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 * machmode.h (is_int_mode): New fuction.
6 * combine.c (find_split_point): Use it.
7 (combine_simplify_rtx): Likewise.
8 (simplify_if_then_else): Likewise.
9 (simplify_set): Likewise.
10 (simplify_shift_const_1): Likewise.
11 (simplify_comparison): Likewise.
12 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13 * cse.c (notreg_cost): Likewise.
15 * cselib.c (cselib_lookup_1): Likewise.
16 * dojump.c (do_jump_1): Likewise.
17 (do_compare_rtx_and_jump): Likewise.
18 * dse.c (get_call_args): Likewise.
19 * dwarf2out.c (rtl_for_decl_init): Likewise.
20 (native_encode_initializer): Likewise.
21 * expmed.c (emit_store_flag_1): Likewise.
22 (emit_store_flag): Likewise.
23 * expr.c (convert_modes): Likewise.
24 (store_field): Likewise.
25 (expand_expr_real_1): Likewise.
26 * fold-const.c (fold_read_from_constant_string): Likewise.
27 * gimple-ssa-sprintf.c (get_format_string): Likewise.
28 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
29 * optabs.c (expand_binop): Likewise.
30 (expand_unop): Likewise.
31 (expand_abs_nojump): Likewise.
32 (expand_one_cmpl_abs_nojump): Likewise.
33 * simplify-rtx.c (mode_signbit_p): Likewise.
34 (val_signbit_p): Likewise.
35 (val_signbit_known_set_p): Likewise.
36 (val_signbit_known_clear_p): Likewise.
37 (simplify_relational_operation_1): Likewise.
38 * tree.c (vector_type_mode): Likewise.
40 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
41 Alan Hayward <alan.hayward@arm.com>
42 David Sherwood <david.sherwood@arm.com>
44 * machmode.h (smallest_mode_for_size): Fix formatting.
45 (smallest_int_mode_for_size): New function.
46 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
47 instead of smallest_mode_for_size.
48 * combine.c (make_extraction): Likewise.
49 * config/arc/arc.c (arc_expand_movmem): Likewise.
50 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
51 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
52 * config/s390/s390.c (s390_expand_insv): Likewise.
53 * config/sparc/sparc.c (assign_int_registers): Likewise.
54 * config/spu/spu.c (spu_function_value): Likewise.
55 (spu_function_arg): Likewise.
56 * coverage.c (get_gcov_type): Likewise.
57 (get_gcov_unsigned_t): Likewise.
58 * dse.c (find_shift_sequence): Likewise.
59 * expmed.c (store_bit_field_1): Likewise.
60 * expr.c (convert_move): Likewise.
61 (store_field): Likewise.
62 * internal-fn.c (expand_arith_overflow): Likewise.
63 * optabs-query.c (get_best_extraction_insn): Likewise.
64 * optabs.c (expand_twoval_binop_libfunc): Likewise.
65 * stor-layout.c (layout_type): Likewise.
66 (initialize_sizetypes): Likewise.
67 * targhooks.c (default_get_mask_mode): Likewise.
68 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
70 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * machmode.h (opt_mode::else_blk): New function.
75 (int_mode_for_mode): Declare.
76 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
77 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
79 * cfgexpand.c (expand_debug_expr): Likewise.
80 * combine.c (gen_lowpart_or_truncate): Likewise.
81 (gen_lowpart_for_combine): Likewise.
82 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
83 * config/avr/avr.c (avr_to_int_mode): Likewise.
84 (avr_out_plus_1): Likewise.
85 (avr_out_plus): Likewise.
86 (avr_out_round): Likewise.
87 * config/i386/i386.c (ix86_split_to_parts): Likewise.
88 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
89 (s390_expand_vcond): Likewise.
90 * config/spu/spu.c (spu_split_immediate): Likewise.
91 (spu_expand_mov): Likewise.
92 * dse.c (get_stored_val): Likewise.
93 * expmed.c (store_bit_field_1): Likewise.
94 (convert_extracted_bit_field): Use int_mode_for_mode instead of
96 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
97 (extract_low_bits): Likewise.
98 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
99 handling rather than repeating the check.
100 (emit_group_store): Likewise.
101 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
102 * optabs.c (expand_absneg_bit): Likewise.
103 (expand_copysign_absneg): Likewise.
104 (expand_copysign_bit): Likewise.
105 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
106 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
107 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
108 * var-tracking.c (prepare_call_arguments): Likewise.
109 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
110 int_mode_for_mode instead of mode_for_size.
111 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
114 Alan Hayward <alan.hayward@arm.com>
115 David Sherwood <david.sherwood@arm.com>
117 * machmode.h (int_mode_for_size): New function.
118 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
119 instead of mode_for_size.
120 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
122 * combine.c (expand_field_assignment): Use int_mode_for_size
123 instead of mode_for_size.
124 (make_extraction): Likewise.
125 (simplify_shift_const_1): Likewise.
126 (simplify_comparison): Likewise.
127 * dojump.c (do_jump): Likewise.
128 * dwarf2out.c (mem_loc_descriptor): Likewise.
129 * emit-rtl.c (init_derived_machine_modes): Likewise.
130 * expmed.c (flip_storage_order): Likewise.
131 (convert_extracted_bit_field): Likewise.
132 * expr.c (copy_blkmode_from_reg): Likewise.
133 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
134 * internal-fn.c (expand_mul_overflow): Likewise.
135 * lower-subreg.c (simple_move): Likewise.
136 * optabs-libfuncs.c (init_optabs): Likewise.
137 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
138 * tree.c (vector_type_mode): Likewise.
139 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
140 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
141 * tree-vect-generic.c (expand_vector_parallel): Likewise.
142 * tree-vect-stmts.c (vectorizable_load): Likewise.
143 (vectorizable_store): Likewise.
145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
146 Alan Hayward <alan.hayward@arm.com>
147 David Sherwood <david.sherwood@arm.com>
149 * coretypes.h (pod_mode): New type.
150 (scalar_int_mode_pod): New typedef.
151 * machmode.h (pod_mode): New class.
152 (int_n_data_t::m): Change type to scalar_int_mode_pod.
153 * genmodes.c (emit_mode_int_n): Update accordingly.
154 * lower-subreg.h (target_lower_subreg): Change type to
156 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
160 Alan Hayward <alan.hayward@arm.com>
161 David Sherwood <david.sherwood@arm.com>
163 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
164 machine_mode to scalar_int_mode.
165 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
166 (rs6000_option_override_internal): Remove cast to int.
167 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
168 machine_mode to scalar_int_mode.
169 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
170 (rs6000_option_override_internal): Remove cast to int.
171 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
172 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
174 * config/s390/s390.c (s390_expand_builtin): Likewise.
175 * coretypes.h (scalar_int_mode): New type.
176 (opt_scalar_int_mode): New typedef.
177 * machmode.h (scalar_int_mode): New class.
178 (scalar_int_mode::includes_p): New function.
179 (byte_mode): Change type to scalar_int_mode.
180 (word_mode): Likewise.
181 (ptr_mode): Likewise.
182 * emit-rtl.c (byte_mode): Likewise.
183 (word_mode): Likewise.
184 (ptr_mode): Likewise.
185 (init_derived_machine_modes): Update accordingly.
186 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
187 and MODE_PARTIAL_INT.
188 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
192 Alan Hayward <alan.hayward@arm.com>
193 David Sherwood <david.sherwood@arm.com>
195 * target.def (libgcc_floating_mode_supported_p): Take a
197 * doc/tm.texi: Regenerate.
198 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
200 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
201 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
205 Alan Hayward <alan.hayward@arm.com>
206 David Sherwood <david.sherwood@arm.com>
208 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
209 * doc/tm.texi: Regenerate.
210 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
211 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
212 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
213 * targhooks.h (default_floatn_mode): Likewise.
214 * targhooks.c (default_floatn_mode): Likewise.
215 * tree.c (build_common_tree_nodes): Update accordingly.
217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
218 Alan Hayward <alan.hayward@arm.com>
219 David Sherwood <david.sherwood@arm.com>
221 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
222 (mode_iterator::iterate_p): Likewise.
223 (mode_iterator::get_wider): Likewise.
224 * expr.c (init_expr_target): Use opt_scalar_float_mode.
226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
227 Alan Hayward <alan.hayward@arm.com>
228 David Sherwood <david.sherwood@arm.com>
230 * coretypes.h (opt_scalar_float_mode): New typedef.
231 * machmode.h (float_mode_for_size): New function.
232 * emit-rtl.c (double_mode): Delete.
233 (init_emit_once): Use float_mode_for_size.
234 * stor-layout.c (layout_type): Likewise.
235 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
238 Alan Hayward <alan.hayward@arm.com>
239 David Sherwood <david.sherwood@arm.com>
241 * output.h (assemble_real): Take a scalar_float_mode.
242 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
243 * config/arm/arm.md (consttable_4): Likewise.
244 (consttable_8): Likewise.
245 (consttable_16): Likewise.
246 * config/mips/mips.md (consttable_float): Likewise.
247 * config/s390/s390.c (s390_output_pool_entry): Likewise.
248 * varasm.c (assemble_real): Take a scalar_float_mode.
249 (output_constant_pool_2): Update accordingly.
250 (output_constant): Likewise.
252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
253 Alan Hayward <alan.hayward@arm.com>
254 David Sherwood <david.sherwood@arm.com>
256 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
257 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
258 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
259 (native_encode_real): Likewise.
260 (native_interpret_real): Likewise.
261 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
262 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
265 Alan Hayward <alan.hayward@arm.com>
266 David Sherwood <david.sherwood@arm.com>
268 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
269 <scalar_float_mode>. Simplify.
270 (gen_extend_conv_libfunc): Likewise.
272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
273 Alan Hayward <alan.hayward@arm.com>
274 David Sherwood <david.sherwood@arm.com>
276 * coretypes.h (scalar_float_mode): New type.
277 * machmode.h (mode_traits::from_int): Use machine_mode if
278 USE_ENUM_MODES is defined.
279 (is_a): New function.
281 (dyn_cast): Likewise.
282 (scalar_float_mode): New class.
283 (scalar_float_mode::includes_p): New function.
284 (is_float_mode): Likewise.
285 * gdbhooks.py (MachineModePrinter): New class.
286 (build_pretty_printer): Use it for scalar_float_mode.
287 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
288 (format_helper::format_helper): Turn into a template.
289 * genmodes.c (get_mode_class): New function.
290 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
291 or machine_mode if none.
292 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
293 as_a <scalar_float_mode>.
294 * dwarf2out.c (mem_loc_descriptor): Likewise.
295 (insert_float): Likewise.
296 (add_const_value_attribute): Likewise.
297 * simplify-rtx.c (simplify_immed_subreg): Likewise.
298 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
299 (expand_unop): Update accordingly.
300 (expand_abs_nojump): Likewise.
301 (expand_copysign_absneg): Take a scalar_float_mode.
302 (expand_copysign_bit): Likewise.
303 (expand_copysign): Update accordingly.
305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
306 Alan Hayward <alan.hayward@arm.com>
307 David Sherwood <david.sherwood@arm.com>
309 * coretypes.h (opt_mode): New class.
310 * machmode.h (opt_mode): Likewise.
311 (opt_mode::else_void): New function.
312 (opt_mode::require): Likewise.
313 (opt_mode::exists): Likewise.
314 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
315 (GET_MODE_2XWIDER_MODE): Likewise.
316 (mode_iterator::get_wider): Update accordingly.
317 (mode_iterator::get_2xwider): Likewise.
318 (mode_iterator::get_known_wider): Likewise, turning into a template.
319 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
320 forcing a wider mode to exist.
321 * config/cr16/cr16.h (LONG_REG_P): Likewise.
322 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
323 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
324 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
325 * lower-subreg.c (init_lower_subreg): Likewise.
326 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
327 on the final iteration.
328 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
329 a wider mode exists before asking for a move pattern.
330 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
331 forcing a wider mode to exist.
332 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
333 returning false if no such mode exists.
334 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
335 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
336 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
337 Avoid checking for a MODE_INT if we already know the mode is not a
339 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
340 forcing a wider mode to exist.
341 (expmed_mult_highpart_optab): Likewise.
342 (expmed_mult_highpart): Likewise.
343 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
345 * lto-streamer-in.c (lto_input_mode_table): Likewise.
346 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
347 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
348 * internal-fn.c (expand_mul_overflow): Update use of
349 GET_MODE_2XWIDER_MODE.
350 * omp-low.c (omp_clause_aligned_alignment): Likewise.
351 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
353 (convert_plusminus_to_widen): Likewise.
354 * tree-switch-conversion.c (array_value_type): Likewise.
355 * var-tracking.c (emit_note_insn_var_location): Likewise.
356 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
357 Return false inside rather than outside the loop if no wider mode
359 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
360 and GET_MODE_2XWIDER_MODE
361 (can_compare_p): Use else_void.
362 * gdbhooks.py (OptMachineModePrinter): New class.
363 (build_pretty_printer): Use it for opt_mode.
365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
366 Alan Hayward <alan.hayward@arm.com>
367 David Sherwood <david.sherwood@arm.com>
369 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
370 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
373 Alan Hayward <alan.hayward@arm.com>
374 David Sherwood <david.sherwood@arm.com>
376 * machmode.h (mode_traits): New structure.
377 (get_narrowest_mode): New function.
378 (mode_iterator::start): Likewise.
379 (mode_iterator::iterate_p): Likewise.
380 (mode_iterator::get_wider): Likewise.
381 (mode_iterator::get_known_wider): Likewise.
382 (mode_iterator::get_2xwider): Likewise.
383 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
384 (FOR_EACH_MODE): Likewise.
385 (FOR_EACH_MODE_FROM): Likewise.
386 (FOR_EACH_MODE_UNTIL): Likewise.
387 (FOR_EACH_WIDER_MODE): Likewise.
388 (FOR_EACH_2XWIDER_MODE): Likewise.
389 * builtins.c (expand_builtin_strlen): Use new mode iterators.
390 * combine.c (simplify_comparison): Likewise
391 * config/i386/i386.c (type_natural_mode): Likewise.
392 * cse.c (cse_insn): Likewise.
393 * dse.c (find_shift_sequence): Likewise.
394 * emit-rtl.c (init_derived_machine_modes): Likewise.
395 (init_emit_once): Likewise.
396 * explow.c (hard_function_value): Likewise.
397 * expmed.c (extract_fixed_bit_field_1): Likewise.
398 (extract_bit_field_1): Likewise.
399 (expand_divmod): Likewise.
400 (emit_store_flag_1): Likewise.
401 * expr.c (init_expr_target): Likewise.
402 (convert_move): Likewise.
403 (alignment_for_piecewise_move): Likewise.
404 (widest_int_mode_for_size): Likewise.
405 (emit_block_move_via_movmem): Likewise.
406 (copy_blkmode_to_reg): Likewise.
407 (set_storage_via_setmem): Likewise.
408 (compress_float_constant): Likewise.
409 * omp-low.c (omp_clause_aligned_alignment): Likewise.
410 * optabs-query.c (get_best_extraction_insn): Likewise.
411 * optabs.c (expand_binop): Likewise.
412 (expand_twoval_unop): Likewise.
413 (expand_twoval_binop): Likewise.
414 (widen_leading): Likewise.
415 (widen_bswap): Likewise.
416 (expand_parity): Likewise.
417 (expand_unop): Likewise.
418 (prepare_cmp_insn): Likewise.
419 (prepare_float_lib_cmp): Likewise.
420 (expand_float): Likewise.
421 (expand_fix): Likewise.
422 (expand_sfix_optab): Likewise.
423 * postreload.c (move2add_use_add2_insn): Likewise.
424 * reg-stack.c (reg_to_stack): Likewise.
425 * reginfo.c (choose_hard_reg_mode): Likewise.
426 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
427 * stor-layout.c (mode_for_size): Likewise.
428 (smallest_mode_for_size): Likewise.
429 (mode_for_vector): Likewise.
430 (finish_bitfield_representative): Likewise.
431 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
432 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
433 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
434 * var-tracking.c (prepare_call_arguments): Likewise.
436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
437 Alan Hayward <alan.hayward@arm.com>
438 David Sherwood <david.sherwood@arm.com>
440 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
441 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
442 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
443 * machmode.h (mode_size): Move earlier in file.
444 (mode_precision): Likewise.
445 (mode_inner): Likewise.
446 (mode_nunits): Likewise.
447 (mode_unit_size): Likewise.
448 (unit_unit_precision): Likewise.
449 (mode_wider): Likewise.
450 (mode_2xwider): Likewise.
451 (machine_mode): New class.
452 (mode_to_bytes): New function.
453 (mode_to_bits): Likewise.
454 (mode_to_precision): Likewise.
455 (mode_to_inner): Likewise.
456 (mode_to_unit_size): Likewise.
457 (mode_to_unit_precision): Likewise.
458 (mode_to_nunits): Likewise.
459 (GET_MODE_SIZE): Use mode_to_bytes.
460 (GET_MODE_BITSIZE): Use mode_to_bits.
461 (GET_MODE_PRECISION): Use mode_to_precision.
462 (GET_MODE_INNER): Use mode_to_inner.
463 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
464 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
465 (GET_MODE_NUNITS): Use mode_to_nunits.
466 * system.h (ALWAYS_INLINE): New macro.
467 * config/powerpcspe/powerpcspe-c.c
468 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
469 int for arg1_mode and arg2_mode.
471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
472 Alan Hayward <alan.hayward@arm.com>
473 David Sherwood <david.sherwood@arm.com>
475 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
476 Prefix mode names with E_ in case statements.
477 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
478 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
479 (aarch64_split_simd_move): Likewise.
480 (aarch64_gen_storewb_pair): Likewise.
481 (aarch64_gen_loadwb_pair): Likewise.
482 (aarch64_gen_store_pair): Likewise.
483 (aarch64_gen_load_pair): Likewise.
484 (aarch64_get_condition_code_1): Likewise.
485 (aarch64_constant_pool_reload_icode): Likewise.
486 (get_rsqrte_type): Likewise.
487 (get_rsqrts_type): Likewise.
488 (get_recpe_type): Likewise.
489 (get_recps_type): Likewise.
490 (aarch64_gimplify_va_arg_expr): Likewise.
491 (aarch64_simd_container_mode): Likewise.
492 (aarch64_emit_load_exclusive): Likewise.
493 (aarch64_emit_store_exclusive): Likewise.
494 (aarch64_expand_compare_and_swap): Likewise.
495 (aarch64_gen_atomic_cas): Likewise.
496 (aarch64_emit_bic): Likewise.
497 (aarch64_emit_atomic_swap): Likewise.
498 (aarch64_emit_atomic_load_op): Likewise.
499 (aarch64_evpc_trn): Likewise.
500 (aarch64_evpc_uzp): Likewise.
501 (aarch64_evpc_zip): Likewise.
502 (aarch64_evpc_ext): Likewise.
503 (aarch64_evpc_rev): Likewise.
504 (aarch64_evpc_dup): Likewise.
505 (aarch64_gen_ccmp_first): Likewise.
506 (aarch64_gen_ccmp_next): Likewise.
507 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
508 (alpha_emit_xfloating_libcall): Likewise.
509 (emit_insxl): Likewise.
510 (alpha_arg_type): Likewise.
511 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
512 (arc_preferred_simd_mode): Likewise.
513 (arc_secondary_reload): Likewise.
514 (get_arc_condition_code): Likewise.
515 (arc_print_operand): Likewise.
516 (arc_legitimate_constant_p): Likewise.
517 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
518 * config/arc/arc.md (casesi_load): Likewise.
519 (casesi_compact_jump): Likewise.
520 * config/arc/predicates.md (proper_comparison_operator): Likewise.
521 (cc_use_register): Likewise.
522 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
523 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
524 (arm_init_iwmmxt_builtins): Likewise.
525 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
526 (neon_expand_vector_init): Likewise.
527 (arm_attr_length_move_neon): Likewise.
528 (maybe_get_arm_condition_code): Likewise.
529 (arm_emit_vector_const): Likewise.
530 (arm_preferred_simd_mode): Likewise.
531 (arm_output_iwmmxt_tinsr): Likewise.
532 (thumb1_output_casesi): Likewise.
533 (thumb2_output_casesi): Likewise.
534 (arm_emit_load_exclusive): Likewise.
535 (arm_emit_store_exclusive): Likewise.
536 (arm_expand_compare_and_swap): Likewise.
537 (arm_evpc_neon_vuzp): Likewise.
538 (arm_evpc_neon_vzip): Likewise.
539 (arm_evpc_neon_vrev): Likewise.
540 (arm_evpc_neon_vtrn): Likewise.
541 (arm_evpc_neon_vext): Likewise.
542 (arm_validize_comparison): Likewise.
543 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
544 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
545 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
546 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
547 (c6x_preferred_simd_mode): Likewise.
548 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
549 (epiphany_rtx_costs): Likewise.
550 * config/epiphany/predicates.md (proper_comparison_operator):
552 * config/frv/frv.c (condexec_memory_operand): Likewise.
553 (frv_emit_move): Likewise.
554 (output_move_single): Likewise.
555 (output_condmove_single): Likewise.
556 (frv_hard_regno_mode_ok): Likewise.
557 (frv_matching_accg_mode): Likewise.
558 * config/h8300/h8300.c (split_adds_subs): Likewise.
559 (h8300_rtx_costs): Likewise.
560 (h8300_print_operand): Likewise.
561 (compute_mov_length): Likewise.
562 (output_logical_op): Likewise.
563 (compute_logical_op_length): Likewise.
564 (compute_logical_op_cc): Likewise.
565 (h8300_shift_needs_scratch_p): Likewise.
566 (output_a_shift): Likewise.
567 (compute_a_shift_length): Likewise.
568 (compute_a_shift_cc): Likewise.
569 (expand_a_rotate): Likewise.
570 (output_a_rotate): Likewise.
571 * config/i386/i386.c (classify_argument): Likewise.
572 (function_arg_advance_32): Likewise.
573 (function_arg_32): Likewise.
574 (function_arg_64): Likewise.
575 (function_value_64): Likewise.
576 (ix86_gimplify_va_arg): Likewise.
577 (ix86_legitimate_constant_p): Likewise.
578 (put_condition_code): Likewise.
579 (split_double_mode): Likewise.
580 (ix86_avx256_split_vector_move_misalign): Likewise.
581 (ix86_expand_vector_logical_operator): Likewise.
582 (ix86_split_idivmod): Likewise.
583 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
584 (ix86_build_const_vector): Likewise.
585 (ix86_build_signbit_mask): Likewise.
586 (ix86_match_ccmode): Likewise.
587 (ix86_cc_modes_compatible): Likewise.
588 (ix86_expand_branch): Likewise.
589 (ix86_expand_sse_cmp): Likewise.
590 (ix86_expand_sse_movcc): Likewise.
591 (ix86_expand_int_sse_cmp): Likewise.
592 (ix86_expand_vec_perm_vpermi2): Likewise.
593 (ix86_expand_vec_perm): Likewise.
594 (ix86_expand_sse_unpack): Likewise.
595 (ix86_expand_int_addcc): Likewise.
596 (ix86_split_to_parts): Likewise.
597 (ix86_vectorize_builtin_gather): Likewise.
598 (ix86_vectorize_builtin_scatter): Likewise.
599 (avx_vpermilp_parallel): Likewise.
600 (inline_memory_move_cost): Likewise.
601 (ix86_tieable_integer_mode_p): Likewise.
602 (x86_maybe_negate_const_int): Likewise.
603 (ix86_expand_vector_init_duplicate): Likewise.
604 (ix86_expand_vector_init_one_nonzero): Likewise.
605 (ix86_expand_vector_init_one_var): Likewise.
606 (ix86_expand_vector_init_concat): Likewise.
607 (ix86_expand_vector_init_interleave): Likewise.
608 (ix86_expand_vector_init_general): Likewise.
609 (ix86_expand_vector_set): Likewise.
610 (ix86_expand_vector_extract): Likewise.
611 (emit_reduc_half): Likewise.
612 (ix86_emit_i387_round): Likewise.
613 (ix86_mangle_type): Likewise.
614 (ix86_expand_round_sse4): Likewise.
615 (expand_vec_perm_blend): Likewise.
616 (canonicalize_vector_int_perm): Likewise.
617 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
618 (expand_vec_perm_1): Likewise.
619 (expand_vec_perm_interleave3): Likewise.
620 (expand_vec_perm_even_odd_pack): Likewise.
621 (expand_vec_perm_even_odd_1): Likewise.
622 (expand_vec_perm_broadcast_1): Likewise.
623 (ix86_vectorize_vec_perm_const_ok): Likewise.
624 (ix86_expand_vecop_qihi): Likewise.
625 (ix86_expand_mul_widen_hilo): Likewise.
626 (ix86_expand_sse2_abs): Likewise.
627 (ix86_expand_pextr): Likewise.
628 (ix86_expand_pinsr): Likewise.
629 (ix86_preferred_simd_mode): Likewise.
630 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
631 * config/i386/sse.md (*andnot<mode>3): Likewise.
632 (<mask_codefor><code><mode>3<mask_name>): Likewise.
633 (*<code><mode>3): Likewise.
634 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
635 (ia64_expand_atomic_op): Likewise.
636 (ia64_arg_type): Likewise.
637 (ia64_mode_to_int): Likewise.
638 (ia64_scalar_mode_supported_p): Likewise.
639 (ia64_vector_mode_supported_p): Likewise.
640 (expand_vec_perm_broadcast): Likewise.
641 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
642 (iq2000_function_arg_advance): Likewise.
643 (iq2000_function_arg): Likewise.
644 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
645 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
646 (m68k_libcall_value): Likewise.
647 (m68k_function_value): Likewise.
648 (sched_attr_op_type): Likewise.
649 * config/mcore/mcore.c (mcore_output_move): Likewise.
650 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
652 (microblaze_function_arg): Likewise.
653 * config/mips/mips.c (mips16_build_call_stub): Likewise.
654 (mips_print_operand): Likewise.
655 (mips_mode_ok_for_mov_fmt_p): Likewise.
656 (mips_vector_mode_supported_p): Likewise.
657 (mips_preferred_simd_mode): Likewise.
658 (mips_expand_vpc_loongson_even_odd): Likewise.
659 (mips_expand_vec_unpack): Likewise.
660 (mips_expand_vi_broadcast): Likewise.
661 (mips_expand_vector_init): Likewise.
662 (mips_expand_vec_reduc): Likewise.
663 (mips_expand_msa_cmp): Likewise.
664 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
665 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
666 (cc_flags_for_mode): Likewise.
667 * config/msp430/msp430.c (msp430_print_operand): Likewise.
668 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
669 (nds32_output_casesi_pc_relative): Likewise.
670 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
671 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
672 (nvptx_gen_unpack): Likewise.
673 (nvptx_gen_pack): Likewise.
674 (nvptx_gen_shuffle): Likewise.
675 (nvptx_gen_wcast): Likewise.
676 (nvptx_preferred_simd_mode): Likewise.
677 * config/pa/pa.c (pa_secondary_reload): Likewise.
678 * config/pa/predicates.md (base14_operand): Likewise.
679 * config/powerpcspe/powerpcspe-c.c
680 (altivec_resolve_overloaded_builtin): Likewise.
681 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
683 (rs6000_preferred_simd_mode): Likewise.
684 (output_vec_const_move): Likewise.
685 (rs6000_expand_vector_extract): Likewise.
686 (rs6000_split_vec_extract_var): Likewise.
687 (reg_offset_addressing_ok_p): Likewise.
688 (rs6000_legitimate_offset_address_p): Likewise.
689 (rs6000_legitimize_address): Likewise.
690 (rs6000_emit_set_const): Likewise.
691 (rs6000_const_vec): Likewise.
692 (rs6000_emit_move): Likewise.
693 (spe_build_register_parallel): Likewise.
694 (rs6000_darwin64_record_arg_recurse): Likewise.
695 (swap_selector_for_mode): Likewise.
696 (spe_init_builtins): Likewise.
697 (paired_init_builtins): Likewise.
698 (altivec_init_builtins): Likewise.
699 (do_load_for_compare): Likewise.
700 (rs6000_generate_compare): Likewise.
701 (rs6000_expand_float128_convert): Likewise.
702 (emit_load_locked): Likewise.
703 (emit_store_conditional): Likewise.
704 (rs6000_output_function_epilogue): Likewise.
705 (rs6000_handle_altivec_attribute): Likewise.
706 (rs6000_function_value): Likewise.
707 (emit_fusion_gpr_load): Likewise.
708 (emit_fusion_p9_load): Likewise.
709 (emit_fusion_p9_store): Likewise.
710 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
711 (fusion_gpr_mem_load): Likewise.
712 (fusion_addis_mem_combo_load): Likewise.
713 (fusion_addis_mem_combo_store): Likewise.
714 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
715 (fusion_gpr_mem_load): Likewise.
716 (fusion_addis_mem_combo_load): Likewise.
717 (fusion_addis_mem_combo_store): Likewise.
718 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
720 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
721 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
722 (rs6000_preferred_simd_mode): Likewise.
723 (output_vec_const_move): Likewise.
724 (rs6000_expand_vector_extract): Likewise.
725 (rs6000_split_vec_extract_var): Likewise.
726 (reg_offset_addressing_ok_p): Likewise.
727 (rs6000_legitimate_offset_address_p): Likewise.
728 (rs6000_legitimize_address): Likewise.
729 (rs6000_emit_set_const): Likewise.
730 (rs6000_const_vec): Likewise.
731 (rs6000_emit_move): Likewise.
732 (rs6000_darwin64_record_arg_recurse): Likewise.
733 (swap_selector_for_mode): Likewise.
734 (paired_init_builtins): Likewise.
735 (altivec_init_builtins): Likewise.
736 (rs6000_expand_float128_convert): Likewise.
737 (emit_load_locked): Likewise.
738 (emit_store_conditional): Likewise.
739 (rs6000_output_function_epilogue): Likewise.
