15831 (rtx_equal_for_cselib_1): Add depth argument. If depth
15832 is 128, don't look up VALUE locs and punt. Increment
15833 depth in recursive calls when walking VALUE locs.
15835 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
15837 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
15838 (make_gcov_file_name): Use the canonical path name for generating
15840 (read_line): Fix handling of files with ascii null bytes.
15842 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
15844 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
15845 to initialise a vector register instead
15846 of using a const_int.
15848 2017-03-30 Jakub Jelinek <jakub@redhat.com>
15850 PR translation/80189
15851 * gimplify.c (omp_default_clause): Use %qs instead of %s in
15852 diagnostic messages.
15854 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
15857 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
15858 (dfp_diex_<mode>): Update mode of operand 1.
15859 * doc/extend.texi (dxex, dxexq): Document change to return type.
15860 (diex, diexq): Document change to argument type.
15862 2017-03-30 Martin Jambor <mjambor@suse.cz>
15865 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
15866 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
15867 it reflects the signature changes performed at the callee side.
15868 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
15869 to cgraph_build_function_type_skip_args.
15870 (build_function_decl_skip_args): Adjust call to the above function.
15872 2017-03-30 Jakub Jelinek <jakub@redhat.com>
15875 * config/i386/sse.md
15876 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
15877 register as dest whenever it is a MEM not rtx_equal_p to the
15878 corresponding dup operand, and when forcing into reg move the
15879 reg into the memory afterwards.
15880 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
15881 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
15882 for the force_reg mode.
15883 (avx512vl_vextractf128<mode>): Use register as dest either
15884 always when a MEM, or when it is a MEM not rtx_equal_p to the
15885 corresponding dup operand, or even not when it is a CONST_VECTOR
15886 depending on the mode and lo vs. hi.
15887 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
15889 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
15890 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
15891 Likewise. Require that operands[2] is even.
15892 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
15893 Remove extraneous parens. Require that operands[2] is a multiple
15895 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
15896 operands[0] is a MEM if <mask_applied>, the predicates/constraints
15897 disallow memory then.
15899 2017-03-30 Richard Biener <rguenther@suse.de>
15901 PR tree-optimization/77498
15902 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
15903 to non-constants over backedges.
15905 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
15907 PR rtl-optimization/80233
15908 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
15909 as last_combined_insn. Do not test for BARRIER_P separately.
15911 2017-03-29 Andreas Schwab <schwab@suse.de>
15914 * calls.c (prepare_call_address): Convert funexp to Pmode before
15915 copying to temp reg.
15917 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15919 PR tree-optimization/80158
15920 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
15921 Handle possible future case of more than one alternate
15923 (replace_rhs_if_not_dup): Likewise.
15924 (replace_one_candidate): Likewise.
15926 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
15928 PR rtl-optimization/80193
15929 * ira.c (ira): Do not check allocation for LRA.
15931 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
15933 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
15934 (nvptx_output_simt_exit): Declare.
15935 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
15936 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
15937 (init_softstack_frame): Move initialization of crtl->is_leaf to...
15938 (nvptx_declare_function_name): ...here. Emit declaration of local
15939 memory space buffer for omp_simt_enter insn.
15940 (nvptx_output_unisimt_switch): New.
15941 (nvptx_output_softstack_switch): New.
15942 (nvptx_output_simt_enter): New.
15943 (nvptx_output_simt_exit): New.
15944 * config/nvptx/nvptx.h (struct machine_function): New fields
15945 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
15946 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
15947 (UNSPECV_SIMT_EXIT): Ditto.
15948 (omp_simt_enter_insn): New insn.
15949 (omp_simt_enter): New expansion.
15950 (omp_simt_exit): New insn.
15951 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
15953 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
15954 (expand_GOMP_SIMT_ENTER_ALLOC): New.
15955 (expand_GOMP_SIMT_EXIT): New.
15956 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
15957 (GOMP_SIMT_ENTER_ALLOC): Ditto.
15958 (GOMP_SIMT_EXIT): Ditto.
15959 * target-insns.def (omp_simt_enter): New insn.
15960 (omp_simt_exit): Ditto.
15961 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
15963 (lower_rec_simd_input_clauses): Implement SIMT privatization.
15964 (lower_rec_input_clauses): Likewise.
15965 (lower_lastprivate_clauses): Handle SIMT privatization.
15967 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
15968 (ompdevlow_adjust_simt_enter): New.
15969 (find_simtpriv_var_op): New.
15970 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
15971 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
15973 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
15974 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
15975 (copy_decl_for_dup_finish): Ditto.
15977 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
15979 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
15982 * config/i386/i386.c (ix86_option_override_internal): Always
15983 allow -mpreferred-stack-boundary=3 for 64-bit targets.
15985 2017-03-28 Bin Cheng <bin.cheng@arm.com>
15987 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
15989 2017-03-28 Bin Cheng <bin.cheng@arm.com>
15991 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
15992 mark new edge's irreducible flag accordign to it.
15993 (vect_do_peeling): Check loop preheader edge's irreducible flag
15994 and pass it to function slpeel_add_loop_guard.
15996 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
15998 PR tree-optimization/80218
15999 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16000 Update block frequencies and counts.
16002 2017-03-28 Richard Biener <rguenther@suse.de>
16004 PR tree-optimization/78644
16005 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16006 of a simplification result we may not use it at all.
16008 2017-03-28 Richard Biener <rguenther@suse.de>
16011 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16012 without arguments, generate default definition of a SSA name.
16014 2017-03-28 Richard Biener <rguenther@suse.de>
16016 PR middle-end/80222
16017 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16018 TYPE_REF_CAN_ALIAS_ALL references.
16019 * fold-const.c (fold_indirect_ref_1): Likewise.
16021 2017-03-28 Martin Liska <mliska@suse.cz>
16024 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16025 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16027 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16028 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16030 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16031 (EXTRA_SPECS): Define.
16032 (SUBTARGET_EXTRA_SPECS): Likewise.
16033 (SUBTARGET_CPP_SPEC): Likewise.
16034 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16035 SUBTARGET_EXTRA_SPECS.
16036 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16038 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16040 * config/arc/simdext.md (vst64_insn): Update pattern.
16041 (vld32wh_insn): Likewise.
16042 (vld32wl_insn): Likewise.
16043 (vld64_insn): Likewise.
16044 (vld32_insn): Likewise.
16046 2017-03-28 Marek Polacek <polacek@redhat.com>
16049 * fold-const.c (fold_comparison): Use protected_set_expr_location
16050 instead of SET_EXPR_LOCATION.
16052 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16054 * tree.c (add_expr): Avoid name lookup warning.
16056 2017-03-27 Jeff Law <law@redhat.com>
16058 PR tree-optimization/80216
16059 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16060 function name. Limit recursion depth.
16061 (record_temporary_equivalences): Corresponding changes.
16063 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16065 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16068 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16071 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16073 * cfgcleanup.c (reg_note_cfa_p): New array.
16074 (insns_have_identical_cfa_notes): New function.
16075 (old_insns_match_p): Don't cross-jump in between /f
16076 and non-/f instructions. If both i1 and i2 are frame related,
16077 verify all CFA notes, their order and content.
16079 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16082 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16083 HImode and SImode with zero extend to DImode to one insn.
16084 (bswap<mode>2_extenddi): Likewise.
16085 (bswapsi2_extenddi): Likewise.
16086 (bswaphi2_extendsi): Likewise.
16087 (bswaphi2): Combine bswap HImode and SImode into one insn.
16088 Separate memory insns from swapping register.
16089 (bswapsi2): Likewise.
16090 (bswap<mode>2): Likewise.
16091 (bswaphi2_internal): Delete, no longer used.
16092 (bswapsi2_internal): Likewise.
16093 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16094 store, and gpr<-gpr swap insns.
16095 (bswap<mode>2_store): Likewise.
16096 (bswaphi2_reg): Register only splitter, combine with the splitter.
16097 (bswaphi2 splitter): Likewise.
16098 (bswapsi2_reg): Likewise.
16099 (bswapsi2 splitter): Likewise.
16100 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16101 the insns into load, store, and register/register insns.
16102 (bswapdi2_ldbrx): Likewise.
16103 (bswapdi2_load): Likewise.
16104 (bswapdi2_store): Likewise.
16105 (bswapdi2_reg): Likewise.
16107 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16109 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16110 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16112 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16115 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16117 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16118 special handling for target option conflicts between dform
16119 options (-mpower9-dform, -mpower9-dform-vector,
16120 -mpower9-dform-scalar) and -mno-direct-move.
16122 2017-03-27 Richard Biener <rguenther@suse.de>
16124 PR tree-optimization/80181
16125 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16127 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16129 * config/arc/predicates.md (move_double_src_operand): Replace the
16130 call to move_double_src_operand with a call to address_operand.
16132 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16134 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16135 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16136 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16138 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16140 * config/arc/predicates.md (long_immediate_loadstore_operand):
16141 Consider scaled addresses cases.
16143 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16145 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16146 restored when in interrupt.
16147 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16148 doesn't have delay slot.
16150 2017-03-27 Richard Biener <rguenther@suse.de>
16153 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16154 inlined thunk clones.
16156 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16159 * asan.c (instrument_derefs): Copy over last operand from
16160 original COMPONENT_REF to the new COMPONENT_REF with
16161 DECL_BIT_FIELD_REPRESENTATIVE.
16162 * ubsan.c (instrument_object_size): Likewise.
16164 2017-03-27 Richard Biener <rguenther@suse.de>
16166 PR tree-optimization/80170
16167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16168 sure DR/SCEV didnt fold in constants we do not see when looking
16169 at the reference base alignment.
16171 2017-03-27 Richard Biener <rguenther@suse.de>
16173 PR middle-end/80171
16174 * gimple-fold.c (fold_ctor_reference): Properly guard against
16175 NULL return value from canonicalize_constructor_val.
16177 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16180 * config/i386/i386.c (ix86_expand_builtin)
16181 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16182 flags reg setting and flags reg using instructions.
16183 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16184 clobbering instructions to zero extend op2.
16186 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16188 * doc/install.texi (Configuration) <--with-aix-soname>:
16189 Update link to AIX ld.
16191 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16193 PR rtl-optimization/80160
16194 PR rtl-optimization/80159
16195 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16196 reg_alternate_class into account.
16198 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16201 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16202 to consider in curr_insn_transform.
16204 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16206 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16207 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16208 and emit_mode_inner.
16210 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16212 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16213 argument to the overloaded builtin variants. Use the new flag to
16214 deprecate certain builtin variants.
16215 * config/s390/s390-builtin-types.def: Add new builtin types.
16216 * config/s390/s390-builtins.h: Support new flags field for
16217 overloaded builtins.
16218 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16219 (s390_macro_to_expand): Enable vector float data type.
16220 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16221 builtins by incrementing the __VEC__ version number.
16222 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16224 (s390_resolve_overloaded_builtin): Emit error messages depending
16225 on the builtin flags.
16226 * config/s390/s390.c (s390_expand_builtin): Support additional
16227 flags argument. Change error message to match the messages
16228 emitted in s390-c.c.
16229 * config/s390/s390.md: New UNSPEC_* constants.
16230 (op_type): Add new instruction types.
16231 * config/s390/vecintrin.h: Add new builtins and test data class
16233 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16234 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16235 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16236 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16237 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16238 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16240 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16241 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16242 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16243 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16245 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16246 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16247 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16248 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16250 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16251 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16252 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16253 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16254 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16255 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16256 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16258 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16259 ("vec_scatter_element<V_HW_4:mode>_DI")
16260 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16261 ("vec_fpint<mode>", "vflls")
16262 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16263 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16264 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16265 ("*vec_cmphe<mode>_cc"): ... these.
16267 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16268 mode constant instead of magic value.
16270 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16272 * config/s390/s390.c (s390_expand_vec_compare): Support other
16273 vector floating point modes than just V2DF.
16274 (s390_expand_vcond): Likewise.
16275 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16276 (s390_cannot_change_mode_class): Prevent mode changes between TF
16277 and V1TF in vector registers.
16278 * config/s390/s390.md (DF, SF): New mode attributes.
16279 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16280 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16281 SFmode support for VRs.
16282 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16284 (VFT, VF_HW): New mode iterators.
16285 (vw, sdx): New mode attributes.
16286 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16287 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16288 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16289 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16290 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16291 also the new vector floating point modes. Renaming to ...
16293 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16294 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16295 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16296 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16297 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16298 ("vec_unordered<mode>"): ... these.
16300 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16301 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16302 ("*vec_extendv2df"): New insn definitions.
16304 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16306 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16307 ("mulditi3_2", "*muldi3_sign"): New patterns.
16308 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16309 rename the pattern definition.
16311 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16313 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16315 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16317 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16319 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16320 instruction if possible.
16321 * config/s390/vector.md (vec_halfnumelts): New mode
16323 ("*vec_vllezlf<mode>"): New pattern.
16325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16327 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16328 ("popcountv4si2", "popcountv2di2"): Rename to ...
16329 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16330 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16332 ("popcount<mode>2_vxe"): New pattern.
16334 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16336 * common/config/s390/s390-common.c (processor_flags_table): Add
16338 * config.gcc: Add arch12.
16339 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16340 Default to arch12 for unknown CPU model numbers.
16341 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16342 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16343 PROCESSOR_max sanity check.
16344 * config/s390/s390-opts.h (enum processor_type): Add
16346 * config/s390/s390.c (processor_table): Add arch12.
16347 (s390_expand_builtin): Add check for B_VXE flag.
16348 (s390_issue_rate): Add PROCESSOR_ARCH12.
16349 (s390_get_sched_attrmask): Likewise.
16350 (s390_get_unit_mask): Likewise.
16351 (s390_sched_score): Enable z13 scheduling for arch12.
16352 (s390_sched_reorder): Likewise.
16353 (s390_sched_variable_issue): Likewise.
16354 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16356 (s390_tune_attr): Use z13 scheduling also for arch12.
16357 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16358 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16359 (TARGET_VXE_P): New macros.
16360 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16361 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16362 * config/s390/s390.opt: Add arch12 as processor_type.
16364 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16366 * config/s390/s390.md
16367 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16368 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16369 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16371 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16372 Rename expanders to ...
16374 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16375 ("fixuns_truncdddi2_emu"): ... these.
16377 ("fixuns_trunc<mode>si2_emu"): New expander.
16379 ("*fixuns_truncdfdi2_z13"): Rename to ...
16380 ("*fixuns_truncdfdi2_vx"): ... this.
16382 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16384 * config/s390/2964.md: Remove the single element vector compare
16385 instructions which are no longer used.
16386 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16388 (s390_canonicalize_comparison): Remove handling of DFmode
16390 (s390_expand_vec_compare_scalar): Remove function.
16391 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16392 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16394 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16396 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16398 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16400 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
16401 will anyway by matched by mov<mode>_64dfp.
16403 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16405 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
16406 vlef/vstef. Add missing operand to vleif.
16408 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16410 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
16411 pair for all vector types with 64 bit elements.
16412 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
16413 * config/s390/vector.md (V_HW_64): ... here.
16414 (V_128_NOSINGLE): New mode iterator.
16415 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
16416 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
16417 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
16418 ("*vec_load_pairv2di"): Change to ...
16419 ("*vec_load_pair<mode>"): ... this one.
16421 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16423 * config/s390/constraints.md: Add comments.
16424 (jKK): Reject element sizes > 8 bytes.
16425 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
16427 * config/s390/s390.md: Add the s_operand checks formerly in
16428 s390_split_ok_p to various splitters where they are still
16430 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
16431 for 128 bit vectors. Plus two splitters.
16433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16435 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
16438 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16441 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
16442 error if the boundary argument is not constant.
16444 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16446 PR rtl-optimization/80112
16447 * loop-doloop.c (doloop_condition_get): Don't check condition
16448 if cmp isn't SET with IF_THEN_ELSE src.
16450 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16452 PR tree-optimization/80158
16453 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
16454 replacing a candidate statement, also replace it for the
16455 candidate's alternate interpretation.
16456 (replace_rhs_if_not_dup): Likewise.
16457 (replace_one_candidate): Likewise.
16459 2017-03-24 Richard Biener <rguenther@suse.de>
16461 PR tree-optimization/80167
16462 * graphite-isl-ast-to-gimple.c
16463 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
16465 (translate_isl_ast_to_gimple::get_rename): Likewise.
16467 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16469 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
16470 handling of certain combinations of target options, including the
16471 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
16472 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
16474 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16477 * config/arm/arm.md (*load_multiple): Add reload_completed to
16478 matching condition.
16480 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16481 Richard Biener <rguenth@suse.de>
16483 PR tree-optimization/79908
16484 PR tree-optimization/80136
16485 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16486 been cast away, gimplify_and_add suffices.
16488 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
16490 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
16492 2017-03-23 Richard Biener <rguenther@suse.de>
16494 PR tree-optimization/80032
16495 * gimplify.c (gimple_push_cleanup): Forced unconditional
16496 cleanups still have to go to the conditional_cleanups
16499 2017-03-22 Jakub Jelinek <jakub@redhat.com>
16501 PR tree-optimization/80072
16502 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
16504 (next_operand_entry_id): Change type to unsigned int.
16505 (sort_by_operand_rank): Make sure to return the right return value
16506 even if unsigned fields are bigger than INT_MAX.
16507 (struct oecount): Change cnt and id type to unsigned int.
16508 (oecount_hasher::equal): Formatting fix.
16509 (oecount_cmp): Make sure to return the right return value
16510 even if unsigned fields are bigger than INT_MAX.
16511 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
16514 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
16515 TREE_READONLY on result if writing it more than once.
16518 * doc/invoke.texi (-fsanitize=thread): Document that with
16519 -fnon-call-exceptions atomics are not able to throw
16523 * tsan.c: Include tree-eh.h.
16524 (instrument_builtin_call): Call maybe_clean_eh_stmt or
16525 maybe_clean_or_replace_eh_stmt where needed.
16526 (instrument_memory_accesses): Add cfg_changed argument.
16527 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
16528 if it returned true.
16529 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
16531 PR rtl-optimization/63191
16532 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
16533 wrapper function, moved the whole old content into ...
16534 (ix86_delegitimize_address_1): ... this. New inline function.
16535 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
16536 true as last argument instead of ix86_delegitimize_address.
16538 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16540 * config/aarch64/aarch64.c (generic_branch_cost): Copy
16541 cortexa57_branch_cost.
16543 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16545 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
16547 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16550 * doc/md.texi (Constraints): Document wA constraint.
16551 * config/rs6000/constraints.md (wA): New.
16552 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
16553 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
16554 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
16555 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
16557 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
16560 * gimplify.c (is_oacc_declared): New function.
16561 (oacc_default_clause): Use it to set default flags for acc declared
16562 variables inside parallel regions.
16563 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
16564 declared variables.
16565 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
16566 declare attribute to any decl as necessary.
16568 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16571 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
16572 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
16573 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
16574 (arm_arch_lpae): This.
16575 * config/arm/arm.c (arm_arch7ve): Rename into ...
16576 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
16577 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
16580 2017-03-22 Martin Liska <mliska@suse.cz>
16583 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
16584 error message instead of an ICE.
16586 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16588 * doc/extend.texi (6.11 Additional Floating Types): Revise.
16590 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
16592 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
16594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16597 2017-03-21 Martin Sebor <msebor@redhat.com>
16599 * doc/extend.texi: Use "cannot" instead of "can't."
16600 * doc/hostconfig.texi: Same.
16601 * doc/install.texi: Same.
16602 * doc/invoke.texi: Same.
16603 * doc/loop.texi: Same.
16604 * doc/md.texi: Same.
16605 * doc/objc.texi: Same.
16606 * doc/rtl.texi: Same.
16607 * doc/tm.texi: Same.
16608 * doc/tm.texi.in: Same.
16609 * doc/trouble.texi: Same.
16611 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
16614 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
16615 (collect_checksum_attributes): Set it.
16616 (die_checksum_ordered): Use it.
16618 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16620 PR tree-optimization/79908
16621 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
16622 change: For a VA_ARG whose LHS has been cast away, use
16623 force_gimple_operand to construct the side effects.
16625 2017-03-21 David Malcolm <dmalcolm@redhat.com>
16627 PR translation/80001
16628 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
16629 more amenable to translation.
16630 (oacc_loop_auto_partitions): Likewise.
16632 2017-03-21 Marek Polacek <polacek@redhat.com>
16633 Martin Sebor <msebor@redhat.com>
16635 PR tree-optimization/80109
16636 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
16637 on INTEGRAL_TYPE_P.
16639 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16640 Segher Boessenkool <segher@kernel.crashing.org>
16643 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
16644 check reg_used_between_p between insn and one of succ or succ2
16645 depending on if succ is artificial insn not inserted into insn
16648 2017-03-21 Martin Liska <mliska@suse.cz>
16650 PR gcov-profile/80081
16651 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
16652 * doc/gcc.texi: Include gcov-dump stuff.
16653 * doc/gcov-dump.texi: New file.
16655 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
16657 PR rtl-optimization/79150
16658 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
16659 conditional jump, if the jump is the last insn of the loop.
16661 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16662 Richard Biener <rguenth@suse.de>
16664 PR tree-optimization/79908
16665 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16666 been cast away, use force_gimple_operand to construct the side
16669 2017-03-21 Martin Liska <mliska@suse.cz>
16671 PR libfortran/79956
16672 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
16675 2017-03-21 Brad Spengler <spender@grsecurity.net>
16678 * plugin.c (htab_hash_plugin): New function.
16679 (add_new_plugin): Use it and adjust.
16680 (parse_plugin_arg_opt): Adjust.
16681 (init_one_plugin): Likewise.
16683 2017-03-21 Richard Biener <rguenther@suse.de>
16685 PR tree-optimization/80032
16686 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
16687 if set force the cleanup to happen unconditionally.
16688 (gimplify_target_expr): Push inserted clobbers with force_uncond
16689 to avoid them being removed by control-dependent DCE.
16691 2017-03-21 Richard Biener <rguenther@suse.de>
16693 PR tree-optimization/80122
16694 * tree-inline.c (copy_bb): Do not expans va-arg packs or
16695 va_arg_pack_len when the inlined call stmt requires pack
16697 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
16699 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16702 * tsan.c (instrument_builtin_call): If the memory model argument
16703 is not a constant, assume it is valid.
16706 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
16709 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
16711 PR rtl-optimization/79910
16712 * combine.c (can_combine_p): Do not allow combining an I0 or I1
16713 if its dest is used by an insn before I2 (other than the combined
16714 insns themselves, which are properly handled already).
16716 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
16719 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
16721 * combine.c (record_used_regs): New static function.
16722 (try_combine): Handle situations where there is an additional
16723 instruction between I2 and I3 which needs to have a LOG_LINK
16727 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
16729 * combine.c (try_combine): Delete redundant i1 test. Call
16730 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
16732 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16735 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
16736 alternatives 13/14.
16738 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16740 PR tree-optimization/80054
16741 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
16742 the optimization if a PHI or any of its arguments is not dominated
16743 by the candidate's basis. Use gphi* rather than gimple* as
16745 (replace_profitable_candidates): Clean up a gimple* variable that
16746 should be a gphi* variable.
16748 2017-03-20 Martin Sebor <msebor@redhat.com>
16751 * doc/extend.texi (attribute constructor): Document present limitation.
16753 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
16756 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
16757 __POWER9_VECTOR__ #ifdef control, change template definition to
16758 use Power9-specific built-in function.
16759 (vec_any_eq): Likewise.
16760 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
16761 to control outcomes from this test.
16762 (vector_ae_<mode>p): For VEC_F modes, likewise.
16764 2017-03-20 Ian Lance Taylor <iant@google.com>
16766 * config/i386/i386.c (ix86_function_regparm): Save an extra
16767 register for -fsplit-stack with DECL_STATIC_CHAIN.
16769 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
16772 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
16773 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
16775 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
16777 * config/riscv/riscv.c (riscv_print_operand): Use "fence
16779 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
16782 2017-03-20 Marek Polacek <polacek@redhat.com>
16785 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
16787 2017-03-20 Richard Biener <rguenther@suse.de>
16789 PR tree-optimization/80113
16790 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
16791 allocate extra SSA name for PHI def.
16792 (add_close_phis_to_outer_loops): Likewise.
16793 (add_close_phis_to_merge_points): Likewise.
16794 (copy_loop_close_phi_args): Likewise.
16795 (copy_cond_phi_nodes): Likewise.
16797 2017-03-20 Martin Liska <mliska@suse.cz>
16799 PR middle-end/79753
16800 * tree-chkp.c (chkp_build_returned_bound): Do not build
16801 returned bounds for a LHS that's not a BOUNDED_P type.
16803 2017-03-20 Martin Liska <mliska@suse.cz>
16807 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
16808 COMPLEX_CST and VECTOR_CST.
16810 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16813 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
16814 target operand. A new splitter adds the clobber statement in case
16815 the target operand is dead anyway.
16817 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
16819 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
16820 to age-old versions of binutils and glibc.
16822 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
16824 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
16826 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
16828 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
16830 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
16832 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
16833 requirement for binutils 2.13.
16835 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
16837 * combine.c (try_combine): Delete redundant i1 test. Call
16838 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
16840 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
16842 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
16843 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
16845 <riscv64-*-elf>: Re-arrange section
16846 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
16847 <riscv32-*-linux>: Likewise.
16848 <riscv64-*-elf>: Likewise
16849 <riscv64-*-linux>: Likewise.
16851 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
16854 * aarch64.opt(verbose-cost-dump): Fix typo.
16856 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
16859 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
16860 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
16862 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
16864 * reload.c (find_reloads): When reloading a nonoffsettable address,
16865 use RELOAD_OTHER for it and its address reloads.
16867 PR rtl-optimization/79910
16868 * combine.c (record_used_regs): New static function.
16869 (try_combine): Handle situations where there is an additional
16870 instruction between I2 and I3 which needs to have a LOG_LINK
16873 2017-03-17 Jeff Law <law@redhat.com>
16875 PR tree-optimization/71437
16876 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
16877 conditional in the hash table first.
16878 (vrp_dom_walker::before_dom_children): Extract condition from
16879 ASSERT_EXPR. Record condition, its inverion and any implied
16880 conditions as well.
16882 2017-03-17 Marek Polacek <polacek@redhat.com>
16883 Markus Trippelsdorf <markus@trippelsdorf.de>
16885 PR tree-optimization/80079
16886 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
16889 2017-03-17 Richard Biener <rguenther@suse.de>
16891 PR middle-end/80075
16892 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
16893 Properly verify the LHS before the RHS possibly claims to be
16895 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
16898 2017-03-17 Martin Jambor <mjambor@suse.cz>
16900 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
16901 (List of -O2 options): Likewise.
16902 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
16905 2017-03-17 Tom de Vries <tom@codesourcery.com>
16907 * gcov-dump.c (print_usage): Print bug_report_url.
16909 2017-03-17 Richard Biener <rguenther@suse.de>
16911 PR middle-end/80050
16912 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
16913 (parser::peek): Likewise.
16915 2017-03-17 Richard Biener <rguenther@suse.de>
16917 PR tree-optimization/80048
16918 * sese.c (free_sese_info): Properly release rename_map and
16919 copied_bb_map elements.
16921 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
16923 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
16924 Add linked-list forward and backlinks. Insert on
16925 construction, remove on destruction.
16926 (class pass_store_merging): Add m_stores_head field.
16927 (pass_store_merging::terminate_and_process_all_chains):
16928 Iterate over m_stores_head list.
16929 (pass_store_merging::terminate_all_aliasing_chains):
16931 (pass_store_merging::execute): Check for debug stmts first.
16932 Push new chains onto the m_stores_head stack.
16934 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
16937 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
16938 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
16939 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
16941 2017-03-16 Jeff Law <law@redhat.com>
16943 PR tree-optimization/71437
16944 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
16945 member function. Implementation moved into after_dom_children
16946 member function and into the threader's thread_outgoing_edges
16948 (dom_opt_dom_walker::after_dom_children): Simplify by moving
16949 some code into new thread_outgoing_edges.
16950 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
16951 definition. Simplify marker handling (do it here). Assume we always
16952 have the available expression and the const/copies tables.
16953 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
16955 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
16956 * tree-vrp.c (equiv_stack): No longer file scoped.
16957 (vrp_dom_walker): New class.
16958 (vrp_dom_walker::before_dom_children): New member function.
16959 (vrp_dom_walker::after_dom_children): Likewise.
16960 (identify_jump_threads): Setup domwalker. Use it rather than
16961 walking edges in a random order by hand. Simplify setup/finalization.
16962 (finalize_jump_threads): Remove.
16963 (vrp_finalize): Do not call identify_jump_threads here.
16964 (execute_vrp): Do it here instead and call thread_through_all_blocks
16967 PR tree-optimization/71437
16968 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
16970 (simplify_stmt_for_jump_threading): Add basic_block argument. All
16972 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
16973 (dom_opt_dom_walker::thread_across_edge): Remove
16974 handle_dominating_asserts argument. All callers changed.
16975 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
16976 changes. Remove calls to lhs_of_dominating_assert. Other
16977 uses of handle_dominating_asserts turn into unconditional code
16978 (simplify_control_stmt_condition_1): Likewise.
16979 (simplify_control_stmt_condition): Likewise.
16980 (thread_through_normal_block, thread_across_edge): Likewise.
16981 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
16982 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
16983 object if it is not an SSA_NAME.
16984 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
16985 before calling into the VRP specific simplifiers.
16986 (identify_jump_threads): Remove handle_dominating_asserts
16989 2017-03-16 Jakub Jelinek <jakub@redhat.com>
16992 * tree-diagnostic.c (default_tree_printer): No longer static.
16993 * tree-diagnostic.h (default_tree_printer): New prototype.
16995 2017-03-16 Tamar Christina <tamar.christina@arm.com>
16997 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
16998 Change ins into fmov.
17000 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17002 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17003 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17004 Use h_con constraint for operand 1.
17005 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17006 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17008 2017-03-15 Jeff Law <law@redhat.com>
17010 PR tree-optimization/71437
17011 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17012 (record_temporary_equivalences): Use it.
17014 PR tree-optimization/71437
17015 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17016 tree-ssa-scopedtables.
17017 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17018 (record_conditions, record_cond, vuse_eq): Likewise.
17019 (record_edge_info): Adjust to API tweak of record_conditions.
17020 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17021 (record_temporary_equivalences, optimize_stmt): Likewise.
17022 (eliminate_redundant_computations): Likewise.
17023 (record_equivalences_from_stmt): Likewise.
17024 * tree-ssa-scopedtables.c: Include options.h and params.h.
17025 (vuse_eq): New function, moved from tree-ssa-dom.c
17026 (build_and_record_new_cond): Likewise.
17027 (record_conditions): Likewise. Accept vector of conditions rather
17028 than edge_equivalence structure for first argument.
17029 for the first argument.
17030 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17031 from tree-ssa-dom.c.
17032 (avail_exprs_stack::record_cond): Likewise.
17033 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17034 from tree-ssa-dom.c.
17035 (avail_exprs_stack): Add new member functions lookup_avail_expr
17037 (record_conditions): Declare.
17039 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17042 * lra-constraints.c (process_alt_operands): Increase reject for
17043 reloading an input/output operand.
17045 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17048 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17049 insns to convert from signed/unsigned char/short to IEEE 128-bit
17051 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17053 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17056 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17057 subreg of inner mode for values already in registers.
17059 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17061 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17062 iteration reg is used after the loop.
17064 2017-03-14 Martin Sebor <msebor@redhat.com>
17066 PR tree-optimization/79800
17067 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17068 precision in negative-positive range.
17069 (format_floating): Call non-const overload with adjusted precision.
17071 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17074 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17077 2017-03-14 Martin Sebor <msebor@redhat.com>
17079 PR middle-end/80020
17080 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17081 * builtins.def (aligned_alloc): Use it.
17084 * Makefile.in (GTFILES): Add calls.c.
17085 * calls.c: Include "gt-calls.h".
17087 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17089 PR rtl-optimization/79728
17090 * regs.h (struct target_regs): New field
17091 x_contains_allocatable_regs_of_mode.
17092 (contains_allocatable_regs_of_mode): New macro.
17093 * reginfo.c (init_reg_sets_1): Initialize it, and change
17094 contains_reg_of_mode so it includes global regs as well.
17095 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17096 rather than contains_regs_of_mode.
17098 2017-03-14 Martin Liska <mliska@suse.cz>
17100 * doc/invoke.texi: Document options that can't be combined with
17101 -fcheck-pointer-bounds.
17103 2017-03-14 Martin Liska <mliska@suse.cz>
17105 PR middle-end/79831
17106 * doc/invoke.texi (-Wchkp): Document the option.
17108 2017-03-14 Martin Liska <mliska@suse.cz>
17110 * Makefile.in: Install gcov-dump.
17112 2017-03-14 Martin Liska <mliska@suse.cz>
17114 * multiple_target.c (expand_target_clones): Bail out for
17115 an invalid attribute.
17117 2017-03-14 Richard Biener <rguenther@suse.de>
17119 * alias.c (struct alias_set_entry): Pack properly.
17120 * cfgloop.h (struct loop): Likewise.
17121 * cse.c (struct set): Likewise.
17122 * ipa-utils.c (struct searchc_env): Likewise.
17123 * loop-invariant.c (struct invariant): Likewise.
17124 * lra-remat.c (struct cand): Likewise.
17125 * recog.c (struct change_t): Likewise.
17126 * rtl.h (struct address_info): Likewise.
17127 * symbol-summary.h (function_summary): Likewise.
17128 * tree-loop-distribution.c (struct partition): Likewise.
17129 * tree-object-size.c (struct object_size_info): Likewise.
17130 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17131 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17132 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17133 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17134 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17135 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17136 (struct _stmt_vec_info): Likewise.
17138 2017-03-14 Martin Liska <mliska@suse.cz>
17141 * multiple_target.c (create_dispatcher_calls): Check that
17142 a target can create a function dispatcher.
17144 2017-03-14 Martin Liska <mliska@suse.cz>
17147 * multiple_target.c (expand_target_clones): Drop local.local
17148 flag for default implementation.
17150 2017-03-14 Richard Biener <rguenther@suse.de>
17152 PR tree-optimization/80030
17153 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17155 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17157 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17158 gcc_fallthrough() instead of __attribute__((fallthrough));
17160 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17162 * doc/gcc.texi: Remove "up" link to (DIR).
17163 * doc/gccint.texi: Ditto.
17165 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17167 * doc/install.texi (Specific) <avr>: Remove reference to
17170 2017-03-13 Jeff Law <law@redhat.com>
17172 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17173 attribute rather than comments.
17175 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17176 match_scratch operand is highest.
17178 2017-03-13 Martin Liska <mliska@suse.cz>
17180 PR middle-end/78339
17181 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17182 is a CHKP clone, use original declaration.
17184 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17186 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17187 (arc_conditional_register_usage): Use a different allocation order
17188 when optimizing for size.
17189 * common/config/arc/arc-common.c (arc_option_optimization_table):
17190 Section anchors default on when optimizing for size.
17192 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17194 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17196 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17198 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17199 * config/arc/arc.md (cpu_facility): Add cd variant.
17200 (*movqi_insn): Add code density variant.
17201 (*movhi_insn): Likewise.
17202 (*movqi_insn): Likewise.
17203 (*addsi3_mixed): Likewise.
17204 (subsi3_insn): Likewise.
17206 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17208 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17210 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17212 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17213 expressions with MINUS and UNARY ops.
17215 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17218 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17220 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17221 between vec_select and vector argument.
17222 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17223 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17224 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17225 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17226 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17227 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17229 2017-03-13 Richard Biener <rguenther@suse.de>
17232 * params.def (vect-max-peeling-for-alignment): Fix typo.
17234 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17236 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17237 issue that only occurred with binutils below 2.18.
17239 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17241 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17242 refer to binutils 2.11/2.12 minimum.
17244 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17246 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17247 ftp.kernel.org and simplify binutils requirement.
17249 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17251 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17253 (Optimize Options): Ditto. Also remove redundancy.
17255 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17257 PR translation/79848
17258 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17260 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17261 to G_ to avoid double translation.
17263 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17265 PR translation/79923
17266 * auto-profile.c (get_combined_location): Convert leading
17267 character of diagnostics to lower case and remove trailing period.
17268 (read_profile): Likewise for various diagnostics.
17269 * config/arm/arm.c (arm_option_override): Remove trailing period
17270 from various diagnostics.
17271 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17272 (msp430_expand_delay_cycles): Likewise.
17274 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17277 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17278 full command-line argument, rather than just "str".
17279 (aarch64_validate_march): Likewise.
17280 (aarch64_validate_mtune): Likewise.
17282 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17284 PR rtl-optimization/78911
17285 * lra-assigns.c (must_not_spill_p): New function.
17286 (spill_for): Use it.
17288 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17290 PR tree-optimization/79981
17291 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17292 ATOMIC_COMPARE_EXCHANGE ifn result.
17293 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17294 IFN_ATOMIC_COMPARE_EXCHANGE.
17296 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17299 * opts.c (parse_sanitizer_options): Add missing question mark to
17300 "did you mean" message.
17302 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17304 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17306 (VMULEUH_UNS): Likewise.
17307 (VMULOUB_UNS): Likewise.
17308 (VMULOUH_UNS): Likewise.
17309 * config/rs6000/rs6000.c (builtin_function_type): Remove
17310 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17312 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17315 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17316 x with result of extra_parsing_for_operand_code_0.
17317 (function_reader::extra_parsing_for_operand_code_0): Convert
17318 return type from void to rtx, returning x. When reading
17319 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17320 larger size containing struct block_symbol.
17322 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17324 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17325 -mfloat128-hardware without -m64.
17327 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17330 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17331 entries to the case statement that marks unsigned arguments to
17332 overloaded functions.
17334 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17337 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17339 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17342 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17343 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17345 2017-03-10 Martin Liska <mliska@suse.cz>
17349 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17350 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17353 2017-03-10 Olivier Hainque <hainque@adacore.com>
17355 * tree-switch-conversion (array_value_type): Start by resetting
17356 candidate type to it's main variant.
17358 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17360 PR rtl-optimization/79909
17361 * combine.c (try_combine): Use simplify_replace_rtx on individual
17362 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17363 of the whole CALL_INSN_FUNCTION_USAGE.
17365 PR tree-optimization/79972
17366 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17367 get_range_info on SSA_NAMEs. Formatting fixes.
17369 2017-03-10 Richard Biener <rguenther@suse.de>
17370 Jakub Jelinek <jakub@redhat.com>
17372 PR tree-optimization/77975
17373 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17374 edge to be constant.
17375 (get_val_for): For constant x return it. Formatting fix.
17376 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17377 would use the same bases as the current one.
17379 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17381 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17382 instead of vec_select for V1TImode.
17383 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17385 (VSX_LE_128): Add V1TI to this mode iterator.
17386 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17387 (*vsx_le_perm_store_<mode>): Likewise.
17388 (pre-reload splitter for VSX stores): Likewise.
17389 (post-reload splitter for VSX stores): Likewise.
17390 (*vsx_xxpermdi2_le_<mode>): Likewise.
17391 (*vsx_lxvd2x2_le_<mode>): Likewise.
17392 (*vsx_stxvd2x2_le_<mode>): Likewise.
17394 2017-03-09 Michael Eager <eager@eagercon.com>
17396 Correct failures with --enable-checking=yes,rtl.
17398 * config/microblaze/microblaze.c (microblaze_expand_shift):
17399 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17400 test for const0_rtx.
17401 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
17402 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
17404 2017-03-09 Richard Biener <rguenther@suse.de>
17406 PR tree-optimization/79977
17407 * graphite-scop-detection.c (scop_detection::merge_sese):
17408 Handle the case of extra exits to blocks dominating the entry.
17410 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
17412 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17415 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
17417 PR rtl-optimization/79949
17418 * lra-constraints.c (process_alt_operands): Check memory when
17419 trying to predict a cycle. Print about the overall increase.
17421 2017-03-09 Richard Biener <rguenther@suse.de>
17423 PR middle-end/79971
17424 * gimple-expr.c (useless_type_conversion_p): Preserve
17425 TYPE_SATURATING for fixed-point types.
17427 2017-03-09 Richard Biener <rguenther@suse.de>
17430 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
17431 alignment of BLKmode params.
17433 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17436 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
17437 (VALL_NO_V2Q): Likewise.
17439 * config/aarch64/aarch64-simd.md
17440 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
17442 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
17443 VALL_NO_V2Q mode iterator.
17444 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
17446 2017-03-09 Martin Liska <mliska@suse.cz>
17448 PR tree-optimization/79631
17449 * tree-chkp-opt.c (chkp_is_constant_addr): Call
17450 tree_int_cst_sign_bit just for INTEGER constants.
17452 2017-03-09 Martin Liska <mliska@suse.cz>
17456 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
17459 2017-03-09 Marek Polacek <polacek@redhat.com>
17462 * tree.c (inchash::add_expr): Handle TREE_VEC.
17464 2017-03-09 Martin Liska <mliska@suse.cz>
17467 (chkp_narrow_size_and_offset): New function.
17468 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
17469 (void chkp_parse_bit_field_ref): New function.
17470 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
17471 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
17473 2017-03-09 Martin Liska <mliska@suse.cz>
17476 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
17477 (chkp_find_bounds_1): Remove gcc_unreachable.
17479 2017-03-09 Jakub Jelinek <jakub@redhat.com>
17482 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
17483 BUILT_IN_SYNC*, determine the access type from the size suffix and
17484 always build a MEM_REF with that type. Handle forgotten
17485 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
17488 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
17489 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
17490 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
17491 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
17492 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
17493 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
17494 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
17495 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
17496 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
17497 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
17498 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
17499 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
17500 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
17501 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
17502 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
17503 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
17504 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
17505 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
17506 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
17507 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
17508 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
17509 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
17510 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
17511 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
17512 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
17513 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
17514 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
17515 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
17516 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
17517 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
17518 definitions outside of __OPTIMIZE__ guarded section.
17521 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
17522 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
17523 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
17524 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
17527 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17529 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
17530 ("vfenez<mode>"): Add missing constraints.
17532 2017-03-08 Martin Sebor <msebor@redhat.com>
17535 * config/nds32/nds32.c (nds32_option_override):
17536 Fix misspelled diagnostic.
17538 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17541 * gimplify.c (gimplify_omp_for): Replace index var in outer
17542 taskloop statement with an artificial variable and add
17543 OMP_CLAUSE_PRIVATE clause for it.
17545 2017-03-08 Richard Biener <rguenther@suse.de>
17547 PR tree-optimization/79955
17548 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
17549 for accesses that are completely outside of the variable.
17551 2017-03-08 Andrew Haley <aph@redhat.com>
17553 PR tree-optimization/79943
17554 * tree-ssa-loop-split.c (compute_new_first_bound): When
17555 calculating the new upper bound, (END-BEG) should be added, not
17558 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17560 * config/avr/avr.md (setmemhi): Make sure match_dup
17561 operand number comes before match_scratch.
17563 2017-03-08 Richard Biener <rguenther@suse.de>
17565 PR tree-optimization/79920
17566 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
17567 with ncopies == 1 to ...
17568 (vect_transform_slp_perm_load): ... here. Properly compute
17569 all element loads by iterating VF times over the group. Do
17570 not handle ncopies (computed in a broken way) in
17571 vect_create_mask_and_perm.
17573 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17576 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
17577 is a uniform vector, use uniform_vector_p return value instead of
17578 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
17580 2017-03-07 Marek Polacek <polacek@redhat.com>
17582 PR middle-end/79809
17583 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
17584 (alloca_call_type): Likewise.
17586 2017-03-07 Martin Liska <mliska@suse.cz>
17588 * gcov.c (process_args): Put comment to correct location.
17590 2017-03-07 Martin Liska <mliska@suse.cz>
17592 PR middle-end/68270
17593 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
17594 Use array_at_struct_end_p instead of DECL_CHAIN (field).
17595 (chkp_narrow_bounds_for_field): Likewise.
17596 (chkp_parse_array_and_component_ref): Pass one more argument to
17599 2017-03-07 Richard Biener <rguenther@suse.de>
17601 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
17604 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17606 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
17607 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
17609 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17612 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
17613 to end of description.
17614 (PARAM_MAX_STORES_TO_MERGE): Likewise.
17616 2017-03-07 Jakub Jelinek <jakub@redhat.com>
17618 PR rtl-optimization/79901
17619 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
17621 (*avx512f_<code><mode>3<mask_name>): ... this.
17622 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
17623 iterator instead of VI8_AVX2_AVX512BW.
17625 PR rtl-optimization/79901
17626 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
17627 min/max expander, expand it using expand_vec_cond_expr.
17630 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
17633 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17636 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
17637 to void * for PCH reasons.
17638 * dwarf2out.c (output_loc_operands, output_die): Cast
17639 v.val_vec.array to unsigned char *.
17641 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
17644 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
17647 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
17649 PR rtl-optimization/79571
17650 * lra-constraints.c (process_alt_operands): Calculate static
17651 reject and subtract it from overall when only addresses will be
17654 2017-03-06 Julia Koval <julia.koval@intel.com>
17657 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
17658 incoming stack boundary to 128 for 64-bit targets.
17660 2017-03-06 Richard Biener <rguenther@suse.de>
17662 PR tree-optimization/79894
17663 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
17664 to NULL after folding it.
17666 2017-03-06 Richard Biener <rguenther@suse.de>
17668 PR tree-optimization/79824
17669 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
17670 check disabling peeling for gaps.
17672 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
17674 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17675 attributes): Document gettimeofday.
17677 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17679 * config/s390/s390.c (s390_option_override_internal): Set
17680 PARAM_MIN_VECT_LOOP_BOUND
17682 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17684 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
17685 * config/s390/s390.md: Likewise.
17687 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17690 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
17691 (<avx2_avx512>_perm<mode>): Rename to ...
17692 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
17694 (<avx512>_perm<mode>_mask): Rename to ...
17695 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
17697 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
17698 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
17699 instead of VI8F_256_512.
17700 (avx512f_perm<mode>): New define_expand.
17701 (avx512f_perm<mode>_mask): Likewise.
17702 (avx512f_perm<mode>_1<mask_name>): New define_insn.
17703 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
17705 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17707 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
17708 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
17710 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
17712 2017-03-06 Martin Liska <mliska@suse.cz>
17715 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
17716 when having a SSA NAME w/o VAR_DECL assigned to it.
17718 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17720 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
17721 msa_dpsub_<su>_d): Fix MODE for vec_select.
17723 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17725 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
17727 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
17729 2017-03-06 Richard Biener <rguenther@suse.de>
17731 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
17732 * plugin.c (register_plugin_info): Likewise.
17733 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
17735 2017-03-05 Jakub Jelinek <jakub@redhat.com>
17737 * config/i386/sse.md (sse_storehps, sse_storelps,
17738 avx_<castmode><avxsizesuffix>_<castmode>,
17739 avx512f_<castmode><avxsizesuffix>_<castmode>,
17740 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
17741 in condition that at least one operand is not a MEM.
17743 2017-03-03 Jakub Jelinek <jakub@redhat.com>
17745 PR middle-end/79805
17746 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
17747 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
17749 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
17750 gimple_call_nothrow_p flag based on whether original builtin can throw.
17751 If it can, emit following stmts on the fallthrough edge.
17752 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
17753 don't create new bb if inserting just debug stmts on the edge, try to
17754 insert them on the fallthru bb or just reset debug stmts.
17756 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
17759 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
17760 restore recog_data (including the operand rtxes inside it) around
17761 the call to get_insn_template.
17763 2017-03-03 Martin Sebor <msebor@redhat.com>
17765 PR tree-optimization/79699
17766 * context.c (context::~context): Free MPFR caches to avoid
17767 a memory leak on program exit.
17769 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17771 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
17772 Use wide_int::ulow () instead of .elt (0).
17774 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
17776 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
17777 (*pushxf): Limit oF constraint to 32bit targets and add oC
17778 constraint for 64bit targets.
17779 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
17780 (*pushdf): Change rmF constraint to rmC.
17782 2017-03-03 Martin Liska <mliska@suse.cz>
17784 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
17785 Remove unused variable.
17787 2017-03-03 Jakub Jelinek <jakub@redhat.com>
17790 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
17791 is a memory operand, increase num_memory.
17792 (ix86_expand_args_builtin): Likewise.
17794 2017-03-03 Jan Hubicka <jh@suse.cz>
17797 * ipa-devirt.c (maybe_record_node): Properly handle
17798 __cxa_pure_virtual visibility.
17800 2017-03-03 Martin Liska <mliska@suse.cz>
17802 PR tree-optimization/79803
17803 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
17805 (pass_loop_prefetch::execute): Disabled optimization if an
17806 assumption about L1 cache size is not met.
17808 2017-03-03 Martin Liska <mliska@suse.cz>
17810 PR rtl-optimization/79574
17811 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
17812 (hash_scan_set): Likewise.
17813 (dump_hash_table): Likewise.
17814 (hoist_code): Likewise.