740 (rs6000_handle_altivec_attribute): Likewise.
741 (rs6000_function_value): Likewise.
742 (emit_fusion_gpr_load): Likewise.
743 (emit_fusion_p9_load): Likewise.
744 (emit_fusion_p9_store): Likewise.
745 * config/rx/rx.c (rx_gen_move_template): Likewise.
746 (flags_from_mode): Likewise.
747 * config/s390/predicates.md (s390_alc_comparison): Likewise.
748 (s390_slb_comparison): Likewise.
749 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
750 (s390_vector_mode_supported_p): Likewise.
751 (s390_cc_modes_compatible): Likewise.
752 (s390_match_ccmode_set): Likewise.
753 (s390_canonicalize_comparison): Likewise.
754 (s390_emit_compare_and_swap): Likewise.
755 (s390_branch_condition_mask): Likewise.
756 (s390_rtx_costs): Likewise.
757 (s390_secondary_reload): Likewise.
758 (__SECONDARY_RELOAD_CASE): Likewise.
759 (s390_expand_cs): Likewise.
760 (s390_preferred_simd_mode): Likewise.
761 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
762 * config/sh/sh.c (sh_print_operand): Likewise.
763 (dump_table): Likewise.
764 (sh_secondary_reload): Likewise.
765 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
766 * config/sh/sh.md (casesi_worker_1): Likewise.
767 (casesi_worker_2): Likewise.
768 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
769 (fcc_comparison_operator): Likewise.
770 * config/sparc/sparc.c (sparc_expand_move): Likewise.
771 (emit_soft_tfmode_cvt): Likewise.
772 (sparc_preferred_simd_mode): Likewise.
773 (output_cbranch): Likewise.
774 (sparc_print_operand): Likewise.
775 (sparc_expand_vec_perm_bmask): Likewise.
776 (vector_init_bshuffle): Likewise.
777 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
778 (spu_vector_mode_supported_p): Likewise.
779 (spu_expand_insv): Likewise.
780 (spu_emit_branch_or_set): Likewise.
781 (spu_handle_vector_attribute): Likewise.
782 (spu_builtin_splats): Likewise.
783 (spu_builtin_extract): Likewise.
784 (spu_builtin_promote): Likewise.
785 (spu_expand_sign_extend): Likewise.
786 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
787 (tilegx_simd_int): Likewise.
788 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
789 (tilepro_simd_int): Likewise.
790 * config/v850/v850.c (const_double_split): Likewise.
791 (v850_print_operand): Likewise.
792 (ep_memory_offset): Likewise.
793 * config/vax/vax.c (vax_rtx_costs): Likewise.
794 (vax_output_int_move): Likewise.
795 (vax_output_int_add): Likewise.
796 (vax_output_int_subtract): Likewise.
797 * config/visium/predicates.md (visium_branch_operator): Likewise.
798 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
799 (visium_print_operand_address): Likewise.
800 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
801 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
802 (xtensa_expand_conditional_branch): Likewise.
803 (xtensa_copy_incoming_a7): Likewise.
804 (xtensa_output_literal): Likewise.
805 * dfp.c (decimal_real_maxval): Likewise.
806 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
809 Alan Hayward <alan.hayward@arm.com>
810 David Sherwood <david.sherwood@arm.com>
812 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
813 (mode_nunits_inline): Likewise.
814 (mode_inner_inline): Likewise.
815 (mode_unit_size_inline): Likewise.
816 (mode_unit_precision_inline): Likewise.
817 (emit_insn_modes_h): Likewise. Also emit a #define of the
819 (emit_mode_wider): Add an E_ prefix to mode names.
820 (emit_mode_complex): Likewise.
821 (emit_mode_inner): Likewise.
822 (emit_mode_adjustments): Likewise.
823 (emit_mode_int_n): Likewise.
824 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
825 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
826 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
827 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
828 (CRC32_BUILTIN, ENTRY): Likewise.
829 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
830 (aarch64_pop_regs): Likewise.
831 (aarch64_process_components): Likewise.
832 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
833 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
834 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
835 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
836 * config/arm/arm.c (arm_init_libfuncs): Likewise.
837 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
839 * config/i386/i386-builtin.def (pcmpestr): Likewise.
840 (pcmpistr): Likewise.
841 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
842 * config/mmix/mmix.c (mmix_output_condition): Likewise.
843 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
845 * config/rl78/rl78.c (mduc_regs): Likewise.
846 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
847 (htm_expand_builtin): Likewise.
848 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
849 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
850 * config/xtensa/xtensa.c (print_operand): Likewise.
851 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
852 (NUM_MODE_VECTOR_INT): Likewise.
853 * genoutput.c (null_operand): Likewise.
854 (output_operand_data): Likewise.
855 * genrecog.c (print_parameter_value): Likewise.
856 * lra.c (debug_operand_data): Likewise.
858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
860 * dwarf2out.c (scompare_loc_descriptor_wide)
861 (scompare_loc_descriptor_narrow): New functions, split out from...
862 (scompare_loc_descriptor): ...here.
863 * expmed.c (emit_store_flag_int): New function, split out from...
864 (emit_store_flag): ...here.
866 2017-08-30 Richard Biener <rguenther@suse.de>
868 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
869 (dwarf2out_early_finish): Move setting of AT_pubnames from
870 early debug output to early finish.
872 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
874 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
875 and -mdata-region to the assembler.
877 2017-08-30 Richard Biener <rguenther@suse.de>
879 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
881 (gen_subprogram_die): Add DW_AT_object_pointer only early.
882 (dwarf2out_early_global_decl): Only generate a DIE for the
883 abstract origin if it doesn't already exist or is a declaration DIE.
884 (resolve_addr): Do not add the linkage name twice when
885 generating a stub DIE for the DW_TAG_GNU_call_site target.
887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
889 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
890 Use machine_mode rather than int for arg1_mode.
892 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
895 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
896 that the second argument of the built-in functions to unpack
897 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
898 switch statement instead a lot of if statements.
899 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
900 Allow 64-bit values to be in Altivec registers as well as
901 traditional floating point registers.
902 (pack<mode>, FMOVE128_VSX iterator): Likewise.
904 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
906 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
907 MAX_REGS_PER_ADDRESS == 1.
909 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
911 * config/i386/i386.opt (flag_fentry): Do not init to -1.
912 * config/i386/i386.c (ix86_option_override_internal): Simplify
913 setting of opts->x_flag_entry.
915 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
916 Jakub Jelinek <jakub@redhat.com>
917 Richard Biener <rguenther@suse.de>
919 PR tree-optimization/81503
920 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
921 folded constant fits in the target type; reorder tests for clarity.
923 2017-08-29 Martin Liska <mliska@suse.cz>
925 * passes.def: Include pass_lower_switch.
926 * stmt.c (dump_case_nodes): Remove and move to
927 tree-switch-conversion.
928 (case_values_threshold): Likewise.
929 (expand_switch_as_decision_tree_p): Likewise.
930 (emit_case_decision_tree): Likewise.
931 (expand_case): Likewise.
932 (balance_case_nodes): Likewise.
933 (node_has_low_bound): Likewise.
934 (node_has_high_bound): Likewise.
935 (node_is_bounded): Likewise.
936 (emit_case_nodes): Likewise.
937 (struct simple_case_node): New struct.
938 (add_case_node): Remove.
939 (emit_case_dispatch_table): Use vector instead of case_list.
940 (reset_out_edges_aux): Remove.
941 (compute_cases_per_edge): Likewise.
942 (expand_case): Build list of simple_case_node.
943 (expand_sjlj_dispatch_table): Use it.
944 * tree-switch-conversion.c (struct case_node): Moved from
946 (emit_case_nodes): Likewise.
947 (node_has_low_bound): Likewise.
948 (node_has_high_bound): Likewise.
949 (node_is_bounded): Likewise.
950 (case_values_threshold): Likewise.
951 (reset_out_edges_aux): Likewise.
952 (compute_cases_per_edge): Likewise.
953 (add_case_node): Likewise.
954 (dump_case_nodes): Likewise.
955 (balance_case_nodes): Likewise.
956 (expand_switch_as_decision_tree_p): Likewise.
957 (emit_jump): Likewise.
958 (emit_case_decision_tree): Likewise.
959 (try_switch_expansion): Likewise.
960 (do_jump_if_equal): Likewise.
961 (emit_cmp_and_jump_insns): Likewise.
962 (fix_phi_operands_for_edge): New function.
963 (record_phi_operand_mapping): Likewise.
964 (class pass_lower_switch): New pass.
965 (pass_lower_switch::execute): New function.
966 (make_pass_lower_switch): Likewise.
967 (conditional_probability):
968 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
969 * tree-pass.h: Add make_pass_lower_switch.
971 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
974 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
977 2017-08-29 Richard Biener <rguenther@suse.de>
979 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
980 we do not add a DW_AT_inline attribute twice.
981 (gen_subprogram_die): Remove code setting DW_AT_inline on
982 DECL_ABSTRACT_P nodes.
984 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
986 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
987 calls to internal functions.
988 (gimplify_modify_expr): Likewise.
989 * tree-call-cdce.c (use_internal_fn): Likewise.
990 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
991 (convert_to_divmod): Set the nothrow flag.
992 * tree-if-conv.c (predicate_mem_writes): Likewise.
993 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
994 (vectorizable_call): Likewise.
995 (vectorizable_store): Likewise.
996 (vectorizable_load): Likewise.
997 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
998 (vect_recog_mask_conversion_pattern): Likewise.
1000 2017-08-29 Martin Liska <mliska@suse.cz>
1003 * gcc.c (driver_handle_option): Add new argument.
1004 * opts-common.c (handle_option): Pass
1005 target_option_override_hook.
1006 * opts-global.c (lang_handle_option): Add new option.
1007 (set_default_handlers): Add new argument.
1008 (decode_options): Likewise.
1009 * opts.c (target_handle_option): Likewise.
1010 (common_handle_option): Call target_option_override_hook.
1011 * opts.h (struct cl_option_handler_func): Add hook for
1012 target option override.
1013 (struct cl_option_handlers): Likewise.
1014 (set_default_handlers): Add new argument.
1015 (decode_options): Likewise.
1016 (common_handle_option): Likewise.
1017 (target_handle_option): Likewise.
1018 * toplev.c (toplev::main): Pass targetm.target_option.override
1021 2017-08-29 Richard Biener <rguenther@suse.de>
1022 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1024 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1025 life to the active subtree.
1027 2017-08-28 Jeff Law <law@redhat.com>
1029 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1030 derive_equivalences.
1031 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1033 (edge_info::derive_equivalences): New private member function
1035 * tree-ssa-dom.c (class edge_info): Changed from a struct
1036 to a class. Add ctor/dtor, methods and data members.
1037 (edge_info::edge_info): Renamed from allocate_edge_info.
1038 Initialize additional members.
1039 (edge_info::~edge_info): New.
1040 (free_dom_edge_info): Delete the edge info.
1041 (record_edge_info): Use new class & associated member functions.
1042 Tighten forms for testing for edge equivalences.
1043 (record_temporary_equivalences): Iterate over the simple
1044 equivalences rather than assuming there's only one per edge.
1045 (cprop_into_successor_phis): Iterate over the simple
1046 equivalences rather than assuming there's only one per edge.
1047 (optimize_stmt): Use operand_equal_p rather than pointer
1048 equality for mini-DSE code.
1050 2017-08-28 Nathan Sidwell <nathan@acm.org>
1052 * gcc.c (execute): Fold SIGPIPE handling into switch
1053 statement. Adjust internal error message.
1055 2017-08-28 Richard Biener <rguenther@suse.de>
1058 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1059 Do nothing for removed DIEs.
1061 2017-08-28 Richard Biener <rguenther@suse.de>
1063 PR tree-optimization/81977
1064 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1067 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1070 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1071 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1072 using targetm.gen_mem_thread_fence.
1074 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1077 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1078 predicate to register_operand. Reorder operands.
1079 (*btr<mode>): Ditto.
1080 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1081 (*btr<mode>_mask): Ditto.
1083 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1085 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1086 * config/rs6000/xmmintrin.h: New file.
1087 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1089 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1092 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1093 parameter and_insn and return it.
1094 (recombine_lvx_pattern): Insert a copy to ensure availability of
1095 the base register of the copied masking operation at the point of
1096 the instruction replacement.
1097 (recombine_stvx_pattern): Likewise.
1099 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1101 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1102 undocumented switches.
1103 (-mpower9-dform-vector): Likewise.
1104 (-mpower9-dform): Likewise.
1105 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1106 comments to delete references to -mpower9-dform* switches.
1107 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1108 Delete reference to -mpower9-dform* switches, test for
1109 -mpower9-vector instead.
1110 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1111 (OTHER_P9_VECTOR_MASKS): Likewise.
1112 (POWERPC_MASKS): Likewise.
1113 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1114 tests against -mpower9-dform* to -mpower9-vector. Delete code
1115 that checked for -mpower9-dform* consistancy with other options.
1116 Add test for -mpower9-misc to enable other power9 switches.
1117 (rs6000_init_hard_regno_mode_ok): Likewise.
1118 (rs6000_option_override_internal): Likewise.
1119 (rs6000_emit_prologue): Likewise.
1120 (rs6000_emit_epilogue): Likewise.
1121 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1122 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1123 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1125 (emit_fusion_p9_store): Likewise.
1126 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1127 resetting these macros if the assembler does not support ISA 3.0
1129 (TARGET_P9_DFORM_VECTOR): Likewise.
1130 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1131 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1133 2017-08-25 Alan Modra <amodra@gmail.com>
1135 PR rtl-optimization/81747
1136 * cse.c (cse_extended_basic_block): Don't attempt to record
1137 equivalences for degenerate conditional branches that branch
1138 to their fall-through.
1140 2017-08-24 Martin Sebor <msebor@redhat.com>
1143 * gimple-fold.c (size_must_be_zero_p): New function.
1144 (gimple_fold_builtin_memory_op): Call it.
1146 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1148 * config/rs6000/mm_malloc.h: New file.
1150 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1152 PR tree-optimization/81913
1153 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1154 analysis when either IVs in condition can wrap.
1156 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1158 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1159 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1161 2017-08-24 Richard Biener <rguenther@suse.de>
1164 * targhooks.c (default_target_can_inline_p): Properly
1165 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1166 is present and always compare.
1167 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1168 infer -mfpmath=sse from TARGET_SSE_P.
1169 (ix86_can_inline_p): Properly use target_option_default_node when
1170 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1172 2017-08-24 Richard Biener <rguenther@suse.de>
1175 * dwarf2out.c (output_die): Handle flag_generate_offload like
1177 (output_comp_unit): Likewise.
1178 (gen_array_type_die): Likewise.
1179 (dwarf2out_early_finish): Likewise.
1180 (note_variable_value_in_expr): Likewise.
1181 (dwarf2out_finish): Likewise. Adjust assert.
1182 * cgraphunit.c (symbol_table::compile): Move setting of
1183 flag_generate_offload earlier ...
1184 (symbol_table::finalize_compilation_unit): ... here, before
1185 early debug finalization.
1187 2017-08-24 Richard Biener <rguenther@suse.de>
1189 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1190 and ipa-fnsummary.h.
1191 (ix86_can_inline_p): When ix86_fpmath flags do not match
1192 check whether the callee uses FP math at all.
1194 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1197 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1198 instead of TYPE_PRECISION.
1200 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1201 Alan Hayward <alan.hayward@arm.com>
1202 David Sherwood <david.sherwood@arm.com>
1204 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1206 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1208 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1210 (ix86_frame::stack_realign_allocate): New field.
1211 (struct machine_frame_state): Modify comments.
1212 (machine_frame_state::sp_realigned_fp_end): New field.
1213 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1215 (sp_valid_at): Add assertion to assure no attempt to access invalid
1216 offset of a realigned stack.
1217 (fp_valid_at): Likewise.
1218 (choose_baseaddr): Modify comments.
1219 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1220 ix86_expand_prologue.
1221 (ix86_expand_prologue): Modify stack realignment and allocation.
1222 (ix86_expand_epilogue): Modify comments.
1223 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1224 avx2_runtime, avx512f, and avx512f_runtime.
1226 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1228 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1229 (mstackrealign): Do not init to -1.
1230 * config/i386/i386.c (ix86_option_override_internal):
1231 Check opts_set, not opts when setting default value of
1232 opts->x_ix86_force_align_arg_pointer.
1234 2017-08-23 Richard Biener <rguenther@suse.de>
1236 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1237 lang_hooks.decl_printable_name.
1238 * print-rtl-function.c (print_rtx_function): Likewise.
1239 * tree-pretty-print.c (dump_function_header): Likewise.
1241 2017-08-23 Richard Biener <rguenther@suse.de>
1244 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1245 -g0 at compile-time.
1247 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1250 * doc/sourcebuild.texi (Other hardware attributes):
1253 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1256 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1257 Fix single-use check.
1259 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1261 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1263 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1265 * doc/install.texi: Modify to add more details on running selected
1268 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1270 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1271 is combined with -mabi=ms.
1272 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1275 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1277 PR tree-optimization/81488
1278 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1279 and cached_basis fields.
1280 (MAX_SPREAD): New constant.
1281 (alloc_cand_and_find_basis): Initialize new fields.
1282 (clear_visited): New function.
1283 (create_phi_basis_1): Rename from create_phi_basis, set visited
1284 and cached_basis fields.
1285 (create_phi_basis): New wrapper function.
1286 (phi_add_costs_1): Rename from phi_add_costs, add spread
1287 parameter, set visited field, short-circuit when limits reached.
1288 (phi_add_costs): New wrapper function.
1289 (record_phi_increments_1): Rename from record_phi_increments, set
1291 (record_phi_increments): New wrapper function.
1292 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1293 (phi_incr_cost): New wrapper function.
1294 (all_phi_incrs_profitable_1): Rename from
1295 all_phi_incrs_profitable, set visited field.
1296 (all_phi_incrs_profitable): New wrapper function.
1298 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1299 Alan Hayward <alan.hayward@arm.com>
1300 David Sherwood <david.sherwood@arm.com>
1302 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1303 that takes the outer and inner modes.
1304 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1305 comparison as the canonical test for a paradoxical subreg.
1306 * combine.c (simplify_set): Use paradoxical_subreg_p.
1307 (make_extraction): Likewise.
1308 (force_to_mode): Likewise.
1309 (rtx_equal_for_field_assignment_p): Likewise.
1310 (gen_lowpart_for_combine): Likewise.
1311 (simplify_comparison): Likewise.
1312 * cse.c (equiv_constant): Likewise.
1313 * expmed.c (store_bit_field_1): Likewise.
1314 * final.c (alter_subreg): Likewise.
1315 * fwprop.c (propagate_rtx): Likewise.
1316 (forward_propagate_subreg): Likewise.
1317 * ira-conflicts.c (ira_build_conflicts): Likewise.
1318 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1319 * lra-constraints.c (curr_insn_transform): Likewise.
1320 (split_reg): Likewise.
1321 * lra-eliminations.c (move_plus_up): Likewise.
1322 (lra_eliminate_regs_1): Likewise.
1323 * recog.c (general_operand): Likewise.
1324 * ree.c (combine_reaching_defs): Likewise.
1325 * reload.c (push_reload): Likewise.
1326 (find_reloads): Likewise.
1327 * reload1.c (elimination_effects): Likewise.
1328 (compute_reload_subreg_offset): Likewise.
1329 (choose_reload_regs): Likewise.
1330 * rtlanal.c (subreg_lsb_1): Likewise.
1331 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1332 (simplify_subreg): Likewise.
1333 * var-tracking.c (track_loc_p): Likewise.
1334 * emit-rtl.c (byte_lowpart_offset): Likewise.
1335 (paradoxical_subreg_p): Delete out-of-line definition.
1337 2017-08-22 Jeff Law <law@redhat.com>
1339 PR tree-optimization/81741
1340 PR tree-optimization/71947
1341 * tree-ssa-dom.c: Include tree-inline.h.
1342 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1343 equivalences if one is more expensive to compute than the other.
1344 * tree-ssa-scopedtables.h (class const_or_copies): Make
1345 record_const_or_copy_raw method private.
1346 (class avail_exprs_stack): New method simplify_binary_operation.
1347 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1348 avail_exprs_stack::simplify_binary_operation as needed.
1349 (avail_exprs_stack::simplify_binary_operation): New function.
1351 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
1353 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
1354 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
1355 (DOT_SYMBOLS): Likewise.
1356 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
1357 (RELOCATABLE_NEEDS_FIXUP): Likewise.
1358 (RS6000_ABI_NAME): Likewise.
1359 (TARGET_CMODEL): Likewise.
1360 (TOC_SECTION_ASM_OP): Likewise.
1361 (SET_CMODEL): New macro.
1362 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
1364 2017-08-22 Richard Biener <rguenther@suse.de>
1366 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
1367 to allow for free-lang-data replacements similar to verify_type_variant.
1369 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1370 Alan Hayward <alan.hayward@arm.com>
1371 David Sherwood <david.sherwood@arm.com>
1373 * config/aarch64/aarch64.md (casesi): Use DImode rather than
1374 VOIDmode for the LABEL_REF.
1376 2017-08-22 Richard Biener <rguenther@suse.de>
1378 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
1380 2017-08-22 Richard Biener <rguenther@suse.de>
1382 * common.opt (feliminate-dwarf2-dups): Ignore.
1383 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
1384 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
1385 same_die_p_wrap, compute_section_prefix,
1386 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
1387 (comdat_symbol_id, comdat_symbol_number): Likewise.
1388 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
1390 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
1391 (output_die): Mark unreachable path unreachable.
1392 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
1393 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
1394 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
1395 (dwarf2out_early_finish): Likewise.
1397 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
1399 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
1401 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
1404 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
1405 not VAR_P. Filter attribute warnings with OPT_Wattributes.
1406 (avr_attribute_table) <io, io_low, address>: Initialize
1407 .decl_required with true.
1409 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
1411 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
1412 undocumented debugging options.
1413 (-mvsx-scalar-double): Likewise.
1414 (-mallow-df-permute): Likewise.
1415 (-mvectorize-builtins): Likewise.
1416 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1417 (rs6000_builtin_vectorized_function): Likewise.
1418 (rs6000_builtin_md_vectorized_function): Likewise.
1419 (rs6000_opt_vars): Likewise.
1421 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
1424 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
1425 (*btrq_imm): Rename from *btrq.
1426 (*btcq_imm): Rename from *btcq.
1427 (btsc): New code attribute.
1428 (*<btsc><mode>): New insn pattern.
1429 (*btr<mode>): Ditto.
1430 (*<btsc><mode>_mask): New insn_and_split pattern.
1431 (*btr<mode>_mask): Ditto.
1433 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1434 Alan Hayward <alan.hayward@arm.com>
1435 David Sherwood <david.sherwood@arm.com>
1437 * function.c (pad_below): Simplify padding calculation.
1439 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1440 Alan Hayward <alan.hayward@arm.com>
1441 David Sherwood <david.sherwood@arm.com>
1443 * target.def (function_prologue): Remove frame size argument.
1444 (function_epilogue): Likewise.
1445 * doc/tm.texi: Regenerate.
1446 * final.c (final_start_function): Update call to function_prologue.
1447 (final_end_function): Update call to function_epilogue.
1448 (default_function_pro_epilogue): Remove frame size argument.
1449 * output.h (default_function_pro_epilogue): Likewise.
1450 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
1451 (arm_output_function_prologue): Likewise.
1452 * config/frv/frv.c (frv_function_prologue): Likewise.
1453 (frv_function_epilogue): Likewise.
1454 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
1455 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
1456 (ia64_output_function_epilogue): Likewise.
1457 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
1458 (m32r_output_function_epilogue): Likewise.
1459 * config/microblaze/microblaze.c (microblaze_function_prologue)
1460 (microblaze_function_epilogue): Likewise.
1461 * config/mips/mips.c (mips_output_function_prologue): Likewise.
1462 (mips_output_function_epilogue): Likewise.
1463 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
1464 (mmix_target_asm_function_epilogue): Likewise.
1465 * config/msp430/msp430.c (msp430_start_function): Likewise.
1466 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
1467 (nds32_asm_function_epilogue): Likewise.
1468 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
1469 * config/pa/pa.c (pa_output_function_prologue): Likewise.
1470 (pa_output_function_epilogue): Likewise.
1471 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
1472 (rs6000_output_function_epilogue): Likewise.
1473 * config/rl78/rl78.c (rl78_start_function): Likewise.
1474 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
1475 (rs6000_output_function_epilogue): Likewise.
1476 * config/rx/rx.c (rx_output_function_prologue): Likewise.
1477 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
1478 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
1479 (sparc_asm_function_epilogue): Likewise.
1481 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1483 * tree.h (type_has_mode_precision_p): New function.
1484 * convert.c (convert_to_integer_1): Use it.
1485 * expr.c (expand_expr_real_2): Likewise.
1486 (expand_expr_real_1): Likewise.
1487 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
1488 * match.pd: Likewise.
1489 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
1490 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
1491 * tree-tailcall.c (process_assignment): Likewise.
1492 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1493 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
1494 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
1495 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1496 (vectorizable_assignment): Likewise.
1497 (vectorizable_shift): Likewise.
1498 (vectorizable_operation): Likewise.
1499 * tree-vrp.c (register_edge_assert_for_2): Likewise.
1501 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
1503 * match.pd: Add pow (C, x) simplification.
1505 2017-08-21 Richard Biener <rguenther@suse.de>
1507 PR tree-optimization/81900
1508 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
1509 for blocks with abnormal predecessors.
1510 (compute_antic): Do not set visited flag prematurely.
1512 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
1515 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
1517 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1519 * stor-layout.h (vector_type_mode): Move to...
1520 * tree.h (vector_type_mode): ...here.
1521 * stor-layout.c (vector_type_mode): Move to...
1522 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
1524 2017-08-21 Richard Biener <rguenther@suse.de>
1526 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
1527 register_external_die hooks.
1528 (debug_false_tree_charstarstar_uhwistar): Declare.
1529 (debug_nothing_tree_charstar_uhwi): Likewise.
1530 * debug.c (do_nothing_debug_hooks): Adjust.
1531 (debug_false_tree_charstarstar_uhwistar): New do nothing.
1532 (debug_nothing_tree_charstar_uhwi): Likewise.
1533 * dbxout.c (dbx_debug_hooks): Adjust.
1534 (xcoff_debug_hooks): Likewise.
1535 * sdbout.c (sdb_debug_hooks): Likewise.
1536 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1537 * dwarf2out.c (macinfo_label_base): New global.
1538 (dwarf2out_register_external_die): New function for the
1539 register_external_die hook.
1540 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
1541 (dwarf2_debug_hooks): Use them.
1542 (dwarf2_lineno_debug_hooks): Adjust.
1543 (struct die_struct): Add with_offset flag.
1544 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
1545 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
1546 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
1547 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
1548 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
1549 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
1550 defining section names for the early LTO debug variants.
1551 (reset_indirect_string): New helper.
1552 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
1553 (print_dw_val): Add support for offsetted symbol references.
1554 (get_ultimate_context): Split out from is_cxx.
1555 (is_cxx): Use get_ultimate_context.
1556 (is_fortran): Add decl overload.
1557 (compute_comp_unit_symbol): Split out worker from
1558 compute_section_prefix.
1559 (compute_section_prefix): Call compute_comp_unit_symbol and
1560 set comdat_type_p here.
1561 (output_die): Skip DIE symbol output for the LTO added one.
1562 Handle DIE symbol references with offset.
1563 (output_comp_unit): Guard section name mangling properly.
1564 For LTO debug sections emit a symbol at the section beginning
1565 which we use to refer to its DIEs.
1566 (add_abstract_origin_attribute): For DIEs registered via
1567 dwarf2out_register_external_die directly refer to the early
1568 DIE rather than indirectly through the shadow one we created.
1569 Remove obsolete call to dwarf2out_abstract_function for
1570 non-function/block origins.
1571 (gen_array_type_die): When generating early LTO debug do
1572 not emit DW_AT_string_length.
1573 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
1574 late when in LTO. As suggested place a gcc_unreachable for
1575 the DECL_ABSTRACT_P case.
1576 (gen_subprogram_die): Avoid another specification DIE
1577 for early built declarations/definitions for the late LTO case.
1578 (gen_variable_die): Add type references for late duplicated VLA dies
1580 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
1581 we have the abstract instance already.
1582 (process_scope_var): Adjust decl DIE contexts in LTO which
1583 first puts them in limbo.
1584 (gen_decl_die): Do not generate type DIEs late apart from
1585 types for VLAs or for decls we do not yet have a DIE. Do not
1586 call dwarf2out_abstract_function late.
1587 (dwarf2out_early_global_decl): Make sure to create DIEs
1588 for abstract instances of a decl first.
1589 (dwarf2out_late_global_decl): Adjust comment.
1590 (output_macinfo_op): With multiple macro sections use
1591 macinfo_label_base to distinguish labels.
1592 (output_macinfo): Likewise. Update macinfo_label_base.
1593 Pass in the line info label.
1594 (note_variable_value_in_expr): When generating LTO resolve
1595 all variable values here by generating DIEs as needed.
1596 (init_sections_and_labels): Add early LTO debug flag parameter
1597 and generate different sections and names if set. Add generation
1598 counter for the labels so we can have multiple of them.
1599 (reset_dies): Helper to allow DIEs to be output multiple times.
1600 (dwarf2out_finish): When outputting DIEs to the fat part of an
1601 LTO object first reset DIEs.
1602 (dwarf2out_early_finish): Output early DIEs when generating LTO.
1603 (modified_type_die): Check for decl_ultimate_origin being self
1605 (gen_type_die_with_usage): Likewise.
1606 (gen_typedef_die): Allow decl_ultimate_origin being self.
1607 (set_decl_abstract_flags): Remove.
1608 (set_block_abstract_flags): Likewise.