17816 2017-03-03 Richard Biener <rguenther@suse.de>
17818 * fixed-value.c (fixed_from_string): Restore use of elt (1)
17819 in place of uhigh ().
17820 (fixed_convert_from_real): Likewise.
17822 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
17825 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
17827 2017-03-03 Richard Biener <rguenther@suse.de>
17829 PR middle-end/79818
17830 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
17831 TYPE_OVERFLOW_UNDEFINED check.
17833 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17835 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
17837 (vector_ae_<mode>_p): Likewise.
17838 (vector_nez_<mode>_p): Likewise.
17839 (vector_ne_v2di_p): Likewise.
17840 (vector_ae_v2di_p): Likewise.
17841 (vector_ne_<mode>_p): Likewise.
17842 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
17844 (vsx_tsqrt<mode>2_fe): Likewise.
17846 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
17849 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
17851 2017-03-02 Jakub Jelinek <jakub@redhat.com>
17853 PR rtl-optimization/79780
17854 * cprop.c (one_cprop_pass): When second and further conditional trap
17855 in a single basic block is turned into an unconditional trap, turn it
17856 into a deleted note to avoid RTL verification failures.
17858 2017-03-02 Richard Biener <rguenther@suse.de>
17860 * fold-const.c (const_binop): Use ulow () instead of elt (0).
17862 2017-03-02 Richard Biener <rguenther@suse.de>
17864 PR tree-optimization/79345
17866 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
17867 param and abort the walk, returning -1 if it is hit.
17868 (walk_aliased_vdefs): Take a limit param and pass it on.
17869 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
17870 defaulting to 0 and return a signed int.
17871 * tree-ssa-uninit.c (struct check_defs_data): New struct.
17872 (check_defs): New helper.
17873 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
17874 about uninitialized memory.
17875 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
17876 bogus uninitialized warning.
17877 (fixed_convert_from_real): Likewise.
17879 2017-03-02 Bin Cheng <bin.cheng@arm.com>
17881 PR tree-optimization/66768
17882 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
17883 iv_use if base object can't be determined.
17885 2017-03-02 Jakub Jelinek <jakub@redhat.com>
17887 PR tree-optimization/79345
17888 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
17889 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
17890 (get_pattern_stats): Initialize it.
17891 * genemit.c (gen_expand): Verify match_scratch numbers come after
17892 match_operand/match_dup numbers.
17893 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
17894 match_scratch numbers.
17895 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
17897 * config/s390/s390.md (trunctdsd2): Likewise.
17899 2017-03-02 Richard Biener <rguenther@suse.de>
17901 * wide-int.h (wide_int_storage::operator=): Implement in terms
17904 2017-03-02 Richard Biener <rguenther@suse.de>
17906 PR tree-optimization/79777
17907 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
17908 the to insert expression to sth existing.
17910 2017-03-01 Martin Sebor <msebor@redhat.com>
17912 PR middle-end/79692
17913 * gimple-ssa-sprintf.c
17914 (directive::known_width_and_precision): New function.
17915 (format_integer): Use it.
17916 (get_mpfr_format_length): Consider the full range of precision
17917 when computing %g output with the # flag. Set the likely byte
17918 count to 3 rather than 1 when precision is indeterminate.
17919 (format_floating): Correct the lower bound of precision.
17921 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17923 * doc/invoke.texi: Document default code model for 64-bit Linux.
17925 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17928 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
17929 udiv rather than div since input pattern is unsigned.
17931 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
17933 * config/i386/i386.c (print_reg): Warn for values of
17934 unsupported size in integer register.
17936 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
17939 * config/rs6000/predicates.md (current_file_function_operand): Do
17940 not allow self calls to be local if the function is replaceable.
17942 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
17945 * config/rs6000/altivec.h (vec_ctz and others): Change the
17946 preprocessor macro that controls conditional compilation from
17947 _ARCH_PWR9 to __POWER9_VECTOR__.
17948 (vec_all_ne): Change parameterization of __altivec_scalar_pred
17949 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
17950 control (instead of _ARCH_PWR9 control) so that template
17951 definition uses power9-specific function.
17952 (vec_any_eq): Likewise.
17953 (vec_all_ne): Change macro definition to use a power9-specific
17954 expansion under #ifdef __POWER9_VECTOR__ control (instead of
17955 _ARCH_PWR9 control).
17956 (vec_any_eq) Likewise.
17957 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
17958 expansion for CMPNEF to remove support for xvcmpnesp instruction.
17959 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
17960 support for xvcmpnedp instruction.
17961 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
17962 macro expansion so that Power9 implementation of vec_all_ne does
17963 not use the AltiVec predicate framework.
17964 (VCMPNEH_P): Likewise.
17965 (VCMPNEW_P): Likewise.
17966 (VCMPNED_P): Likewise.
17967 (VCMPNEFP_P): Likewise.
17968 (VCMPNEDP_P): Likewise.
17969 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
17970 implementation of vec_any_eq to not use AltiVec predicate
17972 (VCMPAEH_P): Likewise.
17973 (VCMPAEW_P): Likewise.
17974 (VCMPAED_P): Likewise.
17975 (VCMPAEFP_P): Likewise.
17976 (VCMPAEDP_P): Likewise.
17977 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
17978 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
17979 not use the AltiVec predicate framework.
17980 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
17981 of vec_any_eq to not use AltiVec predicate framework.
17982 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17983 support for predefined __POWER9_VECTOR__ macro to indicate that
17984 Power9 instruction selection is enabled.
17985 (altivec_overloaded_builtins): Remove extraneous
17986 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
17987 function argument types RS6000_BTI_bool_V16QI and
17988 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
17989 entry for overloaded function argument types RS6000_BTI_bool_V4SI
17990 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
17991 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
17992 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
17993 Power9 for implementations of vec_cmpne. Change the signature for
17994 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
17995 (representing vec_all_ne) to remove the previously described first
17996 argument of type RS6000_BTI_INTSI, as this was an artifact of
17997 reliance on the AltiVec predicate framework, which is no longer
17998 used in the implementation of these functions. Add
17999 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18000 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18001 since, unlike the AltiVec predicate framework implementation, we
18002 do not share function descriptors between vec_alle and vec_anyeq.
18003 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18004 set of modes that receive special treatment even when
18005 TARGET_P9_VECTOR is true. The special treatment emits code that
18006 does not depend on Power9 instructions.
18007 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18008 define_expand to not rely on AltiVec predicate framework.
18009 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18011 (vector_ne_v2di_p): Change this define_expand to not rely on
18012 AltiVec predicate framework.
18013 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18015 (vector_ne_<mode>_p): Change this define_expand to not rely on
18016 AltiVec predicate framework.
18017 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18019 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18020 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18021 define_insn pattern.
18022 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18023 define_insn pattern because the xvcmpne<VSs>. instruction is not
18025 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18026 instruction is not supported.
18028 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18030 * config/nvptx/nvptx.c: Include intl.h.
18032 2017-03-01 Martin Jambor <mjambor@suse.cz>
18035 * ipa-prop.h (ipa_bits): Removed field known.
18036 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18037 to pointers. Adjusted their comments to warn about their sharing.
18038 (ipcp_transformation_summary): Change bits to a vector of pointers.
18039 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18040 (ipa_get_ipa_bits_for_value): Declare.
18041 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18042 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18043 (ipa_bits_hash_table): Likewise.
18044 (ipa_vr_ggc_hash_traits): Likewise.
18045 (ipa_vr_hash_table): Likewise.
18046 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18047 being pointers and vr_known being removed.
18048 (ipa_set_jf_unknown): Likewise.
18049 (ipa_get_ipa_bits_for_value): New function.
18050 (ipa_set_jfunc_bits): Likewise.
18051 (ipa_get_value_range): New overloaded functions.
18052 (ipa_set_jfunc_vr): Likewise.
18053 (ipa_compute_jump_functions_for_edge): Use the above functions to
18054 construct bits and vr parts of jump functions.
18055 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18056 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18058 (ipcp_grow_transformations_if_necessary): Also allocate
18059 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18061 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18062 them. Fix too long lines.
18063 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18064 vr_known being removed.
18065 (ipa_read_jump_function): Use new setter functions to construct bits
18066 and vr parts of jump functions or set them to NULL.
18067 (write_ipcp_transformation_info): Adjust for bits being pointers.
18068 (read_ipcp_transformation_info): Likewise.
18069 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18071 Include gt-ipa-prop.h.
18072 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18074 (ipcp_store_bits_results): Likewise.
18075 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18076 Do not write to existing jump functions but use a temporary instead.
18078 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18081 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18082 attempt to use its first operand as BIT_FIELD_REF base.
18084 2017-03-01 Richard Biener <rguenther@suse.de>
18086 PR middle-end/79721
18087 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18088 interpolating formula in wrapping arithmetic.
18089 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18091 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18093 PR tree-optimization/79734
18094 * tree-vect-generic.c (expand_vector_condition): Optimize
18095 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18096 Handle VEC_COND_EXPR where comparison has different inner width from
18097 type's inner width.
18099 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18101 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18102 markup, and similar issues. Remove @opindex entries for things
18103 that aren't options. Add missing -mmpy-option entries.
18105 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18107 PR tree-optimization/79737
18108 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18109 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18110 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18111 instead of byte_size. Formatting fix.
18112 (shift_bytes_in_array_right): Formatting fix.
18114 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18117 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18118 condition on optimize for the leaf function test.
18120 2017-02-28 Martin Liska <mliska@suse.cz>
18123 * read-rtl-function.c (function_reader::handle_unknown_directive):
18124 Bail out when one uses -flto.
18126 2017-02-28 Martin Liska <mliska@suse.cz>
18128 * common.opt: Replace space with tabular for options of <number>
18130 * config/i386/i386.opt: Show <number> value for
18131 -mlarge-data-threshold.
18132 * opts.c (print_filtered_help): Do not display number in hexadecimal
18135 2017-02-28 Martin Liska <mliska@suse.cz>
18137 * common.opt: Fix --help=option -Q for options which are of
18140 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18142 * config/i386/i386.c (print_reg): Error out for values
18143 of 8-bit size in invalid integer register.
18145 2017-02-28 Martin Sebor <msebor@redhat.com>
18147 PR tree-optimization/79691
18148 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18150 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18153 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18154 gcc_unreachable with output_operand_lossage.
18156 2017-02-28 Richard Biener <rguenther@suse.de>
18158 PR tree-optimization/79740
18159 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18161 (visit_nary_op): Insert the nary into the hashtable if we
18162 pattern-matched sth.
18163 * tree-ssa-pre.c (eliminate_insert): Robustify.
18165 2017-02-28 Richard Biener <rguenther@suse.de>
18167 PR middle-end/79731
18168 * fold-const.c (decode_field_reference): Reject out-of-bound
18171 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18173 * config/i386/i386.c: Include intl.h.
18174 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18175 instead of just cond ? "..." : "...".
18176 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18177 * coverage.c (read_counts_file): Likewise.
18178 * omp-offload.c: Include intl.h.
18179 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18180 of just cond ? "..." : "...".
18181 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18182 of just cond ? "..." : "...".
18184 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18187 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18189 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18190 'tune for' CPU name.
18191 * config/arm/arm-cpu-data.h: Regenerated.
18193 2017-02-28 Richard Biener <rguenther@suse.de>
18195 PR tree-optimization/79732
18196 * tree-inline.c (expand_call_inline): Do not shadow var.
18198 2017-02-28 Richard Biener <rguenther@suse.de>
18200 PR tree-optimization/79723
18201 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18202 address-space properly.
18204 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18206 * doc/optinfo.texi (Optimization groups): Fix option used for
18208 * doc/invoke.texi (-fopt-info): Document "omp".
18209 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18210 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18211 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18212 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18213 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18215 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18217 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18218 OPTGROUP_OMP with "omp".
18220 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18223 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18224 for arithmetic shift of unsigned V2DI.
18226 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18228 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18229 arc/linux.h headers.
18230 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18231 (LINK_SPEC): Likewise.
18232 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18233 (EXTRA_SPECS): Likewise.
18234 (STARTFILE_SPEC): Likewise.
18235 (ENDFILE_SPEC): Likewise.
18236 (LIB_SPEC): Likewise.
18237 (TARGET_SDATA_DEFAULT): Likewise.
18238 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18239 (MULTILIB_DEFAULTS): Likewise.
18240 (DWARF2_UNWIND_INFO): Likewise.
18241 * config/arc/big.h: New file.
18242 * config/arc/elf.h: Likewise.
18243 * config/arc/linux.h: Likewise.
18244 * config/arc/t-uClibc: Remove.
18246 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18248 PR tree-optimization/77536
18249 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18250 (tree_transform_and_unroll_loop): Use above function to compute the
18251 estimated niter of unrolled loop and use it when scaling profile.
18252 Also use count info rather than frequency if it's non-zero.
18253 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18254 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18255 (vect_transform_loop): Call above function.
18257 2017-02-27 Richard Biener <rguenther@suse.de>
18259 PR tree-optimization/45397
18260 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18261 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18262 (visit_nary_op): Add pattern matching for CSEing sign-changed
18263 or truncated operations with wider ones.
18265 2017-02-27 Richard Biener <rguenther@suse.de>
18267 PR tree-optimization/79690
18268 * tree-vect-stmts.c (vectorizable_store): Use vector type
18269 built from the DR with address-space.
18271 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18273 * doc/invoke.texi (Optimize Options): Refine the description
18274 of asan-use-after-return.
18276 2017-02-25 Alan Modra <amodra@gmail.com>
18278 PR rtl-optimization/79584
18279 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18280 base, not ad->base_term, the reg within base. Remove assertion
18281 that ad->base == ad->base_term. Replace gen_int_mode using
18282 bogus mode with const0_rtx.
18284 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18286 PR middle-end/79396
18287 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18288 FMA_EXPR like tcc_binary or tcc_unary.
18290 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18293 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18295 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18296 (output_loc_operands): Handle DW_OP_call_ref and
18297 DW_OP_GNU_variable_value.
18298 (struct variable_value_struct): New type.
18299 (struct variable_value_hasher): Likewise.
18300 (variable_value_hash): New variable.
18301 (string_types): Remove.
18302 (copy_loc_descr): New function.
18303 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18304 (prepend_loc_descr_to_each): New function.
18305 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18306 instead of add_loc_descr_to_each if the first argument is single
18307 location list and the second has multiple.
18308 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18309 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18310 when looking for variable value which doesn't have other location info.
18311 (loc_list_from_tree): Formatting fix.
18312 (gen_array_type_die): Simplify DW_AT_string_length handling.
18313 (adjust_string_types): Remove.
18314 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18315 string_types. Call resolve_variable_values.
18316 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18317 (resolve_addr_in_expr): Likewise. Add A argument.
18318 (copy_deref_exprloc): Remove deref argument. Adjust for the
18319 original expression being DW_OP_GNU_variable_value with optionally
18320 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18321 optionally after it.
18322 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18323 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18324 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18325 (variable_value_hasher::hash, variable_value_hasher::equal): New
18327 (resolve_variable_value_in_expr, resolve_variable_value,
18328 resolve_variable_values, note_variable_value_in_expr,
18329 note_variable_value): New functions.
18330 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18333 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18336 * opts.h (handle_generated_option): Add GENERATED_P argument.
18337 * opts-common.c (handle_option): Adjust function comment.
18338 (handle_generated_option): Add GENERATED_P argument, pass it to
18340 (control_warning_option): Pass false to handle_generated_option
18342 * opts.c (maybe_default_option): Pass true to handle_generated_option
18344 * optc-gen.awk: Likewise.
18346 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18348 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18349 a REG, look at the REG it is a SUBREG of.
18350 (splitter for cmpeqsi_t): Ditto.
18352 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18354 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18355 the special USEs with the pattern of the insn, not the insn itself.
18357 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18360 * doc/invoke.texi: Document -mload-store-pairs.
18362 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18363 Sandra Loosemore <sandra@codesourcery.com>
18365 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18366 argument isn't a CONST_INT.
18367 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18368 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18369 (nios2_validate_compare): Bypass alternate compare logic if *op2
18370 is not a CONST_INT.
18371 (ldstwm_operation_p): Return false if first_base is not a REG or
18372 if first_offset is not a CONST_INT.
18374 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18376 * config/cris/cris.md: Use correct operand in a define_peephole2.
18378 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18380 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18382 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18384 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18385 this_insn if it is an INSN or JUMP_INSN.
18386 (force_offsettable): Look at base, not at addr.
18387 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18388 on things that aren't necessarily CONST_INTs.
18390 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18392 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18393 -mfpmath=sse is the default also for x86-32 targets with SSE2
18394 instruction set when @option{-ffast-math} is enabled
18396 2017-02-24 Jeff Law <law@redhat.com>
18398 PR rtl-optimizatoin/79286
18399 * ira.c (update_equiv_regs): Drop may_trap_p exception to
18402 2017-02-24 Richard Biener <rguenther@suse.de>
18404 PR tree-optimization/79389
18405 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
18408 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
18410 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
18411 function comment to reflect reality.
18412 (loop_exits_before_overflow): Fix typo in function description.
18414 2017-02-24 Richard Biener <rguenther@suse.de>
18416 PR tree-optimization/79389
18417 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
18418 properly that a threading opportunity exists. Detect conditional
18419 copy/constant propagation opportunities.
18421 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
18423 * config/visium/visium.md (type): Add trap.
18424 (b): New mode attribute.
18425 (*btst): Rename into...
18426 (*btst<mode>): ...this and adjust.
18427 (*cbranchsi4_btst_insn): Rename into...
18428 (*cbranch<mode>4_btst_insn): ...this and adjust.
18429 (trap): New define_insn.
18431 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18433 PR tree-optimization/79389
18434 * ifcvt.c (struct noce_if_info): Add rev_cond field.
18435 (noce_reversed_cond_code): New function.
18436 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
18437 reversed_comparison_code. Formatting fix.
18438 (noce_try_store_flag): Test rev_cond != NULL in addition to
18439 reversed_comparison_code.
18440 (noce_try_store_flag_constants): Likewise.
18441 (noce_try_store_flag_mask): Likewise.
18442 (noce_try_addcc): Use rev_cond if non-NULL instead of
18443 reversed_comparison_code.
18444 (noce_try_cmove_arith): Likewise. Formatting fixes.
18445 (noce_try_minmax, noce_try_abs): Clear rev_cond.
18446 (noce_find_if_block): Initialize rev_cond.
18447 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
18448 instead of false as last argument never attempt to reverse it
18451 2017-02-23 Bin Cheng <bin.cheng@arm.com>
18453 PR tree-optimization/79663
18454 * tree-predcom.c (combine_chains): Process refs in reverse order
18455 only for ZERO length chains, and add explaining comment.
18457 2017-02-23 Jeff Law <law@redhat.com>
18459 PR tree-optimization/79578
18460 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
18461 in call to operand_equal_p.
18463 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
18466 * config/i386/cpuid.h: Fix another undefined behavior.
18468 2017-02-23 Richard Biener <rguenther@suse.de>
18470 PR tree-optimization/79683
18471 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
18472 vector types for data-refs.
18474 2017-02-23 Martin Liska <mliska@suse.cz>
18476 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
18478 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18480 PR middle-end/79665
18481 * internal-fn.c (get_range_pos_neg): Moved to ...
18482 * tree.c (get_range_pos_neg): ... here. No longer static.
18483 * tree.h (get_range_pos_neg): New prototype.
18484 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
18485 are known to be in between 0 and signed maximum inclusive, try to
18486 expand both unsigned and signed divmod and use the cheaper one from
18489 2017-02-22 Jeff Law <law@redhat.com>
18491 PR tree-optimization/79578
18492 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
18493 to compare base operands.
18495 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
18498 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
18499 gpc_reg_operand instead of fpr_reg_operand.
18501 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
18503 * config/mips/mips.c (mips_return_in_memory): Force FP
18504 vector types to be returned in memory for o32 ABI.
18506 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18508 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
18509 instead of DW_TAG_member for static data member declarations and don't
18510 set no_linkage_name for static inline data members.
18511 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
18514 2017-02-22 Martin Liska <mliska@suse.cz>
18516 * doc/invoke.texi: Replace inequality signs with square brackets
18519 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18521 PR tree-optimization/68644
18522 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
18524 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
18527 * lra-constraints.c (simplify_operand_subreg): Handle
18528 WORD_REGISTER_OPERATIONS targets.
18530 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18533 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
18534 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
18535 elimination by swapping fld*.
18537 2017-02-22 Richard Biener <rguenther@suse.de>
18539 PR tree-optimization/79673
18540 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
18541 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
18542 irrelevant address-space qualifiers and avoiding a
18543 ADDR_SPACE_CONVERT_EXPR from fold_convert.
18545 2017-02-22 Richard Biener <rguenther@suse.de>
18547 PR tree-optimization/79666
18548 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
18549 to not symbolically negate if that may introduce undefined
18552 2017-02-22 Martin Liska <mliska@suse.cz>
18555 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
18556 * data-streamer-out.c (streamer_write_gcov_count_stream):
18558 * value-prof.c (stream_out_histogram_value): Make assert more
18559 precise based on type of counter.
18561 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
18564 * config/i386/i386.md (standard_x87sse_constant_load splitter):
18565 Use nonimmediate_operand instead of memory_operand for operand 1.
18566 (float-extend standard_x87sse_constant_load splitter): Ditto.
18568 2017-02-21 Jeff Law <law@redhat.com>
18570 PR tree-optimization/79621
18571 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
18572 blocks with edges to themselves.
18574 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18577 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
18578 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
18579 Use gimple_call_builtin_p.
18582 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
18583 on temporarily removed DEBUG_INSNs.
18585 PR tree-optimization/79649
18586 * tree-loop-distribution.c (classify_partition): Give up on
18587 non-generic address space loads/stores.
18589 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
18591 * doc/loop.texi (Loop manipulation): Remove nonexistent
18592 tree_ssa_loop_version from the documentation.
18593 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
18595 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18598 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
18599 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
18600 * config/rs6000/rs6000.c: Include except.h.
18601 (rs6000_expand_split_stack_prologue): Call
18602 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
18604 2017-02-21 Martin Jambor <mjambor@suse.cz>
18607 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
18608 have been analyzed.
18610 2017-02-21 Martin Jambor <mjambor@suse.cz>
18612 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
18613 for backward compatibility only.
18614 * doc/invoke.texi (Option Summary): Remove all references to
18615 -fipa-cp-alignment.
18617 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
18621 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18623 * lra-constraints.c (curr_insn_transform): Handle
18624 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18626 2017-02-21 Martin Liska <mliska@suse.cz>
18628 * config/i386/i386.opt: Replace -masm-dialect with -masm.
18630 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
18632 PR translation/79638
18633 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
18635 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
18638 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
18639 (arm_function_ok_for_sibcall): Return false for an indirect call by
18640 descriptor if all the argument registers are used.
18641 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
18642 alignment of the function.
18644 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18646 PR tree-optimization/61441
18647 * simplify-rtx.c (simplify_const_unary_operation): For
18648 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
18649 the sNaN unmodified.
18651 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18653 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18654 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18655 instead of SYSTEM_HEADER_DIR.
18657 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
18658 Martin Liška <mliska@suse.cz>
18660 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
18661 Fix typos and grammar, use active voice, and clarify.
18663 2017-02-20 Marek Polacek <polacek@redhat.com>
18665 PR middle-end/79537
18666 * gimplify.c (gimplify_expr): Handle unused *&&L;.
18669 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
18671 2017-02-20 Jakub Jelinek <jakub@redhat.com>
18674 * config/i386/i386.c (ix86_expand_builtin): Handle
18675 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
18676 ix86_builtins_isa[fcode].isa as a requirement of those
18677 flags and any other flag in the bitmask.
18678 (ix86_init_mmx_sse_builtins): Use 0 instead of
18679 ~OPTION_MASK_ISA_64BIT as mask.
18680 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
18681 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
18682 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
18683 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
18685 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18688 * lra-constraints.c (split_reg): Check requested split mode
18689 is supported by the register.
18691 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18693 * lra-constraints.c (simplify_operand_subreg): Remove early
18696 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18699 * lra-constraints.c (curr_insn_transform): Tighten condition
18700 for converting SUBREG reloads from OP_OUT to OP_INOUT.
18702 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18705 * lra-constraints.c (curr_insn_transform): Handle
18706 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18708 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
18711 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
18713 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
18715 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
18718 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
18721 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18723 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
18726 2017-02-18 Jakub Jelinek <jakub@redhat.com>
18728 * final.c (last_columnnum, override_columnnum): New variables.
18729 (final_start_function): Set last_columnnum, pass it to begin_prologue
18730 hook and pass 0 to dwarf2out_begin_prologue.