1609 (dwarf2out_abstract_function): Treat the early generated DIEs
1610 as the abstract copy and only add DW_AT_inline and
1611 DW_AT_artificial here and call set_decl_origin_self.
1612 If the DIE has an abstract origin don't do anything.
1613 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
1614 if we have none yet (Go fails to build one, PR78628).
1615 (variably_modified_type_p): Prevent endless recursion for Ada
1616 cyclic pointer types.
1617 * lto-streamer-in.c: Include debug.h.
1618 (dref_queue): New global.
1619 (lto_read_tree_1): Stream in DIE references.
1620 (lto_input_tree): Register DIE references.
1621 (input_function): Stream DECL_DEBUG_ARGS.
1622 * lto-streamer-out.c: Include debug.h.
1623 (lto_write_tree_1): Output DIE references.
1624 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
1625 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
1626 (output_function): Stream DECL_DEBUG_ARGS.
1627 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
1628 Stream DECL_ABSTRACT_ORIGIN.
1629 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
1630 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
1631 DECL_CONTEXT for file-scope decls.
1632 * lto-streamer.h (struct dref_entry): Declare.
1633 (dref_queue): Likewise.
1634 * cfgexpand.c (pass_expand::execute): Do not call the
1635 outlining_inline_function hook here.
1636 * lto-wrapper.c (debug_obj): New global.
1637 (tool_cleanup): Unlink it if required.
1638 (debug_objcopy): New function.
1639 (run_gcc): Handle early debug sections in the IL files by
1640 extracting them to separate files, partially linkin them and
1641 feeding the result back as result to the linker.
1642 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
1643 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
1644 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
1645 sections into a separate segment.
1646 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
1648 (darwin_asm_dwarf_section): Likewise.
1649 (darwin_asm_output_dwarf_offset): Likewise.
1650 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
1652 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1653 Alan Hayward <alan.hayward@arm.com>
1654 David Sherwood <david.sherwood@arm.com>
1656 * read-md.h (md_reader::record_potential_iterator_use): Replace
1657 pointer argument with an rtx and an index.
1658 * read-rtl.c (iterator_group::apply_iterator): Likewise.
1659 (apply_mode_iterator): Likewise.
1660 (apply_code_iterator): Likewise.
1661 (apply_int_iterator): Likewise.
1662 (apply_subst_iterator): Likewise.
1663 (record_iterator_use): Likewise.
1664 (record_attribute_use): Likewise.
1665 (md_reader::record_potential_iterator_use): Likewise. Update calls
1666 to record_iterator_use and apply_iterator.
1667 (iterator_use): Replace ptr with x and index.
1668 (attribute_use): Likewise.
1669 (apply_attribute_uses): Update calls to apply_iterator.
1670 (apply_iterators): Likewise. Update initialization of iterator_use.
1671 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
1672 and record_potential_iterator_use.
1673 (rtx_reader::read_rtx_operand): Likewise.
1675 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
1676 Alan Hayward <alan.hayward@arm.com>
1677 David Sherwood <david.sherwood@arm.com>
1679 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
1682 2017-08-21 Richard Biener <rguenther@suse.de>
1685 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
1686 at struct end conservatively when comparing common bases.
1688 2017-08-21 Richard Biener <rguenther@suse.de>
1690 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
1691 (mem_ref_in_stmt): Remove.
1692 (determine_max_movement): Use ref index to get at the reference.
1693 (invariantness_dom_walker::before_dom_children): Deal with
1694 lim data already initialized.
1695 (gather_mem_refs_stmt): Initialize lim data and record ref index.
1697 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
1699 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
1700 (TARGET_ISA_ROUND): Ditto.
1701 (TARGET_ROUND): Ditto.
1702 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
1703 * config/i386/i386.md: Ditto.
1704 * config/i386/sse.md: Ditto.
1705 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
1706 with OPTION_MASK_ISA_SSE4_1.
1708 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
1711 * doc/extend.texi (x86 Built-in Functions): Correct the name of
1712 __builtin_ia32_lzcnt_u16.
1714 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1717 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
1718 (rs6000_set_current_function): Rewrite function to use it.
1720 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
1723 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
1724 and TYPE_WARN_IF_NOT_ALIGN.
1725 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
1726 (handle_warn_if_not_align): New.
1727 (place_union_field): Call handle_warn_if_not_align.
1728 (place_field): Call handle_warn_if_not_align.
1729 Copy TYPE_WARN_IF_NOT_ALIGN.
1730 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
1731 (layout_type): Likewise.
1732 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
1734 (tree_decl_common): Add warn_if_not_align.
1735 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
1736 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
1737 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
1738 (DECL_WARN_IF_NOT_ALIGN): Likewise.
1739 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
1740 * doc/extend.texi: Document warn_if_not_aligned attribute.
1741 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
1743 2017-08-17 Martin Liska <mliska@suse.cz>
1746 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
1747 (get_data_dependence): Use it as pointer type.
1748 (distribute_loop): Likewise.
1750 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1752 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
1753 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
1754 (p8_vmrgow_<mode>_direct): New define_insn.
1755 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
1756 handle endianness for vmrgew and vmrgow permute patterns.
1758 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
1760 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
1761 * config/rs6000/rs6000-cpus.def: Remove comment.
1762 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
1763 (POWERPC_MASKS): Likewise.
1764 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
1765 use of TARGET_VSX_TIMODE.
1766 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
1767 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
1768 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
1769 (rs6000_option_override_internal): Remove dead code.
1770 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
1771 (rs6000_legitimize_reload_address): Likewise.
1772 (rs6000_legitimate_address_p): Likewise.
1773 (rs6000_opt_masks): Delete "vsx-timode".
1774 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
1775 from function comment.
1776 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
1777 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
1778 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
1780 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
1781 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
1782 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
1785 (define_peephole2): Likewise.
1787 2017-08-17 Martin Sebor <msebor@redhat.com>
1790 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
1791 past the end of an array.
1792 (test_pp_format): Add test cases.
1794 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
1796 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
1797 missing ECF_NOTHROW flags.
1799 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
1802 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
1803 operands residing in integer registers.
1804 (*vsx_le_perm_load_<mode>): Likewise.
1805 (*vsx_le_perm_store_<mode>): Likewise.
1806 (define_peephole2): Add peepholes to optimize the above.
1808 2017-08-17 Marek Polacek <polacek@redhat.com>
1811 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
1812 to mimic what shorten_compare did. Change the return type to bool.
1813 (fold_cond_expr_with_comparison): Update call to
1814 operand_equal_for_comparison_p.
1815 (fold_ternary_loc): Likewise.
1817 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
1819 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
1821 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
1823 2017-08-17 Richard Biener <rguenther@suse.de>
1825 * tree-ssa-structalias.c (solve_graph): When propagating
1826 to successors update the graphs succ edges and avoid duplicate work.
1828 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
1831 * config/i386/i386.c (ix86_option_override_internal): Save target
1832 specific options after ix86_stack_protector_guard_reg was changed.
1834 2017-08-17 Richard Biener <rguenther@suse.de>
1836 PR tree-optimization/81827
1837 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
1838 (new_var_info): Initialize it conservatively.
1839 (get_call_vi): Mark register vars.
1840 (new_scalar_tmp_constraint_exp): Likewise.
1841 (handle_rhs_call): Likewise.
1842 (handle_const_call): Likewise.
1843 (create_function_info_for): Likewise.
1844 (solve_constraints): Sort varinfos to separate register from
1845 non-register vars to pack points-to solution bitmaps during
1848 2017-08-17 Marek Polacek <polacek@redhat.com>
1850 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
1852 2017-08-17 Richard Biener <rguenther@suse.de>
1854 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
1855 to true when overflow is undefined and we saturated the result.
1857 2017-08-17 Alan Modra <amodra@gmail.com>
1860 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
1861 Don't use store multiple if only one reg needs saving.
1862 (interesting_frame_related_regno): New function.
1863 (rs6000_frame_related): Don't emit frame info for regs that
1865 (rs6000_emit_epilogue): Likewise.
1867 2017-08-16 Nathan Sidwell <nathan@acm.org>
1869 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
1870 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
1871 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
1872 * tree.c (free_lang_data_in_type): Use else-if chain. Always
1873 clear TYPE_LANG_1. Remove obsolete member-function stripping.
1874 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
1875 (verify_type): Adjust for TYPE_BINFO move.
1876 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
1877 process TYPE_BINFO directly.
1878 (hash_tree): Likewise.
1879 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1881 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1884 2017-08-16 David Malcolm <dmalcolm@redhat.com>
1886 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
1888 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
1891 * config/i386/i386.md (*anddi_1_btr): Change predicates of
1892 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
1893 Add ix86_binary_operator_ok to insn constraint.
1894 (*iordi_1_bts): Ditto.
1895 (*xordi_1_btc): Ditto.
1896 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
1897 Update corresponding peephole2 pattern.
1901 2017-08-16 Bin Cheng <bin.cheng@arm.com>
1903 PR tree-optimization/81832
1904 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
1905 copy loop header which has IFN_LOOP_DIST_ALIAS call.
1907 2017-08-16 Marek Polacek <polacek@redhat.com>
1910 * fold-const.c (fold_indirect_ref_1): Restore original behavior
1911 regarding size_zero_node.
1913 2017-08-16 Martin Liska <mliska@suse.cz>
1916 * config.gcc: Respect previously set extra_objs in case
1919 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
1921 PR tree-optimization/81835
1922 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
1923 the phi SSA_NAME. Check that the condition in a COND_EXPR does
1924 not depend on the phi.
1926 2017-08-16 Alan Modra <amodra@gmail.com>
1928 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
1931 2017-08-16 Alan Modra <amodra@gmail.com>
1933 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
1934 (save_reg_p): ..into this. Update all callers.
1935 (first_reg_to_save): Simplify.
1937 2017-08-16 Alan Modra <amodra@gmail.com>
1939 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
1942 2017-08-15 Joseph Myers <joseph@codesourcery.com>
1946 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
1947 constant count if that count is less than 32.
1949 2017-08-15 Nathan Sidwell <nathan@acm.org>
1951 * gcc.c (execute): Emit friendlier message if inferior is killed
1952 by an external cause.
1954 2017-08-15 Richard Biener <rguenther@suse.de>
1956 PR tree-optimization/81790
1957 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
1958 CONSTRUCTORs from simplifying and VN.
1960 2017-08-14 Martin Sebor <msebor@redhat.com>
1962 * builtin-attrs.def: Add comments.
1964 2017-08-14 Martin Sebor <msebor@redhat.com>
1967 * doc/extend.texi (attribute nonstring): Document new attribute.
1969 2017-08-14 Martin Sebor <msebor@redhat.com>
1972 * tree-diagnostic.c (default_tree_printer): Handle %G.
1973 * gimple-pretty-print.h (percent_G_format): Declare new function.
1974 * gimple-pretty-print.c (percent_G_format): Define.
1975 * tree-pretty-print.c (percent_K_format): Add argument.
1977 2017-08-14 Martin Sebor <msebor@redhat.com>
1979 PR translation/79998
1980 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
1981 Remove a stray space.
1983 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
1986 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
1987 (*iordi_1_bts): Ditto.
1988 (*xordi_1_btc): Ditto.
1990 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1993 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
1994 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1996 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
1997 quoted strings, and make more translator-friendly.
1998 (darwin_rs6000_override_options): Likewise.
1999 (rs6000_option_override_internal): Likewise.
2000 (rs6000_return_in_memory): Fix overlong line.
2001 (init_cmulative_args): Use quoted strings, and make more
2002 translator-friendly.
2003 (rs6000_pass_by_reference): Fix overlong line.
2004 (def_builtin): Use quoted strings.
2005 (altivec_expand_predicate_builtin): Use quoted strings, and make
2006 more translator-friendly.
2007 (htm_expand_builtin): Use quoted strings.
2008 (cpu_expand_builtin): Use quoted strings, and make more
2009 translator-friendly.
2010 (altivec_expand_builtin): Likewise.
2011 (paired_expand_predicate_builtin): Likewise.
2012 (rs6000_invalid_builtin): Likewise.
2013 (builtin_function_type): Use quoted strings.
2014 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2015 more translator-friendly.
2016 (rs6000_trampoline_init): Likewise.
2017 (rs6000_handle_altivec_attribute): Likewise.
2018 (rs6000_inner_target_options): Use quoted strings.
2019 (rs6000_disable_incompatible_switches): Likewise.
2020 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2021 strings, and make more translator-friendly.
2022 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2024 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2026 PR tree-optimization/81799
2027 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2028 cond_expr to simple gimple operand.
2030 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2033 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2035 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2039 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2040 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2041 TARGET_GASISR_PROLOGUES.
2042 * config/avr/avr.c (avr_option_override): Same.
2043 (avr_pass_pre_proep::execute): Same.
2045 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2048 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2049 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2051 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2053 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2054 define_insn_and_split. Split to a memory load from 0 in
2055 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2056 using PTR mode iterator.
2057 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2058 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2059 (*add_tp_<mode>): Redefine as define_insn_and_split.
2060 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2061 address space. Merge with *add_tp_x32 using PTR mode iterator.
2062 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2063 Split to an add with a memory load from 0 in
2064 DEFAULT_TLS_SEG_REG address space.
2066 2017-08-12 Andrew Pinski <apinski@cavium.com>
2068 * config/aarch64/aarch64-option-extensions.def (rdma):
2069 Fix feature string to what Linux prints out in /proc/cpuinfo.
2071 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2074 * dwarf2out.c (modified_type_die): For C typedef types that have
2075 an ultimate origin, process the ultimate origin instead of the
2077 (gen_typedef_die): Assert that input DECLs have no ultimate
2079 (gen_type_die_with_usage): For typedef variants that have an
2080 ultimate origin, just call gen_decl_die on the original DECL.
2081 (process_scope_var): Avoid creating DIEs for local typedefs and
2082 concrete static variables.
2084 2017-08-12 Alan Modra <amodra@gmail.com>
2088 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2089 match gnu-user.h startfile.
2090 (ENDFILE_LINUX_SPEC): Similarly.
2092 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2095 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2097 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2099 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2100 * config/aarch64/aarch64.md (mov<mode>): Change.
2101 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2102 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2103 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2105 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2107 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2108 for storage order barriers.
2110 2017-08-11 Martin Liska <mliska@suse.cz>
2113 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2114 variables of void type.
2116 2017-08-11 Martin Liska <mliska@suse.cz>
2118 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2119 TARGET_SUPPORTS_ALIASES.
2120 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2121 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2122 (optimize_weakref): Likewise.
2123 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2124 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2125 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2127 2017-08-11 Martin Liska <mliska@suse.cz>
2130 * config/i386/i386.c (make_resolver_func): Do complete
2131 refactoring of the function.
2133 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2136 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2137 * config/i386/i386.c (ix86_stack_protect_guard): Use
2138 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2139 * doc/invoke.texi (x86 Options): Document
2140 -mstack-protector-guard-symbol= option.
2142 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2144 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2145 * config/i386/i386.c (ix86_split_stack_guard): New function.
2146 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2147 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2148 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2149 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2150 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2151 (split_stack_space_check): Call ix86_split_stack_guard.
2153 2017-08-10 Martin Sebor <msebor@redhat.com>
2155 * print-tree.c (print_node): Print location using the established
2157 Replace spaces with colons.
2158 (debug_raw, debug): Ditto.
2160 2017-08-10 Martin Sebor <msebor@redhat.com>
2163 * pretty-print.c (pp_format): Correct the handling of %s precision.
2165 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2168 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2170 (ix86_finalize_stack_frame_flags): This. Also clear
2171 frame_pointer_needed if -fno-omit-frame-pointer is used without
2173 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2174 with ix86_finalize_stack_frame_flags.
2175 (ix86_expand_epilogue): Likewise.
2176 (ix86_expand_split_stack_prologue): Likewise.
2177 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2179 2017-08-10 Martin Liska <mliska@suse.cz>
2182 * c-attribs.c (handle_target_attribute):
2183 Report warning for an empty string argument of target attribute.
2185 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2188 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2190 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2191 or DECL_NONLOCAL labels.
2192 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2193 or DECL_NONLOCAL labels here.
2195 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2197 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2198 to indicate when early gimple folding has been disabled.
2199 (rs6000_gimple_fold_builtin): Add debug content.
2200 (rs6000_invalid_builtin): Fix whitespace.
2201 (rs6000_expand_builtin): Fix whitespace.
2202 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2204 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2207 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2208 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2210 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2212 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2213 (qdf24xx): Likewise.
2214 * config/aarch64/aarch64.md: Include falkor.md.
2215 * config/aarch64/falkor.md: New.
2217 2017-08-09 Marek Polacek <polacek@redhat.com>
2220 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2221 * diagnostic.c (emit_diagnostic): Add a comment.
2222 (emit_diagnostic_valist): New function.
2224 2017-08-09 Marek Polacek <polacek@redhat.com>
2227 * input.c (make_location): New overload.
2228 * input.h (make_location): Declare.
2230 2017-08-08 Alan Modra <amodra@gmail.com>
2231 H.J. Lu <hongjiu.lu@intel.com>
2234 * gcc.c (NO_PIE_SPEC): Delete.
2235 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2237 (LINK_PIE_SPEC): ..to here.
2238 (LINK_COMMAND_SPEC): Support -no-pie.
2239 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2240 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2242 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2243 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2244 (ENDFILE_CRTEND_SPEC): Similarly.
2246 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2249 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2250 (mstack-protector-guard-offset=): Ditto.
2251 * config/i386/i386.c (ix86_option_override): Handle
2252 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2254 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2255 ix86_stack_protect_guard_offset variables.
2256 (TARGET_STACK_PROTECT_GUARD): Always define.
2257 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2258 and -mstack-protector-guard-offset= options.
2260 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2262 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2263 boundary case for the last candidate.
2265 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2267 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2268 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2270 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2273 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2274 * config/aarch64/aarch64-builtins.c
2275 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2276 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2277 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2279 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2280 Andrew Pinski <pinskia@gmail.com>
2283 * internal-fn.def (XORSIGN): New.
2284 * optabs.def (xorsign_optab): New.
2285 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2286 (convert_expand_mult_copysign): New.
2287 (pass_optimize_widening_mul::execute): Call
2288 convert_expand_mult_copysign.
2290 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2292 PR tree-optimization/81354
2293 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2294 Insert on edges rather than explicitly creating landing pads.
2295 (analyze_candidates_and_replace): Commit edge inserts.
2297 2017-08-08 Richard Biener <rguenther@suse.de>
2300 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2301 (expand_simple_operations): Also look through ADDR_EXPRs with
2302 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2304 2017-08-08 Richard Biener <rguenther@suse.de>
2306 PR tree-optimization/81723
2307 * tree-vect-slp.c (struct bst_traits): New hash traits.
2308 (bst_fail): New global.
2309 (vect_build_slp_tree_2): New worker, split out from ...
2310 (vect_build_slp_tree): ... this now wrapping it with using
2311 bst_fail set to cache SLP tree build fails. Properly handle
2313 (vect_analyze_slp_instance): Allocate and free bst_fail.
2315 2017-08-08 Martin Liska <mliska@suse.cz>
2318 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2319 LABEL_DECLs that can be from a different function.
2321 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2323 PR tree-optimization/81744
2324 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2325 loop's number of iterations.
2327 2017-08-08 Martin Liska <mliska@suse.cz>
2329 * asan.c: Include header files.
2330 * attribs.c (build_decl_attribute_variant): New function moved
2332 (build_type_attribute_qual_variant): Likewise.
2333 (cmp_attrib_identifiers): Likewise.
2334 (simple_cst_list_equal): Likewise.
2335 (omp_declare_simd_clauses_equal): Likewise.
2336 (attribute_value_equal): Likewise.
2337 (comp_type_attributes): Likewise.
2338 (build_type_attribute_variant): Likewise.
2339 (lookup_ident_attribute): Likewise.
2340 (remove_attribute): Likewise.
2341 (merge_attributes): Likewise.
2342 (merge_type_attributes): Likewise.
2343 (merge_decl_attributes): Likewise.
2344 (merge_dllimport_decl_attributes): Likewise.
2345 (handle_dll_attribute): Likewise.
2346 (attribute_list_equal): Likewise.
2347 (attribute_list_contained): Likewise.
2348 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2349 (lookup_attribute_by_prefix): Likewise.
2350 * bb-reorder.c: Include header files.
2351 * builtins.c: Likewise.
2352 * calls.c: Likewise.
2353 * cfgexpand.c: Likewise.
2354 * cgraph.c: Likewise.
2355 * cgraphunit.c: Likewise.
2356 * convert.c: Likewise.
2357 * dwarf2out.c: Likewise.
2358 * final.c: Likewise.
2359 * fold-const.c: Likewise.
2360 * function.c: Likewise.
2361 * gimple-expr.c: Likewise.
2362 * gimple-fold.c: Likewise.
2363 * gimple-pretty-print.c: Likewise.
2364 * gimple.c: Likewise.
2365 * gimplify.c: Likewise.
2366 * hsa-common.c: Likewise.
2367 * hsa-gen.c: Likewise.
2368 * internal-fn.c: Likewise.
2369 * ipa-chkp.c: Likewise.
2370 * ipa-cp.c: Likewise.
2371 * ipa-devirt.c: Likewise.
2372 * ipa-fnsummary.c: Likewise.
2373 * ipa-inline.c: Likewise.
2374 * ipa-visibility.c: Likewise.
2376 * lto-cgraph.c: Likewise.
2377 * omp-expand.c: Likewise.
2378 * omp-general.c: Likewise.
2379 * omp-low.c: Likewise.
2380 * omp-offload.c: Likewise.
2381 * omp-simd-clone.c: Likewise.
2382 * opts-global.c: Likewise.
2383 * passes.c: Likewise.
2384 * predict.c: Likewise.
2385 * sancov.c: Likewise.
2386 * sanopt.c: Likewise.
2387 * symtab.c: Likewise.
2388 * toplev.c: Likewise.
2389 * trans-mem.c: Likewise.
2390 * tree-chkp.c: Likewise.
2391 * tree-eh.c: Likewise.
2392 * tree-into-ssa.c: Likewise.
2393 * tree-object-size.c: Likewise.
2394 * tree-parloops.c: Likewise.
2395 * tree-profile.c: Likewise.
2396 * tree-ssa-ccp.c: Likewise.
2397 * tree-ssa-live.c: Likewise.
2398 * tree-ssa-loop.c: Likewise.
2399 * tree-ssa-sccvn.c: Likewise.
2400 * tree-ssa-structalias.c: Likewise.
2401 * tree-ssa.c: Likewise.
2402 * tree-streamer-in.c: Likewise.
2403 * tree-vectorizer.c: Likewise.
2404 * tree-vrp.c: Likewise.
2406 * ubsan.c: Likewise.
2407 * varasm.c: Likewise.
2408 * varpool.c: Likewise.
2409 * tree.c: Remove functions moved to attribs.[ch].
2411 * config/aarch64/aarch64.c: Add attrs.h header file.
2412 * config/alpha/alpha.c: Likewise.
2413 * config/arc/arc.c: Likewise.
2414 * config/arm/arm.c: Likewise.
2415 * config/avr/avr.c: Likewise.
2416 * config/bfin/bfin.c: Likewise.
2417 * config/c6x/c6x.c: Likewise.
2418 * config/cr16/cr16.c: Likewise.
2419 * config/cris/cris.c: Likewise.
2420 * config/darwin.c: Likewise.
2421 * config/epiphany/epiphany.c: Likewise.
2422 * config/fr30/fr30.c: Likewise.
2423 * config/frv/frv.c: Likewise.
2424 * config/ft32/ft32.c: Likewise.
2425 * config/h8300/h8300.c: Likewise.
2426 * config/i386/winnt.c: Likewise.
2427 * config/ia64/ia64.c: Likewise.
2428 * config/iq2000/iq2000.c: Likewise.
2429 * config/lm32/lm32.c: Likewise.
2430 * config/m32c/m32c.c: Likewise.
2431 * config/m32r/m32r.c: Likewise.
2432 * config/m68k/m68k.c: Likewise.
2433 * config/mcore/mcore.c: Likewise.
2434 * config/microblaze/microblaze.c: Likewise.
2435 * config/mips/mips.c: Likewise.
2436 * config/mmix/mmix.c: Likewise.
2437 * config/mn10300/mn10300.c: Likewise.
2438 * config/moxie/moxie.c: Likewise.
2439 * config/msp430/msp430.c: Likewise.
2440 * config/nds32/nds32-isr.c: Likewise.
2441 * config/nds32/nds32.c: Likewise.
2442 * config/nios2/nios2.c: Likewise.
2443 * config/nvptx/nvptx.c: Likewise.
2444 * config/pa/pa.c: Likewise.
2445 * config/pdp11/pdp11.c: Likewise.
2446 * config/powerpcspe/powerpcspe.c: Likewise.
2447 * config/riscv/riscv.c: Likewise.
2448 * config/rl78/rl78.c: Likewise.
2449 * config/rx/rx.c: Likewise.
2450 * config/s390/s390.c: Likewise.
2451 * config/sh/sh.c: Likewise.
2452 * config/sol2.c: Likewise.
2453 * config/sparc/sparc.c: Likewise.
2454 * config/spu/spu.c: Likewise.
2455 * config/stormy16/stormy16.c: Likewise.
2456 * config/tilegx/tilegx.c: Likewise.
2457 * config/tilepro/tilepro.c: Likewise.
2458 * config/v850/v850.c: Likewise.
2459 * config/vax/vax.c: Likewise.
2460 * config/visium/visium.c: Likewise.
2461 * config/xtensa/xtensa.c: Likewise.
2463 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2466 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
2467 constraints since the -mupper-regs-* switches have been
2469 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
2470 into a vector from a double word element that was extracted from
2471 another vector, and eliminate extra XXPERMDI instructions.
2472 (vsx_concat_<mode>_2): Likewise.
2473 (vsx_concat_<mode>_3): Likewise.
2474 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
2475 concat to allow optimizing inserts from previous extracts.
2477 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
2479 * config/i386/i386.c (ix86_stack_protect_guard): Generate
2480 memory reference to a SSP offset in TLS address space.
2481 (ix86_print_operand) <case '@'>: Remove.
2482 (ix86_print_operand_punct_valid_p): Remove '@' code.
2483 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
2485 (stack_tls_protect_set_<mode>): Remove.
2486 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
2487 (stack_tls_protect_test_<mode>): Remove.
2488 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
2490 2017-08-07 Olivier Hainque <hainque@adacore.com>
2493 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
2495 2017-08-07 Douglas Rupp <rupp@adacore.com>
2497 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
2498 variable was referenced as multidir in command.
2500 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2503 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
2506 2017-08-07 Martin Liska <mliska@suse.cz>
2508 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
2509 * config/rl78/rl78.c: Add include of attribs.h.
2510 * config/sh/sh.c: Likewise.
2511 * config/v850/v850.c: Likewise.
2513 2017-08-07 Tom de Vries <tom@codesourcery.com>
2516 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
2518 2017-08-07 Martin Liska <mliska@suse.cz>
2520 * config/mips/mips.c: Include attribs.h.
2522 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
2525 * doc/invoke.texi: Document change in behvaior for -Ofast for
2528 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
2530 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
2532 (aarch64_pop_regs): Likewise.
2533 (aarch64_gen_load_pair): Likewise.
2534 (aarch64_save_callee_saves): Likewise.
2535 (aarch64_restore_callee_saves): Likewise.
2537 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2539 * config/i386/i386.c: Revert the last change.
2541 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2544 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2546 (ix86_finalize_stack_frame_flags): This. Also clear
2547 frame_pointer_needed if -fno-omit-frame-pointer is used without
2549 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2550 with ix86_finalize_stack_frame_flags.
2551 (ix86_expand_epilogue): Likewise.
2552 (ix86_expand_split_stack_prologue): Likewise.
2554 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
2557 * config/i386/i386.c (get_builtin_code_for_version): Set priority
2558 to P_AES for Westmere.
2560 2017-08-07 Jonathan Yong <10walls@gmail.com>
2562 * config/i386/mingw.opt (fset-stack-executable): Removed.
2563 * config/i386/cygming.opt (fset-stack-executable): Moved
2565 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
2567 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
2569 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
2571 2017-08-07 Marek Polacek <polacek@redhat.com>
2574 * fold-const.c (fold_indirect_ref_1): Check type_domain.
2576 2017-08-07 Martin Liska <mliska@suse.cz>
2578 * attribs.h (canonicalize_attr_name): New function.
2579 (cmp_attribs): Move from c-format.c and adjusted.
2580 (is_attribute_p): Moved from tree.h.
2581 * tree-inline.c: Add new includes.
2582 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
2583 (private_is_attribute_p): Remove.
2584 (private_lookup_attribute): Likewise.
2585 (private_lookup_attribute_by_prefix): Simplify.
2586 (remove_attribute): Use is_attribute_p.
2587 * tree.h: Remove removed declarations.
2589 2017-08-07 Jakub Jelinek <jakub@redhat.com>
2592 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
2593 instead of computing it in the function. Formatting fix.
2594 (expand_case): Don't rely on default_edge being the first edge,
2595 clear it if removing it, pass default_edge to
2596 emit_case_dispatch_table.
2597 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
2600 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
2602 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
2603 insn in the function, emit NOP after the insn.
2605 2017-08-06 Tom de Vries <tom@codesourcery.com>
2607 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
2610 2017-08-06 Tom de Vries <tom@codesourcery.com>
2612 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
2615 2017-08-04 Yury Gribov <tetra2005@gmail.com>
2617 PR tree-optimization/57371
2618 * match.pd: New pattern.
2620 2017-08-04 Marek Polacek <polacek@redhat.com>
2623 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
2624 perform the computation in offset_int.
2626 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2628 PR tree-optimization/81136
2629 * tree-vectorizer.h: Include tree-hash-traits.h.
2630 (vec_base_alignments): New typedef.
2631 (vec_info): Add a base_alignments field.
2632 (vect_record_base_alignments): Declare.
2633 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
2635 (DR_IS_CONDITIONAL_IN_STMT): New macro.
2636 (create_data_ref): Add an is_conditional_in_stmt argument.
2637 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
2638 the is_conditional_in_stmt field.
2639 (data_ref_loc): Add an is_conditional_in_stmt field.
2640 (get_references_in_stmt): Set the is_conditional_in_stmt field.
2641 (find_data_references_in_stmt): Update call to create_data_ref.
2642 (graphite_find_data_references_in_stmt): Likewise.
2643 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
2644 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
2645 (vect_record_base_alignment): New function.
2646 (vect_record_base_alignments): Likewise.
2647 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
2648 for nested statements even if we fail to compute a misalignment.
2649 Use pooled base alignments for unconditional references.
2650 (vect_find_same_alignment_drs): Compare base addresses instead
2652 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
2653 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
2655 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2657 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
2658 Add an explicit name for the enum. Use auto_vec for slp_instances
2660 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
2662 (_bb_vec_info): Add a constructor and destructor.
2663 (vinfo_for_stmt): Return NULL for uids of -1 as well.
2664 (destroy_loop_vec_info): Delete.
2665 (vect_destroy_datarefs): Likewise.
2666 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
2667 (vec_info::vec_info): New function.
2668 (vec_info::~vec_info): Likewise.
2669 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
2670 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
2671 destroy_loop_vec_info.
2672 * tree-vect-loop.c (new_loop_vec_info): Replace with...
2673 (_loop_vec_info::_loop_vec_info): ...this.
2674 (destroy_loop_vec_info): Replace with...
2675 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
2676 the stmt_vec_infos. Leave handling of vec_info information to its
2677 destructor. Remove explicit vector releases.
2678 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
2679 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
2680 * tree-vect-slp.c (new_bb_vec_info): Replace with...
2681 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
2682 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
2683 (destroy_bb_vec_info): Replace with...
2684 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
2685 information to its destructor.
2686 (vect_slp_analyze_bb_1): Use new and delete instead of
2687 new_bb_vec_info and destroy_bb_vec_info.
2688 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
2691 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
2693 * tree-data-ref.h (subscript): Add access_fn field.
2694 (data_dependence_relation): Add could_be_independent_p.
2695 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
2696 (same_access_functions): Move to tree-data-ref.c.
2697 * tree-data-ref.c (ref_contains_union_access_p): New function.
2698 (access_fn_component_p): Likewise.
2699 (access_fn_components_comparable_p): Likewise.
2700 (dr_analyze_indices): Add a reference to access_fn_component_p.
2701 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
2703 (constant_access_functions): Likewise.
2704 (add_other_self_distances): Likewise.
2705 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
2706 (initialize_data_dependence_relation): Use XCNEW and remove
2707 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
2708 of access functions that have the same type. Allow the
2709 subsequence to end with different bases in some circumstances.
2710 Record the chosen access functions in SUB_ACCESS_FN.
2711 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
2712 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
2713 (subscript_dependence_tester_1): Likewise dra and drb.
2714 (build_classic_dist_vector): Update calls accordingly.
2715 (subscript_dependence_tester): Likewise.
2716 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
2717 DDR_COULD_BE_INDEPENDENT_P.
2718 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
2719 comp_alias_ddrs instead of may_alias_ddrs.
2720 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
2722 (vect_analyze_data_ref_dependence): Use it if
2723 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
2724 distance vectors if that fails.
2725 (dependence_distance_ge_vf): New function.
2726 (vect_prune_runtime_alias_test_list): Use it. Don't clear
2727 LOOP_VINFO_MAY_ALIAS_DDRS.
2729 2017-08-04 Richard Biener <rguenther@suse.de>
2732 * fold-const.c (fold_binary_loc): Properly restrict
2733 minus_var0 && minus_var1 case when associating undefined overflow
2736 2017-08-04 Tom de Vries <tom@codesourcery.com>
2738 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
2740 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2742 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2743 Don't start diagnostic messages with a capital letter.
2744 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2746 (rs6000_invalid_builtin): Likewise.
2747 (rs6000_trampoline_init): Likewise.
2749 2017-08-03 Jakub Jelinek <jakub@redhat.com>
2752 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
2753 after setting changeable df flags.
2755 2017-08-03 Richard Biener <rguenther@suse.de>
2757 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
2758 up if the use is in USE - X.
2760 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
2762 * toplev.c (dumpfile.h): New include.
2763 (internal_error_reentered): New static function. Use it...
2764 (internal_error_function): ...here to handle reentered internal_error.
2766 2017-08-03 Richard Biener <rguenther@suse.de>
2769 * fold-const.c (split_tree): Add minus_var and minus_con
2770 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
2771 here but always use minus_*.
2772 (associate_trees): Assert we never associate with MINUS_EXPR
2773 and NULL first operand. Do not recurse for PLUS_EXPR operands
2774 when associating as MINUS_EXPR either.
2775 (fold_binary_loc): Track minus_var and minus_con.
2777 2017-08-03 Tom de Vries <tom@codesourcery.com>
2780 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
2781 ACCEL_COMPILER, apply finish_options on
2782 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
2784 2017-08-03 Tom de Vries <tom@codesourcery.com>
2787 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
2788 function_entry_patch_area_size > 0.
2790 2017-08-03 Jakub Jelinek <jakub@redhat.com>
2793 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
2794 instead of 10??LU, perform unit multiplication in wide_int,
2795 don't change alloc_object_size_limit if the limit is larger
2798 PR tree-optimization/81655
2799 PR tree-optimization/81588
2800 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
2801 the case when ranges[i].low and high are 1 for unsigned type with
2805 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
2806 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
2808 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2810 * tree-vrp.h: Add include guard.
2812 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
2815 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
2816 (ud2): New insn pattern.
2817 * config/i386/i386.c (ix86_expand_epilogue):
2818 For naked functions, generate ud2 instead of trap insn.
2820 2017-08-02 Marek Polacek <polacek@redhat.com>
2823 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
2825 2017-08-02 Tom de Vries <tom@codesourcery.com>
2826 Cesar Philippidis <cesar@codesourcery.com>
2828 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
2829 Add missing edge probabilities.
2831 2017-08-02 Tamar Christina <tamar.christina@arm.com>
2833 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
2836 2017-08-02 Jakub Jelinek <jakub@redhat.com>
2839 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
2840 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
2841 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
2842 prologue_seq sequences - if any.
2844 2017-08-02 Richard Biener <rguenther@suse.de>
2846 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
2847 via vectors if supported, integer extracts via punning if supported
2848 or otherwise vector extracts.
2850 2017-08-02 Richard Biener <rguenther@suse.de>
2852 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
2854 (bitmap_insert_into_set): ... this.
2856 2017-08-02 Richard Biener <rguenther@suse.de>
2858 PR tree-optimization/81633
2860 2015-08-17 Alan Hayward <alan.hayward@arm.com>
2862 PR tree-optimization/71752
2863 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
2865 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
2867 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
2868 (machine_function::call_ms2sysv_pad_out): Remove field.
2869 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
2870 (ix86_compute_frame_layout): Likewise.
2872 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
2875 * config/i386/i386.c (ix86_set_func_type): Disallow naked
2876 attribute with interrupt attribute.
2878 2017-08-01 Andrew Pinski <apinski@cavium.com>
2880 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
2881 BIT_INSERT_EXPR's operand 1
2882 to see if the types precision matches.
2884 2017-08-01 Martin Liska <mliska@suse.cz>
2887 * builtins.c (expand_builtin_memcpy_args): Remove.
2888 (expand_builtin_memcpy): Call newly added function
2889 expand_builtin_memory_copy_args.
2890 (expand_builtin_memcpy_with_bounds): Likewise.
2891 (expand_builtin_mempcpy): Remove last argument.
2892 (expand_builtin_mempcpy_with_bounds): Likewise.
2893 (expand_builtin_memory_copy_args): New function created from
2894 expand_builtin_mempcpy_args with small modifications.
2895 (expand_builtin_mempcpy_args): Remove.
2896 (expand_builtin_stpcpy): Remove unused argument.
2897 (expand_builtin): Likewise.
2898 (expand_builtin_with_bounds): Likewise.
2900 2017-08-01 Martin Liska <mliska@suse.cz>
2903 Make mempcpy more optimal (PR middle-end/70140).
2905 2017-08-01 Jakub Jelinek <jakub@redhat.com>
2908 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2909 __builtin_vec_cmpne verify both arguments are compatible vectors
2910 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
2911 verify arg1_type is a pointer or array type. For __builtin_vec_st,
2912 move computation of aligned to after checking the argument types.
2916 * config/rs6000/vsx.md (vextract_fp_from_shorth,
2917 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
2920 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
2921 Doug Rupp <rupp@adacore.com>
2922 Olivier Hainque <hainque@adacore.com>
2924 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
2925 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
2927 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
2928 for TARGET_OS_CPP_BUILTIN.
2929 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
2930 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
2932 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
2933 passing required abi options to the assembler for EABI configurations.
2934 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
2935 of .text.hot and .text.unlikely sections for kernel modules when
2936 using ARM style exceptions.
2937 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
2938 options. Add EXTRA_CC1_SPEC.
2939 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
2941 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
2943 (ARM_TARGET2_DWARF_FORMAT): Define.
2944 * config/arm/t-vxworks: Adjust multilib control to removal of the
2945 Diab command line options.
2947 2017-08-01 Martin Liska <mliska@suse.cz>
2949 PR gcov-profile/81561
2950 * gcov.c (unblock): Make unblocking safe as we need to preserve
2951 index correspondence of blocks and block_lists.
2953 2017-08-01 Richard Biener <rguenther@suse.de>
2955 PR tree-optimization/81181
2956 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2957 (compute_antic): ... end of iteration here.
2959 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
2961 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
2962 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
2963 (ftree-slp-vectorize): Likewise.
2964 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
2965 can no longer be set independent of flag_tree_loop_vectorize.
2966 * omp-general.c (emp_max_vf): Likewise.
2967 * opts.c (enable_fdo_optimizations): Remove references to
2968 flag_tree_vectorize, these are now implicit.
2969 (common_handle_option): Remove handling for OPT_ftree_vectorize,
2970 and leave it for the options machinery.
2972 2017-08-01 Martin Liska <mliska@suse.cz>
2975 * builtins.c (expand_builtin_memcpy_args): Remove.
2976 (expand_builtin_memcpy): Call newly added function
2977 expand_builtin_memory_copy_args.
2978 (expand_builtin_memcpy_with_bounds): Likewise.
2979 (expand_builtin_mempcpy): Remove last argument.
2980 (expand_builtin_mempcpy_with_bounds): Likewise.
2981 (expand_builtin_memory_copy_args): New function created from
2982 expand_builtin_mempcpy_args with small modifications.
2983 (expand_builtin_mempcpy_args): Remove.
2984 (expand_builtin_stpcpy): Remove unused argument.
2985 (expand_builtin): Likewise.
2986 (expand_builtin_with_bounds): Likewise.
2988 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
2991 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
2992 print "ds:" only for immediates in generic address space.
2994 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
2997 * config/i386/i386.c (ix86_funciton_naked): New prototype.
2998 (ix86_function_ok_for_sibcall): Return false for naked functions.
3000 2017-08-01 Richard Biener <rguenther@suse.de>
3002 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3003 (compute_antic): Seed worklist with exit block predecessors.
3004 * cfganal.c (dfs_find_deadend): For a cycle return the source
3005 of the edge closing it.
3007 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3009 * config/aarch64/aarch64.c
3010 (aarch64_can_const_movi_rtx_p): Move 0 check.
3012 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3014 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3015 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3017 * match.pd: Ditto in address comparison pattern.
3019 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3021 PR tree-optimization/81627
3022 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3023 closed ssa form for store-store chain.
3025 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3027 PR tree-optimization/81620
3028 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3029 for store-store chain.
3031 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3033 PR tree-optimization/81588
3034 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3035 ranges[i].in_p, invert comparison code ccode. For >/>=,
3036 swap rhs1 and rhs2 and comparison code unconditionally,
3037 for </<= don't do that. Don't swap rhs1/rhs2 again if
3038 ranges[i].in_p, instead invert comparison code ccode if
3039 opcode or oe->rank is BIT_IOR_EXPR.
3042 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3043 a direct optab to conversion optab.
3044 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3045 with GET_MODE_INNER as last argument instead of optab_handler.
3046 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3047 vector extraction if possible and optab is available.
3048 * expr.c (store_constructor): Use convert_optab_handler instead
3049 of optab_handler. Use vector initialization from smaller
3050 vectors if possible and optab is available.
3051 * tree-vect-stmts.c (vectorizable_load): Likewise.
3052 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3054 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3055 of vec_init from half-sized vectors with the same element mode.
3056 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3057 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3058 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3059 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3060 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3061 after mode in gen_vec_extract* calls.
3062 (vec_extract<mode>): Renamed to ...
3063 (vec_extract<mode><ssescalarmodelower>): ... this.
3064 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3065 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3066 element mode after mode in gen_vec_init* calls.
3067 (VEC_INIT_HALF_MODE): New mode iterator.
3068 (vec_init<mode>): Renamed to ...
3069 (vec_init<mode><ssescalarmodelower>): ... this.
3070 (vec_init<mode><ssehalfvecmodelower>): New expander.
3071 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3072 (vec_extractv2sfsf): ... this.
3073 (vec_initv2sf): Renamed to ...
3074 (vec_initv2sfsf): ... this.
3075 (vec_extractv2si): Renamed to ...
3076 (vec_extractv2sisi): ... this.
3077 (vec_initv2si): Renamed to ...
3078 (vec_initv2sisi): ... this.
3079 (vec_extractv4hi): Renamed to ...
3080 (vec_extractv4hihi): ... this.
3081 (vec_initv4hi): Renamed to ...
3082 (vec_initv4hihi): ... this.
3083 (vec_extractv8qi): Renamed to ...
3084 (vec_extractv8qiqi): ... this.
3085 (vec_initv8qi): Renamed to ...
3086 (vec_initv8qiqi): ... this.
3087 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3088 (vec_init<mode>): Renamed to ...
3089 (vec_init<mode><VEC_base_l>): ... this.
3090 (vec_extract<mode>): Renamed to ...
3091 (vec_extract<mode><VEC_base_l>): ... this.
3092 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3093 (vec_initv2sfsf): ... this.
3094 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3095 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3096 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3097 element mode after mode in gen_vec_init* calls.
3098 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3099 (vec_init<mode><Vel>): ... this.
3100 (vec_extract<mode>): Renamed to ...
3101 (vec_extract<mode><Vel>): ... this.
3102 * config/aarch64/iterators.md (Vel): New mode attribute.
3103 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3104 Add element mode after mode in gen_vec_extract* calls.
3105 * config/s390/vector.md (non_vec_l): New mode attribute.
3106 (vec_extract<mode>): Renamed to ...
3107 (vec_extract<mode><non_vec_l>): ... this.
3108 (vec_init<mode>): Renamed to ...
3109 (vec_init<mode><non_vec_l>): ... this.
3110 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3111 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3113 * config/arm/iterators.md (V_elem_l): New mode attribute.
3114 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3115 (vec_extract<mode><V_elem_l>): ... this.
3116 (vec_extractv2di): Renamed to ...
3117 (vec_extractv2didi): ... this.
3118 (vec_init<mode>): Renamed to ...
3119 (vec_init<mode><V_elem_l>): ... this.
3120 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3121 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3122 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3123 Add element mode after gen_vec_extract* calls.
3124 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3125 (vec_init<mode><unitmode>): ... this.
3126 (vec_extract<mode>): Renamed to ...
3127 (vec_extract<mode><unitmode>): ... this.
3128 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3129 (vec_init<mode><unitmode>): ... this.
3130 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3131 (vec_initv2sfsf): ... this.
3132 (vec_extractv2sf): Renamed to ...
3133 (vec_extractv2sfsf): ... this.
3134 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3135 Add element mode after gen_vec_extract* calls.
3136 * config/mips/mips.md (unitmode): New mode iterator.
3137 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3138 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3139 * config/spu/spu.md (inner_l): New mode attribute.
3140 (vec_init<mode>): Renamed to ...
3141 (vec_init<mode><inner_l>): ... this.
3142 (vec_extract<mode>): Renamed to ...
3143 (vec_extract<mode><inner_l>): ... this.
3144 * config/sparc/sparc.md (veltmode): New mode iterator.
3145 (vec_init<VMALL:mode>): Renamed to ...
3146 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3147 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3148 (vec_initv2sisi): ... this.
3149 (vec_initv2sf): Renamed to ...
3150 (vec_initv2sfsf): ... this.
3151 (vec_extractv2sf): Renamed to ...
3152 (vec_extractv2sfsf): ... this.
3153 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3154 (vec_init<mode>): Renamed to ...
3155 (vec_init<mode><VEC_base_l>): ... this.
3156 (vec_extract<mode>): Renamed to ...
3157 (vec_extract<mode><VEC_base_l>): ... this.
3158 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3159 (vec_initv2sfsf): ... this.
3160 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3161 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3162 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3163 gen_vec_init* calls.
3165 2017-08-01 Richard Biener <rguenther@suse.de>
3167 PR tree-optimization/81297
3168 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3169 TREE_OVERFLOW from INTEGER_CSTs.
3171 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3173 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3175 2017-07-31 Carl Love <cel@us.ibm.com>
3177 * config/rs6000/rs6000-c: Add support for built-in functions
3178 vector signed char vec_xl_be (signed long long, signed char *);
3179 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3180 vector signed int vec_xl_be (signed long long, signed int *);
3181 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3182 vector signed long long vec_xl_be (signed long long, signed long long *);
3183 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3184 vector signed short vec_xl_be (signed long long, signed short *);
3185 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3186 vector double vec_xl_be (signed long long, double *);
3187 vector float vec_xl_be (signed long long, float *);
3188 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3189 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3190 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3192 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3193 (altivec_expand_builtin): Add switch statement to call
3194 altivec_expand_xl_be for each builtin.
3195 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3196 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3197 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3198 __builtin_vsx_le_be_v16qi.
3199 * doc/extend.texi: Update the built-in documentation file for the
3200 new built-in functions.
3202 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3205 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3207 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3209 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3211 * config.gcc: Add z14.
3212 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3213 CPU model numbers for z13s and z14.
3214 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3216 * config/s390/s390-opts.h (enum processor_type): Rename
3217 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3218 * config/s390/s390.c (processor_table): Add field for CPU name to
3219 be passed to Binutils.
3220 (s390_asm_output_machine_for_arch): Use the new field in
3221 processor_table for Binutils.
3222 (s390_expand_builtin): Replace arch12 with z14.
3223 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3224 (s390_get_sched_attrmask): Likewise.
3225 (s390_get_unit_mask): Likewise.
3226 * config/s390/s390.opt: Add z14 to processor_type enum.
3228 2017-07-31 Martin Jambor <mjambor@suse.cz>
3231 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3232 regardless of optimization level.
3234 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3235 Martin Liska <mliska@suse.cz>
3237 * predict.def: Remove old comment and adjust probability.
3238 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3241 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3244 * config/i386/i386.c (ix86_function_naked): New function.
3245 (ix86_can_use_return_insn_p): Return false for naked functions.
3246 (ix86_expand_prologue): Skip prologue for naked functions.
3247 (ix86_expand_epilogue): Skip epilogue for naked functions
3248 and emit trap instruction.
3249 (ix86_warn_func_return): New function.
3250 (ix86_attribute_table): Add "naked" attribute specification.
3251 (TARGET_WARN_FUNC_RETURN): Define.
3252 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3254 2017-07-31 Martin Liska <mliska@suse.cz>
3256 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3257 (dump_gimple_bb_header): Always dump BB info.
3258 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3260 2017-07-31 Martin Liska <mliska@suse.cz>
3263 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3264 also with current_function_decl non-null equality.
3266 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3269 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3270 change type to the element type, instead add eltype variable and
3271 use it where we are interested in the element type.
3273 PR tree-optimization/81603
3274 * ipa-polymorphic-call.c
3275 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3276 offset arithmetic in offset_int, bail out if the resulting bit offset
3277 doesn't fit into shwi.
3279 2017-07-31 Martin Liska <mliska@suse.cz>
3281 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3282 (gimplify_save_expr): Fix comment.
3284 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3287 * config/i386/i386.c (ix86_function_arg): Update arguments for
3289 (ix86_compute_frame_layout): Set the initial stack offset to
3290 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3291 INCOMING_FRAME_SP_OFFSET.
3292 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3293 stack before exception handler returns.
3294 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3295 the 'ERROR_CODE' for exception handler.
3297 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3299 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3300 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3301 (ASM_OUTPUT_REG_POP): Ditto.
3302 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3303 instead of asm_fprintf to output pure string.
3305 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3307 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3308 to imported_module_or_decl hook.
3309 (debug_nothing_tree_tree_tree_bool): Remove.
3310 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3311 * debug.c (do_nothing_debug_hooks): Use
3312 debug_nothing_tree_tree_tree_bool_bool instead of
3313 debug_nothing_tree_tree_tree_bool.
3314 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3315 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3316 * sdbout.c (sdb_debug_hooks): Likewise.
3317 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3318 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3320 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3321 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3322 attribute, don't add anything.
3324 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3326 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3327 (fold_build2_stat_loc): Likewise.
3328 (fold_build3_stat_loc): Likewise.
3329 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3330 (fold_build1_loc): Remove macro.
3331 (fold_build2_loc): Likewise.
3332 (fold_build3_loc): Likewise.
3334 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3336 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3337 (gimple_build_debug_bind_source_stat): Likewise.
3338 * gimple.h (gimple_build_debug_bind): Remove macro.
3339 (gimple_build_debug_bind_source): Likewise.
3341 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3343 * bitmap.c (bitmap_alloc): Adjust.
3344 (bitmap_gc_alloc): Likewise.
3345 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3347 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3349 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3350 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3351 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
3352 (bitmap_gc_alloc_stat): Likewise.
3353 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
3355 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3357 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
3358 * rtl.h (shallow_copy_rtx): Remove macro.
3360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3362 * emit-rtl.c (gen_raw_REG): Adjust.
3363 * gengenrtl.c (gendef): Likewise.
3364 * rtl.c (rtx_alloc_stat): Remove _stat from name.
3365 * rtl.h (rtx_alloc): Remove macro.
3367 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3369 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
3370 (build_tree_list_stat): Likewise.
3371 * tree.h (build_tree_list): Remove macro.
3372 (build_tree_list_vec): Likewise.
3374 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3376 * tree.c (make_vector_stat): Remove _stat from name.
3377 (build_vector_stat): Likewise.
3378 * tree.h (make_vector_stat): Remove macro.
3379 (build_vector_stat): Likewise.
3381 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3383 * tree.h (build_var_debug_value): Remove prototype.
3385 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3387 * tree.c (tree_cons_stat): Remove _stat from name.
3388 * tree.h (tree_cons): Remove macro.
3390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3392 * tree.c (build_vl_exp_stat): Remove _stat from name.
3393 * tree.h (build_vl_exp): Remove macro.
3395 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3397 * tree.c (build_decl_stat): Remove _stat from name.
3398 * tree.h (build_decl): Remove macro.
3400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3402 * gimple.c (gimple_build_with_ops_stat): Adjust.
3403 (gimple_alloc_stat): Remove _stat from name.
3404 * gimple.h (gimple_alloc): Remove macro.
3406 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3408 * tree.c (make_tree_vec_stat): Remove _stat from name.
3409 (grow_tree_vec_stat): Likewise.
3410 * tree.h (make_tree_vec_stat): Adjust prototype.
3411 (grow_tree_vec_stat): Likewise.
3412 (make_tree_vec): Remove macro.
3413 (grow_tree_vec): Likewise.
3415 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3417 * fold-const.c (fold_build1_stat_loc): Adjust.
3418 (fold_build2_stat_loc): Likewise.
3419 (fold_build3_stat_loc): Likewise.
3420 * tree.c (build0_stat): Remove _stat from name.
3421 (build1_stat): Likewise.
3422 (build2_stat): Likewise.
3423 (build3_stat): Likewise.
3424 (build4_stat): Likewise.
3425 (build5_stat): Likewise.
3426 * tree.h (build1_loc): Remove macro, and rename _stat function
3428 (build2_loc): Likewise.
3429 (build3_loc): Likewise.
3430 (build4_loc): Likewise.
3431 (build5_loc): Likewise.
3433 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3435 * tree.c (make_int_cst_stat): Remove _stat from name.
3436 * tree.h (make_int_cst_stat): Adjust prototype.
3437 (make_int_cst): Remove macro.
3439 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3441 * tree.c (make_tre_binfo_stat): Remove _stat from name.
3442 * tree.h (make_tree_binfo_stat): Adjust prototype.
3443 (make_tree_binfo): Remove.
3445 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3447 * tree.c (copy_node_stat): Rename to copy_node.
3448 (build_distinct_type_copy): Adjust.
3449 * tree.h (copy_node_stat): Adjust prototype.
3450 (copy_node): Remove macro.
3452 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3454 * tree.c (make_node_stat): rename to make_node.
3455 (build_tree_list_stat): Adjust.
3456 (build0_stat): Likewise.
3457 (build2_stat): Likewise.
3458 (build3_stat): Likewise.
3459 (build4_stat): Likewise.
3460 (build5_stat): Likewise.
3461 (build_decl_stat): Likewise.
3462 * tree.h (make_node_stat): Adjust prototype.
3463 (make_node): remove macro.
3465 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
3467 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
3468 (PPC_FEATURE2_SCV): Likewise.
3469 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
3471 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3473 * config/aarch64/aarch64.c
3474 (aarch64_internal_mov_immediate): Add new special pattern.
3475 * config/aarch64/aarch64.md (*movdi_aarch64):
3476 Add reg/32bit const mov case.
3478 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3479 Richard Sandiford <richard.sandiford@linaro.org>
3481 * config/aarch64/aarch64.md (mov<mode>): Generalize.
3482 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3483 Add integer and movi cases.
3484 (movi-split-hf-df-sf split, fp16): New.
3485 (enabled): Added TARGET_FP_F16INST.
3486 * config/aarch64/iterators.md (GPF_HF): New.
3487 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
3489 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3491 * config/aarch64/aarch64.c
3492 (aarch64_simd_container_mode): Add prototype.
3493 (aarch64_expand_mov_immediate): Add HI support.
3494 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
3495 (aarch64_can_const_movi_rtx_p): New.
3496 (aarch64_preferred_reload_class):
3497 Remove restrictions of using FP registers for certain SIMD operations.
3498 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
3499 (aarch64_valid_floating_const): Add integer move validation.
3500 (aarch64_simd_imm_scalar_p): Remove.
3501 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
3502 (aarch64_legitimate_constant_p): Expand list of supported cases.
3503 * config/aarch64/aarch64-protos.h
3504 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
3505 (aarch64_reinterpret_float_as_int): New.
3506 (aarch64_simd_imm_scalar_p): Remove.
3507 * config/aarch64/constraints.md (Uvi): New.
3508 (Dd): Split into Ds and new Dd.
3509 * config/aarch64/aarch64.md (*movsi_aarch64):
3511 (*movdi_aarch64): Add SIMD mov case.
3513 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3515 * tree-predcom.c: (struct chain): Handle store-store chain in which
3516 stores for elimination only store loop invariant values.
3517 (execute_pred_commoning_chain): Ditto.
3518 (prepare_initializers_chain_store_elim): Ditto.
3519 (prepare_finalizers): Ditto.
3520 (is_inv_store_elimination_chain): New function.
3521 (initialize_root_vars_store_elim_1): New function.
3523 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3525 * tree-predcom.c: Revise general description of the pass.
3526 (enum chain_type): New enum type for store elimination.
3527 (struct chain): New field supporting store elimination.
3528 (struct component): Ditto.
3529 (dump_chain): Dump store-stores chain.
3530 (release_chain): Release resources.
3531 (split_data_refs_to_components): Compute and create component
3532 contains only stores for elimination.
3533 (get_chain_last_ref_at): New function.
3534 (make_invariant_chain): Initialization.
3535 (make_rooted_chain): Specify chain type in parameter and record it.
3536 (add_looparound_copies): Skip for store-stores chain.
3537 (determine_roots_comp): Compute type of chain and pass it to
3539 (initialize_root_vars_store_elim_2): New function.
3540 (finalize_eliminated_stores): New function.
3541 (remove_stmt): Handle store for elimination.
3542 (execute_pred_commoning_chain): Execute predictive commoning on
3544 (determine_unroll_factor): Skip unroll for store-stores chain.
3545 (prepare_initializers_chain_store_elim): New function.
3546 (prepare_initializers_chain): Hanlde store-store chain.
3547 (prepare_finalizers_chain, prepare_finalizers): New function.
3548 (tree_predictive_commoning_loop): Return integer value indicating
3549 if loop is unrolled or lcssa form is corrupted.
3550 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
3552 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3554 * tree-predcom.c (initialize_root): Delete.
3555 (execute_pred_commoning_chain): Initialize root vars and replace
3556 reference of non-combined chain directly, rather than call above
3559 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3561 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
3562 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
3564 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3566 * tree-predcom.c (struct chain): New field init_seq.
3567 (release_chain): Release init_seq.
3568 (prepare_initializers_chain): Record intialization stmts in above
3570 (insert_init_seqs): New function.
3571 (tree_predictive_commoning_loop): Call insert_init_seqs.
3573 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3575 * tree-predcom.c (determine_roots_comp): Skip trivial components.
3577 2017-07-28 Richard Biener <rguenther@suse.de>
3579 * match.pd: Remove superfluous :c.
3580 * genmatch.c (simplify::id): Add member.
3581 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
3583 (current_id): New global.
3584 (dt_node::parent): Move from ...
3585 (dt_operand::parent): ... here. Add for_id member.