18731 (final_scan_insn): Update override_columnnum. Pass last_columnnum
18732 to source_line debug hook.
18733 (notice_source_line): Compute last_columnnum and for debug_column_info
18734 return true on column changes.
18735 * debug.h (struct gcc_debug_hooks): Add column argument to
18736 source_line and begin_prologue hooks.
18737 (debug_nothing_int_charstar_int_bool): Remove prototype.
18738 (debug_nothing_int_int_charstar,
18739 debug_nothing_int_int_charstar_int_bool): New prototypes.
18740 (dwarf2out_begin_prologue): Add column argument.
18741 * debug.c (do_nothing_debug_hooks): Adjust source_line and
18742 begin_prologue hooks.
18743 (debug_nothing_int_charstar_int_bool): Remove.
18744 (debug_nothing_int_int_charstar,
18745 debug_nothing_int_int_charstar_int_bool): New functions.
18746 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
18747 through to dwarf2out_source_line.
18748 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
18749 (dwarf2out_source_line): Add column argument, emit it if requested.
18750 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
18752 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
18753 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
18754 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
18755 through to dwarf2out_begin_prologue.
18756 (vmsdbgout_source_line): Add column argument, pass it through to
18757 dwarf2out_source_line.
18758 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
18759 dbxout_source_line caller.
18760 (dbxout_source_line): Add column argument.
18762 * common.opt (gno-column-info, gcolumn-info): New options.
18763 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
18764 (check_die): Also test for multiple DW_AT_decl_column attributes.
18765 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
18766 DW_AT_decl_column if requested.
18767 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
18769 (gen_variable_die): Likewise.
18770 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
18771 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
18774 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
18775 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
18776 (ix86_handle_option): Handle OPT_m3dnowa.
18777 * doc/invoke.texi (-m3dnowa): Document.
18778 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
18779 -m3dnowa instead of -m3dnow -march=athlon.
18782 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
18783 instead of gcc_assert for K, r and R code checks. Formatting fixes.
18785 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18788 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
18789 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
18790 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
18791 generator for vsx_xxpermdi_<mode>_be.
18792 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
18793 force big-endian semantics.
18794 (vsx_xxpermdi_<mode>_be): New define_expand with same
18795 implementation as previous version of vsx_xxpermdi_<mode>.
18797 2017-02-17 Jakub Jelinek <jakub@redhat.com>
18799 PR tree-optimization/79327
18800 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
18801 variable, its initialization and use.
18803 2017-02-17 Julia Koval <julia.koval@intel.com>
18805 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
18806 (OPTION_MASK_ISA_PKU_UNSET): New.
18807 (ix86_handle_option): Handle -mrdpid.
18808 * config/i386/cpuid.h (bit_RDPID): New.
18809 * config/i386/driver-i386.c (host_detect_local_cpu):
18810 Detect RDPID feature.
18811 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
18812 * config/i386/i386-c.c (ix86_target_macros_internal):
18814 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
18815 (ix86_valid_target_attribute_inner_p): Add "rdpid".
18816 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
18817 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
18818 * config/i386/i386.md (define_insn "rdpid"): New.
18819 * config/i386/i386.opt Add -mrdpid.
18820 * config/i386/immintrin.h (_rdpid_u32): New.
18822 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
18824 PR rtl-optimization/79541
18825 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
18826 instead of transforming it into USE.
18828 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
18830 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
18831 If HONOR_SNANS (SFmode) force the input to a register.
18832 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
18833 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
18834 an frsp or similar insn.
18836 2017-02-17 Martin Liska <mliska@suse.cz>
18838 PR rtl-optimization/79577
18839 * params.def (selsched-max-sched-times): Increase minimum to 1.
18841 2017-02-17 Martin Liska <mliska@suse.cz>
18843 PR rtl-optimization/79574
18844 * gcse.c (want_to_gcse_p): Prevent integer overflow.
18846 2017-02-17 Martin Liska <mliska@suse.cz>
18848 PR tree-optimization/79529
18849 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
18850 ssa_defined_default_def_p to handle cases which are implicitly
18852 * tree-ssa.c (ssa_defined_default_def_p): New function.
18853 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
18854 which are implicitly defined.
18855 * tree-ssa.h (ssa_defined_default_def_p): Declare.
18857 2017-02-17 Richard Biener <rguenther@suse.de>
18859 PR middle-end/79576
18860 * params.def (max-ssa-name-query-depth): Limit to 10.
18862 2017-02-17 Richard Biener <rguenther@suse.de>
18864 PR tree-optimization/79552
18865 * tree-ssa-structalias.c (visit_loadstore): Properly verify
18868 2017-02-17 Richard Biener <rguenther@suse.de>
18871 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
18873 2017-02-17 Marek Polacek <polacek@redhat.com>
18875 PR middle-end/79536
18876 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
18877 (fold_negate_expr): New wrapper.
18879 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
18881 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
18882 Correct terminology and de-emphasize pre-standard behavior.
18884 2017-02-16 Alan Modra <amodra@gmail.com>
18886 PR rtl-optimization/79286
18887 * ira.c (def_dominates_uses): New function.
18888 (update_equiv_regs): Don't create an equivalence for insns that
18889 may trap where the register def does not dominate the use.
18891 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
18893 PR rtl-optimization/78127
18894 * lra.c (lra): Call lra_eliminate before finish the loop after
18897 2017-02-16 Richard Biener <rguenther@suse.de>
18899 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
18901 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
18902 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
18903 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
18904 (isl_val_int_from_wi): New function.
18905 (extract_affine_gmp): Rename to ...
18906 (extract_affine_wi): ... this, take a widest_int.
18907 (extract_affine_int): Just wrap extract_affine_wi.
18908 (add_param_constraints): Use isl_val_int_from_wi.
18909 (add_loop_constraints): Likewise, and extract_affine_wi.
18911 2017-02-15 Jeff Law <law@redhat.com>
18913 PR middle-end/79521
18914 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
18915 ira_init_register_move_cost_if_necessary.
18917 2017-02-15 Martin Sebor <msebor@redhat.com>
18919 PR middle-end/32003
18920 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
18921 removed in a prior commit.
18923 2017-02-15 Bin Cheng <bin.cheng@arm.com>
18925 PR tree-optimization/79347
18926 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
18927 counters during peeling.
18929 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
18931 * Makefile.in (site.exp): Remove "set ISLVER".
18933 2017-02-15 Jakub Jelinek <jakub@redhat.com>
18936 * real.c (real_from_integer): Call real_convert even for decimal.
18938 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
18941 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
18943 2017-02-14 Andrew Pinski <apinski@cavium.com>
18945 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
18946 cores and change the partno/implementer to be correct.
18947 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
18948 the 'B" as the implementer.
18949 * config/aarch64/aarch64-tune.md: Regenerate.
18951 2017-02-14 Carl Love <cel@us.ibm.com>
18953 * config/rs6000/rs6000.c: Add case statement entry to make the
18954 xvcvuxdsp built-in argument unsigned.
18955 * config/rs6000/vsx.md: Fix the source and return operand types so they
18956 match the instruction definitions from the ISA document. Fix typo
18957 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
18960 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
18963 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
18964 member early_clobber_alts.
18965 * lra-lives.c (reg_early_clobber_p): New.
18966 (process_bb_lives): Use it.
18967 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
18968 (debug_operand_data): Initialize early_clobber_alts.
18969 (setup_operand_alternative): Set up early_clobber_alts.
18970 (collect_non_operand_hard_regs): Ditto. Pass early clobber
18971 alternatives to new_insn_reg.
18972 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
18974 (lra_update_insn_regno_info): Pass the new arg.
18976 2017-02-14 Jakub Jelinek <jakub@redhat.com>
18978 PR middle-end/79505
18979 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
18980 (new_oacc_loop_raw): Don't clear already cleared fields.
18983 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
18984 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
18985 _mm512_prefetch_i64gather_ps): New inline functions and macros.
18987 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
18990 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
18992 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
18995 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
18996 the extra instruction to the right place to store 128-bit constant
18999 2017-02-14 Martin Sebor <msebor@redhat.com>
19001 PR middle-end/79448
19002 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19003 warning for strings of unknown length.
19005 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19007 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19009 2017-02-14 Jeff Law <law@redhat.com>
19012 * ira-costs.c (scan_one_insn): Initialize register move costs
19013 for pseudos seen in USE/CLOBBER insns.
19015 PR tree-optimization/79095
19016 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19017 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19018 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19019 if the operands are known to be not equal, then the resulting range
19021 (intersect_ranges): If the new range is ~[0,0] and the old range is
19022 wide, then prefer ~[0,0].
19023 * tree-vrp.c (overflow_comparison_p_1): New function.
19024 (overflow_comparison_p): New function.
19025 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19026 if NAME is used in an overflow test.
19027 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19028 overflow check that can be expressed as an equality test, then adjust
19029 ops to be that equality test.
19031 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19033 * config/s390/s390-builtin-types.def: Remove flags argument.
19034 * config/s390/s390.c (s390_init_builtins): Likewise.
19036 2017-02-14 Martin Liska <mliska@suse.cz>
19038 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19039 vector. Fix trailing white spaces.
19041 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19043 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19046 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19048 PR rtl-optimization/68664
19049 * config/arm/arm.c (arm_sched_can_speculate_insn):
19050 New function. Declare prototype.
19051 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19053 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19055 PR rtl-optimization/68664
19056 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19058 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19060 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19062 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19063 max skip bytes for function, loop and jump.
19065 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19067 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19068 ABS_EXPR for gimple dump.
19070 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19073 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19075 PR tree-optimization/79408
19076 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19077 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19078 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19079 also if rhs1 is INTEGER_CST.
19081 2017-02-14 Richard Biener <rguenther@suse.de>
19083 PR middle-end/79432
19084 * tree-into-ssa.c (insert_phi_nodes): When the function can
19085 have abnormal edges rewrite SSA names with broken use-def
19086 dominance out of SSA and register them for PHI insertion.
19088 2017-02-13 Martin Sebor <msebor@redhat.com>
19090 PR middle-end/79496
19091 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19092 clearing info.nowrite flag when snprintf size argument is a range.
19094 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19096 * cprop.c (cprop_jump): Add missing space in string literal.
19097 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19098 (get_constraint_for_component_ref): Likewise.
19099 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19100 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19101 * lra-constraints.c (process_alt_operands): Likewise.
19102 * ipa-inline.c (inline_small_functions): Likewise.
19103 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19104 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19105 * trans-mem.c (diagnose_tm_1_op): Likewise.
19106 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19107 (grid_parallel_clauses_gridifiable): Likewise.
19109 * config/nvptx/mkoffload.c (process): Add space in between
19112 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19113 "MOD4_SSE_REGS" and "ALL_REGS".
19115 * spellcheck.c (test_data): Add , in between "foo" and "food".
19117 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19120 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19121 boundary crossing check and subsequent code generation agree.
19123 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19125 * config/aarch64/aarch64.c (has_memory_op): Delete.
19126 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19129 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19131 PR rtl-optimization/79388
19132 PR rtl-optimization/79450
19133 * combine.c (distribute_notes): When removing TEM_INSN for which
19134 corresponding dest has last value recorded, invalidate that last
19137 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19139 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19140 of explicit '@'. Add missing assembly comment marker on branch costs
19143 2017-02-13 Nathan Sidwell <nathan@acm.org>
19145 * gengtype-lex.l (<in_struct>): Add '/'.
19147 2017-02-13 Martin Liska <mliska@suse.cz>
19150 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19152 2017-02-13 Richard Biener <rguenther@suse.de>
19154 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19156 * configure: Re-generate.
19157 * config.in: Likewise.
19158 * graphite-dependences.c: Simplify as if
19159 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19160 * graphite-isl-ast-to-gimple.c: Likewise.
19161 * graphite-optimize-isl.c: Likewise.
19162 * graphite-poly.c: Likewise.
19163 * graphite-sese-to-poly.c: Likewise.
19164 * graphite.h: Likewise.
19165 * toplev.c: Include isl/version.h and use isl_version () for
19166 printing the ISL version.
19167 * doc/install.texi: Update ISL requirement.
19169 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19171 * doc/standards.texi (Standards): Update reference to
19174 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19176 * doc/extend.texi (Named Address Spaces): sourceware.org now
19178 * doc/install.texi (Binaries): Ditto.
19181 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19183 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19184 standard terminology "stringize"/"stringizing" throughout.
19185 * doc/cppinternals.texi: Likewise.
19187 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19189 * doc/extend.texi: Fix some spelling mistakes and typos.
19190 * doc/invoke.texi: Likewise.
19192 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19195 * params.def (inline-min-speedup) Change from 10 to 8.
19197 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19199 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19202 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19205 * ipa-inline-analysis.c (get_minimal_bb): New function.
19206 (record_modified): Use it.
19207 (remap_edge_change_prob): Handle also ancestor functions.
19209 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19211 * doc/contrib.texi (Contributors): Remove broken link into
19212 the Mauve CVS repository.
19214 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19216 PR middle-end/79454
19217 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19218 result computation whenever lhs doesn't have vector mode, not
19219 just when it has BLKmode.
19221 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19223 * doc/makefile.texi (profiledbootstrap): Refer to the
19224 installation instructions only in textual form.
19226 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19229 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19231 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19233 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19234 (Specific): Update mingw-w64 reference.
19236 (Specific): Remove broken link to Renesas RX processor.
19238 2017-02-10 Richard Biener <rguenther@suse.de>
19240 * toplev.c (process_options): Do not mention obsolete graphite
19241 options when printing sorry message about missing graphite support.
19242 Mention -floop-nest-optimize.
19244 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19246 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19247 (vtst_p16): Likewise.
19248 (vtstq_p8): Likewise.
19249 (vtstq_p16): Likewise.
19251 (vtstq_p64): Likewise.
19252 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19253 (vset_lane_p64): New.
19254 (vsetq_lane_p64): New.
19256 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19258 PR tree-optimization/79411
19259 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19260 stmt operands are SSA_NAMEs used in abnormal phis.
19261 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19264 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19267 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19270 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19272 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19274 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19276 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19280 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19281 not arbitrary TREE_CONSTANT.
19284 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19285 "omp declare target link" attribute unless is_global_var.
19286 * omp-offload.c (find_link_var_op): Likewise.
19288 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19289 Chung-Lin Tang <cltang@codesourcery.com>
19291 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19293 (gimplify_adjust_omp_clauses): Don't delete TILE.
19294 (gimplify_omp_for): Deal with TILE.
19295 * internal-fn.c (expand_GOACC_TILE): New function.
19296 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19298 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19299 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19301 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19302 avoid DIV for outermost collapse var.
19303 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19304 Remove out of date comments, fix whitespace.
19305 * omp-general.c (omp_extract_for_data): Deal with tiling.
19306 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19307 adjust OLF_DIM_BASE value.
19308 (struct omp_for_data): Add tiling field.
19309 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19310 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19311 for auto loops. Remove default auto determining, moved to
19312 oacc_loop_fixed_partitions.
19313 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19314 stmts, add e_mask field.
19315 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19316 (oacc_thread_numbers): Use oacc_dim_call.
19317 (oacc_xform_tile): New.
19318 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19319 (finish_oacc_loop): Adjust for ifns vector.
19320 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19321 add case for GOACC_TILE fns.
19322 (oacc_loop_xform_loop): Delete.
19323 (oacc_loop_process): Iterate over call list directly, and add
19324 handling for GOACC_TILE fns.
19325 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19327 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19328 vector partitioning to outer loops. Assign 2 partitions to loops
19329 when available. Add TILE handling.
19330 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19331 (execite_oacc_device_lower): Process GOACC_TILE fns,
19332 ignore unknown specs.
19333 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19334 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19335 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19337 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19339 * configure.ac (ACX_BUGURL): Update.
19340 * configure: Regenerate.
19342 2017-02-09 Richard Biener <rguenther@suse.de>
19344 PR tree-optimization/69823
19345 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19346 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19348 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19350 * config/arc/arc-c.def: Add __NPS400__ definition.
19351 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19352 (TARGET_NPS400): Define.
19354 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19356 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19358 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19359 pointer, arch_info.
19360 (arc_cpu_types): Fill the arch_info field with a pointer into the
19361 arc_arch_types table.
19362 (arc_selected_cpu): Declare.
19363 * config/arc/arc.c (arc_selected_cpu): Make global.
19364 (arc_selected_arch): Delete.
19365 (arc_base_cpu): Delete.
19366 (arc_override_options): Remove references to deleted variables,
19367 update access to arch information.
19368 (ARC_OPT): Update access to arch information.
19369 (ARC_OPTX): Likewise.
19370 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19371 (TARGET_ARC600): Update access to arch information.
19372 (TARGET_ARC601): Likewise.
19373 (TARGET_ARC700): Likewise.
19374 (TARGET_EM): Likewise.
19375 (TARGET_HS): Likewise.
19376 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19379 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19382 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19383 condition/operands for integer GE/LE/GEU/LEU operations.
19385 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19387 PR translation/79397
19388 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19391 2017-02-08 Martin Jambor <mjambor@suse.cz>
19394 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19395 whether allocation happened.
19396 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19397 nothing was allocated.
19399 2017-02-08 Jakub Jelinek <jakub@redhat.com>
19401 PR tree-optimization/79408
19402 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
19403 constant, but SSA_NAME with a known integer range, use the minimum
19404 of that range instead of op1 to determine if modulo can be replaced
19405 with its first operand.
19407 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19409 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
19411 2017-02-08 Richard Biener <rguenther@suse.de>
19413 PR tree-optimization/71824
19414 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19415 Check all loops contained in the merged region.
19417 2017-02-07 Andrew Pinski <apinski@cavium.com>
19419 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
19421 2017-02-07 Andrew Pinski <apinski@cavium.com>
19423 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
19424 (thunderxt88): Likewise.
19425 (thunderxt81): Disable LSE and change v8.1 to v8.
19426 (thunderxt83): Likewise.
19428 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19429 Richard Biener <rguenther@suse.de>
19431 PR middle-end/79399
19432 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
19433 type from int to size_t.
19434 * ira-costs.c (struct_costs_size): Change type from int to size_t.
19436 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19438 PR rtl-optimization/79386
19439 * cprop.c (bypass_conditional_jumps): Initialize
19440 bypass_last_basic_block already before splitting bbs after
19441 unconditional traps...
19442 (bypass_conditional_jumps): ... rather than here.
19445 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
19446 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
19447 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
19448 fix -masm=intel patterns.
19450 2017-02-07 Richard Biener <rguenther@suse.de>
19452 PR tree-optimization/79256
19453 PR middle-end/79278
19454 * builtins.c (get_object_alignment_2): Use min_align_of_type
19455 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
19456 and ADJUST_FIELD_ALIGN.
19458 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
19460 * doc/tm.texi: Regenerate.
19461 * stor-layout.c (layout_decl): Adjust.
19462 (update_alignment_for_field): Likewise.
19463 (place_field): Likewise.
19464 (min_align_of_type): Likewise.
19465 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
19466 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
19467 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
19468 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
19469 * config/frv/frv.c (frv_adjust_field_align): Likewise.
19470 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
19471 * config/i386/i386.c (x86_field_alignment): Likewise.
19472 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
19473 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
19474 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
19475 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
19476 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
19477 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
19481 2017-01-30 Richard Biener <rguenther@suse.de>
19483 PR tree-optimization/79256
19484 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
19485 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
19488 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
19490 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
19491 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
19492 builtins to SImode and emit a zero-extend, if necessary.
19494 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19496 * docs/invoke.texi (RISC-V Options): Alphabetize.
19498 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19500 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
19503 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19505 * config/riscv/riscv.c: New file.
19506 * common/config/riscv/riscv-common.c: Likewise.
19507 * config.gcc: Likewise.
19508 * config/riscv/constraints.md: Likewise.
19509 * config/riscv/elf.h: Likewise.
19510 * config/riscv/generic.md: Likewise.
19511 * config/riscv/linux.h: Likewise.
19512 * config/riscv/multilib-generator: Likewise.
19513 * config/riscv/peephole.md: Likewise.
19514 * config/riscv/pic.md: Likewise.
19515 * config/riscv/predicates.md: Likewise.
19516 * config/riscv/riscv-builtins.c: Likewise.
19517 * config/riscv/riscv-c.c: Likewise.
19518 * config/riscv/riscv-ftypes.def: Likewise.
19519 * config/riscv/riscv-modes.def: Likewise.
19520 * config/riscv/riscv-opts.h: Likewise.
19521 * config/riscv/riscv-protos.h: Likewise.
19522 * config/riscv/riscv.h: Likewise.
19523 * config/riscv/riscv.md: Likewise.
19524 * config/riscv/riscv.opt: Likewise.
19525 * config/riscv/sync.md: Likewise.
19526 * config/riscv/t-elf-multilib: Likewise.
19527 * config/riscv/t-linux: Likewise.
19528 * config/riscv/t-linux-multilib: Likewise.
19529 * config/riscv/t-riscv: Likewise.
19530 * configure.ac: Likewise.
19531 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
19532 Waterman as RISC-V maintainers.
19533 * doc/install.texi: Add RISC-V entries.
19534 * doc/invoke.texi: Add RISC-V options section.
19535 * doc/md.texi: Add RISC-V constraints section.
19536 * configure: Regenerated.
19538 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19541 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
19542 false values to be constant vectors with all 0 or all 1 bits set.
19543 (vcondu<mode><mode>): Likewise.
19544 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
19546 (fpmask_comparison_operator): Update comment.
19547 (vecint_comparison_operator): New predicate.
19548 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
19549 vector conditionals when the true and false values are constant
19550 vectors with all 0 bits or all 1 bits set.
19552 2017-02-06 Martin Sebor <msebor@redhat.com>
19554 PR tree-optimization/79376
19555 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
19557 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
19559 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
19560 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
19561 to simplify split condition.
19563 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19565 * omp-expand.c (oxpand_omp_atomic_fetch_op,
19566 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
19569 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
19571 PR rtl-optimization/68664
19572 * target.def (can_speculate_insn): New hook.
19573 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
19574 * doc/tm.texi: Regenerate.
19575 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
19576 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
19577 (rs6000_sched_can_speculate_insn): New function.
19579 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19581 PR tree-optimization/79284
19582 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
19583 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
19584 vectorizable_mask_load_store, vectorizable_operation,
19585 vect_is_simple_cond, get_same_sized_vectype): Use it instead
19586 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
19587 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
19588 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
19589 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
19590 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19591 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
19592 is_gimple_assign (stmt). Replace another such test with
19593 is_gimple_assign (stmt).
19595 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19598 * config/avr/avr.c (rtl-iter.h): Include it.
19599 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
19600 (avr_legitimate_combined_insn): ...and implementation.
19602 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
19604 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
19605 * config/s390/s390.c (s390_const_operand_ok)
19606 (s390_canonicalize_comparison, s390_extract_part)
19607 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
19608 (s390_contiguous_bitmask_p, s390_rtx_costs)
19609 (legitimize_pic_address): Likewise.
19610 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
19611 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
19612 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
19613 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
19614 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
19616 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19618 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
19619 REGNO($0) == REGNO($1).
19621 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19623 * config/s390/linux.h(SIZE_TYPE): Add comment.
19625 2017-02-06 Julian Brown <julian@codesourcery.com>
19626 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19627 Virendra Pathak <virendra.pathak@broadcom.com>
19629 * config/aarch64/aarch64-cores.def: Change the scheduler
19631 * config/aarch64/aarch64.md: Include thunderx2t99.md.
19632 * config/aarch64/thunderx2t99.md: New file.
19634 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19636 * doc/standards.texi (Go Language): Update link to language
19639 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19641 * tree-eh.c (lower_resx): Sanitize profile.
19642 (cleanup_empty_eh_move_lp): Likewise.
19644 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19647 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
19649 * cfgloopmanip.h (loop_version): Update prototype.
19650 * modulo-sched.c (sms_schedule): Update call of loop_version.
19651 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
19652 * tree-parloops.c (gen_parallel_loop): Likewise.
19653 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
19654 * tree-ssa-loop-split.c (split_loop): Likewise.
19655 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
19656 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19658 2017-02-05 Martin Liska <mliska@suse.cz>
19661 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
19663 (print_operand_address): Initialize a struct to zero.
19665 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19667 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
19668 garbage collector only in textual form.
19670 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19672 * doc/extend.texi (x86 specific memory model extensions for
19673 transactional memory): Simplify a phrase.
19675 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
19678 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
19679 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
19680 (atomic_storedi_1): Likewise.