3586 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
3587 (decision_tree::find_node): Relax order requirement when
3588 merging DT_TRUE nodes to ones inbetween the current simplify
3589 and the one we try to merge with. Add diagnostic whenever
3590 we need to enforce pattern order by not merging.
3591 (decision_tree::insert): Set current_id.
3592 (decision_tree::print_node): Dump parent node and for_id.
3593 (parser::last_id): Add member.
3594 (parser::push_simplify): Assign unique id.
3595 (parser::parser): Initialize last_id.
3597 2017-07-28 Martin Liska <mliska@suse.cz>
3600 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
3601 gimple_build_debug_bind.
3603 2017-07-28 Richard Biener <rguenther@suse.de>
3605 PR tree-optimization/81502
3606 * match.pd: Add pattern combining BIT_INSERT_EXPR with
3608 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
3610 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
3611 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
3612 for BIT_FIELD_REF args.
3613 * fold-const.c (make_bit_field_ref): Likewise.
3614 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3616 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3619 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
3620 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
3621 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
3622 Or it into SANITIZER_UNDEFINED.
3623 * ubsan.c: Include gimple-fold.h and varasm.h.
3624 (ubsan_expand_ptr_ifn): New function.
3625 (instrument_pointer_overflow): New function.
3626 (maybe_instrument_pointer_overflow): New function.
3627 (instrument_object_size): Formatting fix.
3628 (pass_ubsan::execute): Call instrument_pointer_overflow
3629 and maybe_instrument_pointer_overflow.
3630 * internal-fn.c (expand_UBSAN_PTR): New function.
3631 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
3632 * sanitizer.def (__ubsan_handle_pointer_overflow,
3633 __ubsan_handle_pointer_overflow_abort): New builtins.
3634 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
3635 * internal-fn.def (UBSAN_PTR): New internal function.
3636 * opts.c (sanitizer_opts): Add pointer-overflow.
3637 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
3638 * fold-const.c (build_range_check): Compute pointer range check in
3639 integral type if pointer arithmetics would be needed. Formatting
3642 2017-07-28 Martin Liska <mliska@suse.cz>
3645 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
3646 parameters that are of a variable-length.
3648 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
3650 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
3652 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
3653 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
3654 (CRT_CALL_STATIC_FUNCTION): Likewise.
3655 (ASM_DEFAULT_SPEC): New define.
3656 (ASM_SPEC32): Likewise.
3657 (ASM_SPEC64): Likewise.
3658 (ASM_SPEC_COMMON): Likewise.
3659 (ASM_SPEC): Likewise.
3660 (INVALID_64BIT): Likewise.
3661 (LINK_OS_DEFAULT_SPEC): Likewise.
3662 (LINK_OS_SPEC32): Likewise.
3663 (LINK_OS_SPEC64): Likewise.
3664 (POWERPC_LINUX): Likewise.
3665 (PTRDIFF_TYPE): Likewise.
3666 (RESTORE_FP_PREFIX): Likewise.
3667 (RESTORE_FP_SUFFIX): Likewise.
3668 (SAVE_FP_PREFIX): Likewise.
3669 (SAVE_FP_SUFFIX): Likewise.
3670 (SIZE_TYPE): Likewise.
3671 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
3672 (TARGET_64BIT): Likewise.
3673 (TARGET_64BIT): Likewise.
3674 (TARGET_AIX): Likewise.
3675 (WCHAR_TYPE_SIZE): Likewise.
3676 (WCHAR_TYPE): Undefine.
3677 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
3678 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
3679 (CPP_OS_RTEMS_SPEC): Delete.
3680 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
3681 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
3683 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
3685 2017-07-28 Jakub Jelinek <jakub@redhat.com>
3687 PR tree-optimization/81578
3688 * tree-parloops.c (build_new_reduction): Bail out if
3689 reduction_code isn't one of the standard OpenMP reductions.
3690 Move the details printing after that decision.
3692 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3694 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
3695 related to reload_in_progress.
3696 (splat_input_operand): Likewise.
3697 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
3699 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
3701 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
3702 (TARGET_INSTANTIATE_DECLS): Likewise.
3703 (legitimate_indexed_address_p): Delete reload_in_progress code.
3704 (rs6000_debug_legitimate_address_p): Likewise.
3705 (rs6000_eliminate_indexed_memrefs): Likewise.
3706 (rs6000_emit_le_vsx_store): Likewise.
3707 (rs6000_emit_move_si_sf_subreg): Likewise.
3708 (rs6000_emit_move): Likewise.
3709 (register_to_reg_type): Likewise.
3710 (rs6000_pre_atomic_barrier): Likewise.
3711 (rs6000_machopic_legitimize_pic_address): Likewise.
3712 (rs6000_allocate_stack_temp): Likewise.
3713 (rs6000_address_for_fpconvert): Likewise.
3714 (rs6000_address_for_altivec): Likewise.
3715 (rs6000_secondary_memory_needed_rtx): Delete function.
3716 (rs6000_check_sdmode): Likewise.
3717 (rs6000_alloc_sdmode_stack_slot): Likewise.
3718 (rs6000_instantiate_decls): Likewise.
3719 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
3720 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
3721 Delete reload_in_progress.
3722 (*vec_reload_and_plus_<mptrsize>): Likewise.
3723 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
3724 (vsx_div_v2di): Likewise.
3725 (vsx_udiv_v2di): Likewise.
3727 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3729 * config/rs6000/rs6000.opt (mlra): Replace with stub.
3730 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
3731 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
3732 (rs6000_debug_reg_global): Delete print of LRA status.
3733 (rs6000_option_override_internal): Delete dead LRA related code.
3734 (rs6000_lra_p): Delete function.
3735 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
3737 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
3739 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
3740 * config/riscv/rtems.h: New file.
3742 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3743 Sudakshina Das <sudi.das@arm.com>
3745 * config/aarch64/aarch64.md
3746 (define_split for and<mode>3nr_compare): Move
3747 non aarch64_logical_operand to a register.
3748 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
3749 register immediate operand to a register.
3750 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
3752 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
3755 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
3757 2017-07-27 Richard Biener <rguenther@suse.de>
3759 PR tree-optimization/81573
3760 PR tree-optimization/81494
3761 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
3762 multi defuse cycle case.
3764 2017-07-27 Richard Biener <rguenther@suse.de>
3766 PR tree-optimization/81571
3767 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
3770 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
3772 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
3773 earlier and only if MASK_FPU is set. Adjust formatting.
3775 2017-07-27 Martin Liska <mliska@suse.cz>
3777 * opt-functions.awk: Add validation of value of Init.
3778 * optc-gen.awk: Pass new argument.
3780 2017-07-27 Martin Liska <mliska@suse.cz>
3782 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
3783 Fix wrong condition.
3785 2017-07-27 Martin Liska <mliska@suse.cz>
3787 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
3788 BBs and edges seen by autoFDO.
3790 2017-07-27 Richard Biener <rguenther@suse.de>
3792 PR tree-optimization/81502
3793 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
3794 with incompatible but same sized type.
3795 (execute_update_addresses_taken): Likewise.
3797 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
3799 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
3800 flag_tree_loop_vectorize rather than flag_tree_vectorize.
3802 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3805 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
3806 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
3807 Change s_operand to memory_operand.
3809 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
3811 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
3812 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
3813 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
3814 Emit instructions rather than returning an expression. Handle TFmode
3815 and KFmode by casting to TImode.
3816 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
3817 (rs6000_emit_le_vsx_store): Likewise.
3818 * config/rs6000/vsx.md (VSX_TI): New iterator.
3819 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
3820 (*vsx_le_undo_permute_<mode>): Likewise.
3821 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
3822 emit the split sequence.
3823 (*vsx_le_perm_store_<mode>): Likewise.
3825 2017-07-27 Jakub Jelinek <jakub@redhat.com>
3827 PR tree-optimization/81555
3828 PR tree-optimization/81556
3829 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
3830 if true, force CHANGED for the recursive invocation.
3831 (reassociate_bb): Remember original length of ops array, pass
3832 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
3834 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
3835 attributes for noipa attribute. For naked attribute use
3836 lookup_attribute first before lookup_attribute_spec.
3837 * final.c (rest_of_handle_final): Disable IPA RA for functions with
3839 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
3840 for functions with noipa attribute.
3841 (cgraph_externally_visible_p): Return true for functions with noipa
3843 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
3844 for functions with noipa attribute.
3845 * doc/extend.texi: Document noipa function attribute.
3846 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
3847 also for functions with noipa attribute.
3848 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
3850 2017-07-26 Andrew Pinski <apinski@cavium.com>
3852 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
3853 vec_unalign_load_cost and vec_unalign_store_cost.
3855 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3857 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
3858 -mvsx-small-integer option.
3859 (ISA_3_0_MASKS_IEEE): Likewise.
3860 (OTHER_VSX_VECTOR_MASKS): Likewise.
3861 (POWERPC_MASKS): Likewise.
3862 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
3863 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
3864 code, only testing for DImode being allowed in non-VSX floating
3866 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
3867 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
3869 (rs6000_option_override_internal): Delete -mvsx-small-integer.
3870 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
3871 TARGET_P8_VECTOR test.
3872 (rs6000_secondary_reload_simple_move): Likewise.
3873 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
3874 since TARGET_P9_VECTOR was already tested.
3875 (rs6000_opt_masks): Remove -mvsx-small-integer.
3876 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
3877 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
3879 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
3880 test for TARGET_VEXTRACTUB was used, and that uses
3882 (p9 extract splitter): Likewise.
3883 (vsx_extract_<mode>_di_p9): Likewise.
3884 (vsx_extract_<mode>_store_p9): Likewise.
3885 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
3886 for TARGET_P9_VECTOR was used. Delete code that is now dead with
3887 the elimination of TARGET_VSX_SMALL_INTEGER.
3888 (vsx_extract_<mode>_p8): Likewise.
3889 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
3890 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
3891 (vsx_set_<mode>_p9): Likewise.
3892 (vsx_set_v4sf_p9): Likewise.
3893 (vsx_set_v4sf_p9_zero): Likewise.
3894 (vsx_insert_extract_v4sf_p9): Likewise.
3895 (vsx_insert_extract_v4sf_p9_2): Likewise.
3896 * config/rs6000/rs6000.md (sign extend splitter): Change
3897 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
3898 (floatsi<mode>2_lfiwax_mem): Likewise.
3899 (floatunssi<mode>2_lfiwzx_mem): Likewise.
3900 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
3901 since a test for TARGET_P9_VECTOR was used.
3902 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
3903 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
3904 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
3905 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
3906 TARGET_P8_VECTOR test.
3907 (fix_trunc<mode>si2_stfiwx): Likewise.
3908 (fix_trunc<mode>si2_internal): Likewise.
3909 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
3910 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
3912 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
3913 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
3914 TARGET_P8_VECTOR test.
3915 (fixuns_trunc<mode>si2_stfiwx): Likewise.
3916 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
3917 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
3919 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
3920 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
3921 since a test for TARGET_P9_VECTOR was used.
3922 (splitter for loading small constants): Likewise.
3924 2017-07-26 Andrew Pinski <apinski@cavium.com>
3926 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
3929 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
3932 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
3933 (fp_valid_at): Likewise.
3935 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
3937 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
3938 (qdf24xx_addrcost_table): Likewise.
3939 (cortexa57_tunings): Update to use generic_branch_cost.
3940 (cortexa72_tunings): Likewise.
3941 (cortexa73_tunings): Likewise.
3942 (qdf24xx_tunings): Likewise.
3944 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
3946 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
3947 (thunderx2t99_branch_cost): Likewise.
3948 (cortexa35_tunings): Update to use generic_branch_cost.
3949 (cortexa53_tunings): Likewise.
3950 (cortexa57_tunings): Likewise.
3951 (cortexa72_tunings): Likewise.
3952 (cortexa73_tunings): Likewise.
3953 (thunderx2t99_tunings): Likewise.
3955 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
3957 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
3958 (sparc_option_override): Honour MASK_FSMULD.
3959 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
3960 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
3961 * config/sparc/sparc.opt (mfsmuld): New option.
3962 * doc/invoke.texi (mfsmuld): Document option.
3964 2017-07-26 Marek Polacek <polacek@redhat.com>
3967 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
3969 2017-07-26 Richard Biener <rguenther@suse.de>
3971 * gimple-match-head.c (do_valueize): Return OP if valueize
3973 (get_def): New helper to get at the def stmt of a SSA name
3975 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
3976 do_valueize to get at the def stmt.
3977 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
3979 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
3982 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
3984 2017-07-26 Martin Liska <mliska@suse.cz>
3987 * function.c (expand_function_start): Make expansion of
3988 nonlocal_goto_save_area after parm_birth_insn.
3990 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
3992 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
3993 from all CPU target flags enable members.
3995 2017-07-26 Richard Biener <rguenther@suse.de>
3997 * genmatch.c (dt_simplify::gen): Make iterator vars const.
3998 (decision_tree::gen): Make 'type' const.
3999 (write_predicate): Likewise.
4001 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4003 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4004 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4005 (rs6000_option_override_internal): Likewise.
4006 (rs6000_expand_vector_set): Likewise.
4007 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4008 (TARGET_UPPER_REGS_SF): Likewise.
4009 (TARGET_UPPER_REGS_DI): Likewise.
4010 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4011 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4012 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4013 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4014 (Splitters for DI constants in Altivec registers): Likewise.
4015 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4016 (vsx_set_v4sf_p9): Likewise.
4017 (vsx_set_v4sf_p9_zero): Likewise.
4018 (vsx_insert_extract_v4sf_p9): Likewise.
4019 (vsx_insert_extract_v4sf_p9_2): Likewise.
4021 2017-07-25 Carl Love <cel@us.ibm.com>
4023 * doc/extend.texi: Update the built-in documentation file for the
4024 existing built-in functions
4025 vector signed char vec_cnttz (vector signed char);
4026 vector unsigned char vec_cnttz (vector unsigned char);
4027 vector signed short vec_cnttz (vector signed short);
4028 vector unsigned short vec_cnttz (vector unsigned short);
4029 vector signed int vec_cnttz (vector signed int);
4030 vector unsigned int vec_cnttz (vector unsigned int);
4031 vector signed long long vec_cnttz (vector signed long long);
4032 vector unsigned long long vec_cnttz (vector unsigned long long);
4034 2017-07-25 Andrew Pinski <apinski@cavium.com>
4036 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4037 accesses where the use is for the first operand of a BIT_INSERT.
4039 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4042 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4043 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4045 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4047 * config/i386/gstabs.h: Delete.
4048 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4050 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4052 * config/i386/i386.c (ix86_decompose_address): Do not check for
4053 register RTX when looking at index_reg or base_reg.
4054 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4056 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4058 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4059 to update EH info here.
4061 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4063 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4065 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4067 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4069 2017-07-25 Torsten Duwe <duwe@suse.de>
4071 * common.opt: Introduce -fpatchable-function-entry
4072 command line option, and its variables function_entry_patch_area_size
4073 and function_entry_patch_area_start.
4074 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4075 including a two-value parser.
4076 * target.def (print_patchable_function_entry): New target hook.
4077 * targhooks.h (default_print_patchable_function_entry): New function.
4078 * targhooks.c (default_print_patchable_function_entry): Likewise.
4079 * toplev.c (process_options): Switch off IPA-RA if
4080 patchable function entries are being generated.
4081 * varasm.c (assemble_start_function): Look at the
4082 patchable-function-entry command line switch and current
4083 function attributes and maybe generate NOP instructions by
4084 calling the print_patchable_function_entry hook.
4085 * doc/extend.texi: Document patchable_function_entry attribute.
4086 * doc/invoke.texi: Document -fpatchable_function_entry
4087 command line option.
4088 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4090 * doc/tm.texi: Re-generate.
4092 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4095 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4096 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4098 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4100 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4102 2017-07-25 Richard Biener <rguenther@suse.de>
4104 PR tree-optimization/81455
4105 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4106 not walk in cycles when looking for guards.
4108 2017-07-25 Richard Biener <rguenther@suse.de>
4110 PR tree-optimization/81529
4111 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4112 when optimizing backedge uses.
4114 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4116 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4118 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4119 to dl_section_ref. On AIX, append an expression to subtract
4120 the size of the section length to dl_section_ref.
4122 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4124 * configure.ac: If any of the config.* scripts fail, exit 1.
4125 * configure: Regenerate.
4127 2017-07-25 Richard Biener <rguenther@suse.de>
4130 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4131 of immediate uses, be more verbose on errors.
4133 2017-07-25 Richard Biener <rguenther@suse.de>
4135 PR tree-optimization/81510
4136 * tree-vect-loop.c (vect_is_simple_reduction): When the
4137 reduction stmt is not inside the loop bail out.
4139 2017-07-25 Richard Biener <rguenther@suse.de>
4141 PR tree-optimization/81303
4142 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4143 profitability check against LOOP_VINFO_NITERSM1.
4145 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4147 * domwalk.c (cmp_bb_postorder): Simplify.
4148 (sort_bbs_postorder): New function. Use it...
4149 (dom_walker::walk): ...here to optimize common cases.
4151 2017-07-25 Martin Liska <mliska@suse.cz>
4154 * ipa-visibility.c (function_and_variable_visibility): Make the
4155 redirection just on target that supports aliasing.
4156 Fix GNU coding style.
4158 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4161 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4163 * config/arm/rtems.h: Likewise.
4164 * config/bfin/rtems.h: Likewise.
4165 * config/i386/rtemself.h: Likewise.
4166 * config/lm32/rtems.h: Likewise.
4167 * config/m32c/rtems.h: Likewise.
4168 * config/m68k/rtemself.h: Likewise.
4169 * config/microblaze/rtems.h: Likewise.
4170 * config/mips/rtems.h: Likewise.
4171 * config/moxie/rtems.h: Likewise.
4172 * config/nios2/rtems.h: Likewise.
4173 * config/powerpcspe/rtems.h: Likewise.
4174 * config/rs6000/rtems.h: Likewise.
4175 * config/rtems.h: Likewise.
4176 * config/sh/rtems.h: Likewise.
4177 * config/sh/rtemself.h: Likewise.
4178 * config/sparc/rtemself.h: Likewise.
4180 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4183 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4184 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4185 * tree-ssa-structalias.c (alias_get_name): Same.
4187 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4190 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4191 instructions if no du chain is found.
4193 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4195 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4197 2017-07-25 Richard Biener <rguenther@suse.de>
4200 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4203 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4205 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4207 (ISA_2_7_MASKS_SERVER): Likewise.
4208 (ISA_3_0_MASKS_IEEE): Likewise.
4209 (OTHER_P8_VECTOR_MASKS): Likewise.
4210 (OTHER_VSX_VECTOR_MASKS): Likewise.
4211 (POWERPC_MASKS): Likewise.
4212 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4213 duplicate list of options.
4214 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4215 explicit -mupper-regs options.
4216 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4217 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4218 alias for -mupper-regs-df.
4219 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4220 (rs6000_init_hard_regno_mode_ok): Likewise.
4221 (rs6000_option_override_internal): Likewise.
4222 (rs6000_opt_masks): Likewise.
4223 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4224 options in terms of whether -mvsx or -mpower8-vector was used.
4225 (TARGET_UPPER_REGS_DI): Likewise.
4226 (TARGET_UPPER_REGS_SF): Likewise.
4227 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4228 -mupper-regs-* options.
4230 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4232 * passes.c (emergency_dump_function): Print some empty lines and a
4233 header before the RTL dump.
4235 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4237 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4239 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4242 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4243 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4245 2017-07-24 Carl Love <cel@us.ibm.com>
4247 * config/rs6000/rs6000-c.c: Add support for built-in functions
4248 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4249 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4250 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4251 vec_extract_fp_from_shortl): Add defines for the two builtins.
4252 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4253 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4255 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4256 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4257 * doc/extend.texi: Update the built-in documentation file for the
4258 new built-in function.
4260 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4263 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4265 * doc/generic.texi: Likewise.
4266 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4267 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4269 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4271 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4272 (aarch64_mls_elt_merge<mode>): Likewise.
4274 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4276 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4277 having __cxa_atexit.
4279 2017-07-23 Michael Collison <michael.collison@arm.com>
4281 * config/arm/arm.c (arm_option_override): Deprecate
4282 use of -mstructure-size-boundary.
4283 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4284 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4286 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4289 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4290 Reduce cost estimate for direct moves.
4292 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4295 * config/i386/i386.c (ix86_option_override_internal): Disable
4296 BMI, BMI2 and TBM instructions for -m16.
4298 2017-07-21 Carl Love <cel@us.ibm.com>
4300 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4301 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4302 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4303 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4304 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4305 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4306 VMULOSW): New enum "unspec" values.
4307 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4308 altivec_vmulosw): New patterns.
4309 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4310 VMULOSW): Add definitions.
4312 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4314 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4315 (qdf24xx): Likewise.
4316 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4317 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4318 (AARCH64_FL_V8_1): Renumber.
4319 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4320 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4321 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4322 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4323 rdma to feature modifiers list.
4325 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4328 * ipa-visibility (function_and_variable_visibility): Convert
4329 recursive PLT call to direct call if appropriate.
4331 2017-07-21 Andrew Pinski <apinski@cavium.com>
4333 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4334 operand 1 to see if the types precision matches.
4335 * fold-const.c (operand_equal_p): Likewise.
4337 2017-07-21 Richard Biener <rguenther@suse.de>
4339 PR tree-optimization/81303
4340 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4341 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4342 (vect_peeling_hash_get_lowest_cost): Adjust.
4343 (vect_enhance_data_refs_alignment): Likewise. Use
4344 vect_get_peeling_costs_all_drs to compute the penalty for no
4345 peeling to match up costs.
4347 2017-07-21 Richard Biener <rguenther@suse.de>
4349 PR tree-optimization/81500
4350 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4351 we didn't identify a reduction path.
4353 2017-07-21 Tom de Vries <tom@codesourcery.com>
4354 Cesar Philippidis <cesar@codesourcery.com>
4356 PR gcov-profile/81442
4357 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
4360 2017-07-21 Tom de Vries <tom@codesourcery.com>
4363 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
4365 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
4366 nvptx_override_options_after_change.
4368 2017-07-21 Ulrich Drepper <drepper@redhat.com>
4370 * dwarf2out.c (output_file_names): Avoid double testing for
4373 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
4375 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
4377 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4379 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
4381 (try_crossjump_to_edge): Do not punt on partitioned functions.
4383 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
4385 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
4386 Put all BBs reachable only via paths crossing cold region to cold
4388 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
4390 2016-07-21 Richard Biener <rguenther@suse.de>
4392 PR tree-optimization/81303
4393 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
4394 into account prologue and epilogue iterations when raising
4395 min_profitable_iters to sth at least covering one vector iteration.
4397 2017-07-21 Tamar Christina <tamar.christina@arm.com>
4399 * config/arm/arm.c (arm_test_cpu_arch_dat):
4402 2017-07-20 Nathan Sidwell <nathan@acm.org>
4404 Remove TYPE_METHODS.
4405 * tree.h (TYPE_METHODS): Delete.
4406 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
4407 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
4408 (dbxout_type_methods): Scan TYPE_FIELDS.
4409 (dbxout_type): Don't check TYPE_METHODS here.
4410 * function.c (use_register_for_decl): Always ignore register for
4411 class types when not optimizing.
4412 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
4413 * tree.c (free_lang_data_in_type): Stitch out member functions and
4414 templates from TYPE_FIELDS.
4415 (build_distinct_type_copy, verify_type_variant,
4416 verify_type): Member fns are on TYPE_FIELDS.
4417 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
4418 * tree-pretty-print.c (dump_generic_node): Likewise.
4420 2017-07-20 Jakub Jelinek <jakub@redhat.com>
4423 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
4424 V2TImode and V4TImode.
4425 (ix86_expand_vector_extract): Likewise.
4426 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
4427 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
4428 (ssescalarmode): Handle V4TImode and V2TImode.
4429 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
4430 (*vec_extractv2ti, *vec_extractv4ti): New insns.
4431 (VEXTRACTI128_MODE): New mode iterator.
4432 (splitter for *vec_extractv?ti first element): New.
4433 (VEC_INIT_MODE): New mode iterator.
4434 (vec_init<mode>): Consolidate 3 expanders into one using
4435 VEC_INIT_MODE mode iterator.
4437 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4439 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
4440 non_spilled_static_chain_regno_p.
4442 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
4444 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
4446 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
4448 * bb-reorder.c (connect_traces): Allow copying of blocks within
4451 2017-07-20 Richard Biener <rguenther@suse.de>
4453 * gimple.h (gimple_phi_result): Add gphi * overload.
4454 (gimple_phi_result_ptr): Likewise.
4455 (gimple_phi_arg): Likewise. Adjust index assert to only
4456 allow actual argument accesses rather than all slots available
4458 (gimple_phi_arg_def): Add gphi * overload.
4459 * tree-phinodes.c (make_phi_node): Initialize only actual
4461 (resize_phi_node): Clear memory not covered by old node,
4462 do not initialize excess argument slots.
4463 (reserve_phi_args_for_new_edge): Initialize new argument slot
4466 2017-07-20 Bin Cheng <bin.cheng@arm.com>
4468 PR tree-optimization/81388
4470 2016-07-21 Bin Cheng <bin.cheng@arm.com>
4472 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
4473 by removing computation of may_be_zero.
4475 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4476 Tom de Vries <tom@codesourcery.com>
4479 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
4480 when gimple level profile disagrees with what RTL expander did.
4482 2017-07-20 Richard Biener <rguenther@suse.de>
4484 PR tree-optimization/61171
4485 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
4486 (vect_analyze_stmt): Add slp instance parameter.
4487 (vectorizable_reduction): Likewise.
4488 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
4489 (vect_is_simple_reduction): Deal with chains not detected
4490 as SLP reduction chain, specifically not properly associated
4491 chains containing a mix of plus/minus.
4492 (get_reduction_op): Remove.
4493 (get_initial_defs_for_reduction): Simplify, pass in whether
4494 this is a reduction chain, pass in the SLP node for the PHIs.
4495 (vect_create_epilog_for_reduction): Get the SLP instance as
4497 (vectorizable_reduction): Get the SLP instance as arg.
4498 During analysis remember the SLP node with the PHIs in the
4499 instance. Simplify getting at the vectorized reduction PHIs.
4500 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
4501 through SLP instance.
4502 (vect_slp_analyze_operations): Likewise.
4503 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
4504 (vect_transform_stmt): Likewise.
4506 2017-07-20 Tom de Vries <tom@codesourcery.com>
4508 PR tree-optimization/81489
4509 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
4510 read of phi arg location to before loop that modifies phi.
4512 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4514 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
4517 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4520 * except.c (execute): Fix ordering issue.
4522 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4524 PR rtl-optimization/81423
4525 * combine.c (make_compound_operation_int): Don't try to optimize
4526 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
4528 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
4530 PR rtl-optimization/81423
4531 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
4532 with a constant that is -1 in the truncated to mode.
4534 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4536 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
4537 (determine_unlikely_bbs): ... here.
4538 * predict.h (propagate_unlikely_bbs_forward): Declare.
4539 * cfgexpand.c (pass_expand::execute): Use it.
4540 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
4542 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
4543 propagate_unlikely_bbs_forward.
4545 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
4548 * except.c (maybe_add_nop_after_section_switch): New function.
4551 2017-07-19 Tom de Vries <tom@codesourcery.com>
4553 * gimple.h (gimple_phi_set_arg): Make assert more strict.
4555 2017-07-19 Tom de Vries <tom@codesourcery.com>
4557 * gimple.h (gimple_phi_arg): Make assert more strict.
4559 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
4561 * config.gcc (powerpc*-*-*): Add mmintrin.h.
4562 * config/rs6000/mmintrin.h: New file.
4563 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
4565 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4567 PR tree-optimization/81346
4568 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
4570 2017-07-19 Tom de Vries <tom@codesourcery.com>
4572 * config/nvptx/nvptx.md (VECIM): Add V2DI.
4574 2017-07-19 Tom de Vries <tom@codesourcery.com>
4576 * config/nvptx/nvptx-modes.def: Add V2DImode.
4577 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
4578 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
4579 (nvptx_output_mov_insn): Handle lack of mov.b128.
4580 (nvptx_print_operand): Handle 'H' and 'L' codes.
4581 (nvptx_vector_mode_supported): Allow V2DImode.
4582 (nvptx_preferred_simd_mode): New function.
4583 (nvptx_data_alignment): New function.
4584 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
4585 nvptx_preferred_simd_mode.
4586 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
4588 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
4590 2017-07-19 Tom de Vries <tom@codesourcery.com>
4592 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
4593 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
4594 (nvptx_vector_mode_supported): New function. Allow V2SImode.
4595 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
4596 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
4597 (mov<VECIM>_insn): New define_insn.
4598 (define_expand "mov<VECIM>): New define_expand.
4600 2017-07-19 Tom de Vries <tom@codesourcery.com>
4602 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
4604 2017-07-19 Jakub Jelinek <jakub@redhat.com>
4606 PR tree-optimization/81346
4607 * fold-const.h (fold_div_compare, range_check_type): Declare.
4608 * fold-const.c (range_check_type): New function.
4609 (build_range_check): Use range_check_type.
4610 (fold_div_compare): No longer static, rewritten into
4611 a match.pd helper function.
4612 (fold_comparison): Don't call fold_div_compare here.
4613 * match.pd (X / C1 op C2): New optimization using fold_div_compare
4616 2017-07-19 Nathan Sidwell <nathan@acm.org>
4618 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
4619 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
4620 * tree.c (find_decls_types_r, verify_type): Use
4621 TYPE_{MIN,MAX}_VALUE_RAW.
4622 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
4623 (hash_tree): Likewise.
4624 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4626 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4629 2017-07-18 Tom de Vries <tom@codesourcery.com>
4632 * omp-expand.c (expand_omp_for_static_chunk): Handle
4633 equal-argument loop exit phi.
4635 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
4638 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
4639 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
4640 operand 2 predicate.
4641 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
4642 operand 2 predicate.
4643 (ror,rol -> rorx splitters): Use const_int_operand as
4644 operand 2 predicate.
4646 2017-06-18 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/81410
4649 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
4650 the gap in the ! slp_perm SLP case after each group.
4652 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4655 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
4658 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
4661 * predict.c (set_even_probabilities): Cleanup; do not affect
4662 probabilities that are already known.
4663 (combine_predictions_for_bb): Call even when count is set.
4665 2017-07-18 Nathan Sidwell <nathan@acm.org>
4667 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
4670 2017-07-18 Bin Cheng <bin.cheng@arm.com>
4673 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
4674 optimization for loop niter analysis.
4676 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
4679 * config/avr/avr.c (avr_optimize_casesi): Don't use
4680 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
4682 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4684 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
4685 body_cost_vec from _vect_peel_extended_info.
4686 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
4687 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
4690 2017-07-18 Bin Cheng <bin.cheng@arm.com>
4692 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
4694 2017-07-18 Richard Biener <rguenther@suse.de>
4696 PR tree-optimization/80620
4697 PR tree-optimization/81403
4698 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
4699 info when re-using a VN table entry.
4701 2017-07-18 Richard Biener <rguenther@suse.de>
4703 PR tree-optimization/81418
4704 * tree-vect-loop.c (vectorizable_reduction): Properly compute
4705 vectype_in. Verify that with lane-reducing reduction operations
4706 we have a single def-use cycle.
4708 2017-07-17 Carl Love <cel@us.ibm.com>
4710 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
4712 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4713 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4714 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4715 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4716 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4717 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4718 VMULOSW): New enum "unspec" values.
4719 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4720 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4721 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4722 altivec_vmulosw): New patterns.
4723 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4724 VMULOSW): Add definitions.
4725 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
4727 * config/alpha/alpha.c: Include predict.h.
4729 2017-07-17 Yury Gribov <tetra2005@gmail.com>
4731 * tree-vrp.c (compare_assert_loc): Fix comparison function
4732 to return predictable results.
4734 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4736 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
4739 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
4740 * doc/invoke.texi (mexpand-adddi): Update text.
4742 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4744 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
4745 that also clobbers the CC register. The old expand code is moved
4747 (*arc_clzsi2): ... here.
4748 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
4749 the CC register. The old expand code is moved to ...
4750 (arc_ctzsi2): ... here.
4752 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4754 * config/arc/arc.opt (mindexed-loads): Use initial value
4755 TARGET_INDEXED_LOADS_DEFAULT.
4756 (mauto-modify-reg): Use initial value
4757 TARGET_AUTO_MODIFY_REG_DEFAULT.
4758 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
4759 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
4760 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
4761 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
4763 2017-07-17 Martin Liska <mliska@suse.cz>
4766 * opts.c (finish_options): Do not allow -fgnu-tm
4767 w/ -fsanitize={kernel-,}address. Say sorry.
4769 2017-07-17 Bin Cheng <bin.cheng@arm.com>
4772 * tree-loop-distribution.c (classify_partition): Only assert on
4773 numer of iterations.
4774 (merge_dep_scc_partitions): Delete prameter. Update function call.
4775 (distribute_loop): Remove code handling loop with unknown niters.
4776 (pass_loop_distribution::execute): Skip loop with unknown niters.
4778 2017-07-17 Bin Cheng <bin.cheng@arm.com>
4781 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
4782 function sort_partitions_by_post_order.
4784 2017-07-17 Bin Cheng <bin.cheng@arm.com>
4786 PR tree-optimization/81374
4787 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
4788 the max index of basic blocks, rather than number of basic blocks.
4790 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4792 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
4794 (arc_legitimate_pic_operand_p): Likewise.
4795 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
4797 (arc_needs_pcl_p): Likewise.
4798 (arc_legitimate_pc_offset_p): Likewise.
4799 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
4800 function is also used in constrains.md.
4801 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
4802 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
4803 PLUS. Only return true/false in known cases, otherwise assert.
4804 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
4805 is already called in arc_legitimate_constant_p.
4806 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
4808 (LEGITIMATE_PIC_OPERAND_P): Use
4809 arc_raw_symbolic_reference_mentioned_p function.
4810 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
4815 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4816 Andrew Burgess <andrew.burgess@embecosm.com>
4818 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
4819 (arc_return_address_register): New function.
4820 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
4821 (arc_handle_fndecl_attribute): Add naked attribute.
4822 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4823 (TARGET_WARN_FUNC_RETURN): Likewise.
4824 (arc_allocate_stack_slots_for_args): New function.
4825 (arc_warn_func_return): Likewise.
4826 (machine_function): Change type fn_type.
4827 (arc_compute_function_type): Consider new naked function type,
4828 change function return type.
4829 (arc_must_save_register): Adapt to handle new
4830 arc_compute_function_type's return type.
4831 (arc_expand_prologue): Likewise.
4832 (arc_expand_epilogue): Likewise.
4833 (arc_return_address_regs): Delete.
4834 (arc_return_address_register): New function.
4835 (arc_epilogue_uses): Use above function.
4836 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
4837 (arc_function_type): Change encoding, add naked type.
4838 (ARC_INTERRUPT_P): Change to handle the new encoding.
4839 (ARC_FAST_INTERRUPT_P): Likewise.
4840 (ARC_NORMAL_P): Define.
4841 (ARC_NAKED_P): Likewise.
4842 (arc_compute_function_type): Delete prototype.
4843 * config/arc/arc.md (in_ret_delay_slot): Use
4844 arc_return_address_register function.
4845 (simple_return): Likewise.
4846 (p_return_i): Likewise.
4848 2017-07-17 Jakub Jelinek <jakub@redhat.com>
4850 PR tree-optimization/81428
4851 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
4852 can't be built for those types.
4854 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
4856 Remove stuff dead since r239246.
4858 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
4859 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
4860 (avr_inform_devices): Remove dead stuff.
4862 2017-07-17 Tamar Christina <tamar.christina@arm.com>
4864 * config/arm/arm_neon.h: Fix softp typo.
4866 2017-07-17 Jakub Jelinek <jakub@redhat.com>
4868 PR tree-optimization/81365
4869 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
4870 aggregate moves onto bb predecessor edges, make sure there are no
4871 loads that could alias the lhs in between the start of bb and the
4874 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
4877 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
4878 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
4879 [LSHIFTRT, outer_code = TRUNCATE]: Same.
4881 2017-07-17 Jakub Jelinek <jakub@redhat.com>
4883 PR tree-optimization/81396
4884 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
4885 (init_symbolic_number): Initialize it to 1.
4886 (perform_symbolic_merge): Add n_ops from both operands into the new
4888 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
4889 without base_addr as useless if they need more than one operation.
4890 (bswap_replace): Handle !bswap case for NULL base_addr.
4892 2017-07-17 Tom de Vries <tom@codesourcery.com>
4895 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
4898 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
4900 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
4901 conditional builtin define __FIX_LEON3FT_B2BST.
4903 2017-07-17 Daniel Cederman <cederman@gaisler.com>
4905 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
4906 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
4909 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
4911 PR rtl-optimization/81424
4912 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
4913 to remove potential trapping from operands if -fnon-call-exceptions.
4915 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4917 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
4918 profile_proability for scalling.
4919 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
4921 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4923 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
4925 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4927 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
4928 fixpoint arithmetics.
4930 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4932 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
4933 fixpoint arithmetics.
4935 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4937 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
4938 fixpoint arithmetics.
4940 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
4942 * profile-count.h (profile_probability::from_reg_br_prob_note,
4943 profile_probability::to_reg_br_prob_note): New functions.
4944 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
4945 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
4946 * predict.c (probability_reliable_p): Update.
4947 (edge_probability_reliable_p): Update.
4948 (br_prob_note_reliable_p): Update.
4949 (invert_br_probabilities): Update.
4950 (add_reg_br_prob_note): New function.
4951 (combine_predictions_for_insn): Update.
4952 * asan.c (asan_clear_shadow): Update.
4953 * cfgbuild.c (compute_outgoing_frequencies): Update.
4954 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
4955 (update_br_prob_note): Update.
4956 (rtl_verify_edges): Update.
4957 (purge_dead_edges): Update.
4958 (fixup_reorder_chain): Update.
4959 * emit-rtl.c (try_split): Update.
4960 * ifcvt.c (cond_exec_process_insns): Update.
4961 (cond_exec_process_if_block): Update.
4962 (dead_or_predicable): Update.
4963 * internal-fn.c (expand_addsub_overflow): Update.
4964 (expand_neg_overflow): Update.
4965 (expand_mul_overflow): Update.
4966 * loop-doloop.c (doloop_modify): Update.
4967 * loop-unroll.c (compare_and_jump_seq): Update.
4968 * optabs.c (emit_cmp_and_jump_insn_1): Update.
4969 * predict.h: Update.
4970 * reorg.c (mostly_true_jump): Update.
4972 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
4973 * config/alpha/alpha.c (emit_unlikely_jump): Update.
4974 * config/arc/arc.c: (emit_unlikely_jump): Update.
4975 * config/arm/arm.c: (emit_unlikely_jump): Update.
4976 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
4977 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
4978 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
4979 (ix86_print_operand): Update.
4980 (ix86_split_fp_branch): Update.
4981 (predict_jump): Update.
4982 * config/ia64/ia64.c (ia64_print_operand): Update.
4983 * config/mmix/mmix.c (mmix_print_operand): Update.
4984 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
4985 (rs6000_expand_split_stack_prologue): Update.
4986 * config/rs6000/rs6000.c: Update.
4987 * config/s390/s390.c (s390_expand_vec_strlen): Update.
4988 (s390_expand_vec_movstr): Update.
4989 (s390_expand_cs_tdsi): Update.
4990 (s390_expand_split_stack_prologue): Update.
4991 * config/sh/sh.c (sh_print_operand): Update.
4992 (expand_cbranchsi4): Update.
4993 (expand_cbranchdi4): Update.
4994 * config/sparc/sparc.c (output_v9branch): Update.
4995 * config/spu/spu.c (get_branch_target): Update.
4996 (ea_load_store_inline): Update.
4997 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
4998 * config/tilepro/tilepro.c: Update.
5000 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5002 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5003 (gimplify_save_expr): Likewise.
5005 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5007 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5009 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5011 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5013 * timevar.def (TV_IPA_FNSUMMARY): Define.
5015 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5017 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5018 to back store errata sensitive sequence from being generated.
5019 (sqrtdf2_fix): Likewise.
5021 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5023 * tree-ssa-threadupdate.c (compute_path_counts,
5024 update_joiner_offpath_counts): Use profile_probability.
5026 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5029 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5031 * config/arm/arm-c.c (arm_cpu_builtins): Define
5032 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5034 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5036 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5037 array entries to represent __ieee128 versions of the
5038 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5039 scalar_extract_sig, and scalar_insert_exp built-in functions.
5040 (altivec_resolve_overloaded_builtin): Add special case handling
5041 for the __builtin_scalar_insert_exp function, as represented by
5042 the P9V_BUILTIN_VEC_VSIEDP constant.
5043 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5044 exponent support for __ieee128 argument.
5045 (VSESQP): Add scalar extract signature support for __ieee128
5047 (VSTDCNQP): Add scalar test negative support for __ieee128
5049 (VSIEQP): Add scalar insert exponent support for __int128 argument
5050 with __ieee128 result.
5051 (VSIEQPF): Add scalar insert exponent support for __ieee128
5052 argument with __ieee128 result.
5053 (VSTDCQP): Add scalar test data class support for __ieee128
5055 (VSTDCNQP): Add overload support for scalar test negative with
5057 (VSTDCQP): Add overload support for scalar test data class
5059 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5061 (UNSPEC_VSX_SIEXPQP): New constant.
5062 (xsxexpqp): New insn for VSX scalar extract exponent quad
5064 (xsxsigqp): New insn for VSX scalar extract significand quad
5066 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5067 precision with floating point argument.
5068 (xststdcqp): New expand for VSX scalar test data class quad
5070 (xststdcnegqp): New expand for VSX scalar test negative quad
5072 (xststdcqp): New insn to match expansions for VSX scalar test data
5073 class quad precision and VSX scalar test negative quad precision.
5074 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5075 special case operand checking to enforce that second operand of
5076 VSX scalar test data class with quad precision argument is a 7-bit
5078 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5079 prototypes and descriptions of __ieee128 versions of
5080 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5081 scalar_test_data_class, and scalar_test_neg built-in functions.
5083 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5085 PR tree-optimization/81162
5086 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5087 replace a negate with an add.
5089 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5091 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5093 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5095 * config/arm/arm-c.c (arm_cpu_builtins): Define
5096 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5098 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5100 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5101 (armv8-r): Set ARM Cortex-R52 as default CPU.
5102 * config/arm/arm-tables.opt: Regenerate.
5103 * config/arm/arm-tune.md: Regenerate.
5104 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5106 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5107 extension for -mcpu=cortex-r52.
5109 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5111 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5112 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5113 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5114 (fp-armv8): Define it as FP_ARMv8 only.
5115 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5116 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5118 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5119 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5120 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5121 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5122 than TARGET_FPU_ARMV8.
5123 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5124 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5125 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5127 * config/arm/neon.md (neon_vrint): Likewise.
5128 (neon_vcvt): Likewise.
5129 (neon_<fmaxmin_op><mode>): Likewise.
5130 (<fmaxmin><mode>3): Likewise.
5131 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5132 * config/arm/predicates.md (arm_cond_move_operator): Check against
5133 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5135 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5137 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5140 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5142 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5143 loop in comment with memset.
5145 2017-07-14 Martin Liska <mliska@suse.cz>
5147 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5148 * dwarf2out.c (is_java): Remove the function.
5149 (output_pubname): Remove usage of the function.
5150 (lower_bound_default): Remove usage of DW_LANG_Java.
5151 (gen_compile_unit_die): Likewise.
5152 * gcc.c: Remove compiler defaults for .java and .zip files.
5153 * gimple-expr.c (remove_suffix): Change as there's no longer
5154 extension than 4-letter one.
5155 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5156 (gimplify_save_expr): Likewise.
5157 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5158 as it's possible even for other languages than Java.
5159 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5160 * lto-opts.c (lto_write_options): Remove reference to Java.
5161 * opts.c (strip_off_ending): Update file extension handling.
5162 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5163 * tree-eh.c (lower_resx): Likewise.
5164 * tree.c (free_lang_data_in_type): Remove dead code.
5165 (find_decls_types_r): Likewise.
5166 (build_common_builtin_nodes): Remove Java from a comment.
5167 (verify_type): Remove dead code.
5168 * varasm.c (assemble_external): Remove Java from a comment.
5170 2017-07-14 Martin Liska <mliska@suse.cz>
5172 * opts.c (finish_options): Add quotes.
5173 (common_handle_option): Likewise.
5175 2017-07-14 Martin Liska <mliska@suse.cz>
5177 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5178 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5180 * dwarf2out.c (lower_bound_default): Do not handle
5182 (gen_compile_unit_die): Likewise.
5183 * gcc.c: Remove default extension binding for GNU Pascal.
5184 * stmt.c: Remove Pascal language from a comment.
5185 * xcoffout.c: Likewise.
5187 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5190 * diagnostic-show-locus.c (fixit_cmp): New function.
5191 (layout::layout): Sort m_fixit_hints.
5192 (column_range::column_range): Assert that the values are valid.
5193 (struct char_span): New struct.
5194 (correction::overwrite): New method.
5195 (struct source_line): New struct.
5196 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5197 calls in terms of classes source_line and char_span, and
5198 correction::overwrite.
5199 (selftest::test_overlapped_fixit_printing_2): New function.
5200 (selftest::diagnostic_show_locus_c_tests): Call it.
5202 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5204 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5205 early if there is no lhs.
5207 2017-07-13 Martin Liska <mliska@suse.cz>
5209 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5210 (gen_reference_type_die): Likewise.
5211 * stor-layout.c: Remove Pascal-related comment.
5213 2017-07-13 Martin Liska <mliska@suse.cz>
5215 * opts.c (finish_options): Add quotes to error messages.
5216 (parse_sanitizer_options): Likewise.
5218 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5220 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5222 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5224 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5226 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5228 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5230 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5232 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5235 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5236 provides the hardware capability bits, define the macro
5237 __BUILTIN_CPU_SUPPORTS__.
5238 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5239 if GLIBC does not provide the hardware capability bits. Add a
5240 gcc_unreachable call if the built-in cpu function is neither
5241 __builtin_cpu_is nor __builtin_cpu_supports.
5242 (rs6000_get_function_versions_dispatcher): Change the warning
5243 that an old GLIBC is used which does not export the capability
5244 bits to be an error.
5245 * doc/extend.texi (target_clones attribute): Document the
5246 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5247 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5248 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5249 the macros defined by GCC if the newer GLIBC is available.
5251 2017-07-12 Jeff Law <law@redhat.com>
5253 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5254 remaining includes slightly.
5255 * config/riscv/riscv-builtins.c: Include profile-count.h.
5257 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5260 * config/avr/avr.c (avr_set_current_function): In diagnostic
5261 messages: Quote keywords and (parts of) identifiers.
5262 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5265 2017-07-12 Carl Love <cel@us.ibm.com>
5267 * config/rs6000/rs6000-c.c: Add support for built-in functions
5268 vector bool char vec_revb (vector bool char);
5269 vector bool short vec_revb (vector short char);
5270 vector bool int vec_revb (vector bool int);
5271 vector bool long long vec_revb (vector bool long long);
5272 * doc/extend.texi: Update the built-in documentation file for the
5273 new built-in functions.
5275 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5277 * config/s390/s390.md: Remove movcc splitter.
5279 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5281 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5282 load/store on condition.
5284 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5287 * config/avr/avr.c (avr_encode_section_info)
5288 [progmem && !TREE_READONLY]: Error if progmem object needs
5291 2017-07-11 Michael Collison <michael.collison@arm.com>
5293 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5296 2017-07-11 Carl Love <cel@us.ibm.com>
5298 * config/rs6000/rs6000-c.c: Add support for builtins
5299 vector unsigned int vec_parity_lsbb (vector signed int);
5300 vector unsigned int vec_parity_lsbb (vector unsigned int);
5301 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5302 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5303 vector unsigned long long vec_parity_lsbb (vector signed long long);
5304 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5305 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5306 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5307 * doc/extend.texi: Update the built-in documentation file for the
5308 new built-in functions.
5310 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5312 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5313 (layout::m_primary_loc): New field.
5314 (layout::layout): Initialize new field. Move location filtering
5315 logic from here to...
5316 (layout::maybe_add_location_range): ...this new method. Add
5317 support for filtering to just the lines already specified by other
5319 (layout::will_show_line_p): New method.
5320 (gcc_rich_location::add_location_if_nearby): New method.
5321 (selftest::test_add_location_if_nearby): New test function.
5322 (selftest::diagnostic_show_locus_c_tests): Call it.
5323 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5326 2017-07-11 Tom de Vries <tom@codesourcery.com>
5328 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5329 (bb_first_real_insn): New function.
5330 (nvptx_single): Add extra initialization of broadcasted condition
5333 2017-07-11 Nathan Sidwell <nathan@acm.org>
5335 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5337 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5339 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5340 Binutils doc as TEXI will mess them up.
5341 * doc/invoke.texi (AVR Options): Same here.
5343 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5345 * config/sparc/sparc.opt (mfix-ut700): New option.
5346 (mfix-gr712rc): Likewise.
5347 (sparc_fix_b2bst): New variable.
5348 * doc/invoke.texi (SPARC options): Document them.
5349 (ARM options): Fix warnings.
5350 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5351 instructions to prevent sequences that can trigger the store-store
5352 errata for certain LEON3FT processors.
5353 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
5354 (sparc_option_override): Set sparc_fix_b2bst appropriately.
5355 * config/sparc/sparc.md (fix_b2bst): New attribute.
5356 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
5358 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
5361 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
5363 (*rsqrtsf2_sse): Ditto.
5365 (div<mode>3): Macroize insn from divdf3 and divsf3
5366 using MODEF mode iterator.
5368 2017-07-10 Martin Sebor <msebor@redhat.com>
5370 PR tree-optimization/80397
5371 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
5372 instead of testing for equality to INTEGER_TYPE.
5374 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
5376 * config.gcc: Remove uclibc from arc target spec.
5378 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
5380 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
5382 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5385 * lto-wrapper.c (remove_option): New function.
5386 (merge_and_complain): Merge PIC/PIE options more realistically.
5388 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5390 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
5394 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
5395 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
5396 * config.in: Regenerate.
5397 * configure: Regenerate.
5398 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
5399 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
5400 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
5401 (TARGET_GASISR_PROLOGUES): ...target mask.
5402 * common/config/avr/avr-common.c
5403 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
5404 Set -mgas-isr-prologues.
5405 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
5406 INSERT_PASS_BEFORE for it.
5407 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
5408 * config/avr/avr.c (avr_option_override)
5409 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
5410 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
5411 (avr_attribute_table) <no_gccisr>: Add new function attribute.
5412 (avr_set_current_function) <is_no_gccisr>: Init machine field.
5413 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
5415 (make_avr_pass_pre_proep): New function.
5416 (emit_push_sfr) <treg>: Add argument to function and use it
5418 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
5419 and set machine->gasisr.yes.
5420 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
5421 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
5422 __gcc_isr.n_pushed to .L__stack_usage.
5423 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
5424 (avr_asm_final_postscan_insn): ...this new static function.
5425 * config/avr/avr.h (machine_function)
5426 <is_no_gccisr, use_L__stack_usage>: New fields.
5427 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
5428 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
5429 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
5430 (gasisr, *gasisr): New expander and insn.
5431 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5432 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
5433 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
5435 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
5437 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
5440 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
5442 Move jump-tables out of .text again.
5445 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
5446 (ASM_OUTPUT_ADDR_VEC): New function.
5447 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
5448 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
5449 INSN_ADDRESSes as asm comment.
5450 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
5451 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
5452 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
5453 * config/avr/avr.md (*tablejump): Adjust comment.
5454 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
5455 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
5457 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
5458 (avr_output_addr_vec): New proto.
5459 (avr_log_t) <insn_addresses>: New field.
5461 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
5464 * config/i386/i386.c (ix86_function_arg_advance): Set
5465 outgoing_args_on_stack to true if there are outgoing arguments
5467 (ix86_function_arg): Likewise.
5468 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
5469 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
5470 * config/i386/i386.h (machine_function): Add
5471 outgoing_args_on_stack.
5473 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
5475 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5477 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
5479 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
5481 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
5482 (REAL_H): Remove $(MACHMODE_H).
5483 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
5485 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
5486 $(MACHMODE_H) and double-int.h.
5487 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
5489 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
5492 2017-07-07 Andrew Pinski <apinski@cavium.com>
5494 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
5495 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
5497 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5499 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
5500 Add warning if GCC was not configured to link against a GLIBC that
5501 exports the hardware capability bits.
5502 (make_resolver_func): Make resolver function private and not a
5503 COMDAT function. Create the name with clone_function_name instead
5504 of make_unique_name.
5507 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
5508 correct operand in doing the split.
5510 2017-07-07 Carl Love <cel@us.ibm.com>
5512 * config/rs6000/rs6000-c: Add support for built-in function
5513 vector unsigned short vec_pack_to_short_fp32 (vector float,
5515 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
5516 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
5517 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
5518 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
5519 (convert_4f32_8i16): Add define_expand.
5520 * doc/extend.texi: Update the built-in documentation file for the
5521 new built-in function.
5523 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5525 * config/sparc/m8.md: New file.
5526 * config/sparc/sparc.md: Include m8.md.
5528 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5530 * config/sparc/sparc.opt: New option -mvis4b.
5531 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
5532 (sparc_option_override): Handle VIS4B.
5533 (enum sparc_builtins): Define
5534 SPARC_BUILTIN_DICTUNPACK{8,16,32},
5535 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
5536 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
5537 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
5538 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
5539 (check_constant_argument): New function.
5540 (sparc_vis_init_builtins): Define builtins
5541 __builtin_vis_dictunpack{8,16,32},
5542 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
5543 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
5544 __builtin_vis_fpcmpde{8,16,32}shl and
5545 __builtin_vis_fpcmpur{8,16,32}shl.
5546 (sparc_expand_builtin): Check that the constant operands to
5547 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
5548 constant and in range.
5549 * config/sparc/sparc-c.c (sparc_target_macros): Handle
5551 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
5552 (SPARC_IMM5_P): Likewise.
5553 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
5554 (enabled): Handle vis4b.
5555 (UNSPEC_DICTUNPACK): New unspec.
5556 (UNSPEC_FPCMPSHL): Likewise.
5557 (UNSPEC_FPUCMPSHL): Likewise.
5558 (UNSPEC_FPCMPDESHL): Likewise.
5559 (UNSPEC_FPCMPURSHL): Likewise.
5560 (cpu_feature): New CPU feature `vis4b'.
5561 (dictunpack{8,16,32}): New insns.
5562 (FPCSMODE): New mode iterator.
5563 (fpcscond): New code iterator.
5564 (fpcsucond): Likewise.
5565 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
5566 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
5567 (fpcmpde{8,16,32}{si,di}shl): Likewise.
5568 (fpcmpur{8,16,32}{si,di}shl): Likewise.
5569 * config/sparc/constraints.md: Define constraints `q' for unsigned
5570 2-bit integer constants and `t' for unsigned 5-bit integer
5572 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
5574 (imm5_operand_dictunpack16): Likewise.
5575 (imm5_operand_dictunpack32): Likewise.
5576 (imm2_operand): Likewise.
5577 * doc/invoke.texi (SPARC Options): Document -mvis4b.
5578 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
5579 ditunpack* and fpcmp*shl builtins.
5581 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5583 * config.gcc: Handle m8 in --with-{cpu,tune} options.
5584 * config.in: Add HAVE_AS_SPARC6 define.
5585 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
5587 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
5589 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
5590 (CPP_CPU_SPEC): Handle m8.
5591 (ASM_CPU_SPEC): Likewise.
5592 * config/sparc/sparc-opts.h (enum processor_type): Add
5594 * config/sparc/sparc.c (m8_costs): New struct.
5595 (sparc_option_override): Handle TARGET_CPU_m8.
5596 (sparc32_initialize_trampoline): Likewise.
5597 (sparc64_initialize_trampoline): Likewise.
5598 (sparc_issue_rate): Likewise.
5599 (sparc_register_move_cost): Likewise.
5600 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
5601 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
5602 (ASM_CPU64_DEFAULT_SPEC): Likewise.
5603 (CPP_CPU_SPEC): Handle M8.
5604 (ASM_CPU_SPEC): Likewise.
5605 (AS_M8_FLAG): Define.
5606 * config/sparc/sparc.md: Add m8 to the cpu attribute.
5607 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
5608 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
5610 * configure: Regenerate.
5611 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
5614 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5616 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
5618 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
5619 ("*movdi_insn_sp32"): Do not set v3pipe.
5620 ("*movsi_insn"): Likewise.
5621 ("*movdi_insn_sp64"): Likewise.
5622 ("*movsf_insn"): Likewise.
5623 ("*movdf_insn_sp32"): Likewise.
5624 ("*movdf_insn_sp64"): Likewise.
5625 ("*zero_extendsidi2_insn_sp64"): Likewise.
5626 ("*sign_extendsidi2_insn"): Likewise.
5627 ("*mov<VM32:mode>_insn"): Likewise.
5628 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5629 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5630 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5631 ("<vlop:code><VL:mode>3"): Likewise.
5632 ("*not_<vlop:code><VL:mode>3"): Likewise.
5633 ("*nand<VL:mode>_vis"): Likewise.
5634 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
5635 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
5636 ("one_cmpl<VL:mode>2"): Likewise.
5637 ("faligndata<VM64:mode>_vis"): Likewise.
5638 ("alignaddrsi_vis"): Likewise.
5639 ("alignaddrdi_vis"): Likweise.
5640 ("alignaddrlsi_vis"): Likewise.
5641 ("alignaddrldi_vis"): Likewise.
5642 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5643 ("bmaskdi_vis"): Likewise.
5644 ("bmasksi_vis"): Likewise.
5645 ("bshuffle<VM64:mode>_vis"): Likewise.
5646 ("cmask8<P:mode>_vis"): Likewise.
5647 ("cmask16<P:mode>_vis"): Likewise.
5648 ("cmask32<P:mode>_vis"): Likewise.
5649 ("pdistn<P:mode>_vis"): Likewise.
5650 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5652 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5654 * config/sparc/sparc.md ("subtype"): New insn attribute.
5655 ("*wrgsr_sp64"): Set insn subtype.
5656 ("*rdgsr_sp64"): Likewise.
5657 ("alignaddrsi_vis"): Likewise.
5658 ("alignaddrdi_vis"): Likewise.
5659 ("alignaddrlsi_vis"): Likewise.
5660 ("alignaddrldi_vis"): Likewise.
5661 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
5662 ("fexpand_vis"): Likewise.
5663 ("fpmerge_vis"): Likewise.
5664 ("faligndata<VM64:mode>_vis"): Likewise.
5665 ("bshuffle<VM64:mode>_vis"): Likewise.
5666 ("cmask8<P:mode>_vis"): Likewise.
5667 ("cmask16<P:mode>_vis"): Likewise.
5668 ("cmask32<P:mode>_vis"): Likewise.
5669 ("fchksm16_vis"): Likewise.
5670 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
5671 ("fmean16_vis"): Likewise.
5672 ("fp<plusminus_insn>64_vis"): Likewise.
5673 ("<plusminus_insn>v8qi3"): Likewise.
5674 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
5675 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
5676 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
5677 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
5678 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
5679 ("*movqi_insn"): Likewise.
5680 ("*movhi_insn"): Likewise.
5681 ("*movsi_insn"): Likewise.
5682 ("movsi_pic_gotdata_op"): Likewise.