19682 2017-02-04 Jakub Jelinek <jakub@redhat.com>
19684 PR tree-optimization/79338
19685 * tree-parloops.c (gather_scalar_reductions): Don't call
19686 vect_analyze_loop_form for loop->inner before destroying loop's
19689 2017-02-03 Martin Sebor <msebor@redhat.com>
19691 PR tree-optimization/79327
19692 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
19693 when precision has resulted in leading zeros.
19694 (format_integer): Adjust the likely counter to assume an unknown
19695 argument that may be zero is non-zero.
19697 2017-02-03 Jason Merrill <jason@redhat.com>
19700 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
19701 avoid copying non-taken branch.
19703 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19705 PR tree-optimization/79340
19706 * tree-vect-loop.c (vectorizable_reduction): Release
19707 vec_defs elements after safe_splicing them into other vectors.
19710 PR tree-optimization/79327
19711 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
19712 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
19714 (format_integer): Use wide_int_to_tree instead of build_int_cst
19715 + to_?hwi. If argmin is NULL, just set argmin and argmax to
19716 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
19717 of shortest and longest sequence.
19719 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
19721 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
19722 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
19724 2017-02-03 Walter Lee <walt@tilera.com>
19727 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
19728 after initial stackframe link reg save.
19729 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
19731 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19734 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
19735 wu for stxssp alternative.
19737 2017-02-03 Martin Sebor <msebor@redhat.com>
19739 PR tree-optimization/79352
19740 * gimple-fold.c (get_range_strlen): Add argument.
19741 (get_range_strlen): Change return type to bool.
19742 (get_maxval_strlen): Pass in a dummy argument.
19743 * gimple-fold.h (get_range_strlen): Change return type to bool.
19744 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
19745 * tree.h (array_at_struct_end_p): Add argument.
19746 * tree.c (array_at_struct_end_p): Handle it.
19748 2017-02-03 Martin Liska <mliska@suse.cz>
19751 * multiple_target.c (create_dispatcher_calls): Redirect edge
19752 from a caller of a dispatcher.
19753 (expand_target_clones): Make the clones local.
19754 (ipa_target_clone): Do both target clones and resolvers.
19755 (ipa_dispatcher_calls): Remove the pass.
19756 (pass_dispatcher_calls::gate): Likewise.
19757 (make_pass_dispatcher_calls): Likewise.
19758 * passes.def (pass_target_clone): Put as very first IPA early
19761 2017-02-03 Martin Liska <mliska@suse.cz>
19763 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
19764 in case of a function with ifunc attribute.
19766 2017-02-03 Martin Liska <mliska@suse.cz>
19768 * cgraph.c (cgraph_node::dump): Dump function version info.
19769 * symtab.c (symtab_node::dump_base): Add missing new line.
19771 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
19773 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
19774 (ifcombine_ifandif): Use it.
19776 2017-02-03 Martin Liska <mliska@suse.cz>
19778 * doc/invoke.texi: Document default value for
19779 use-after-scope-direct-emission-threshold.
19781 2017-02-03 Martin Liska <mliska@suse.cz>
19783 PR tree-optimization/79339
19784 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
19785 (format_floating): Likewise.
19787 2017-02-03 Martin Liska <mliska@suse.cz>
19790 * ipa-prop.c (ipa_node_params_t::insert): Remove current
19792 (ipa_node_params_t::remove): Likewise.
19793 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
19794 initialization from removed ipa_node_params_t::insert.
19795 (ipa_node_params::~ipa_node_params): Move from removed
19796 ipa_node_params_t::release.
19797 * symbol-summary.h (symbol_summary::m_released): New member.
19798 Do not release a summary twice. Do not allow to call finalizer
19799 for types of a summary that live in GGC memory.
19801 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19803 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
19806 2017-02-02 Martin Sebor <msebor@redhat.com>
19808 PR middle-end/79275
19809 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
19810 (format_string): Tighten up the range of output for non-constant
19811 strings and correct the expected range for wide non-constant strings.
19813 2017-02-02 Martin Sebor <msebor@redhat.com>
19815 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
19817 PR middle-end/32003
19818 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
19820 (-fdump-tree-@var): Add to index and document how to come up
19821 with pass-specific option and dump file names.
19822 (-fdump-passes): Clarify where to look for output.
19824 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
19826 PR middle-end/77445
19827 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
19828 statistics of the analyzed path; allow threading for speed when
19829 any of BBs along the path are optimized for speed.
19831 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
19833 PR middle-end/78468
19834 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
19835 settings of the virtual registers.
19838 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
19840 * explow.c (get_dynamic_stack_size): Take known alignment of stack
19841 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
19844 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19846 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
19847 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
19849 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19851 * config/s390/s390.md: Add missing comments with the expanded
19853 * config/s390/vector.md: Likewise.
19854 * config/s390/vx-builtins.md: Likewise.
19856 2017-02-02 Jakub Jelinek <jakub@redhat.com>
19859 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
19860 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
19861 conditions on a single line.
19863 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19865 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
19866 __S390_VX__ to __VX__.
19868 2017-02-01 Andrew Pinski <apinski@cavium.com>
19870 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
19871 stmt_info to record_stmt_cost.
19872 (vect_get_known_peeling_cost): Pass stmt_info if known to
19874 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
19875 cpu_vector_cost field into
19876 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
19877 field into vec_int_stmt_cost and vec_fp_stmt_cost.
19878 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
19879 splitting of scalar_stmt_cost and vec_stmt_cost.
19880 (thunderx_vector_cost): Likewise.
19881 (cortexa57_vector_cost): LIkewise.
19882 (exynosm1_vector_cost): Likewise.
19883 (xgene1_vector_cost): Likewise.
19884 (thunderx2t99_vector_cost): Improve after the splitting of the two
19886 (aarch64_builtin_vectorization_cost): Update for the splitting of
19887 scalar_stmt_cost and vec_stmt_cost.
19889 2017-02-01 Torvald Riegel <triegel@redhat.com>
19890 Richard Henderson <rth@redhat.com>
19892 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
19893 conditional on existance of a fast atomic load.
19894 * optabs-query.c (can_atomic_load_p): New function.
19895 * optabs-query.h (can_atomic_load_p): Declare it.
19896 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
19897 no fast atomic load is available for the particular size of access.
19898 (expand_atomic_compare_and_swap): Likewise.
19899 (expand_atomic_load): Likewise.
19900 (expand_atomic_store): Likewise.
19901 (expand_atomic_fetch_op): Likewise.
19902 * testsuite/lib/target-supports.exp
19903 (check_effective_target_sync_int_128): Remove x86 because it provides
19904 no fast atomic load.
19905 (check_effective_target_sync_int_128_runtime): Likewise.
19907 2017-02-01 Richard Biener <rguenther@suse.de>
19909 * graphite.c: Include tree-vectorizer.h for find_loop_location.
19910 (graphite_transform_loops): Provide opt-info for optimized nests.
19911 * tree-parloop.c (parallelize_loops): Provide opt-info for
19912 parallelized loops.
19914 2017-02-01 Richard Biener <rguenther@suse.de>
19916 PR middle-end/79315
19917 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
19918 was not set before.
19920 2017-02-01 Richard Biener <rguenther@suse.de>
19922 PR tree-optimization/71824
19923 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19924 Verify the loops are valid in the merged SESE region.
19925 (scop_detection::can_represent_loop_1): Check analyzing the
19926 evolution of the number of iterations in the region succeeds.
19928 2017-01-31 Ian Lance Taylor <iant@golang.org>
19930 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
19931 REG_ARGS_SIZE note to 32-bit push insns and call insn.
19933 2017-01-31 David Malcolm <dmalcolm@redhat.com>
19935 PR preprocessor/79210
19936 * input.c (get_substring_ranges_for_loc): Replace line_width
19937 assertion with error-handling.
19939 2017-01-31 Richard Biener <rguenther@suse.de>
19941 PR tree-optimization/77318
19942 * graphite-sese-to-poly.c (extract_affine): Fix assert.
19943 (create_pw_aff_from_tree): Take loop parameter.
19944 (add_condition_to_pbb): Pass loop of the condition to
19945 create_pw_aff_from_tree.
19947 2017-01-31 Jakub Jelinek <jakub@redhat.com>
19949 * config/s390/s390.c (s390_asan_shadow_offset): New function.
19950 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
19952 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
19956 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
19958 (convert_int_to_float128): Likewise.
19959 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
19960 (convert_int_to_float128): Likewise.
19961 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
19962 (UNSPEC_IEEE128_CONVERT): Likewise.
19963 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
19964 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
19965 Use local variables for IBM extended format.
19966 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
19967 (fix_trunc<mode>si2_fprs): Likewise.
19968 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
19969 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
19970 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
19971 to know that we can now have integers of all sizes in vector
19973 (fix<uns>_<mode>di2_hw): Likewise.
19974 (float<uns>_<mode>si2_hw): Likewise.
19975 (fix_<mode>si2_hw): Likewise.
19976 (fixuns_<mode>si2_hw): Likewise.
19977 (float<uns>_<mode>di2_hw): Likewise.
19978 (float_<mode>di2_hw): Likewise.
19979 (float_<mode>si2_hw): Likewise.
19980 (floatuns_<mode>di2_hw): Likewise.
19981 (floatuns_<mode>si2_hw): Likewise.
19982 (xscvqp<su>wz_<mode>): Delete, no longer used.
19983 (xscvqp<su>dz_<mode>): Likewise.
19984 (xscv<su>dqp_<mode>): Likewise.
19985 (ieee128_mfvsrd_64bit): Likewise.
19986 (ieee128_mfvsrd_32bit): Likewise.
19987 (ieee128_mfvsrwz): Likewise.
19988 (ieee128_mtvsrw): Likewise.
19989 (ieee128_mtvsrd_64bit): Likewise.
19990 (ieee128_mtvsrd_32bit): Likewise.
19992 2017-01-31 Martin Liska <mliska@suse.cz>
19995 * ipa-prop.c (ipa_free_all_node_params): Call release method
19996 instead of ~sumbol_summary to not to trigger double times
19999 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20001 PR tree-optimization/71691
20002 * bitmap.h (class auto_bitmap): New.
20003 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20004 is_maybe_undefined instead of ssa_undefined_value_p.
20006 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20008 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20009 __S390_ARCH_LEVEL__ to __ARCH__.
20011 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20013 PR tree-optimization/79267
20014 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20015 if should_remove_lhs_p is true.
20017 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20020 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20021 (add_alignment_attribute): New.
20022 (base_type_die): Add alignment attribute.
20023 (subrange_type_die): Likewise.
20024 (modified_type_die): Likewise.
20025 (gen_array_type_die): Likewise.
20026 (gen_descr_array_type_die: Likewise.
20027 (gen_enumeration_type_die): Likewise.
20028 (gen_subprogram_die): Likewise.
20029 (gen_variable_die): Likewise.
20030 (gen_field_die): Likewise.
20031 (gen_ptr_to_mbr_type_die): Likewise.
20032 (gen_struct_or_union_type_die): Likewise.
20033 (gen_subroutine_type_die): Likewise.
20034 (gen_typedef_die): Likewise.
20035 (base_type_cmp): Compare alignment attribute.
20037 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20040 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20041 (setb_unsigned) New pattern for setb with CCUNS.
20042 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20043 subfc./subfe sequence to avoid overflow problems. Generate a
20044 shorter sequence with cmpld/setb for power9.
20045 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20046 for generating subfc. instruction.
20047 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20048 now uses this instruction.
20050 2017-01-30 Ian Lance Taylor <iant@google.com>
20053 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20054 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20056 2017-01-30 Martin Sebor <msebor@redhat.com>
20058 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20059 Move constant to the right of a relational operator.
20060 (get_mpfr_format_length, format_character, format_string): Ditto.
20061 (should_warn_p, maybe_warn): Same.
20063 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20065 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20068 * asan.c (get_translation_unit_decl): Remove function.
20069 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20071 2017-01-30 Martin Liska <mliska@suse.cz>
20073 PR gcov-profile/79259
20074 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20075 -fprofile-generate.
20077 2017-01-30 Martin Liska <mliska@suse.cz>
20080 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20081 Initialize variables with NULL value.
20083 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20086 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20088 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20090 2017-01-30 Richard Biener <rguenther@suse.de>
20092 PR tree-optimization/79276
20093 * tree-vrp.c (process_assert_insertions): Properly adjust common
20094 when removing a duplicate.
20096 * gcc.dg/torture/pr79276.c: New testcase.
20098 2017-01-30 Richard Biener <rguenther@suse.de>
20100 PR tree-optimization/79256
20101 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20102 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20104 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20106 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20109 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20110 ("*r<noxa>sbg_<mode>_sll_bitmask")
20111 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20112 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20113 Use contiguous_bitmask_nowrap_operand.
20115 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20118 * config/rs6000/altivec.h (vec_xl): Revise #define.
20119 (vec_xst): Likewise.
20121 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20123 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20125 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20127 PR rtl-optimization/79194
20128 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20129 traps before call to bypass_conditional_jumps.
20131 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20133 PR tree-optimization/71374
20134 * lra-constraints.c (check_conflict_input_operands): New.
20135 (match_reload): Use it.
20137 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20140 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20141 account to calculate conflict_set.
20143 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20145 PR rtl-optimization/78559
20146 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20147 other_insn in combine.
20149 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20151 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20152 uint16_type_node for BT_UINT16.
20154 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20156 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20157 "RTL Tests" to menu.
20158 (GIMPLE Tests): New node.
20159 (RTL Tests): New node.
20161 2017-01-27 Richard Biener <rguenther@suse.de>
20163 PR tree-optimization/79245
20164 * tree-loop-distribution.c (distribute_loop): Apply cost
20165 modeling also to detected patterns.
20167 2017-01-27 Richard Biener <rguenther@suse.de>
20169 PR tree-optimization/71433
20170 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20171 (compare_assert_loc): New function.
20172 (process_assert_insertions): Sort and optimize assert locations
20173 to remove duplicates and push down identical assertions on
20174 edges to their destination block.
20176 2017-01-27 Richard Biener <rguenther@suse.de>
20178 PR tree-optimization/79244
20179 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20180 out SSA names even if abnormal.
20182 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20184 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20185 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20186 instead of MPFR_RNDN.
20188 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20191 * arm.c (arm_option_override): Don't call build_target_option_node
20192 until after doing all option overrides.
20193 (arm_valid_target_attribute_tree): Likewise.
20195 2017-01-27 Martin Liska <mliska@suse.cz>
20197 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20198 for {cd}tors and C++ {cd}tors.
20200 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20202 * config/s390/s390.md ("*setmem_long_and")
20203 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20205 2017-01-26 Martin Sebor <msebor@redhat.com>
20207 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20210 2017-01-26 Martin Sebor <msebor@redhat.com>
20212 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20213 HAVE_DFmode before using XFmode or DFmode.
20214 (parse_directive): Avoid using the z length modifier to avoid
20215 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20217 PR middle-end/78703
20218 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20219 to accept adjustment as an array.
20220 (get_int_range): New function.
20221 (struct directive): Make width and prec arrays.
20222 (directive::set_width, directive::set_precision): Call get_int_range.
20223 (format_integer, format_floating): Handle width and precision ranges.
20224 (format_string, parse_directive): Same.
20226 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20229 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20230 comdat_type_p set, just clone them, but keep the children in the
20234 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20235 which have direct callers with -fvar-tracking-assignments enabled
20237 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20238 inside of type units.
20240 2017-01-26 Martin Sebor <msebor@redhat.com>
20242 PR middle-end/78703
20243 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20245 (struct format_result): Replace number_chars, number_chars_min,
20246 and number_chars_max with a single member of struct result_range.
20248 (format_result::operator+=): Adjust.
20249 (struct fmtresult): Remove bounded. Handle likely and unlikely
20251 (fmtresult::adjust_for_width_or_precision): New function.
20252 (fmtresult:type_max_digits): New function.
20253 (bytes_remaining): Handle likely and unlikely counters.
20254 (min_bytes_remaining): Remove.
20255 (format_percent): Simplify.
20256 (format_integer, format_floating): Set likely and unlikely counters.
20257 (get_string_length, format_character, format_string): Same.
20258 (format_plain, should_warn_p): New function.
20259 (maybe_warn): Call should_warn_p. Update diagnostic messages
20260 and handle those for all directives, including plain strings.
20261 (format_directive): Handle likely and unlikely counters.
20262 Remove unnecessary quoting from diagnostics. Add an informational
20264 (add_bytes): Remove.
20265 (pass_sprintf_length::compute_format_length): Simplify.
20266 (try_substitute_return_value): Handle likely and unlikely counters.
20268 2017-01-26 Carl Love <cel@us.ibm.com>
20270 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20271 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20273 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20276 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20277 endianess for subregs into account.
20278 * lra-constraints.c (lra_constraints): Do risky transformations
20279 always on the first iteration.
20280 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20281 last_call_used_reg_set.
20282 (process_bb_lives): Define and use last_call_used_reg_set.
20283 * lra.c (lra): Always continue after lra_constraints on the first
20286 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20288 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20290 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20292 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20294 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20295 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20296 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20297 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20298 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20299 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20300 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20301 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20302 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20304 2017-01-26 Marek Polacek <polacek@redhat.com>
20307 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20308 for the third operand.
20310 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20312 PR middle-end/79236
20313 * omp-low.c (struct omp_context): Add simt_stmt field.
20314 (scan_omp_for): Return omp_context *.
20315 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20316 context to the _simt_ SIMD stmt.
20317 (lower_omp_for): For combined SIMD with sibling _simt_
20318 SIMD, make sure to use the same decls in _looptemp_
20319 clauses as in the sibling.
20321 2017-01-26 David Sherwood <david.sherwood@arm.com>
20323 PR middle-end/79212
20324 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20327 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20330 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20331 emit fld b; fld a; if possible.
20333 * brig-builtins.def: Update copyright years.
20334 * config/arm/arm_acle_builtins.def: Update copyright years.
20336 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20339 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20340 constraint instead of o for the stxsd instruction.
20342 2017-01-25 Carl Love <cel@us.ibm.com>
20344 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20345 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20347 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20349 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20351 2017-01-25 Richard Biener <rguenther@suse.de>
20353 PR tree-optimization/69264
20354 * target.def (vector_alignment_reachable): Improve documentation.
20355 * doc/tm.texi: Regenerate.
20356 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20358 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20359 earlier changes with respect to TYPE_USER_ALIGN.
20360 (vector_alignment_reachable_p): Likewise. Improve dumping.
20362 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20365 * config/arm/arm.md (xordi3): Force constant operand into a register
20368 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20370 * doc/invoke.texi (-fstore-merging): Correct default optimization
20371 levels at which it is enabled.
20372 (-O): Move -fstore-merging from list to...
20375 2017-01-25 Richard Biener <rguenther@suse.de>
20378 * omp-expand.c: Include debug.h.
20379 (expand_omp_taskreg): Make sure to generate early debug before
20380 outlining anything from a function.
20381 (expand_omp_target): Likewise.
20382 (grid_expand_target_grid_body): Likewise.
20384 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20387 * asan.c (get_translation_unit_decl): New function.
20388 (asan_add_global): Extract modules file name from globals
20389 TRANSLATION_UNIT_DECL name.
20391 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20394 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20395 for long calls with APCS frame and VFP.
20397 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20399 * cfg.c (original_copy_tables_initialized_p): New function.
20400 * cfg.h (original_copy_tables_initialized_p): New decl.
20401 * cfgrtl.c (relink_block_chain): Guard the call to
20402 free_original_copy_tables with a call to
20403 original_copy_tables_initialized_p.
20404 * cgraph.h (symtab_node::native_rtl_p): New decl.
20405 * cgraphunit.c (symtab_node::native_rtl_p): New function.
20406 (symtab_node::needed_p): Don't assert for early assembly output
20407 for __RTL functions.
20408 (cgraph_node::finalize_function): Set "force_output" for __RTL
20410 (cgraph_node::analyze): Bail out early for __RTL functions.
20411 (analyze_functions): Update assertion to support __RTL functions.
20412 (cgraph_node::expand): Bail out early for __RTL functions.
20413 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
20415 * function.h (struct function): Update comment for field
20417 * gimple-expr.c: Include "tree-pass.h".
20418 (gimple_has_body_p): Return false for __RTL functions.
20419 * Makefile.in (OBJS): Add run-rtl-passes.o.
20420 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
20422 (gcc::pass_manager::get_clean_slate): New accessor.
20423 * passes.c: Include "insn-addr.h".
20424 (should_skip_pass_p): Add logging. Update logic for running
20425 "expand" to be compatible with both __GIMPLE and __RTL. Guard
20426 property-provider override so it is only done for gimple passes.
20428 (skip_pass): New function.
20429 (execute_one_pass): Call skip_pass when skipping passes.
20430 * read-md.c (md_reader::read_char): Support filtering
20431 the input to a subset of line numbers.
20432 (md_reader::md_reader): Initialize fields
20433 m_first_line and m_last_line.
20434 (md_reader::read_file_fragment): New function.
20435 * read-md.h (md_reader::read_file_fragment): New decl.
20436 (md_reader::m_first_line): New field.
20437 (md_reader::m_last_line): New field.
20438 * read-rtl-function.c (function_reader::create_function): Only
20439 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
20440 curr_properties. Set DECL_INITIAL to a dummy block.
20441 (read_rtl_function_body_from_file_range): New function.
20442 * read-rtl-function.h (read_rtl_function_body_from_file_range):
20444 * run-rtl-passes.c: New file.
20445 * run-rtl-passes.h: New file.
20447 2017-01-24 Jeff Law <law@redhat.com>
20449 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
20452 2017-01-24 Bin Cheng <bin.cheng@arm.com>
20454 PR tree-optimization/79159
20455 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
20456 (record_nonwrapping_iv): Improve boundary using above function if no
20457 value range information.
20459 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
20460 Martin Jambor <mjambor@suse.cz>
20462 * brig-builtins.def: New file.
20463 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
20464 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
20465 (DEF_HSAIL_SAT_BUILTIN): Likewise.
20466 (DEF_HSAIL_INTR_BUILTIN): Likewise.
20467 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
20468 * builtin-types.def (BT_INT8): New.
20469 (BT_INT16): Likewise.
20470 (BT_UINT8): Likewise.
20471 (BT_UINT16): Likewise.
20472 (BT_FN_ULONG): Likewise.
20473 (BT_FN_UINT_INT): Likewise.
20474 (BT_FN_UINT_ULONG): Likewise.
20475 (BT_FN_UINT_LONG): Likewise.
20476 (BT_FN_UINT_PTR): Likewise.
20477 (BT_FN_ULONG_PTR): Likewise.
20478 (BT_FN_INT8_FLOAT): Likewise.
20479 (BT_FN_INT16_FLOAT): Likewise.
20480 (BT_FN_UINT32_FLOAT): Likewise.
20481 (BT_FN_UINT16_FLOAT): Likewise.
20482 (BT_FN_UINT8_FLOAT): Likewise.
20483 (BT_FN_UINT64_FLOAT): Likewise.
20484 (BT_FN_UINT16_UINT32): Likewise.
20485 (BT_FN_UINT32_UINT16): Likewise.
20486 (BT_FN_UINT16_UINT16_UINT16): Likewise.
20487 (BT_FN_INT_PTR_INT): Likewise.
20488 (BT_FN_UINT_PTR_UINT): Likewise.
20489 (BT_FN_LONG_PTR_LONG): Likewise.
20490 (BT_FN_ULONG_PTR_ULONG): Likewise.
20491 (BT_FN_VOID_UINT64_UINT64): Likewise.
20492 (BT_FN_UINT8_UINT8_UINT8): Likewise.
20493 (BT_FN_INT8_INT8_INT8): Likewise.
20494 (BT_FN_INT16_INT16_INT16): Likewise.
20495 (BT_FN_INT_INT_INT): Likewise.
20496 (BT_FN_UINT_FLOAT_UINT): Likewise.
20497 (BT_FN_FLOAT_UINT_UINT): Likewise.
20498 (BT_FN_ULONG_UINT_UINT): Likewise.
20499 (BT_FN_ULONG_UINT_PTR): Likewise.
20500 (BT_FN_ULONG_ULONG_ULONG): Likewise.
20501 (BT_FN_UINT_UINT_UINT): Likewise.
20502 (BT_FN_VOID_UINT_PTR): Likewise.
20503 (BT_FN_UINT_UINT_PTR: Likewise.
20504 (BT_FN_UINT32_UINT64_PTR): Likewise.
20505 (BT_FN_INT_INT_UINT_UINT): Likewise.
20506 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
20507 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
20508 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
20509 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
20510 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
20511 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
20512 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
20513 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
20514 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
20515 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
20516 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
20517 * doc/frontends.texi: List BRIG FE.