5683 ("*movdi_insn_sp32"): Likewise.
5684 ("*movdi_insn_sp64"): Likewise.
5685 ("movdi_pic_gotdata_op"): Likewise.
5686 ("*movsf_insn"): Likewise.
5687 ("*movdf_insn_sp32"): Likewise.
5688 ("*movdf_insn_sp64"): Likewise.
5689 ("*zero_extendhisi2_insn"): Likewise.
5690 ("*zero_extendqihi2_insn"): Likewise.
5691 ("*zero_extendqisi2_insn"): Likewise.
5692 ("*zero_extendqidi2_insn"): Likewise.
5693 ("*zero_extendhidi2_insn"): Likewise.
5694 ("*zero_extendsidi2_insn_sp64"): Likewise.
5695 ("ldfsr"): Likewise.
5696 ("prefetch_64"): Likewise.
5697 ("prefetch_32"): Likewise.
5698 ("tie_ld32"): Likewise.
5699 ("tie_ld64"): Likewise.
5700 ("*tldo_ldub_sp32"): Likewise.
5701 ("*tldo_ldub1_sp32"): Likewise.
5702 ("*tldo_ldub2_sp32"): Likewise.
5703 ("*tldo_ldub_sp64"): Likewise.
5704 ("*tldo_ldub1_sp64"): Likewise.
5705 ("*tldo_ldub2_sp64"): Likewise.
5706 ("*tldo_ldub3_sp64"): Likewise.
5707 ("*tldo_lduh_sp32"): Likewise.
5708 ("*tldo_lduh1_sp32"): Likewise.
5709 ("*tldo_lduh_sp64"): Likewise.
5710 ("*tldo_lduh1_sp64"): Likewise.
5711 ("*tldo_lduh2_sp64"): Likewise.
5712 ("*tldo_lduw_sp32"): Likewise.
5713 ("*tldo_lduw_sp64"): Likewise.
5714 ("*tldo_lduw1_sp64"): Likewise.
5715 ("*tldo_ldx_sp64"): Likewise.
5716 ("*mov<VM32:mode>_insn"): Likewise.
5717 ("*mov<VM64:mode>_insn_sp64"): Likewise.
5718 ("*mov<VM64:mode>_insn_sp32"): Likewise.
5720 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5722 * config/sparc/sparc.md ("type"): New insn type viscmp.
5723 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
5725 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
5726 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
5727 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
5728 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
5730 ("n7_vis_logical_11cycle"): Likewise.
5731 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
5732 * config/sparc/niagara2.md ("niag3_vis": Likewise.
5733 * config/sparc/niagara.md ("niag_vis"): Likewise.
5734 * config/sparc/ultra3.md ("us3_fga"): Likewise.
5735 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
5737 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
5739 * config/sparc/sparc.md: New instruction type `bmask'.
5740 (bmaskdi_vis): Use the `bmask' type.
5741 (bmasksi_vis): Likewise.
5742 * config/sparc/ultra3.md (us3_array): Likewise.
5743 * config/sparc/niagara7.md (n7_array): Likewise.
5744 * config/sparc/niagara4.md (n4_array): Likewise.
5745 * config/sparc/niagara2.md (niag2_vis): Likewise.
5746 (niag3_vis): Likewise.
5747 * config/sparc/niagara.md (niag_vis): Likewise.
5749 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5751 * ipa-comdats.c: Remove optimize check from gate.
5752 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
5753 for functions not optimized.
5754 (ipa_fn_summary_read): Skip optimize check.
5755 (ipa_fn_summary_write): Likewise.
5756 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
5758 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
5760 (can_inline_edge_p): Check flag_pcc_struct_return for match.
5761 (check_callers): Give up on caller which is not optimized.
5762 (inline_small_functions): Likewise.
5763 (ipa_inline): Do not give up when not optimizing.
5764 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
5765 away unoptimizes cdtors.
5766 (whole_program_function_and_variable_visibility): Do
5767 ipa_discover_readonly_nonaddressable_vars in LTO mode.
5768 * ipa.c (process_references): Do not check optimize.
5769 (symbol_table::remove_unreachable_nodes): Update optimize check.
5770 (set_writeonly_bit): Update optimize check.
5771 (pass_ipa_cdtor_merge::gate): Do not check optimize.
5772 (pass_ipa_single_use::gate): Remove.
5774 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5776 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
5777 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
5778 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
5779 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
5780 permute_load, permute_store, adjust_extract, adjust_splat,
5781 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
5782 replace_swap_with_copy, dump_swap_insn_table,
5783 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
5784 recombine_lvx_pattern, recombine_stvx_pattern,
5785 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
5786 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
5787 to file rs6000-p8swap.c.
5788 * config/rs6000/rs6000-p8swap.c: New file.
5789 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
5790 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
5791 and rs6000*-*-* targets.
5793 2017-07-06 David Malcolm <dmalcolm@redhat.com>
5795 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
5797 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5799 * lto-wrapper.c (merge_and_complain): Do not merge
5800 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
5801 fsigned_zeros, ftrapping_math, fwrapv.
5802 (append_compiler_options): Do not track these options.
5803 (append_linker_options): Likewie
5805 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5807 * cgraphunit.c (cgraph_node::finalize_function): When
5808 !flag_toplevel_reorde set no_reorder flag.
5809 (varpool_node::finalize_decl): Likewise.
5810 (symbol_table::compile): Drop no toplevel reorder path.
5812 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5814 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
5815 edges; zero probability is not better than uninitialized.
5817 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
5819 * asan.h (asan_sanitize_allocas_p): Declare.
5820 * asan.c (asan_sanitize_allocas_p): New function.
5821 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
5822 (handle_builtin_alloca): Likewise.
5823 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
5824 if !asan_sanitize_allocas_p.
5825 * params.def (asan-instrument-allocas): Add new option.
5826 * params.h (ASAN_PROTECT_ALLOCAS): Define.
5827 * opts.c (common_handle_option): Disable allocas sanitization for
5830 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
5832 * asan.c: Include gimple-fold.h.
5833 (get_last_alloca_addr): New function.
5834 (handle_builtin_stackrestore): Likewise.
5835 (handle_builtin_alloca): Likewise.
5836 (asan_emit_allocas_unpoison): Likewise.
5837 (get_mem_refs_of_builtin_call): Add new parameter, remove const
5838 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
5839 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
5840 (instrument_builtin_call): Pass gimple iterator to
5841 get_mem_refs_of_builtin_call.
5842 (last_alloca_addr): New global.
5843 * asan.h (asan_emit_allocas_unpoison): Declare.
5844 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
5845 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
5846 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
5847 if function calls alloca.
5848 * gimple-fold.c (replace_call_with_value): Remove static keyword.
5849 * gimple-fold.h (replace_call_with_value): Declare.
5850 * internal-fn.c: Include asan.h.
5851 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
5852 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
5854 2017-07-06 David Malcolm <dmalcolm@redhat.com>
5856 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
5857 (C_SELFTEST_FLAGS): New.
5858 (CPP_SELFTEST_FLAGS): New.
5859 (SELFTEST_DEPS): New, from deps of s-selftest.
5860 (C_SELFTEST_DEPS): New, from deps of s-selftest.
5861 (CPP_SELFTEST_DEPS): New.
5862 (selftest): Add dependency on s-selftest-c++.
5863 (s-selftest): Rename to...
5864 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
5865 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
5866 than SELFTEST_FLAGS.
5867 (selftest-gdb): Rename to...
5868 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
5870 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
5871 (selftest-valgrind): Rename to...
5872 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
5874 (selftest-valgrind): Reintroduce as an alias for
5875 selftest-c-valgrind.
5876 (s-selftest-c++): New.
5877 (selftest-c++-gdb): New.
5878 (selftest-c++-valgrind): New.
5880 2017-07-06 Olivier Hainque <hainque@adacore.com>
5882 * gcc.c (process_command): When deciding if undefined variables
5883 should be ignored when processing specs, accept "gcc -v" as well.
5885 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
5887 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
5888 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
5890 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
5892 * config/arm/arm-cpus.in (armv8-r): Add new entry.
5893 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
5894 * config/arm/arm-tables.opt: Regenerate.
5895 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
5897 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
5899 2017-07-06 Carl Love <cel@us.ibm.com>
5901 * ChangeLog: Clean up from mid air collision
5903 2017-07-06 Carl Love <cel@us.ibm.com>
5905 * config/rs6000/rs6000-c.c: Add support for built-in functions
5906 vector signed int vec_subc (vector signed int, vector signed int);
5907 vector signed __int128 vec_subc (vector signed __int128,
5908 vector signed __int128);
5909 vector unsigned __int128 vec_subc (vector unsigned __int128,
5910 vector unsigned __int128);
5911 vector signed int vec_sube (vector signed int, vector signed int,
5913 vector unsigned int vec_sube (vector unsigned int,
5914 vector unsigned int,
5915 vector unsigned int);
5916 vector signed __int128 vec_sube (vector signed __int128,
5917 vector signed __int128,
5918 vector signed__int128);
5919 vector unsigned __int128 vec_sube (vector unsigned __int128,
5920 vector unsigned __int128,
5921 vector unsigned __int128);
5922 vector signed int vec_subec (vector signed int, vector signed int,
5924 vector unsigned int vec_subec (vector unsigned int,
5925 vector unsigned int,
5926 vector unsigned int);
5927 vector signed __int128 vec_subec (vector signed __int128,
5928 vector signed __int128,
5929 vector signed__int128);
5930 vector unsigned __int128 vec_subec (vector unsigned __int128,
5931 vector unsigned __int128,
5932 vector unsigned __int128);
5933 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
5934 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
5935 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
5936 BU_ALTIVEC_OVERLOAD_X definitions.
5937 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
5938 * doc/extend.texi: Update the built-in documentation file for the new
5941 2017-07-06 David Malcolm <dmalcolm@redhat.com>
5944 * diagnostic-show-locus.c (layout::layout): Use start and finish
5945 spelling location for the start and finish of each range.
5946 * genmatch.c (linemap_client_expand_location_to_spelling_point):
5947 Add unused aspect param.
5948 * input.c (expand_location_1): Add "aspect" param, and use it
5949 to access the correct part of the location.
5950 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
5952 (expand_location_to_spelling_point): Likewise.
5953 (linemap_client_expand_location_to_spelling_point): Add "aspect"
5954 param, and pass it to expand_location_1.
5956 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
5958 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
5959 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
5960 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
5961 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
5962 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
5963 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
5964 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
5965 _mm_maskz_getmant_ss): New intrinsics.
5966 (__builtin_ia32_getexpss128_mask): Changed to ...
5967 __builtin_ia32_getexpss128_round ... this.
5968 (__builtin_ia32_getexpsd128_mask): Changed to ...
5969 __builtin_ia32_getexpsd128_round ... this.
5970 * config/i386/i386-builtin-types.def
5971 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
5972 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
5973 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
5974 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
5975 __builtin_ia32_getmantss_mask_round): New builtins.
5976 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
5977 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
5978 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
5979 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
5980 * config/i386/sse.md
5981 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
5982 avx512f_sgetexp<mode><mask_scalar_name>
5983 <round_saeonly_scalar_name> ... this.
5984 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
5985 %0, %1, %2<round_saeonly_op3>}): Changed to ...
5986 vgetexp<ssescalarmodesuffix>
5987 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
5988 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
5989 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
5990 avx512f_vgetmant<mode><mask_scalar_name>
5991 <round_saeonly_scalar_name> ... this.
5992 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
5993 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
5994 vgetmant<ssescalarmodesuffix>
5995 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
5996 %0<mask_scalar_operand4>, %1, %2
5997 <round_saeonly_scalar_mask_op4>, %3} ... this.
5998 * config/i386/subst.md (mask_scalar_operand4,
5999 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6000 round_saeonly_scalar_nimm_predicate): New subst attributes.
6002 2017-07-06 Julia Koval <julia.koval@intel.com>
6004 * config/i386/i386.c (ix86_erase_embedded_rounding):
6005 Remove code for old rounding pattern.
6007 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6009 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6011 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6013 * doc/sourcebuild.texi (Test Directives, Variants of
6014 dg-require-support): Add documentation for dg-require-stack-check.
6016 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6018 * config/i386/subst.md (mask_scalar, round_scalar,
6019 round_saeonly_scalar): New meta-templates.
6020 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6021 round_scalar_mask_operand3, round_scalar_mask_op3,
6022 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6023 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6024 round_saeonly_scalar_constraint,
6025 round_saeonly_scalar_prefix): New subst attribute.
6026 * config/i386/sse.md
6027 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6028 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6029 <round_scalar_name> ... this.
6030 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6031 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6032 <round_scalar_name> ... this.
6033 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6034 <sse>_vm<code><mode>3<mask_scalar_name>
6035 <round_saeonly_scalar_name> ... this.
6036 (v<plusminus_mnemonic><ssescalarmodesuffix>
6037 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6038 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6039 v<plusminus_mnemonic><ssescalarmodesuffix>
6040 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6041 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6042 (v<multdiv_mnemonic><ssescalarmodesuffix>
6043 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6044 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6045 v<multdiv_mnemonic><ssescalarmodesuffix>
6046 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6047 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6048 (v<maxmin_float><ssescalarmodesuffix>
6049 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6050 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6051 v<maxmin_float><ssescalarmodesuffix>
6052 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6053 %0<mask_scalar_operand3>, %1, %<iptr>2
6054 <round_saeonly_scalar_mask_op3>} ... this.
6056 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6058 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6059 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6061 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6062 Alan Hayward <alan.hayward@arm.com>
6063 David Sherwood <david.sherwood@arm.com>
6065 * combine.c (simplify_if_then_else): Remove "enum" before
6067 * compare-elim.c (can_eliminate_compare): Likewise.
6068 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6070 (aarch64_lookup_simd_builtin_type): Likewise.
6071 (aarch64_simd_builtin_type): Likewise.
6072 (aarch64_init_simd_builtin_types): Likewise.
6073 (aarch64_simd_expand_args): Likewise.
6074 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6076 (aarch64_reverse_mask): Likewise.
6077 (aarch64_simd_emit_reg_reg_move): Likewise.
6078 (aarch64_gen_adjusted_ldpstp): Likewise.
6079 (aarch64_ccmp_mode_to_code): Likewise.
6080 (aarch64_operands_ok_for_ldpstp): Likewise.
6081 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6082 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6084 (aarch64_min_divisions_for_recip_mul): Likewise.
6085 (aarch64_reassociation_width): Likewise.
6086 (aarch64_get_condition_code_1): Likewise.
6087 (aarch64_simd_emit_reg_reg_move): Likewise.
6088 (aarch64_simd_attr_length_rglist): Likewise.
6089 (aarch64_reverse_mask): Likewise.
6090 (aarch64_operands_ok_for_ldpstp): Likewise.
6091 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6092 (aarch64_gen_adjusted_ldpstp): Likewise.
6093 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6095 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6096 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6097 (arm_lookup_simd_builtin_type): Likewise.
6098 (arm_simd_builtin_type): Likewise.
6099 (arm_init_simd_builtin_types): Likewise.
6100 (arm_expand_builtin_args): Likewise.
6101 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6102 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6103 (ft32_setup_incoming_varargs): Likewise.
6104 (ft32_function_arg): Likewise.
6105 (ft32_function_arg_advance): Likewise.
6106 (ft32_pass_by_reference): Likewise.
6107 (ft32_arg_partial_bytes): Likewise.
6108 (ft32_valid_pointer_mode): Likewise.
6109 (ft32_addr_space_pointer_mode): Likewise.
6110 (ft32_addr_space_legitimate_address_p): Likewise.
6111 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6113 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6114 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6115 (iamcu_alignment): Likewise.
6116 (canonicalize_vector_int_perm): Likewise.
6117 (ix86_noce_conversion_profitable_p): Likewise.
6118 (ix86_mpx_bound_mode): Likewise.
6119 (ix86_operands_ok_for_move_multiple): Likewise.
6120 * config/microblaze/microblaze-protos.h
6121 (microblaze_expand_conditional_branch_reg): Likewise.
6122 * config/microblaze/microblaze.c
6123 (microblaze_expand_conditional_branch_reg): Likewise.
6124 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6126 (rs6000_reassociation_width): Likewise.
6127 (rs6000_invalid_binary_op): Likewise.
6128 (fusion_p9_p): Likewise.
6129 (emit_fusion_p9_load): Likewise.
6130 (emit_fusion_p9_store): Likewise.
6131 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6133 (riscv_hard_regno_mode_ok_p): Likewise.
6134 (riscv_address_insns): Likewise.
6135 (riscv_split_symbol): Likewise.
6136 (riscv_legitimize_move): Likewise.
6137 (riscv_function_value): Likewise.
6138 (riscv_hard_regno_nregs): Likewise.
6139 (riscv_expand_builtin): Likewise.
6140 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6141 (riscv_build_integer): Likewise.
6142 (riscv_split_integer): Likewise.
6143 (riscv_legitimate_constant_p): Likewise.
6144 (riscv_cannot_force_const_mem): Likewise.
6145 (riscv_regno_mode_ok_for_base_p): Likewise.
6146 (riscv_valid_base_register_p): Likewise.
6147 (riscv_valid_offset_p): Likewise.
6148 (riscv_valid_lo_sum_p): Likewise.
6149 (riscv_classify_address): Likewise.
6150 (riscv_legitimate_address_p): Likewise.
6151 (riscv_address_insns): Likewise.
6152 (riscv_load_store_insns): Likewise.
6153 (riscv_force_binary): Likewise.
6154 (riscv_split_symbol): Likewise.
6155 (riscv_force_address): Likewise.
6156 (riscv_legitimize_address): Likewise.
6157 (riscv_move_integer): Likewise.
6158 (riscv_legitimize_const_move): Likewise.
6159 (riscv_legitimize_move): Likewise.
6160 (riscv_address_cost): Likewise.
6161 (riscv_subword): Likewise.
6162 (riscv_output_move): Likewise.
6163 (riscv_canonicalize_int_order_test): Likewise.
6164 (riscv_emit_int_order_test): Likewise.
6165 (riscv_function_arg_boundary): Likewise.
6166 (riscv_pass_mode_in_fpr_p): Likewise.
6167 (riscv_pass_fpr_single): Likewise.
6168 (riscv_pass_fpr_pair): Likewise.
6169 (riscv_get_arg_info): Likewise.
6170 (riscv_function_arg): Likewise.
6171 (riscv_function_arg_advance): Likewise.
6172 (riscv_arg_partial_bytes): Likewise.
6173 (riscv_function_value): Likewise.
6174 (riscv_pass_by_reference): Likewise.
6175 (riscv_setup_incoming_varargs): Likewise.
6176 (riscv_print_operand): Likewise.
6177 (riscv_elf_select_rtx_section): Likewise.
6178 (riscv_save_restore_reg): Likewise.
6179 (riscv_for_each_saved_reg): Likewise.
6180 (riscv_register_move_cost): Likewise.
6181 (riscv_hard_regno_mode_ok_p): Likewise.
6182 (riscv_hard_regno_nregs): Likewise.
6183 (riscv_class_max_nregs): Likewise.
6184 (riscv_memory_move_cost): Likewise.
6185 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6186 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6187 (rl78_addr_space_address_mode): Likewise.
6188 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6190 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6191 (rs6000_reassociation_width): Likewise.
6192 (rs6000_invalid_binary_op): Likewise.
6193 (fusion_p9_p): Likewise.
6194 (emit_fusion_p9_load): Likewise.
6195 (emit_fusion_p9_store): Likewise.
6196 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6197 (ok_for_simple_move_operands): Likewise.
6198 (ok_for_simple_move_strict_operands): Likewise.
6199 (ok_for_simple_arith_logic_operands): Likewise.
6200 (visium_legitimize_reload_address): Likewise.
6201 (visium_select_cc_mode): Likewise.
6202 (output_cbranch): Likewise.
6203 (visium_split_double_move): Likewise.
6204 (visium_expand_copysign): Likewise.
6205 (visium_expand_int_cstore): Likewise.
6206 (visium_expand_fp_cstore): Likewise.
6207 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6208 (visium_function_arg): Likewise.
6209 (visium_function_arg_advance): Likewise.
6210 (visium_libcall_value): Likewise.
6211 (visium_setup_incoming_varargs): Likewise.
6212 (visium_legitimate_constant_p): Likewise.
6213 (visium_legitimate_address_p): Likewise.
6214 (visium_legitimize_address): Likewise.
6215 (visium_secondary_reload): Likewise.
6216 (visium_register_move_cost): Likewise.
6217 (visium_memory_move_cost): Likewise.
6218 (prepare_move_operands): Likewise.
6219 (ok_for_simple_move_operands): Likewise.
6220 (ok_for_simple_move_strict_operands): Likewise.
6221 (ok_for_simple_arith_logic_operands): Likewise.
6222 (visium_function_value_1): Likewise.
6223 (rtx_ok_for_offset_p): Likewise.
6224 (visium_legitimize_reload_address): Likewise.
6225 (visium_split_double_move): Likewise.
6226 (visium_expand_copysign): Likewise.
6227 (visium_expand_int_cstore): Likewise.
6228 (visium_expand_fp_cstore): Likewise.
6229 (visium_split_cstore): Likewise.
6230 (visium_select_cc_mode): Likewise.
6231 (visium_split_cbranch): Likewise.
6232 (output_cbranch): Likewise.
6233 (visium_print_operand_address): Likewise.
6234 * expmed.c (flip_storage_order): Likewise.
6235 * expmed.h (emit_cstore): Likewise.
6236 (flip_storage_order): Likewise.
6237 * genrecog.c (validate_pattern): Likewise.
6238 * hsa-gen.c (gen_hsa_addr): Likewise.
6239 * internal-fn.c (expand_arith_overflow): Likewise.
6240 * ira-color.c (allocno_copy_cost_saving): Likewise.
6241 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6242 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6243 (process_invariant_for_inheritance): Likewise.
6244 * lra-eliminations.c (move_plus_up): Likewise.
6245 * omp-low.c (lower_oacc_reductions): Likewise.
6246 * simplify-rtx.c (simplify_subreg): Likewise.
6247 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6248 (TARGET_CHKP_BOUND_MODE): Likewise..
6249 * targhooks.c (default_chkp_bound_mode): Likewise.
6250 (default_setup_incoming_vararg_bounds): Likewise.
6251 * targhooks.h (default_chkp_bound_mode): Likewise.
6252 (default_setup_incoming_vararg_bounds): Likewise.
6253 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6254 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6255 (have_whole_vector_shift): Likewise.
6256 * tree-vect-stmts.c (vectorizable_load): Likewise.
6257 * doc/tm.texi: Regenerate.
6259 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6261 Graceful degrade if Binutils PR21472 is not available.
6264 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6265 .rodata in flash test fails.
6266 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6267 * confgure: Regenerate.
6268 * config.in: Regenerate.
6269 * config/avr/avr.c (avr_asm_named_section)
6270 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6271 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6272 (avr_asm_init_sections): Same.
6274 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6276 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6277 (fma<VH:mode>4_intrinsic): Likewise.
6278 (*fmsub<VCVTF:mode>4): Likewise.
6279 (*fmsub<VH:mode>4_intrinsic): Likewise.
6281 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6284 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6285 Don't depend on "optimize > 0".
6286 (out_movhi_r_mr, out_movqi_mr_r): Same.
6287 (out_movhi_mr_r, out_movqi_r_mr): Same.
6288 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6289 io_address_operand on "optimize > 0".
6291 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6293 * tree-loop-distribution.c: Add general explanantion on the pass.
6294 (generate_loops_for_partition): Mark distributed loop.
6295 (pg_add_dependence_edges): New parameter. Handle alias data
6296 dependence specially and record it in the parameter if asked.
6297 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6298 (init_partition_graph_vertices, add_partition_graph_edge): New.
6299 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6300 (free_partition_graph_vdata, build_partition_graph): New.
6301 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6302 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6303 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6304 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6305 (version_for_distribution_p, finalize_partitions): New.
6306 (distribute_loop): Handle alias data dependence specially. Factor
6307 out loop fusion code as functions and call these functions.
6309 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6311 * tree-loop-distribution.c (classify_partition): New parameter and
6312 better handle reduction statement.
6313 (rdg_build_partitions): Revise comment.
6314 (distribute_loop): Compute statements in all partitions and pass it
6315 to classify_partition.
6317 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6319 * tree-loop-distribution.c (enum partition_type): New.
6320 (struct partition): New field type.
6321 (partition_merge_into): Add parameter. Update partition type.
6322 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6323 (build_rdg_partition_for_vertex): Compute partition type.
6324 (rdg_build_partitions): Dump partition type.
6325 (distribute_loop): Update calls to partition_merge_into.
6327 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6329 * tree-loop-distribution.c (struct ddr_hasher): New.
6330 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6332 (classify_partition): Call get_data_dependence.
6333 (pg_add_dependence_edges): Ditto.
6334 (distribute_loop): Release data dependence hash table.
6336 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6338 * tree-loop-distribution.c (ref_base_address): Delete.
6339 (similar_memory_accesses): Rename ...
6340 (share_memory_accesses): ... to this. Check if partitions access
6341 the same memory reference.
6342 (distribute_loop): Call share_memory_accesses.
6344 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6346 * tree-loop-distribution.c (struct partition): New field recording
6348 (partition_alloc, partition_free): Init and release data refs.
6349 (partition_merge_into): Merge data refs.
6350 (build_rdg_partition_for_vertex): Collect data refs for partition.
6351 (pg_add_dependence_edges): Change parameters from vector to bitmap.
6353 (distribute_loop): Remve data refs from vertice data of partition
6356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6358 * tree-loop-distribution.c (params.h): Include header file.
6359 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
6360 (datarefs_vec): New global var.
6361 (create_rdg_vertices): Use datarefs_vec directly.
6362 (free_rdg): Don't free data references.
6363 (build_rdg): Update use. Don't free data references.
6364 (distribute_loop): Compute global variable for data references.
6365 Bail out if there are too many data references.
6367 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6369 * tree-loop-distribution.c (loop_nest): New global var.
6370 (build_rdg): Use loop directly, rather than loop nest.
6371 (pg_add_dependence_edges): Remove loop nest parameter. Use global
6373 (distribute_loop): Compute global variable loop nest. Update use.
6375 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6377 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
6378 (partition_merge_into): New parameter. Dump reason for fusion.
6379 (distribute_loop): Update use of partition_merge_into.
6381 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6383 * tree-loop-distribution.c (bb_top_order_index): New.
6384 (bb_top_order_index_size, bb_top_order_cmp): New.
6385 (stmts_from_loop): Use topological order.
6386 (pass_loop_distribution::execute): Compute and release topological
6387 order for basic blocks.
6389 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6391 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
6394 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6396 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
6397 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
6398 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
6399 * internal-fn.def (LOOP_DIST_ALIAS): New.
6400 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
6401 (fold_loop_internal_call): ... this.
6402 (vect_loop_dist_alias_call): New function.
6403 (set_uid_loop_bbs): Call fold_loop_internal_call.
6404 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
6407 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6410 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
6411 Require dead FLAGS_REG at the beginning of a peephole.
6413 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
6416 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
6417 arguments in the call to __builtin_ia32_sbb_u32.
6418 (_subborrow_u64): Swap _X and _Y arguments in the call to
6419 __builtin_ia32_sbb_u64.
6421 2017-07-04 Jakub Jelinek <jakub@redhat.com>
6424 * tree-vrp.c (compare_assert_loc): Turn into a function template
6425 with stable template parameter. Only test if a->e is NULL,
6426 !a->e == !b->e has been verified already. Use e == NULL or
6427 e != NULL instead of e or ! e tests. If stable is true, don't use
6428 iterative_hash_expr, on the other side allow a or b or both NULL
6429 and sort the NULLs last.
6430 (process_assert_insertions): Sort using compare_assert_loc<false>
6431 instead of compare_assert_loc, later sort using
6432 compare_assert_loc<true> before calling process_assert_insertions_for
6433 in a loop. Use break instead of continue once seen NULL pointer.
6435 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6437 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6438 Cortex-R7 and Cortex-R8 processors.
6440 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6442 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
6443 uninitialized while src is not.
6445 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
6447 * common/config/arm/arm-common.c: Adjust include path for
6449 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
6450 (arm-cpu.h): Create in build directory. Adjust dependency rules.
6451 (arm-cpu-data.h): Likewise.
6452 (arm-cpu-cdata.h): Likewise.
6453 * config/arm/arm-cpu.h: Delete.
6454 * config/arm/arm-cpu-cdata.h: Delete.
6455 * config/arm/arm-cpu-data.h: Delete.
6457 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
6459 * config/arm/arm-cpus.in (cortex-a55): New.
6460 (cortex-a75): Likewise.
6461 (cortex-a75.cortex-a55): Likewise.
6462 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
6464 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
6465 * config/arm/arm-cpu-cdata.h: Regenerate.
6466 * config/arm/arm-cpu-data.h: Regenerate.
6467 * config/arm/arm-cpu.h: Regenerate.
6468 * config/arm/arm-tables.opt: Regenerate.
6469 * config/arm/arm-tune.md: Regenerate.
6471 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6473 * haifa-sched.c (sched_create_recovery_edges): Update profile.
6475 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
6477 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
6480 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
6482 PR tree-optimization/81292
6483 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
6484 full_string_p, also call adjust_related_strinfos if the adjustment
6485 is simple, otherwise invalidate related strinfos.
6487 2017-07-04 Martin Liska <mliska@suse.cz>
6490 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
6491 newly created variable as DECL_IGNORED_P.
6493 2017-07-04 Martin Liska <mliska@suse.cz>
6496 * ipa-inline.c (inline_small_functions):
6497 Use xstrdup_for_dump.
6499 2017-07-04 Tom de Vries <tom@codesourcery.com>
6501 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
6503 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6506 * config/darwin.c (darwin_function_switched_text_sections):
6509 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
6511 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
6513 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6515 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
6517 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6519 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
6520 min_profitable_iters, and th as inclusive lower bounds.