20518 * doc/install.texi (Testing): Add BRIG tesring requirements.
20519 * doc/invoke.texi (Overall Options): Mention BRIG.
20520 * doc/standards.texi (Standards): Doucment BRIG HSA version.
20522 2017-01-24 Richard Biener <rguenther@suse.de>
20524 PR translation/79208
20525 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
20527 2017-01-24 Martin Jambor <mjambor@suse.cz>
20530 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
20531 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
20532 and known_contexts.
20534 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
20536 PR middle-end/79123
20537 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
20538 casts from signed to unsigned really don't have a range.
20540 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
20542 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
20543 GMP_RNDx for compatiblity.
20545 2017-01-24 Martin Liska <mliska@suse.cz>
20548 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
20549 that would prevent us to call alloca with -1 as argument.
20551 2017-01-24 Jakub Jelinek <jakub@redhat.com>
20553 * dwarf2out.c (output_compilation_unit_header, output_file_names):
20554 Avoid -Wformat-security warning.
20556 2017-01-23 Andrew Pinski <apinski@cavium.com>
20558 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
20561 2017-01-23 Martin Sebor <msebor@redhat.com>
20563 PR middle-end/78703
20564 * gimple-ssa-sprintf.c (warn_level): New global.
20565 (format_integer): Use it here and throughout the rest of the file.
20566 Use the same switch to compute sign as base.
20567 (maybe_warn): New function.
20568 (format_directive): Factor out warnings into maybe_warn.
20569 Add debugging output. Use warn_level.
20570 (add_bytes): Use warn_level.
20571 (pass_sprintf_length::compute_format_length): Add debugging output.
20572 (try_substitute_return_value): Same.
20573 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
20575 PR middle-end/78703
20576 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
20577 (struct fmtresult, format_integer, format_floating): Adjust.
20578 (fmtresult::fmtresult): Set max correctly in two argument ctor.
20579 (get_string_length, format_string,format_directive): Same.
20580 (pass_sprintf_length::compute_format_length): Same.
20581 (try_substitute_return_value): Simplify slightly.
20583 PR middle-end/78703
20584 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
20585 (fmtresult::operator+=): Outlined.
20586 (struct fmtresult): Add ctors.
20587 (struct conversion_spec): Rename...
20588 (struct directive): ...to this. Add and remove data members.
20589 (directive::set_width, directive::set_precision): New functions.
20590 (format_percent): Use fmtresult ctor.
20591 (get_width_and_precision): Remove.
20592 (format_integer): Make naming changes. Avoid computing width and
20594 (format_floating): Same. Adjust indentation.
20595 (format_character, format_none): New functions.
20596 (format_string): Moved character handling to format_character.
20597 (format_directive): Remove arguments, change return type.
20598 (parse_directive): New function.
20599 (pass_sprintf_length::compute_format_length): Move directive
20600 parsing to parse_directive.
20602 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20604 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
20605 (assign_assembler_name_if_needed): ... this.
20606 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
20607 (assign_assembler_name_if_needed): ... this.
20608 (free_lang_data_in_cgraph): Adjust callers.
20609 * cgraphunit.c (cgraph_node::analyze): Likewise.
20610 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
20613 2017-01-23 Richard Biener <rguenther@suse.de>
20615 PR tree-optimization/79088
20616 PR tree-optimization/79188
20617 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
20618 resetting loop bounds after last path deletion. Reset loop
20619 bounds of the target loop, make code match the comments.
20620 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
20621 Make sure loops need no fixups.
20623 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20625 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
20626 exponent support with double type for first argument.
20627 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
20628 type returned by __builtin_vec_extract_sig,
20629 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
20630 functions from "vector int" to "vector unsigned int" or from
20631 "vector long long int" to "vector unsigned long long int".
20632 Changed type returned by __builtin_vec_extract_exp,
20633 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
20634 functions from "vector int" to "vector unsigned int" or from
20635 "vector long long int" to "vector unsigned long long int".
20636 Changed return type of __builtin_vec_test_data_class,
20637 __builtin_vec_test_data_class_sp, and
20638 __builtin_vec_test_data_class_dp from "vector int" to
20639 "vector bool int" or from "vector long long int" to "vector bool
20640 long long int" and changed second argument type from "unsigned
20641 int" to "int". Added new overloaded function forms "vector float
20642 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
20643 "vector float __builtin_vec_insert_exp_sp (vector float, vector
20644 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
20645 double, vector unsigned long long int)" and "vector double
20646 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
20647 long int)". Changed return type of
20648 __builtin_scalar_test_data_class and
20649 __builtin_scalar_test_data_class_sp and
20650 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
20651 int" and changed second argument from "unsigned int" to "int".
20652 Changed type returned by __builtin_scalar_test_neg,
20653 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
20654 from "int" to "bool int". Added new overloaded function form
20655 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
20656 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
20657 exponent double-precision with floating point first argument.
20658 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
20659 documentation of scalar_test_data_class, scalar_test_neg,
20660 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
20661 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
20662 vec_test_data_class built-in functions to reflect refinements in
20663 their type signatures.
20665 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
20667 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
20669 (aarch64_elf_asm_destructor): Likewise.
20671 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
20673 PR rtl-optimization/78634
20674 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
20675 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
20676 * ifcvt.c (noce_try_cmove): Add missing cost check.
20678 PR rtl-optimization/71724
20679 * combine.c (if_then_else_cond): Look for situations where it is
20680 beneficial to undo the work of one of the recursive calls.
20682 2017-01-23 Bin Cheng <bin.cheng@arm.com>
20684 PR tree-optimization/70754
20685 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
20686 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
20687 combined stmt before it if not NULL.
20688 (combine_chains): Process refs reversely and compute dominance point
20691 2017-01-23 Martin Liska <mliska@suse.cz>
20693 PR tree-optimization/79196
20694 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
20695 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
20697 (strlen_optimize_stmt): Call the renamed function.
20699 2017-01-23 Michael Matz <matz@suse.de>
20701 PR tree-optimization/78384
20702 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
20704 2017-01-23 Richard Biener <rguenther@suse.de>
20706 PR tree-optimization/79186
20707 * tree-vrp.c (register_new_assert_for): Make sure we've seen
20708 both incoming edges before moving an assert.
20710 2017-01-23 Martin Jambor <mjambor@suse.cz>
20712 * ipa-prop.c (load_from_param_1): Removed.
20713 (load_from_unmodified_param): Bits from load_from_param_1 put back
20715 (load_from_param): Removed.
20716 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
20717 with stmt. Reverted back to use of load_from_unmodified_param.
20719 2017-01-23 Martin Jambor <mjambor@suse.cz>
20722 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
20723 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
20724 field a pointer to garbage collected vector, mark lattices and
20725 ipcp_orig_node with GTY((skip)).
20726 (ipa_get_param_count): Adjust to descriptors being a pointer.
20727 (ipa_get_param): Likewise.
20728 (ipa_get_type): Likewise.
20729 (ipa_get_param_move_cost): Likewise.
20730 (ipa_set_param_used): Likewise.
20731 (ipa_get_controlled_uses): Likewise.
20732 (ipa_set_controlled_uses): Likewise.
20733 (ipa_is_param_used): Likewise.
20734 (ipa_node_params_t): Move into garbage collector. New methods insert
20736 (ipa_node_params_sum): Annotate wth GTY(()).
20737 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
20739 (ipa_load_from_parm_agg): Adjust declaration.
20740 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
20741 * ipa-profile.c (ipa_profile): Likewise.
20742 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
20743 (ipa_populate_param_decls): Make descriptors parameter garbage
20745 (ipa_dump_param): Adjust to descriptors being a pointer.
20746 (ipa_alloc_node_params): Likewise.
20747 (ipa_initialize_node_params): Likewise.
20748 (load_from_param_1): Make descriptors parameter garbage collected.
20749 (load_from_unmodified_param): Likewise.
20750 (load_from_param): Likewise.
20751 (ipa_load_from_parm_agg): Likewise.
20752 (ipa_node_params::~ipa_node_params): Removed.
20753 (ipa_free_all_node_params): Remove call to delete operator.
20754 (ipa_node_params_t::insert): New.
20755 (ipa_node_params_t::remove): Likewise.
20756 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
20757 copy known_csts and known_contexts vectors.
20758 (ipa_read_node_info): Adjust to descriptors being a pointer.
20759 (ipcp_modif_dom_walker): Make m_descriptors field garbage
20761 (ipcp_transform_function): Make descriptors variable garbage
20764 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
20766 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
20767 * config/i386/avx512dqintrin.h: Ditto.
20768 * config/i386/avx512fintrin.h: Ditto.
20769 * config/i386/i386.c: Handle new builtins.
20770 * config/i386/i386-builtin.def: Add new builtins.
20771 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
20772 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
20774 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20775 Martin Liska <mliska@suse.cz>
20777 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
20778 * asan.c (asan_expand_poison_ifn): Support stores and use
20779 appropriate ASAN report function.
20780 * internal-fn.c (expand_ASAN_POISON_USE): New function.
20781 * internal-fn.def (ASAN_POISON_USE): Declare.
20782 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
20783 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
20784 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
20785 ASAN_POISON calls w/o LHS.
20786 * tree-ssa.c (execute_update_addresses_taken): Create clobber
20787 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
20788 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
20789 * gimplify.c (asan_poison_variables): Add attribute
20790 use_after_scope_memory to variables that really needs to live
20792 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
20793 having the attribute.
20795 2017-01-23 Martin Liska <mliska@suse.cz>
20797 * asan.c (create_asan_shadow_var): New function.
20798 (asan_expand_poison_ifn): Likewise.
20799 * asan.h (asan_expand_poison_ifn): New declaration.
20800 * internal-fn.c (expand_ASAN_POISON): Likewise.
20801 * internal-fn.def (ASAN_POISON): New builtin.
20802 * sanopt.c (pass_sanopt::execute): Expand
20803 asan_expand_poison_ifn.
20804 * tree-inline.c (copy_decl_for_dup_finish): Make function
20806 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
20807 * tree-ssa.c (is_asan_mark_p): New function.
20808 (execute_update_addresses_taken): Rewrite local variables
20809 (identified just by use-after-scope as addressable) into SSA.
20811 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
20813 * doc/install.texi (Specific): opensource.apple.com uses https
20814 now. Remove trailing slash.
20816 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
20818 * README.Portability: Remove note on an Irix compatibility issue.
20820 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
20822 * gcov.c (INCLUDE_ALGORITHM): Define.
20823 (INCLUDE_VECTOR): Define.
20824 No longer include <vector> and <algorithm> directly.
20826 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
20828 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
20830 * doc/invoke.texi (Code Gen Options): Ditto.
20832 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
20835 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
20837 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
20839 rtl-optimization/79125
20840 * cprop.c (local_cprop_pass): Handle cases where we make an
20841 unconditional trap.
20843 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
20847 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
20848 read from, for big endian.
20850 2017-01-20 Jiong Wang <jiong.wang@arm.com>
20852 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
20853 register pauth builtins for LP64 only.
20855 2017-01-20 Marek Polacek <polacek@redhat.com>
20858 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
20861 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
20863 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
20865 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
20866 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
20867 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
20869 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20872 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
20873 in big-endian mode when they are not a single duplicated value.
20875 2017-01-20 Richard Biener <rguenther@suse.de>
20877 * BASE-VER: Bump to 7.0.1.
20879 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
20881 * omp-low.c (omplow_simd_context): New struct. Use it...
20882 (lower_rec_simd_input_clauses): ...here and...
20883 (lower_rec_input_clauses): ...here to hold common data. Adjust all
20884 references to idx, lane, max_vf, is_simt.
20886 2017-01-20 Graham Markall <graham.markall@embecosm.com>
20888 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
20891 2017-01-20 Martin Jambor <mjambor@suse.cz>
20893 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
20894 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
20896 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
20897 (GTFILES): Rename hsa.c to hsa-common.c.
20898 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
20899 * hsa-dump.c: Likewise.
20900 * hsa-gen.c: Likewise.
20901 * hsa-regalloc.c: Likewise.
20902 * ipa-hsa.c: Likewise.
20903 * omp-expand.c: Likewise.
20904 * omp-low.c: Likewise.
20905 * toplev.c: Likewise.
20907 2017-01-20 Marek Polacek <polacek@redhat.com>
20910 * doc/invoke.texi: Document -Wduplicated-branches.
20911 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
20912 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
20913 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
20914 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
20915 return 0 only when not OEP_LEXICOGRAPHIC.
20916 (fold_build_cleanup_point_expr): Use the expression
20917 location when building CLEANUP_POINT_EXPR.
20918 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
20919 * tree.c (add_expr): Handle error_mark_node.
20921 2017-01-20 Martin Liska <mliska@suse.cz>
20924 * tree-profile.c (init_ic_make_global_vars): Do not call
20926 (gimple_init_gcov_profiler): Likewise.
20928 2017-01-20 Martin Liska <mliska@suse.cz>
20931 * cgraph.h (maybe_create_reference): Remove argument and
20933 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
20935 * ipa-cp.c (create_specialized_node): Likewise.
20936 * symtab.c (symtab_node::maybe_create_reference): Handle
20937 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
20939 2017-01-20 Martin Liska <mliska@suse.cz>
20941 * read-rtl-function.c (function_reader::create_function): Use
20942 build_decl instread of build_decl_stat.
20944 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
20946 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
20947 * config/i386/avx512dqintrin.h: Ditto.
20948 * config/i386/avx512fintrin.h: Ditto.
20949 * config/i386/i386-builtin-types.def: Add new types.
20950 * config/i386/i386.c: Handle new types.
20951 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
20952 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
20953 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
20954 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
20955 (__builtin_ia32_kshiftridi): New.
20956 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
20958 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
20962 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
20963 define to rs6000_init_stack_protect_guard.
20964 (rs6000_init_stack_protect_guard): New function.
20966 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
20967 Yunqiang Su <yunqiang.su@imgtec.com>
20969 * config.gcc (supported_defaults): Add madd4.
20970 (with_madd4): Add validation.
20971 (all_defaults): Add madd4.
20972 * config/mips/mips.opt (mmadd4): New option.
20973 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
20975 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
20977 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
20978 (ISA_HAS_FUSED_MADD4): Likewise.
20979 * doc/invoke.texi (-mmadd4): Document the new option.
20980 * doc/install.texi (--with-madd4): Document the new option.
20982 2017-01-19 Jiong Wang <jiong.wang@arm.com>
20984 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
20985 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
20986 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
20987 (aarch64_init_pauth_hint_builtins): New.
20988 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
20989 (aarch64_expand_builtin): Expand new builtins.
20991 2017-01-19 Jiong Wang <jiong.wang@arm.com>
20993 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
20994 * combine-stack-adj.c (no_unhandled_cfa): Handle
20995 REG_CFA_TOGGLE_RA_MANGLE.
20996 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
20997 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
20998 info for return address signing.
20999 (aarch64_expand_epilogue): Likewise.
21001 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21003 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21004 * config/aarch64/aarch64-protos.h
21005 (aarch64_return_address_signing_enabled): New declaration.
21006 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21008 (aarch64_expand_prologue): Sign return address before it's pushed onto
21010 (aarch64_expand_epilogue): Authenticate return address fetched from
21012 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21013 (aarch64_attributes): New function attributes for "sign-return-address".
21014 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21015 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21016 ("*do_return"): Generate combined instructions according to key index.
21017 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21018 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21020 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21021 * config/aarch64/aarch64.opt (msign-return-address=): New.
21022 * doc/extend.texi (AArch64 Function Attributes): Documents
21023 "sign-return-address=".
21024 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21026 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21028 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21029 overall option summary.
21031 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21033 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21034 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21035 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21036 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21038 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21040 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21041 -mpower9-minmax by default for -mcpu=power9.
21042 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21043 128-bit floating point.
21045 2017-01-20 Alan Modra <amodra@gmail.com>
21047 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21048 optimizing for size.
21050 2017-01-20 Alan Modra <amodra@gmail.com>
21053 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21054 for strcmp and strncmp from corresponding builtin decl.
21056 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21058 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21059 instead of i386/rtems-64.h.
21060 * config/i386/rtems-64.h: Remove.
21062 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21066 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21068 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21070 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21072 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21073 Change int to HOST_WIDE_INT.
21074 * config/aarch64/aarch64-protos.h
21075 (aarch64_simd_gen_const_vector_dup): Likewise.
21076 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21078 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21080 * langhooks-def.h (lhd_type_for_size): New decl.
21081 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21082 * langhooks.c (lhd_type_for_size): New function, taken from
21085 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21087 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21088 define_bypass for CR latency.
21089 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21090 (power9-alu2): Add define_bypass for CR latency.
21092 (power9-mul): Update insn latency.
21093 (power9-mul-compare): Update insn latency, bypass latency and remove
21096 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21098 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21100 * config/aarch64/aarch64.md
21101 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21102 aarch64_nopcrelative_literal_loads.
21103 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21105 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21107 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21108 TARGET_LOONGSON_3A.
21109 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21111 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21114 * config.gcc (supported_defaults): Add lxc1-sxc1.
21115 (with_lxc1_sxc1): Add validation.
21116 (all_defaults): Add lxc1-sxc1.
21117 * config/mips/mips.opt (mlxc1-sxc1): New option.
21118 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21120 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21121 __mips_no_lxc1_sxc1.
21122 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21123 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21124 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21126 2017-01-19 Richard Biener <rguenther@suse.de>
21128 PR tree-optimization/72488
21129 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21130 sure to restore SSA info.
21131 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21133 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21135 PR rtl-optimization/79121
21136 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21137 of the inner type when shifting an extended value.
21139 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21142 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21143 interposable aliases.
21145 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21148 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21149 Use the evmergelohi instruction.
21150 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21151 (mov_sitf_e500_subreg8_2_be): Likewise.
21152 (mov_sitf_e500_subreg12_2_le): Likewise.
21153 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21154 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21155 (mov_sitf_e500_subreg8_2_le): Likewise.
21156 (mov_sitf_e500_subreg12_2_be): Likewise.
21158 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21160 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21161 attribute from vecsimple to vecperm.
21162 (altivec_vbpermq2): Likewise.
21164 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21167 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21169 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21170 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21171 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21172 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21173 case where N arg is SIZE_MAX.
21174 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21175 (cmpstrsi): Add pattern.
21177 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21179 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21180 __builtin_vec_revb builtins.
21181 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21182 built-in functions to support generation of the ISA 3.0 XXBR<x>
21183 vector byte reverse instructions.
21184 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21185 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21186 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21187 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21188 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21189 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21190 (P9V_BUILTIN_VEC_REVB): Likewise.
21191 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21192 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21193 (p9_xxbrq_v16qi): Likewise.
21194 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21195 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21196 (p9_xxbrh_v8hi): Likewise.
21197 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21198 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21199 vec_revb built-in functions.
21201 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21203 PR rtl-optimization/78952
21204 * config/i386/i386.md (any_extract): New code iterator.
21205 (*insvqi_2): Use any_extract for source operand.
21206 (*insvqi_3): Use any_shiftrt for source operand.
21208 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21210 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21212 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21214 2017-01-18 Matthias Klose <doko@ubuntu.com>
21216 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21218 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21220 * config/rs6000/altivec.h (vec_bperm): Change #define.
21221 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21222 (altivec_vbpermq2): New define_insn.
21223 (altivec_vbpermd): Likewise.
21224 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21225 function interface.
21226 (VBPERMD): Likewise.
21227 (VBPERM): New polymorphic function interface.
21228 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21229 Add entries for P9V_BUILTIN_VEC_VBPERM.
21230 * doc/extend.texi: Add interfaces for vec_bperm.
21232 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21234 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21235 first letter of error messages.
21236 (s390_resolve_overloaded_builtin): Likewise.
21237 * config/s390/s390.c (s390_expand_builtin): Likewise.
21238 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21239 (s390_valid_target_attribute_inner_p): Likewise.
21240 * config/s390/s390.md ("tabort"): Likewise.
21242 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21244 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21245 (ISA_AVOID_DIV_HILO): New macro.
21246 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21247 (ISA_HAS_DDIV): Likewise.
21249 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21251 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21253 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21255 * doc/invoke.texi (fabi-version): Spelling fix.
21257 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21260 * doc/invoke.texi (fabi-version): Mention mangling fix for
21263 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21266 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21268 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21271 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21274 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21276 (stack_protect_test): Ditto.
21277 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21278 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21280 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21281 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21282 -mstack-protector-guard-offset=.
21283 (RS/6000 and PowerPC Options): Ditto.
21285 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21287 * config/i386/i386.h (MASK_CLASS_P): New define.
21288 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21289 there are no registers from different register sets also when
21290 mask registers are used. Update function comment.
21291 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21292 to (*k/*r) and (*k/*km) alternatives.
21294 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21296 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21297 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21298 (EH_RETURN_HANDLER_RTX): New define.
21299 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21300 Force frame pointer in EH return functions.
21301 (aarch64_expand_epilogue): Add barrier for eh_return.
21302 (aarch64_final_eh_return_addr): Remove.
21303 (aarch64_eh_return_handler_rtx): New function.
21304 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21306 (aarch64_eh_return_handler_rtx): New prototype.
21308 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21310 * config/rs6000/altivec.h (vec_rlmi): New #define.
21311 (vec_vrlnm): Likewise.
21312 (vec_rlnm): Likewise.
21313 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21314 (UNSPEC_VRLNM): Likewise.
21315 (VIlong): New mode iterator.
21316 (altivec_vrl<VI_char>mi): New define_insn.
21317 (altivec_vrl<VI_char>nm): Likewise.
21318 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21320 (VRLDNM): Likewise.
21321 (RLNM): New polymorphic function entry.
21322 (VRLWMI): New monomorphic function entry.
21323 (VRLDMI): Likewise.
21324 (RLMI): New polymorphic function entry.
21325 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21326 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21327 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21330 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21333 * dwarf2out.c (field_byte_offset): Restore the
21334 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21335 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21336 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21339 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21342 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21344 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21347 * dwarf2out.c (add_data_member_location_attribute): For constant
21348 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21349 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21350 DW_AT_byte_size attributes.
21352 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21354 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21355 after forcing to constant memory when the code model is medium.
21357 2017-01-17 Julia Koval <julia.koval@intel.com>
21360 * config/i386/avx512fintrin.h
21361 (_mm512_i32gather_ps): Change __addr type to void const*.
21362 (_mm512_mask_i32gather_ps): Ditto.
21363 (_mm512_i32gather_pd): Ditto.
21364 (_mm512_mask_i32gather_pd): Ditto.
21365 (_mm512_i64gather_ps): Ditto.
21366 (_mm512_mask_i64gather_ps): Ditto.
21367 (_mm512_i64gather_pd): Ditto.
21368 (_mm512_mask_i64gather_pd): Ditto.
21369 (_mm512_i32gather_epi32): Ditto.
21370 (_mm512_mask_i32gather_epi32): Ditto.
21371 (_mm512_i32gather_epi64): Ditto.
21372 (_mm512_mask_i32gather_epi64): Ditto.
21373 (_mm512_i64gather_epi32): Ditto.
21374 (_mm512_mask_i64gather_epi32): Ditto.
21375 (_mm512_i64gather_epi64): Ditto.
21376 (_mm512_mask_i64gather_epi64): Ditto.
21377 (_mm512_i32scatter_ps): Change __addr type to void*.
21378 (_mm512_mask_i32scatter_ps): Ditto.
21379 (_mm512_i32scatter_pd): Ditto.
21380 (_mm512_mask_i32scatter_pd): Ditto.
21381 (_mm512_i64scatter_ps): Ditto.
21382 (_mm512_mask_i64scatter_ps): Ditto.
21383 (_mm512_i64scatter_pd): Ditto.
21384 (_mm512_mask_i64scatter_pd): Ditto.
21385 (_mm512_i32scatter_epi32): Ditto.
21386 (_mm512_mask_i32scatter_epi32): Ditto.
21387 (_mm512_i32scatter_epi64): Ditto.
21388 (_mm512_mask_i32scatter_epi64): Ditto.
21389 (_mm512_i64scatter_epi32): Ditto.
21390 (_mm512_mask_i64scatter_epi32): Ditto.
21391 (_mm512_i64scatter_epi64): Ditto.
21392 (_mm512_mask_i64scatter_epi64): Ditto.
21393 * config/i386/avx512pfintrin.h
21394 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21395 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21396 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21397 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21398 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21399 (_mm512_prefetch_i32scatter_ps): Ditto.
21400 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
21401 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
21402 (_mm512_prefetch_i64scatter_pd): Ditto.
21403 (_mm512_prefetch_i64scatter_ps): Ditto.
21404 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
21405 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
21406 * config/i386/avx512vlintrin.h
21407 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
21408 (_mm_mmask_i32gather_ps): Ditto.
21409 (_mm256_mmask_i32gather_pd): Ditto.
21410 (_mm_mmask_i32gather_pd): Ditto.
21411 (_mm256_mmask_i64gather_ps): Ditto.
21412 (_mm_mmask_i64gather_ps): Ditto.
21413 (_mm256_mmask_i64gather_pd): Ditto.
21414 (_mm_mmask_i64gather_pd): Ditto.
21415 (_mm256_mmask_i32gather_epi32): Ditto.
21416 (_mm_mmask_i32gather_epi32): Ditto.
21417 (_mm256_mmask_i32gather_epi64): Ditto.
21418 (_mm_mmask_i32gather_epi64): Ditto.
21419 (_mm256_mmask_i64gather_epi32): Ditto.
21420 (_mm_mmask_i64gather_epi32): Ditto.
21421 (_mm256_mmask_i64gather_epi64): Ditto.
21422 (_mm_mmask_i64gather_epi64): Ditto.
21423 (_mm256_i32scatter_ps): Change __addr type to void*.
21424 (_mm256_mask_i32scatter_ps): Ditto.
21425 (_mm_i32scatter_ps): Ditto.
21426 (_mm_mask_i32scatter_ps): Ditto.
21427 (_mm256_i32scatter_pd): Ditto.
21428 (_mm256_mask_i32scatter_pd): Ditto.
21429 (_mm_i32scatter_pd): Ditto.
21430 (_mm_mask_i32scatter_pd): Ditto.
21431 (_mm256_i64scatter_ps): Ditto.
21432 (_mm256_mask_i64scatter_ps): Ditto.
21433 (_mm_i64scatter_ps): Ditto.
21434 (_mm_mask_i64scatter_ps): Ditto.
21435 (_mm256_i64scatter_pd): Ditto.
21436 (_mm256_mask_i64scatter_pd): Ditto.
21437 (_mm_i64scatter_pd): Ditto.
21438 (_mm_mask_i64scatter_pd): Ditto.
21439 (_mm256_i32scatter_epi32): Ditto.
21440 (_mm256_mask_i32scatter_epi32): Ditto.
21441 (_mm_i32scatter_epi32): Ditto.
21442 (_mm_mask_i32scatter_epi32): Ditto.
21443 (_mm256_i32scatter_epi64): Ditto.
21444 (_mm256_mask_i32scatter_epi64): Ditto.
21445 (_mm_i32scatter_epi64): Ditto.
21446 (_mm_mask_i32scatter_epi64): Ditto.
21447 (_mm256_i64scatter_epi32): Ditto.
21448 (_mm256_mask_i64scatter_epi32): Ditto.
21449 (_mm_i64scatter_epi32): Ditto.
21450 (_mm_mask_i64scatter_epi32): Ditto.
21451 (_mm256_i64scatter_epi64): Ditto.
21452 (_mm256_mask_i64scatter_epi64): Ditto.
21453 (_mm_i64scatter_epi64): Ditto.
21454 (_mm_mask_i64scatter_epi64): Ditto.
21455 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
21456 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
21457 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
21458 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
21459 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
21460 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
21461 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
21462 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
21463 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
21464 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
21465 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
21466 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
21467 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
21468 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
21469 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
21470 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
21471 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
21472 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
21473 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
21474 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
21475 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
21476 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
21477 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
21478 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
21479 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
21480 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
21481 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
21482 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
21483 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
21484 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
21485 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
21486 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
21487 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
21488 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
21489 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
21490 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
21491 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
21492 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
21493 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
21494 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
21495 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
21496 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
21497 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
21498 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
21499 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
21500 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
21501 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
21502 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
21503 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
21504 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
21505 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
21506 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
21507 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
21508 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
21509 definitions accordingly.
21511 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
21512 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21515 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
21518 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
21521 * ira-conflicts.c (ira_build_conflicts): Update total conflict
21522 hard regs for inner regno.
21524 2017-01-17 Martin Liska <mliska@suse.cz>
21527 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
21528 assumption and add comment.
21530 2017-01-17 Nathan Sidwell <nathan@acm.org>
21532 * ipa-visibility.c (localize_node): New function, broken out of ...
21533 (function_and_variable_visibility): ... here. Call it.
21535 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21537 PR middle-end/77445
21538 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
21539 correctly set frequency of oudgoing edge.
21540 (duplicate_thread_path): Fix profile updating.
21542 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21545 * configure.ac: Add GCC_BASE_VER.
21546 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
21547 version from BASE-VER file.
21548 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
21549 (gcc.o): Depend on $(BASEVER).
21550 * common.opt (dumpfullversion): New option.
21551 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
21552 * doc/invoke.texi: Document -dumpfullversion.
21553 * doc/install.texi: Document --with-gcc-major-version-only.
21554 * configure: Regenerated.
21556 2017-01-17 Richard Biener <rguenther@suse.de>
21558 PR tree-optimization/71433
21559 * tree-vrp.c (register_new_assert_for): Merge same asserts
21560 on all incoming edges.
21561 (process_assert_insertions_for): Handle insertions at the
21564 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
21566 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
21567 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
21569 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
21572 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
21575 2017-01-17 Alan Modra <amodra@gmail.com>
21578 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
21579 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
21580 symbolic stack limit when pic.
21582 2017-01-16 Martin Sebor <msebor@redhat.com>
21584 PR tree-optimization/78608
21585 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
21587 2017-01-16 Jeff Law <law@redhat.com>
21590 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
21591 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
21592 for several include directories that may be relative to sysroot.
21593 * config/i386/x-mingw32 (gplus_includedir): Define.
21594 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
21595 (native_system_includedir): Likewise.
21596 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
21597 override if TARGET_SYSTEM_ROOT is defined.
21598 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
21600 PR tree-optimization/79090
21601 PR tree-optimization/33562
21602 PR tree-optimization/61912
21603 PR tree-optimization/77485
21604 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
21605 and computed trims into the dump file.
21607 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21609 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
21611 2017-01-16 Jakub Jelinek <jakub@redhat.com>
21614 * gimplify.c (gimplify_init_constructor): If want_value and
21615 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
21619 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
21620 sequence. Formatting fixes.
21621 (doloop_optimize): Formatting fixes.
21624 * gcc.c (debug_level_greater_than_spec_func): New function.
21625 (static_spec_functions): Add debug-level-gt spec function.
21626 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
21628 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21629 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21630 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
21631 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
21632 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
21633 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
21635 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
21637 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
21638 QImode fixups to general and mask registers only.
21640 2017-01-16 Carl Love <cel@us.ibm.com>
21642 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
21643 for built-in functions
21644 vector signed char vec_nabs (vector signed char)
21645 vector signed short vec_nabs (vector signed short)
21646 vector signed int vec_nabs (vector signed int)
21647 vector signed long long vec_nabs (vector signed long long)
21648 vector float vec_nabs (vector float)
21649 vector double vec_nabs (vector double)
21650 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
21652 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
21653 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
21654 * doc/extend.texi: Update the documentation file for the new built-in
21657 2017-01-16 Martin Sebor <msebor@redhat.com>
21659 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
21662 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21664 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
21665 UNSPEC_VSX__XXSPLTD to require special splat handling.
21667 2017-01-16 David Malcolm <dmalcolm@redhat.com>
21670 * system.h: Poison strndup.
21672 2017-01-16 Alan Modra <amodra@gmail.com>
21675 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
21678 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
21680 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
21682 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
21684 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
21685 call recog here. Assert that INSN_CODE (insn) is non-negative.
21687 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21690 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
21692 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
21693 in the currently scheduled RTL fragment.
21695 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21697 PR rtl-optimization/78751
21698 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
21701 2017-01-14 Jeff Law <law@redhat.com>
21703 PR tree-optimization/79090
21704 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
21705 variable length stores.
21706 (compute_trims): Delete dead assignment to *trim_tail.
21707 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
21710 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
21712 PR rtl-optimization/78626
21713 PR rtl-optimization/78727
21714 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
21715 of a block, and split such blocks after everything else is finished.
21717 2017-01-14 Alan Modra <amodra@gmail.com>
21720 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
21721 target legitimate_combined_insn.
21722 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
21723 (rs6000_legitimate_combined_insn): New function.
21724 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
21726 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
21727 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
21728 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
21730 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
21732 * doc/frontends.texi (G++ and GCC): Remove references to Java.
21734 2017-01-13 Jeff Law <law@redhat.com>
21736 PR tree-optimization/33562
21737 PR tree-optimization/61912
21738 PR tree-optimization/77485
21739 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
21741 (delete_dead_assignment): Likewise.
21742 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
21743 statement to delete_dead_call and delete_dead_assignment.
21745 2017-01-13 David Malcolm <dmalcolm@redhat.com>
21748 * substring-locations.c (format_warning_va): Strengthen case 1 so
21749 that both endpoints of the substring must be within the format
21750 range for just the substring to be printed.
21752 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
21754 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
21755 * config/i386/i386.c (ix86_target_string): Add missing options
21756 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
21757 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
21758 flags_other and ix86_target_other to flags2_other. Display unknown
21760 (ix86_valid_target_attribute_inner_p): Add missing options and
21761 reorder options by implied ISAs, as in ix86_target_string.
21763 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
21765 * hash-table.h (hash_table::too_empty_p): New function.
21766 (hash_table::expand): Use it.
21767 (hash_table::traverse): Likewise.
21768 (hash_table::empty_slot): Use sizeof (value_type) instead of
21769 sizeof (PTR) to convert bytes to elements. Shrink the table
21770 if the current size is excessive for the current number of
21773 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
21775 * ira-costs.c (record_reg_classes): Break from the inner loop
21776 early once alt_fail is known to be true. Update outer loop
21777 handling accordingly.
21779 2017-01-13 Jeff Law <law@redhat.com>
21781 * tree-ssa-dse.c (decrement_count): New function.
21782 (increment_start_addr, maybe_trim_memstar_call): Likewise.
21783 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
21784 when we know the partially dead statement is a mem* function.
21786 PR tree-optimization/61912
21787 PR tree-optimization/77485
21788 * tree-ssa-dse.c: Include expr.h.
21789 (maybe_trim_constructor_store): New function.
21790 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
21792 PR tree-optimization/33562
21793 PR tree-optimization/61912
21794 PR tree-optimization/77485
21795 * doc/invoke.texi: Document new dse-max-object-size param.
21796 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
21797 * tree-ssa-dse.c: Include params.h.
21798 (dse_store_status): New enum.
21799 (initialize_ao_ref_for_dse): New, partially extracted from
21801 (valid_ao_ref_for_dse, normalize_ref): New.
21802 (setup_live_bytes_from_ref, compute_trims): Likewise.
21803 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
21804 (maybe_trim_partially_dead_store): Likewise.
21805 (maybe_trim_complex_store): Likewise.
21806 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
21807 Track what bytes live from the original store. Return tri-state
21808 for dead, partially dead or live.
21809 (dse_dom_walker): Add constructor, destructor and new private members.
21810 (delete_dead_call, delete_dead_assignment): New extracted from
21812 (dse_optimize_stmt): Make a member of dse_dom_walker.
21813 Use initialize_ao_ref_for_dse.
21815 PR tree-optimization/33562
21816 PR tree-optimization/61912
21817 PR tree-optimization/77485
21818 * sbitmap.h (bitmap_count_bits): Prototype.
21819 (bitmap_clear_range, bitmap_set_range): Likewise.
21820 * sbitmap.c (bitmap_clear_range): New function.
21821 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
21823 2017-01-13 Martin Liska <mliska@suse.cz>
21826 * function.c (set_cfun): Add new argument force.
21827 * function.h (set_cfun): Likewise.
21828 * ipa-inline-transform.c (inline_call): Use the function when
21829 strict alising from is dropped for function we inline to.
21831 2017-01-13 Richard Biener <rguenther@suse.de>
21833 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
21834 for dumping GIMPLE INTEGER_CSTs.
21836 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21838 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
21839 to 201112L since C++17.
21841 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21844 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
21845 if -fsanitize=kernel-address is present.
21847 2017-01-13 Richard Biener <rguenther@suse.de>
21849 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
21850 as _Literal ( type ) number in case usual suffixes do not
21851 preserve all information.
21853 2017-01-13 Richard Biener <rguenther@suse.de>
21855 PR tree-optimization/77283
21856 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
21857 and ssa-iterators.h.
21858 (is_feasible_trace): Implement a cost model based on joiner
21861 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21864 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
21865 char or short to __float128/_Float128 directly.
21867 2017-01-12 Martin Sebor <msebor@redhat.com>
21869 to -Wformat-overflow.
21870 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
21871 (min_bytes_remaining): Same.
21872 (get_string_length): Same.
21873 (format_string): Same.
21874 (format_directive): Same.
21876 (pass_sprintf_length::handle_gimple_call): Same.
21878 2017-01-12 Jakub Jelinek <jakub@redhat.com>
21880 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
21881 info.nowrite calls with no lhs that can't throw. Return bool
21882 whether gsi_remove has been called or not.
21883 (pass_sprintf_length::handle_gimple_call): Return bool whether
21884 try_substitute_return_value called gsi_remove. Formatting fix.
21885 (pass_sprintf_length::execute): Don't use gsi_remove if
21886 handle_gimple_call returned true.
21889 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
21890 be removed due to side-effects, don't remove following barrier nor
21891 turn the successor edge into fallthru edge.
21893 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21896 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
21897 element-reversing loads and stores as not swappable.
21899 2017-01-12 Nathan Sidwell <nathan@acm.org>
21900 Nicolai Stange <nicstange@gmail.com>
21902 * combine.c (try_combine): Don't ignore result of overlap checking
21903 loop. Combine overlap & asm check into single loop.
21905 2017-01-12 Richard Biener <rguenther@suse.de>
21907 * tree-pretty-print.c (dump_generic_node): Provide -gimple
21908 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
21910 2017-01-12 Richard Biener <rguenther@suse.de>
21912 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
21913 and TS_TARGET_OPTION directly derive from TS_BASE.
21914 * tree-core.h (tree_optimization_option): Derive from tree_base.
21915 (tree_target_option): Likewise.
21917 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
21919 * config/i386/i386.c (memory_address_length): Increase len
21920 only when rip_relative_addr_p returns false.
21922 2017-01-11 Julia Koval <julia.koval@intel.com>
21924 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
21925 (OPTION_MASK_ISA_SGX_SET): New.
21926 (ix86_handle_option): Handle OPT_msgx.
21927 * config.gcc: Added sgxintrin.h.
21928 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
21929 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
21930 * config/i386/i386.c (ix86_target_string): Add -msgx.
21932 (ix86_option_override_internal): Handle new options.
21933 (ix86_valid_target_attribute_inner_p): Add sgx.
21934 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
21935 * config/i386/i386.opt: Add msgx.
21936 * config/i386/sgxintrin.h: New file.
21937 * config/i386/x86intrin.h: Add sgxintrin.h.
21939 2017-01-11 Jakub Jelinek <jakub@redhat.com>
21942 * fold-const.c (maybe_nonzero_address): Return 1 for function
21944 (tree_single_nonzero_warnv_p): Don't handle function local objects
21948 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
21951 2017-01-11 David Malcolm <dmalcolm@redhat.com>
21954 * opts.c: Include "spellcheck.h"
21955 (struct string_fragment): New struct.
21956 (struct edit_distance_traits<const string_fragment &>): New
21958 (get_closest_sanitizer_option): New function.
21959 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
21961 2017-01-11 Jakub Jelinek <jakub@redhat.com>
21963 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
21965 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
21966 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
21967 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
21968 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
21969 for initial die_offset if dwarf_split_debug_info.
21970 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
21971 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
21973 (output_skeleton_debug_sections): Formatting fix. Use
21974 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
21975 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
21977 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
21979 * config/arm/cortex-a53.md: Add bypasses for
21980 cortex_a53_r2f_cvt.
21981 (cortex_a53_r2f): Only use for transfers.
21982 (cortex_a53_f2r): Likewise.
21983 (cortex_a53_r2f_cvt): Add reservation for conversions.
21984 (cortex_a53_f2r_cvt): Likewise.
21986 2017-01-11 Tamar Christina <tamar.christina@arm.com>
21988 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
21989 to all inlined functions, change static to extern.
21991 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
21994 * config/arm/arm.c (legitimize_pic_address): Handle reference to
21996 (arm_assemble_integer): Likewise.
21998 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22000 * config.gcc: Use new awk script to check CPU, FPU and architecture
22001 parameters for --with-... options.
22002 * config/arm/parsecpu.awk: New file
22003 * config/arm/arm-cpus.in: New file.
22004 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22006 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22008 * config/arm/t-arm: Update dependency rules.
22009 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22010 of processing .def files.
22011 * config/arm/genopt.sh: Deleted.
22012 * config/arm/gentune.sh: Deleted.
22013 * config/arm/arm-cores.def: Deleted.
22014 * config/arm/arm-arches.def: Deleted.
22015 * config/arm/arm-fpus.def: Deleted.
22016 * config/arm/arm-tune.md: Regenerated.
22017 * config/arm/arm-tables.opt: Regenerated.
22018 * config/arm/arm-cpu.h: New generated file.
22019 * config/arm/arm-cpu-data.h: New generated file.
22020 * config/arm/arm-cpu-cdata.h: New generated file.
22022 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22025 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22027 (input_varpool_node): Unpack dynamically_initialized bit.
22029 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22031 PR rtl-optimization/79032
22032 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22033 the alignment of the adjusted memory reference against that of MODE,
22034 instead of the alignment of the original memory reference.
22036 2017-01-11 Martin Jambor <mjambor@suse.cz>
22038 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22040 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22041 decorated functions.
22043 2017-01-11 Richard Biener <rguenther@suse.de>
22045 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22046 set range/nonnull info for PHI results. Do not set it on
22047 stmts marked for removal.
22049 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22051 * expr.c (store_field): In the bitfield case, fetch the return value
22052 from the registers before applying a single big-endian adjustment.
22053 Always do a final load for a BLKmode value not larger than a word.
22055 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22058 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22059 that we correctly handle column numbers greater than
22060 LINE_MAP_MAX_COLUMN_NUMBER.
22062 2017-01-10 Martin Sebor <msebor@redhat.com>
22064 PR middle-end/78245
22065 * gimple-ssa-sprintf.c (get_destination_size): Call
22066 {init,fini}object_sizes.
22067 * tree-object-size.c (addr_object_size): Adjust.
22068 (pass_through_call): Adjust.
22069 (pass_object_sizes::execute): Adjust.
22070 * tree-object-size.h (fini_object_sizes): Declare.
22072 2017-01-10 Martin Sebor <msebor@redhat.com>
22074 PR tree-optimization/78775
22075 * builtins.c (get_size_range): Move...
22076 * calls.c: ...to here.
22077 (alloc_max_size): Accept zero argument.
22078 (operand_signed_p): Remove.
22079 (maybe_warn_alloc_args_overflow): Call get_size_range.
22080 * calls.h (get_size_range): Declare.
22082 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22084 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22085 from TI's devices.csv file as of September 2016.
22086 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22088 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22090 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22091 * doc/invoke.texi: Likewise.
22092 * doc/md.texi: Likewise.
22093 * doc/objc.texi: Likewise.
22095 2017-01-10 Joshua Conner <joshconner@google.com>
22097 * config/arm/fuchsia-elf.h: New file.
22098 * config/fuchsia.h: New file.
22099 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22100 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22102 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22104 2016-01-10 Richard Biener <rguenther@suse.de>
22106 PR tree-optimization/79034
22107 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22108 Propagate out degenerate PHIs in the joiner.
22110 2017-01-10 Martin Liska <mliska@suse.cz>
22112 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22113 (sort_congruence_classes_by_decl_uid): Likewise.
22114 (sort_congruence_class_groups_by_decl_uid): Likewise.
22115 (sem_item_optimizer::merge_classes): Sort class, groups in these
22116 classes and members in the groups by DECL_UID of declarations.
22117 This would make merge operations stable.
22119 2017-01-10 Martin Liska <mliska@suse.cz>
22121 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22122 usage of m_classes_vec.
22123 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22124 (sem_item_optimizer::get_group_by_hash): Likewise.
22125 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22126 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22127 (sem_item_optimizer::verify_classes): Likewise.
22128 (sem_item_optimizer::process_cong_reduction): Likewise.
22129 (sem_item_optimizer::dump_cong_classes): Likewise.
22130 (sem_item_optimizer::merge_classes): Likewise.
22131 * ipa-icf.h (congruence_class_hash): Rename from
22132 congruence_class_group_hash. Remove declaration of m_classes_vec.
22134 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22136 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22137 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22138 * config.gcc: Add avx512vpopcntdqintrin.h.
22139 * config/i386/avx512vpopcntdqintrin.h: New.
22140 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22141 * config/i386/i386-builtin-types.def: Add new types.
22142 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22143 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22144 __builtin_ia32_vpopcountq_v8di_mask): New.
22145 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22146 __AVX512VPOPCNTDQ__.
22147 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22148 (PTA_AVX512VPOPCNTDQ): Define.
22149 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22150 TARGET_AVX512VPOPCNTDQ_P): Define.
22151 * config/i386/i386.opt: Add mavx512vpopcntdq.
22152 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22153 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22155 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22157 PR middle-end/77484
22158 * predict.def (PRED_CALL): Set to 67.
22160 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22162 * expr.c (store_field): In the bitfield case, if the value comes from
22163 a function call and is of an aggregate type returned in registers, do
22164 not modify the field mode; extract the value in all cases if the mode
22165 is BLKmode and the size is not larger than a word.
22167 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22170 * config/i386/cpuid.h: Fix undefined behavior.
22172 2017-01-04 Jeff Law <law@redhat.com>
22174 PR tree-optimization/79007
22175 PR tree-optimization/67955
22176 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22177 conservative for pt.null when flag_non_call_exceptions is on.
22179 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22181 PR translation/79019
22182 PR translation/79020
22183 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22184 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22185 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22187 * config/avr/avr.opt (maccumulate-args): Likewise.
22188 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22189 * common.opt (freport-bug): Likewise.
22190 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22191 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22192 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22193 translatable string.
22194 * config/i386/i386.c (function_value_32): Likewise.
22195 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22196 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22198 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22199 * common/config/msp430/msp430-common.c (msp430_handle_option):
22201 * symtab.c (symtab_node::verify_base): Likewise.
22202 * opts.c (set_debug_level): Likewise.
22203 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22204 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22205 missing whitespace to translatable strings.
22206 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22207 * config/sh/superh.h: Likewise.
22208 * config/i386/xopintrin.h: Likewise.
22209 * config/i386/znver1.md: Likewise.
22210 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22211 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22212 * double-int.h (struct double_int): Likewise.
22213 * double-int.c (div_and_round_double): Likewise.
22214 * wide-int.cc: Likewise.
22215 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22216 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22217 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22218 (crossjumps_occurred): ... this.
22219 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22222 PR tree-optimization/78899
22223 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22224 returning bool return struct loop *, NULL for failure and the new
22226 (versionable_outer_loop_p): Don't version outer loop if it has
22227 dont_vectorized bit set.
22228 (tree_if_conversion): When versioning outer loop, ensure
22229 tree_if_conversion is performed also on the inner loop of the
22230 non-vectorizable outer loop copy.
22231 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22232 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22233 prevent vectorization of it.
22234 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22235 the outer loop vectorization of the non-scalar version is attempted
22236 before vectorization of the inner loop in scalar version. If
22237 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22238 vectorization of its inner loop.
22239 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22240 has 2 inner loops, rename also on edges from bb whose single pred
22241 is outer_loop->header. Fix typo in function comment.
22243 2017-01-09 Martin Sebor <msebor@redhat.com>
22246 * asan.c (asan_emit_stack_protection): Increase local buffer size
22247 to avoid snprintf truncation warning.
22249 2017-01-09 Andrew Pinski <apinski@cavium.com>
22251 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22252 to reference thunderx2t99 for the tuning structure
22253 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22255 (thunderx2t99_extra_costs): This.
22256 * config/aarch64/aarch64-tune.md: Regenerate.
22257 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22258 (vulcan_addrcost_table): This.
22259 (vulcan_regmove_cost): Rename to ...
22260 (thunderx2t99_regmove_cost): This.
22261 (vulcan_vector_cost): Rename to ...
22262 (thunderx2t99_vector_cost): this.
22263 (vulcan_branch_cost): Rename to ...
22264 (thunderx2t99_branch_cost): This.
22265 (vulcan_tunings): Rename to ...
22266 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22267 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22269 2017-01-09 Martin Jambor <mjambor@suse.cz>
22273 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22274 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22275 (propagate_vr_accross_jump_function): Use the above function for all
22276 value range computations for pass-through jump functions and type
22277 converasion from explicit value range values.
22278 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22279 parameters from TYPE_ARG_TYPES.
22280 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22281 (ipa_write_node_info): Stream type of the actual argument.
22282 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22284 2017-01-09 Martin Liska <mliska@suse.cz>
22287 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22288 (lookup_compiler): Do not show error message with have_E.
22290 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22292 PR tree-optimization/78938
22293 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22294 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22295 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22296 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22299 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22301 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22304 2017-01-09 Richard Biener <rguenther@suse.de>
22306 PR tree-optimization/78997
22307 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22308 name condition properly.
22310 2017-01-09 Richard Biener <rguenther@suse.de>
22313 * dwarf2out.c (is_cxx): New overload with context.
22314 (is_naming_typedef_decl): Use it.
22316 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22318 * invoke.texi (Option Summary): Correct spacing in option lists
22319 and add line breaks to fix over-long lines.
22321 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22323 PR middle-end/17660
22325 * extend.texi (Common Variable Attributes): Add xref to GCC
22326 Internals manual to explain mode attribute keywords.
22328 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22331 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22332 and Preprocessor Options.
22333 (Options for Linking): Document -pthread here....
22334 (RS/6000 and PowerPC Options): ...not here.
22335 (Solaris 2 Options): ...or here.
22336 * doc/cppopts.texi: Document -pthread.
22338 2017-01-08 Martin Sebor <msebor@redhat.com>
22340 PR middle-end/77708
22341 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22342 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22343 New member functions.
22344 (format_directive): Used them.
22346 (pass_sprintf_length::handle_gimple_call): Same.
22347 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22348 to avoid truncation for any argument.
22349 (extract_affine_mul): Same.
22350 * tree.c (get_file_function_name): Same.
22352 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22354 PR middle-end/77484
22355 * predict.def (PRED_INDIR_CALL): Set to 86.
22357 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22359 PR preprocessor/54124
22360 * doc/cppopts.texi: Reformat -d subtable to list the full name
22361 of the options. Add cross-reference to the docs for the general
22362 compiler -d options.
22363 * doc/invoke.texi (Developer Options): Add cross-reference to the
22364 preprocessor-specific -d option documentation.
22366 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22368 PR preprocessor/13498
22369 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22370 redudant material, and reflect new command-line options.
22371 (System Headers): Likewise.
22373 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22375 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22376 -isystem, and -idirafter. Copy-edit.
22377 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22378 default for -ftrack-macro-expansion. Delete obsolete and
22379 badly-formatted implementation details about -fdebug-cpp output.
22380 * doc/cppwarnopts.texi: Copy-edit.
22382 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22385 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22386 that the transition from a max line width >= 1<<10 to narrower
22387 lines works correctly.
22389 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22391 * doc/options.texi (PerFunction): New.
22392 * opt-functions.awk (switch_flags): Map both Optimization and
22393 PerFunction to CL_OPTIMIZATION.
22394 * opth-gen.awk: Test for PerFunction flag along with
22396 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22397 it only when the latter is present. Skip those that don't in
22398 the hash function generator.
22399 * common.opt (fvar-tracking): Mark as PerFunction instead of
22401 (fvar-tracking-assignments): Likewise.
22402 (fvar-tracking-assignments-toggle): Likewise.
22403 (fvar-tracking-uninit): Likewise.
22405 2017-01-07 Jakub Jelinek <jakub@redhat.com>
22407 PR translation/79018
22408 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
22411 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
22414 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
22415 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
22416 TARGET_LONG_JUMP_TABLE_OFFSETS.
22417 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
22418 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
22419 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
22420 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
22421 * config/m68k/m68k.md (tablejump expander): Likewise.
22422 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
22423 TARGET_LONG_JUMP_TABLE_OFFSETS.
22424 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
22425 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
22427 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
22428 David Holsgrove <david.holsgrove@xilinx.com>
22430 * common/config/microblaze/microblaze-common.c
22431 (TARGET_EXCEPT_UNWIND_INFO): Remove.
22432 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
22434 * config/microblaze/microblaze.c (microblaze_must_save_register)
22435 (microblaze_expand_epilogue, microblaze_return_addr): Handle
22437 (microblaze_eh_return): New function.
22438 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
22439 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
22440 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
22441 * config/microblaze/microblaze.md (eh_return): New pattern.
22443 2017-01-06 Jakub Jelinek <jakub@redhat.com>
22445 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
22446 GCC_DIAGNOSTIC_STRINGIFY): Define.
22448 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
22450 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22452 * config/arm/arm.md (<mcrr>): New.
22454 * config/arm/arm.c (arm_arch5te): New.
22455 (arm_option_override): Set arm_arch5te.
22456 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
22458 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
22459 (arm_mcrr_qualifiers): ... this. New.
22460 (MRRC_QUALIFIERS): Define to...
22461 (arm_mrrc_qualifiers): ... this. New.
22462 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
22464 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
22465 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
22466 (MRRCI, mrrc, MRRC): New.
22467 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
22468 VUNSPEC_MRRC2): New.
22470 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22472 * config/arm/arm.md (<mcr>): New.
22474 * config/arm/arm.c (arm_coproc_builtin_available): Add
22475 support for mcr, mrc, mcr2 and mrc2.
22476 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
22477 (arm_mcr_qualifiers): ... this. New.
22478 (MRC_QUALIFIERS): Define to ...
22479 (arm_mrc_qualifiers): ... this. New.
22480 (MCR_QUALIFIERS): Define to ...
22481 (arm_mcr_qualifiers): ... this. New.
22482 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
22484 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
22485 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
22486 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
22487 VUNSPEC_MRC2): New.
22489 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22491 * config/arm/arm.md (*ldc): New.
22495 * config/arm/arm.c (arm_coproc_builtin_available): Add
22496 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
22497 (arm_coproc_ldc_stc_legitimate_address): New.
22498 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
22499 'qualifier_const_pointer'.
22500 (LDC_QUALIFIERS): Define to...
22501 (arm_ldc_qualifiers): ... this. New.
22502 (STC_QUALIFIERS): Define to...
22503 (arm_stc_qualifiers): ... this. New.
22504 * config/arm/arm-protos.h
22505 (arm_coproc_ldc_stc_legitimate_address): New.
22506 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
22507 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
22508 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
22509 stc2, stcl, stc2l): New.
22510 * config/arm/constraints.md (Uz): New.
22511 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
22512 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
22513 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
22514 VUNSPEC_STC2L): New.
22516 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22518 * config/arm/arm.md (<cdp>): New.
22519 * config/arm/arm.c (neon_const_bounds): Rename this ...
22520 (arm_const_bounds): ... this.
22521 (arm_coproc_builtin_available): New.
22522 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
22523 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
22524 (CDP_QUALIFIERS): Define to...
22525 (arm_cdp_qualifiers): ... this. New.
22527 (arm_expand_builtin_args): Add case for 6 arguments.
22528 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
22529 (arm_const_bounds): ... this.
22530 (arm_coproc_builtin_available): New.
22531 * config/arm/arm_acle.h (__arm_cdp): New.
22533 * config/arm/arm_acle_builtins.def (cdp): New.
22535 * config/arm/iterators.md (CDPI,CDP,cdp): New.
22536 * config/arm/neon.md: Rename all 'neon_const_bounds' to
22537 'arm_const_bounds'.
22538 * config/arm/types.md (coproc): New.
22539 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
22540 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
22541 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
22542 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
22544 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22546 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
22547 (UBINOP_QUALIFIERS): New.
22549 (acle_builtin_data): New. Change comment.
22550 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
22551 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
22552 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
22553 arm_acle_builtins.def.
22554 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
22555 (arm_init_acle_builtins): New.
22556 (CRC32_BUILTIN): Remove.
22557 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
22558 crc32cb, crc32ch and crc32cw.
22559 (arm_init_crc32_builtins): Remove.
22560 (arm_init_builtins): Use arm_init_acle_builtins rather
22561 than arm_init_crc32_builtins.
22562 (arm_expand_acle_builtin): New.
22563 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
22564 * config/arm/arm_acle_builtins.def: New.
22566 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22568 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
22569 (arm_builtin_datum): ... this.
22570 (arm_init_neon_builtin): Rename to ...
22571 (arm_init_builtin): ... this. Add a new parameters PREFIX
22572 and USE_SIG_IN_NAME.
22573 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
22574 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
22575 'arm_builtin_datum'.
22576 (arm_init_vfp_builtins): Likewise.
22577 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
22578 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
22579 (arm_expand_neon_args): Rename to ...
22580 (arm_expand_builtin_args): ... this. Rename builtin_arg
22581 enum values and differentiate between ARG_BUILTIN_MEMORY
22582 and ARG_BUILTIN_NEON_MEMORY.
22583 (arm_expand_neon_builtin_1): Rename to ...
22584 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
22585 values, arm_expand_builtin_args and add bool parameter NEON.
22586 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
22587 (arm_expand_vfp_builtin): Likewise.
22588 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
22590 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22592 PR middle-end/77484
22593 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
22594 * predict.c (tree_estimate_probability_bb): Reverse direction of
22595 polymorphic call predictor.
22597 2017-01-06 David Malcolm <dmalcolm@redhat.com>
22599 * passes.c (execute_one_pass): Split out pass-skipping logic into...
22600 (determine_pass_name_match): ...this new function and...
22601 (should_skip_pass_p): ...this new function.
22603 2017-01-06 Nathan Sidwell <nathan@acm.org>
22605 * ipa-visibility.c (function_and_variable_visibility): Reformat
22606 comments and long lines. Remove extrneous if.
22607 * symtab.c (symtab_node::make_decl_local): Fix code format.
22608 (symtab_node::set_section_for_node): Fix comment typo.
22610 2017-01-06 Martin Liska <mliska@suse.cz>
22613 * lra-constraints.c: Rename invariant to lra_invariant.
22614 * predict.c (set_even_probabilities): Initialize e to NULL.
22616 2017-01-05 Martin Sebor <msebor@redhat.com>
22618 PR tree-optimization/78910
22619 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
22620 (format_integer): Correct off-by-one error in the handling
22621 of precision with negative numbers in signed conversions..
22623 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
22625 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
22627 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22629 PR tree-optimization/71016
22630 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
22631 factor_out_conditional_conversion. Formatting fix.
22632 (factor_out_conditional_conversion): Add cond_stmt argument.
22633 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
22634 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
22637 2017-01-05 David Malcolm <dmalcolm@redhat.com>
22639 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
22640 read-rtl-function.o, and selftest-rtl.o.
22641 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
22642 (selftest::aarch64_test_loading_full_dump): New function.
22643 (selftest::aarch64_run_selftests): New function.
22644 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
22645 selftest::aarch64_run_selftests.
22646 * config/i386/i386.c
22647 (selftest::ix86_test_loading_dump_fragment_1): New function.
22648 (selftest::ix86_test_loading_call_insn): New function.
22649 (selftest::ix86_test_loading_full_dump): New function.
22650 (selftest::ix86_test_loading_unspec): New function.
22651 (selftest::ix86_run_selftests): Call the new functions.
22652 * emit-rtl.c (maybe_set_max_label_num): New function.
22653 * emit-rtl.h (maybe_set_max_label_num): New decl.
22654 * function.c (instantiate_decls): Guard call to
22655 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
22656 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
22658 * gensupport.c (gen_reader::gen_reader): Pass "false"
22659 for new "compact" param of rtx_reader.
22660 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
22661 rather than an empty string for NULL strings.
22662 * read-md.c: Potentially include config.h rather than bconfig.h.
22663 Wrap include of errors.h with #ifdef GENERATOR_FILE.
22664 (have_error): New global, copied from errors.c.
22665 (md_reader::read_name): Rename to...
22666 (md_reader::read_name_1): ...this, adding "out_loc" param,
22667 and converting "missing name or number" to returning false, rather
22669 (md_reader::read_name): Reimplement in terms of read_name_1.
22670 (md_reader::read_name_or_nil): New function.
22671 (md_reader::read_string): Handle "(nil)" by returning NULL.
22672 (md_reader::md_reader): Add new param "compact".
22673 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
22674 (md_reader::read_file): New method.
22675 * read-md.h (md_reader::md_reader): Add new param "compact".
22676 (md_reader::read_file): New method.
22677 (md_reader::is_compact): New accessor.
22678 (md_reader::read_name): Convert return type from void to file_location.
22679 (md_reader::read_name_or_nil): New decl.
22680 (md_reader::read_name_1): New decl.
22681 (md_reader::m_compact): New field.
22682 (noop_reader::noop_reader): Pass "false" for new "compact" param
22684 (rtx_reader::rtx_reader): Add new "compact" param.
22685 (rtx_reader::read_rtx_operand): Make virtual and convert return
22686 type from void to rtx.
22687 (rtx_reader::read_until): New decl.
22688 (rtx_reader::handle_any_trailing_information): New virtual function.
22689 (rtx_reader::postprocess): New virtual function.
22690 (rtx_reader::finalize_string): New virtual function.
22691 (rtx_reader::m_in_call_function_usage): New field.
22692 (rtx_reader::m_reuse_rtx_by_id): New field.
22693 * read-rtl-function.c: New file.
22694 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
22695 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
22696 (selftest::verify_three_block_rtl_cfg): New decl.
22697 * read-rtl-function.h: New file.
22698 * read-rtl.c: Potentially include config.h rather than bconfig.h.
22699 For host, include function.h, memmodel.h, and emit-rtl.h.
22700 (one_time_initialization): New function.
22701 (struct compact_insn_name): New struct.
22702 (compact_insn_names): New array.
22703 (find_code): Handle insn codes in compact dumps.
22704 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
22705 (bind_subst_iter_and_attr): Likewise.
22706 (add_condition_to_string): Likewise.
22707 (add_condition_to_rtx): Likewise.
22708 (apply_attribute_uses): Likewise.
22709 (add_current_iterators): Likewise.
22710 (apply_iterators): Likewise.
22711 (initialize_iterators): Guard usage of apply_subst_iterator with
22712 #ifdef GENERATOR_FILE.
22713 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
22714 (md_reader::read_mapping): Likewise.
22715 (add_define_attr_for_define_subst): Likewise.
22716 (add_define_subst_attr): Likewise.
22717 (read_subst_mapping): Likewise.
22718 (check_code_iterator): Likewise.
22719 (rtx_reader::read_rtx): Likewise. Move one-time initialization
22721 (one_time_initialization): New function.
22722 (rtx_reader::read_until): New method.
22723 (read_flags): New function.
22724 (parse_reg_note_name): New function.
22725 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
22726 Handle reuse_rtx ids.
22727 Wrap iterator lookup within #ifdef GENERATOR_FILE.
22728 Add parsing support for RTL dumps, mirroring the special-cases in
22729 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
22730 values, and calling handle_any_trailing_information.
22731 (rtx_reader::read_rtx_operand): Convert return type from void
22732 to rtx, returning return_rtx. Handle case 'e'. Call
22733 finalize_string on XSTR and XTMPL fields.
22734 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
22735 "(nil)" values were omitted. Call the postprocess vfunc on the
22737 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
22738 class ctor. Initialize m_in_call_function_usage. Call
22739 one_time_initialization.
22740 * rtl-tests.c (selftest::test_uncond_jump): Call
22741 set_new_first_and_last_insn.
22742 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
22743 * selftest-rtl.c: New file.
22744 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
22745 (selftest::get_insn_by_uid): New decl.
22746 * selftest-run-tests.c (selftest::run_tests): Call
22747 read_rtl_function_c_tests.
22748 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
22749 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
22752 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
22754 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
22755 operands in a special way. Assert that pos+len <= mode precision.
22757 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22759 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
22760 3 argument Alias with unlimited for the negative form.
22761 (fno-vect-cost-model): Removed.
22763 2017-01-05 Martin Liska <mliska@suse.cz>
22765 * hsa-gen.c (gen_hsa_divmod): New function.
22766 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
22768 2017-01-05 Martin Liska <mliska@suse.cz>
22771 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
22774 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22776 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
22777 small constant length operands.
22779 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22781 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
22782 between loop iterations.
22784 2017-01-05 Martin Liska <mliska@suse.cz>
22787 * gimplify.c (gimplify_decl_expr): Compare to
22788 asan_poisoned_variables instread of checking flags.
22789 (gimplify_target_expr): Likewise.
22790 (gimplify_expr): Likewise.
22791 (gimplify_function_tree): Conditionally initialize
22792 asan_poisoned_variables.
22794 2017-01-04 Jeff Law <law@redhat.com>
22796 PR tree-optimizatin/78812
22797 * rtl.h (contains_mem_rtx_p): Prototype.
22798 * ifcvt.c (containts_mem_rtx_p): Move from here to...
22799 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
22800 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
22801 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
22802 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
22804 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22806 * input.c (assert_char_at_range): Default-initialize actual_range.
22808 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22810 * df-scan.c (df_ref_create_structure): Make regno unsigned,
22811 to match the caller.
22813 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22815 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
22816 insns after final jump in test to emit dummy move.
22818 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22820 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
22821 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
22823 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22825 * multiple_target.c (create_dispatcher_calls): Init e_next.
22826 * tree-ssa-loop-split.c (split_loop): Init border.
22827 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
22830 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
22835 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
22836 (altivec_register_operand): Do not return true if the operand
22837 contains a SUBREG mixing SImode and SFmode.
22838 (vsx_register_operand): Likewise.
22839 (vsx_reg_sfsubreg_ok): New predicate.
22840 (vfloat_operand): Do not return true if the operand contains a
22841 SUBREG mixing SImode and SFmode.
22842 (vint_operand): Likewise.
22843 (vlogical_operand): Likewise.
22844 (gpc_reg_operand): Likewise.
22845 (int_reg_operand): Likewise.
22846 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
22847 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
22848 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
22850 (rs6000_emit_move_si_sf_subreg): New helper function.
22851 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
22852 fixup SUBREGs involving SImode and SFmode.
22853 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
22854 numbers for the new peephole2 optimization.
22855 (peephole2 for SFmode unions): New peephole2 to optimize cases in
22856 the GLIBC math library that do AND/IOR/XOR operations on single
22857 precision floating point.
22858 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
22859 target macros to say whether we need to avoid SUBREGs mixing
22861 (TARGET_ALLOW_SF_SUBREG): Likewise.
22862 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
22863 (UNSPEC_SI_FROM_SF): Likewise.
22864 (iorxor): Change spacing.
22865 (and_ior_xor): New iterator for AND, IOR, and XOR.
22866 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
22867 (movdi_from_sf_zero_ext): Likewise.
22868 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
22869 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
22870 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
22871 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
22872 (fms<mode>4): Likewise.
22873 (fnma<mode>4): Likewise.
22874 (fnms<mode>4): Likewise.
22875 (nfma<mode>4): Likewise.
22876 (nfms<mode>4): Likewise.
22878 2017-01-04 Marek Polacek <polacek@redhat.com>
22881 * doc/invoke.texi: Document -Wpointer-compare.
22883 2017-01-04 Jakub Jelinek <jakub@redhat.com>
22885 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
22888 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
22889 descriptions for -gdwarf-5 and emit them as uleb128 instead of
22892 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
22895 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
22896 documentation of the powerpc_popcntb_ok attribute.
22897 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22898 code to issue warning messages if a requested CPU configuration is
22899 not supported by the binary (assembler and loader) toolchain.
22900 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
22901 made to define a built-in function that has been disabled.
22902 (paired_init_builtins): Add assertion to prevent ICE if attempt is
22903 made to define a built-in function that has been disabled.
22904 (altivec_init_builtins): Add comment explaining why definition
22905 of the DST built-in functions is not preceded by an assertion
22906 check. Add assertions to prevent ICE if attempts are made to
22907 define an altivec predicate or an abs* built-in function that has
22909 (htm_init_builtins): Add comment explaining why definition of the
22910 htm built-in functions is not preceded by an assertion check.
22912 2017-01-04 Jeff Law <law@redhat.com>
22914 PR tree-optimizatin/67955
22915 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
22916 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
22917 the points-to solution does not include pt_null. Use DECL_PT_UID
22920 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
22922 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
22923 Use gen_int_mode instead of gen_lopwart for const_int operands.
22925 2017-01-04 Jakub Jelinek <jakub@redhat.com>
22927 PR tree-optimization/71563
22928 * match.pd: Simplify X << Y into X if Y is known to be 0 or
22929 out of range value - has low bits known to be zero.
22931 2017-01-04 Alan Modra <amodra@gmail.com>
22933 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
22934 * configure: Regenerate.
22935 * config.in: Regenerate.
22937 2017-01-04 Jakub Jelinek <jakub@redhat.com>
22940 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
22941 a substring of the message, but strcmp with the whole message. Ifdef
22942 ENABLE_NLS, translate the message first using dgettext.
22944 2017-01-03 Jeff Law <law@redhat.com>
22946 PR tree-optimizatin/78856
22947 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
22948 (mark_threaded_blocks): Remove code to truncate thread paths that
22949 cross multiple loop headers. Instead invalidate the cached loop
22950 iteration information and handle case of a thread path walking
22951 into an irreducible region.
22953 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
22956 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
22957 assertions. Add support for doing the signbit if the IEEE 128-bit
22958 floating point value is in a GPR.
22959 * config/rs6000/rs6000.md (Fsignbit): Delete.
22960 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
22961 Update the length attribute if the value is in a GPR.
22962 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
22963 the sign or zero extension instruction, since the value is always 0/1.
22964 (signbit<mode>2_dm2): Delete using <Fsignbit>.
22967 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
22968 extracting SImode to a GPR register so that we can generate a
22969 store, limit the vector to be in a traditional Altivec register
22970 for the vextuwrx instruction.
22972 2017-01-03 Ian Lance Taylor <iant@google.com>
22974 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
22976 2017-01-03 Martin Sebor <msebor@redhat.com>
22978 PR tree-optimization/78696
22979 * gimple-ssa-sprintf.c (format_floating): Correct handling of
22980 precision. Use MPFR for %f for greater fidelity. Correct handling
22982 (pass_sprintf_length::compute_format_length): Set width and precision
22983 specified by asrerisk to void_node for vararg functions.
22984 (try_substitute_return_value): Adjust dump output.
22986 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
22988 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
22990 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
22992 * doc/invoke.texi (SPARC options): Document -mlra as the default.
22993 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
22994 -mlra/-mno-lra was passed to the compiler.
22996 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
22998 PR rtl-optimization/65618
22999 * emit-rtl.c (try_split): Move initialization of "before" and
23000 "after" to just before the call to emit_insn_after_setloc.
23002 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23004 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23006 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23008 * dwarf2out.c (gen_enumeration_type_die): When
23009 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23011 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23013 PR tree-optimization/78965
23014 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23015 Change first argument from const call_info & to call_info &. For %n
23016 set info.nowrite to false.
23018 PR middle-end/78901
23019 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23020 possibly throwing calls.
23022 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23023 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23024 and fns handling, rather than in a separate case SSA_NAME.
23026 2017-01-02 Jeff Law <law@redhat.com>
23028 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23029 fixes for first_period and second_period variables.
23031 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23034 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23035 (*insvqi_1): New insn pattern.
23036 (*insvqi_1_mem_rex64): Ditto.
23037 (*insvqi_2): Ditto.
23038 (*insvqi_3): Rename from *insvqi.
23040 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23042 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23044 * doc/cfg.texi (Edges): Remove reference to Java.
23045 (Maintaining the CFG): Ditto.
23047 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23049 PR middle-end/77674
23050 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23051 transparent aliases.
23053 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23055 PR middle-end/77484
23056 * predict.def (PRED_CALL): Update hitrate.
23057 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23058 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23059 into direct/indirect/polymorphic variants.
23061 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23063 Update copyright years.
23065 * gcc.c (process_command): Update copyright notice dates.
23066 * gcov-dump.c (print_version): Ditto.
23067 * gcov.c (print_version): Ditto.
23068 * gcov-tool.c (print_version): Ditto.
23069 * gengtype.c (create_file): Ditto.
23070 * doc/cpp.texi: Bump @copying's copyright year.
23071 * doc/cppinternals.texi: Ditto.
23072 * doc/gcc.texi: Ditto.
23073 * doc/gccint.texi: Ditto.
23074 * doc/gcov.texi: Ditto.
23075 * doc/install.texi: Ditto.
23076 * doc/invoke.texi: Ditto.
23078 Copyright (C) 2017 Free Software Foundation, Inc.
23080 Copying and distribution of this file, with or without modification,
23081 are permitted in any medium without royalty provided the copyright
23082 notice and this notice are preserved.