6521 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
6522 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
6523 for min_profitable_iters and min_profitable_estimate.
6524 (vect_transform_loop): Treat th as an inclusive lower bound.
6525 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6527 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
6530 * config/darwin.c (darwin_function_switched_text_sections):
6531 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
6532 in two pieces, and suppress the use of buf.
6534 2017-07-03 Nathan Sidwell <nathan@acm.org>
6536 * hash-table.h (hash_table_mod1): Fix indentation.
6538 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6541 * predict.c (force_edge_cold): Be more careful about propagation
6543 * profile-count.h (profile_probability::guessed,
6544 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
6546 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
6548 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
6550 * doc/invoke.texi (rcpc architecture extension): Document it.
6552 2017-07-03 Richard Biener <rguenther@suse.de>
6554 PR tree-optimization/60510
6555 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
6556 the scalar reduction PHI and use it.
6557 (vectorizable_reduction): Properly guard the single_defuse_cycle
6558 path for non-SLP reduction chains where we cannot use it.
6559 Rework reduc_def/index and vector type deduction. Rework
6560 vector operand gathering during reduction op code-gen.
6561 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
6562 chains dissolve the chain and leave it to non-SLP reduction
6565 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6567 * tree-data-ref.h (dr_alignment): Declare.
6568 * tree-data-ref.c (dr_alignment): New function.
6569 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
6570 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
6572 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
6574 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6576 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
6577 and base_misalignment fields.
6578 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
6579 * tree-data-ref.c: Include builtins.h.
6580 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
6581 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
6582 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
6583 * tree-vect-data-refs.c: Include tree-cfg.h.
6584 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
6585 fields instead of calculating an alignment here.
6586 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
6587 innermost_loop_behavior fields.
6589 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6591 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
6593 (DR_STEP_ALIGNMENT): New macro.
6594 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
6595 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
6596 (create_data_ref): Print it.
6597 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
6598 to tell whether the step preserves vector (mis)alignment.
6599 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6600 Move the check for an integer step and generalise to all INTEGER_CST.
6601 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
6602 Print the outer step alignment.
6604 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6606 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
6607 with offset_alignment.
6608 (DR_ALIGNED_TO): Delete.
6609 (DR_OFFSET_ALIGNMENT): New macro.
6610 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
6611 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
6612 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
6613 (create_data_ref): Likewise.
6614 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
6615 (vect_analyze_data_refs): Likewise.
6616 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
6617 creating dummy innermost behavior.
6619 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6621 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
6622 with a "innermost_loop_behavior *" and refeence tree.
6623 * tree-data-ref.c (dr_analyze_innermost): Likewise.
6624 (create_data_ref): Update call accordingly.
6625 * tree-predcom.c (find_looparound_phi): Likewise.
6627 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6629 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
6630 fields with dr_wrt_vec_loop.
6631 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
6632 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
6633 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
6634 (vect_dr_behavior): New function.
6635 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6636 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
6637 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
6638 track whether the step preserves the misalignment.
6639 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
6640 Use vect_dr_behavior.
6641 (vect_setup_realignment): Update call accordingly.
6642 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
6643 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
6644 call to vect_create_addr_base_for_vector_ref.
6645 (vect_create_cond_for_align_checks): Likewise.
6646 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
6647 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
6648 (vect_recog_mask_conversion_pattern): Likewise.
6649 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
6650 (new_stmt_vec_info): Remove redundant zeroing.
6652 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
6654 * common/config/arm/arm-common.c (arm_be8_option): New function.
6655 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
6656 (ISA_ARMv6): Add isa_bit_be8.
6657 * config/arm/arm.h (arm_be8_option): Add prototype.
6658 (BE8_SPEC_FUNCTION): New define.
6659 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
6660 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
6661 (mlittle-endian): Similarly.
6662 (mbe8, mbe32): New options.
6663 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
6664 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
6666 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6668 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
6670 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6672 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
6673 (cleanup_tree_cfg_bb): Use it.
6674 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
6676 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
6678 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6681 * loop-doloop.c (add_test): Update profile.
6683 2017-07-03 Martin Liska <mliska@suse.cz>
6686 * sanopt.c (rewrite_usage_of_param): New function.
6687 (sanitize_rewrite_addressable_params): Likewise.
6688 (pass_sanopt::execute): Call rewrite_usage_of_param.
6690 2017-07-03 Richard Biener <rguenther@suse.de>
6692 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
6693 back to using VIEW_CONVERT_EXPR.
6695 2017-07-03 Martin Liska <mliska@suse.cz>
6698 * doc/extend.texi: Document when a resolver function is
6699 generated for target_clones.
6701 2017-07-03 Martin Liska <mliska@suse.cz>
6703 * asan.c (asan_emit_stack_protection): Unpoison just red zones
6704 and shadow memory of auto variables which are subject of
6705 use-after-scope sanitization.
6706 (asan_expand_mark_ifn): Add do set only when is_poison.
6708 2016-07-03 Richard Biener <rguenther@suse.de>
6710 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
6712 (vect_force_simple_reduction): Record reduction def -> phi mapping.
6713 (vectorizable_reduction): Perform reduction PHI creation when
6714 visiting a reduction PHI and adjust and simplify code generation
6715 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
6716 (vect_transform_loop): Visit reduction PHIs.
6717 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
6718 defs into the SLP tree.
6719 (vect_build_slp_tree): Reduction defs terminate the recursion.
6720 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
6722 (vect_get_vec_defs_for_stmt_copy): Export.
6723 (vect_get_vec_defs): Likewise.
6724 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
6726 (vect_get_vec_defs_for_stmt_copy): Declare.
6727 (vect_get_vec_defs): Likewise.
6729 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
6731 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
6732 parameter with a "loop" parameter and use it instead of the
6733 loop containing DR_STMT. Don't check simple_iv when doing
6734 BB analysis. Describe the two analysis modes in the comment.
6736 2017-07-03 Tom de Vries <tom@codesourcery.com>
6738 PR tree-optimization/69468
6739 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
6740 (find_same_succ_bb): Handle ignore_edge_flags.
6742 2017-07-03 Tom de Vries <tom@codesourcery.com>
6744 PR tree-optimization/81192
6745 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
6747 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
6749 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
6751 2017-07-03 Tom de Vries <tom@codesourcery.com>
6753 PR tree-optimization/81192
6754 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
6755 BB_SAME_SUCC (bb) == NULL.
6757 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6759 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
6762 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6764 * dumpfile.c: Include profile-count.h
6765 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
6767 (insert_cond_bb): Update profile.
6768 * tree-cfg.h (insert_cond_bb): Update prototype.
6769 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
6770 * tree-dump.c: Do not include tree-cfg.
6772 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6774 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
6776 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6778 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
6781 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
6783 * tree-complex.c (expand_complex_div_wide): update profile.
6785 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6786 Alan Hayward <alan.hayward@arm.com>
6787 David Sherwood <david.sherwood@arm.com>
6789 * Makefile.in (MACHMODE_H): Remove insn-modes.h
6790 (CORETYPES_H): New define.
6791 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
6792 (insn-modes-inline.h, s-modes-inline-h): New rules.
6793 (generated_files): Add insn-modes-inline.h.
6794 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
6795 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
6796 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
6797 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
6798 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
6799 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
6800 (build/gencodes.o, build/genconditions.o): Likewise.
6801 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
6802 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
6803 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
6804 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
6805 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
6806 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
6807 * coretypes.h: Include everything up to real.h for generators.
6808 Include insn-modes.h first. Include wide-int-print.h after
6809 wide-int.h. Include insn-modes-inline.h and then machmode.h.
6810 * machmode.h: Don't include insn-modes.h here.
6811 * function-tests.c: Remove includes of signop.h, machmode.h,
6812 double-int.h and wide-int.h.
6814 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
6816 * optc-save-gen.awk: Likewise.
6817 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
6818 * godump.c: Remove include of wide-int-print.h.
6819 * pretty-print.h: Likewise.
6820 * wide-int-print.cc: Likewise.
6821 * wide-int.cc: Likewise.
6822 * hash-map-tests.c: Remove include of signop.h.
6823 * hash-set-tests.c: Likewise.
6824 * rtl-tests.c: Likewise.
6825 * mkconfig.sh: Remove include of machmode.h.
6826 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
6828 (emit_insn_modes_inline_h): ...this new function. Emit the code
6829 into an insn-modes-inline.h header file, adding appropriate
6830 include guards and end comments.
6831 (emit_insn_modes_c_header): Remove include of machmode.h.
6832 (emit_min_insn_modes_c_header): Include coretypes.h rather than
6834 (main): Handle -i flag and call emit_insn_modes_inline_h when
6837 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6839 * tree-ssa-strlen.c (strinfo): Rename the length field to
6840 nonzero_chars. Add a full_string_p field.
6841 (compare_nonzero_chars, zero_length_string_p): New functions.
6842 (get_addr_stridx): Add an offset_out parameter.
6843 Use compare_nonzero_chars.
6844 (get_stridx): Update accordingly. Use compare_nonzero_chars.
6845 (new_strinfo): Update after above changes to strinfo.
6846 (set_endptr_and_length): Set full_string_p.
6847 (get_string_length): Update after above changes to strinfo.
6848 (unshare_strinfo): Update call to new_strinfo.
6849 (maybe_invalidate): Likewise.
6850 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
6851 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
6852 as a uhwi instead of an shwi. Update after above changes to
6853 strinfo and new_strinfo.
6854 (zero_length_string): Assert that chainsi contains full strings.
6855 Use zero_length_string_p. Update call to new_strinfo.
6856 (adjust_related_strinfos): Update after above changes to strinfo.
6857 Copy full_string_p from origsi.
6858 (adjust_last_stmt): Use zero_length_string_p.
6859 (handle_builtin_strlen): Update after above changes to strinfo and
6860 new_strinfo. Install the lhs as the string length if the previous
6861 entry didn't describe a full string.
6862 (handle_builtin_strchr): Update after above changes to strinfo
6864 (handle_builtin_strcpy): Likewise.
6865 (handle_builtin_strcat): Likewise.
6866 (handle_builtin_malloc): Likewise.
6867 (handle_pointer_plus): Likewise.
6868 (handle_builtin_memcpy): Likewise. Track nonzero characters
6869 that aren't necessarily followed by a nul terminator.
6870 (handle_char_store): Likewise.
6872 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6874 PR tree-optimization/80769
6875 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
6876 for malloc and calloc. Document the new invariant that all related
6877 strinfos have delayed lengths or none do.
6878 (verify_related_strinfos): Move earlier in file.
6879 (set_endptr_and_length): New function, split out from...
6880 (get_string_length): ...here. Also set the lengths of related
6882 (zero_length_string): Assert that chainsi has known (rather than
6884 (adjust_related_strinfos): Likewise.
6886 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
6888 PR tree-optimization/81136
6889 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
6890 assert that two references with the same misalignment have the same
6891 compile-time misalignment if those compile-time misalignments
6894 2017-07-01 Andi Kleen <ak@linux.intel.com>
6896 * print-tree.c (print_node): Print all attributes.
6898 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6900 * cfg.c (scale_bbs_frequencies): New function.
6901 * cfg.h (scale_bbs_frequencies): Declare it.
6902 * cfgloopanal.c (single_likely_exit): Cleanup.
6903 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
6905 (scale_loop_profile): Likewise.
6906 (loop_version): Likewise.
6907 (create_empty_loop_on_edge): Update.
6908 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
6909 scale_loop_frequencies, scale_loop_profile, loopify,
6910 loop_version): Update prototypes.
6911 * modulo-sched.c (sms_schedule): Update.
6912 * predict.c (unlikely_executed_edge_p): Also check probability.
6913 (probably_never_executed_edge_p): Fix typo.
6914 * tree-if-conv.c (version_loop_for_if_conversion): Update.
6915 * tree-parloops.c (gen_parallel_loop): Update.
6916 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
6917 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6918 * tree-ssa-loop-split.c (split_loop): Update.
6919 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
6920 * tree-vect-loop-manip.c (vect_do_peeling): Update.
6921 (vect_loop_versioning): Update.
6922 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6924 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6926 * trans-mem.c (split_bb_make_tm_edge): Update profile.
6928 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6930 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
6931 to keep profile consistent.
6933 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6935 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
6936 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
6937 * profile-count.h (max_safe_multiplier): Make unsigned.
6938 (profile_count::guessed_zero): New.
6940 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6942 * bb-reorder.c (fix_up_crossing_landing_pad,
6943 fix_crossing_conditional_branches): Use make_single_succ_edge
6944 to keep profile consistent.
6946 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
6948 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
6951 2017-07-01 Jakub Jelinek <jakub@redhat.com>
6954 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
6955 the right scopes, make sure cond_jump isn't preserved between multiple
6956 iterations. Search for fallthru edge whenever there are 3+ edges and
6957 use find_fallthru_edge for it.
6959 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
6961 Patch by Alexander Monakov <amonakov@ispras.ru>
6962 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
6963 probabilities consistently.
6965 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
6967 * pa.c (pa_expand_compare_and_swap_loop): Update call of
6968 emit_cmp_and_jump_insns.
6970 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
6973 * tree-inline.c (expand_call_inline): Combine profile statuses.
6975 2017-06-30 Andrew Pinski <apinski@cavium.com>
6977 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
6978 fold_stmt returned true.
6980 2017-06-30 Nathan Sidwell <nathan@acm.org>
6982 * ggc.h (empty_string): Delete.
6983 * cfgexpand.c (expand_asm_stmt): Use plain "".
6984 * optabs.c (expand_asm_memory_barrier): Likewise.
6985 * stringpool.c (empty_string): Delete.
6986 (digit_vector, digit_string): Delete.
6987 (ggc_alloc_string): Use plain "", don't optimize single digit
6988 strings. Use ggc_alloc_atomic.
6990 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
6992 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
6993 comparison set and one other set, use the cost of the non-comparison
6996 2017-06-30 Nathan Sidwell <nathan@acm.org>
6998 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7001 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7003 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7004 loops. Remove now unneeded calls to gimple_switch_set_label() that
7005 just set removed labels to NULL_TREE.
7007 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7009 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7010 (set_range_info): ...here. Only call set_range_info_raw if domain
7012 (set_nonzero_bits): Call set_range_info_raw.
7013 * tree-ssanames.h (set_range_info_raw): New.
7015 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7018 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7019 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7020 of nonimmediate_operand and <store_mask_constraint> instead of m
7021 for the input operand. For V8FI iterator, always split if input
7022 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7023 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7024 <store_mask_predicate> instead of register_operand and
7025 <store_mask_constraint> instead of v for the input operand. Make
7026 sure both operands aren't MEMs for if not <mask_applied>.
7028 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7030 * lto-wrapper.c (copy_file) Close both file descriptors before
7033 2017-06-30 Martin Liska <mliska@suse.cz>
7036 * multiple_target.c (create_dispatcher_calls): Make ifunc
7037 also for function that don't have calls or are not referenced.
7039 2017-06-30 Richard Biener <rguenther@suse.de>
7041 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7042 analyze the first scalar stmt. Move vector type computation
7043 for the BB case here from ...
7044 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7045 live operation processing in the SLP case properly.
7047 2017-06-30 Richard Biener <rguenther@suse.de>
7049 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7051 2017-06-30 Martin Liska <mliska@suse.cz>
7054 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7055 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7057 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7059 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7062 2017-06-30 Richard Biener <rguenther@suse.de>
7064 PR tree-optimization/81249
7065 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7066 condition reduction result to original scalar type.
7068 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7070 * profile-count.h (enum profile_quality): Fix typos and whitespace
7073 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7075 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7076 type for branch probabilities.
7078 2017-06-29 Julian Brown <julian@codesourcery.com>
7079 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7081 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7082 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7083 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7084 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7086 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7088 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7089 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7090 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7091 CC usage from generic code to here.
7092 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7093 CC usage into the target macros.
7095 2017-06-29 Maya Rashish <coypu@sdf.org>
7097 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7100 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7102 * arm/arm-builtins.c: Include profile-count.h
7103 * except.c (sjlj_emit_function_enter): Use
7104 profile_probability::unlikely.
7106 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7108 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7109 and tocrel_offset be pointer args rather than implicitly using
7111 (legitimate_constant_pool_address_p, rs6000_emit_move,
7112 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7113 tocrel_offset and use in toc_relative_expr_p call.
7114 (print_operand, print_operand_address): Use static tocrel_base_oac
7115 and tocrel_offset_oac.
7116 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7119 2017-06-29 Maya Rashish <coypu@sdf.org>
7121 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7123 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7125 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7126 objects, take into account only the alignment of 'op0' and 'mode1' if
7129 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7131 * ccmp.c (ccmp_tree_comparison_p): New function.
7132 (ccmp_candidate_p): Update to use above function.
7133 (get_compare_parts): New function.
7134 (expand_ccmp_next): Update to use new functions.
7135 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7137 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7138 take mode as argument.
7139 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7140 * expr.c (expand_expr_real_1): Pass mode as argument.
7142 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7144 * combine.c (combine_instructions): Print insns to dump_file, together
7147 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7149 * asan.c (asan_emit_stack_protection): Update.
7150 (create_cond_insert_point): Update.
7151 * auto-profile.c (afdo_propagate_circuit): Update.
7152 * basic-block.h (struct edge_def): Turn probability to
7153 profile_probability.
7154 (EDGE_FREQUENCY): Update.
7155 * bb-reorder.c (find_traces_1_round): Update.
7156 (better_edge_p): Update.
7157 (sanitize_hot_paths): Update.
7158 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7159 (make_single_succ_edge): Update.
7160 (check_bb_profile): Update.
7161 (dump_edge_info): Update.
7162 (update_bb_profile_for_threading): Update.
7163 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7164 probabilitycount to 0.
7165 * cfgbuild.c (compute_outgoing_frequencies): Update.
7166 * cfgcleanup.c (try_forward_edges): Update.
7167 (outgoing_edges_match): Update.
7168 (try_crossjump_to_edge): Update.
7169 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7170 (expand_gimple_tailcall): Update.
7171 (construct_init_block): Use make_single_succ_edge.
7172 (construct_exit_block): Use make_single_succ_edge.
7173 * cfghooks.c (verify_flow_info): Update.
7174 (redirect_edge_succ_nodup): Update.
7175 (split_edge): Update.
7176 (account_profile_record): Update.
7177 * cfgloopanal.c (single_likely_exit): Update.
7178 * cfgloopmanip.c (scale_loop_profile): Update.
7179 (set_zero_probability): Remove.
7180 (duplicate_loop_to_header_edge): Update.
7181 * cfgloopmanip.h (loop_version): Update prototype.
7182 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7183 (force_nonfallthru_and_redirect): Update.
7184 (update_br_prob_note): Update.
7185 (rtl_verify_edges): Update.
7186 (purge_dead_edges): Update.
7187 (rtl_lv_add_condition_to_bb): Update.
7188 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7189 * cgraphunit.c (init_lowered_empty_function): Update.
7190 (cgraph_node::expand_thunk): Update.
7191 * cilk-common.c: Include profile-count.h
7192 * dojump.c (inv): Remove.
7193 (jumpifnot): Update.
7194 (jumpifnot_1): Update.
7195 (do_jump_1): Update.
7197 (do_jump_by_parts_greater_rtx): Update.
7198 (do_compare_rtx_and_jump): Update.
7199 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7200 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7201 * dwarf2cfi.c: Include profile-count.h
7202 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7203 (sjlj_emit_dispatch_table): Likewise.
7204 * explow.c: Include profile-count.h
7205 * expmed.c (emit_store_flag_force): Update.
7206 (do_cmp_and_jump): Update.
7207 * expr.c (compare_by_pieces_d::generate): Update.
7208 (compare_by_pieces_d::finish_mode): Update.
7209 (emit_block_move_via_loop): Update.
7210 (store_expr_with_bounds): Update.
7211 (store_constructor): Update.
7212 (expand_expr_real_2): Update.
7213 (expand_expr_real_1): Update.
7214 * expr.h (try_casesi, try_tablejump): Update prototypes.
7215 * gimple-pretty-print.c (dump_probability): Update.
7216 (dump_profile): New.
7217 (dump_gimple_label): Update.
7218 (dump_gimple_bb_header): Update.
7219 * graph.c (draw_cfg_node_succ_edges): Update.
7220 * hsa-gen.c (convert_switch_statements): Update.
7221 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7222 (find_if_case_1): Update.
7223 (find_if_case_2): Update.
7224 * internal-fn.c (expand_arith_overflow_result_store): Update.
7225 (expand_addsub_overflow): Update.
7226 (expand_neg_overflow): Update.
7227 (expand_mul_overflow): Update.
7228 (expand_vector_ubsan_overflow): Update.
7229 * ipa-cp.c (good_cloning_opportunity_p): Update.
7230 * ipa-split.c (split_function): Use make_single_succ_edge.
7231 * ipa-utils.c (ipa_merge_profiles): Update.
7232 * loop-doloop.c (add_test): Update.
7233 (doloop_modify): Update.
7234 * loop-unroll.c (compare_and_jump_seq): Update.
7235 (unroll_loop_runtime_iterations): Update.
7236 * lra-constraints.c (lra_inheritance): Update.
7237 * lto-streamer-in.c (input_cfg): Update.
7238 * lto-streamer-out.c (output_cfg): Update.
7239 * mcf.c (adjust_cfg_counts): Update.
7240 * modulo-sched.c (sms_schedule): Update.
7241 * omp-expand.c (expand_omp_for_init_counts): Update.
7242 (extract_omp_for_update_vars): Update.
7243 (expand_omp_ordered_sink): Update.
7244 (expand_omp_for_ordered_loops): Update.
7245 (expand_omp_for_generic): Update.
7246 (expand_omp_for_static_nochunk): Update.
7247 (expand_omp_for_static_chunk): Update.
7248 (expand_cilk_for): Update.
7249 (expand_omp_simd): Update.
7250 (expand_omp_taskloop_for_outer): Update.
7251 (expand_omp_taskloop_for_inner): Update.
7252 * omp-simd-clone.c (simd_clone_adjust): Update.
7253 * optabs.c (expand_doubleword_shift): Update.
7254 (expand_abs): Update.
7255 (emit_cmp_and_jump_insn_1): Update.
7256 (expand_compare_and_swap_loop): Update.
7257 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7258 * predict.c (predictable_edge_p): Update.
7259 (edge_probability_reliable_p): Update.
7260 (set_even_probabilities): Update.
7261 (combine_predictions_for_insn): Update.
7262 (combine_predictions_for_bb): Update.
7263 (propagate_freq): Update.
7264 (estimate_bb_frequencies): Update.
7265 (force_edge_cold): Update.
7266 * profile-count.c (profile_count::dump): Add missing space into dump.
7267 (profile_count::debug): Add newline.
7268 (profile_count::differs_from_p): Explicitly convert to unsigned.
7269 (profile_count::stream_in): Update.
7270 (profile_probability::dump): New member function.
7271 (profile_probability::debug): New member function.
7272 (profile_probability::differs_from_p): New member function.
7273 (profile_probability::differs_lot_from_p): New member function.
7274 (profile_probability::stream_in): New member function.
7275 (profile_probability::stream_out): New member function.
7276 * profile-count.h (profile_count_quality): Rename to ...
7277 (profile_quality): ... this one.
7278 (profile_probability): New.
7279 (profile_count): Update.
7280 * profile.c (compute_branch_probabilities): Update.
7281 * recog.c (peep2_attempt): Update.
7282 * sched-ebb.c (schedule_ebbs): Update.
7283 * sched-rgn.c (find_single_block_region): Update.
7284 (compute_dom_prob_ps): Update.
7285 (schedule_region): Update.
7286 * sel-sched-ir.c (compute_succs_info): Update.
7287 * stmt.c (struct case_node): Update.
7288 (do_jump_if_equal): Update.
7289 (get_outgoing_edge_probs): Update.
7290 (conditional_probability): Update.
7291 (emit_case_dispatch_table): Update.
7292 (expand_case): Update.
7293 (expand_sjlj_dispatch_table): Update.
7294 (emit_case_nodes): Update.
7295 * targhooks.c: Update.
7296 * tracer.c (better_p): Update.
7297 (find_best_successor): Update.
7298 * trans-mem.c (expand_transaction): Update.
7299 * tree-call-cdce.c: Update.
7300 * tree-cfg.c (gimple_split_edge): Upate.
7301 (move_sese_region_to_fn): Upate.
7302 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7303 * tree-eh.c (lower_resx): Upate.
7304 (cleanup_empty_eh_move_lp): Upate.
7305 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7306 * tree-inline.c (copy_edges_for_bb): Update.
7307 (copy_cfg_body): Update.
7308 * tree-parloops.c (gen_parallel_loop): Update.
7309 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7310 (gimple_gen_time_profiler): Update.
7311 * tree-ssa-dce.c (remove_dead_stmt): Update.
7312 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7313 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7314 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7315 (unloop_loops): Update.
7316 (try_peel_loop): Update.
7317 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7318 * tree-ssa-loop-split.c (connect_loops): Update.
7319 (split_loop): Update.
7320 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7321 (hoist_guard): Update.
7322 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7323 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7324 (value_replacement): Update.
7325 * tree-ssa-reassoc.c (branch_fixup): Update.
7326 * tree-ssa-tail-merge.c (replace_block_by): Update.
7327 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7328 (create_edge_and_update_destination_phis): Update.
7329 (compute_path_counts): Update.
7330 (recompute_probabilities): Update.
7331 (update_joiner_offpath_counts): Update.
7332 (freqs_to_counts_path): Update.
7333 (duplicate_thread_path): Update.
7334 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7335 (struct switch_conv_info): Update.
7336 (gen_inbound_check): Update.
7337 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7338 (vect_do_peeling): Update.
7339 (vect_loop_versioning): Update.
7340 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7341 (optimize_mask_stores): Update.
7342 * ubsan.c (ubsan_expand_null_ifn): Update.
7343 * value-prof.c (gimple_divmod_fixed_value): Update.
7344 (gimple_divmod_fixed_value_transform): Update.
7345 (gimple_mod_pow2): Update.
7346 (gimple_mod_pow2_value_transform): Update.
7347 (gimple_mod_subtract): Update.
7348 (gimple_mod_subtract_transform): Update.
7349 (gimple_ic): Update.
7350 (gimple_stringop_fixed_value): Update.
7351 (gimple_stringops_transform): Update.
7352 * value-prof.h: Update.
7354 2017-06-29 Carl Love <cel@us.ibm.com>
7356 * config/rs6000/rs6000-c.c: Add support for built-in functions
7357 vector signed int vec_signed (vector float);
7358 vector signed long long vec_signed (vector double);
7359 vector signed int vec_signed2 (vector double, vector double);
7360 vector signed int vec_signede (vector double);
7361 vector signed int vec_signedo (vector double);
7362 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
7363 instruction generator.
7364 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7365 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
7366 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
7368 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
7369 vunsignede_v2df): Add define_expands.
7370 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
7371 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
7372 VEC_UNSIGNEDO): Add definitions.
7373 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
7374 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
7375 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
7376 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
7377 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
7378 * config/rs6000/altivec.h (vec_signed, vec_signed2,
7379 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
7380 vec_unsignede, vec_unsignedo): Add builtin defines.
7381 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
7383 * doc/extend.texi: Update the built-in documentation file for the
7384 new built-in functions.
7386 2017-06-29 Richard Biener <rguenther@suse.de>
7388 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
7389 reduction chains to LOOP_VINFO_REDUCTIONS.
7390 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
7391 SLP reductions after processing reduction chains.
7393 2017-06-29 Nathan Sidwell <nathan@acm.org>
7395 * builtins.c (fold_builtin_FUNCTION): Use
7396 lang_hooks.decl_printable_name.
7398 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
7401 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
7402 with only one label.
7403 * stmt.c (expand_case): Assert NCASES is greater than one.
7405 2017-06-29 Richard Biener <rguenther@suse.de>
7407 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
7409 (group_case_labels): Likewise.
7410 (find_taken_edge): Push sanity checking on val to workers...
7411 (find_taken_edge_cond_expr): ... here
7412 (find_taken_edge_switch_expr): ... and here, handle cases
7413 with just a default label.
7414 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
7415 (group_case_labels): Likewise.
7416 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
7417 group_case_labels does anything cleanup the CFG again.
7419 2017-06-29 Bin Cheng <bin.cheng@arm.com>
7421 PR tree-optimization/81196
7422 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
7423 exit condition comparing two IVs.
7425 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
7427 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
7428 profile to the dummy entry at the end of the list of architectures.
7429 * config/arm/arm-cpu-cdata.h: Regenerated.
7431 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7432 Michael Collison <michael.collison@arm.com>
7435 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
7437 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
7438 (*aarch64_reg_<mode>3_minus_mask): New pattern.
7439 (*aarch64_<optab>_reg_di3_mask2): New pattern.
7440 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
7441 of shift when the shift amount is masked with constant equal to
7442 the size of the mode.
7443 * config/aarch64/predicates.md (subreg_lowpart_operator): New
7446 2017-06-29 Martin Liska <mliska@suse.cz>
7448 * config/i386/i386.opt: Change range from [1,5] to [0,5].
7450 2017-06-29 Yury Gribov <tetra2005@gmail.com>
7453 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
7456 (edge_growth_cache_entry::edge_growth_cache_entry): New
7458 (reset_edge_growth_cache): Update to use constructor.
7460 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7462 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
7463 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7464 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
7466 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
7468 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
7469 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
7471 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
7473 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
7474 (*-linux-uclibc*): Add t-uclibc tmake_file.
7475 * config/t-musl: New.
7476 * config/t-uclibc: New.
7478 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
7480 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
7482 (gen_comm_data): Emit architectural setting of arch_prof.
7483 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
7485 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
7486 (armv8-m.base, armv8-m.main): Likewise.
7487 * arm-protos.h (arm_build_target): Add profile field.
7488 (arch_option): Likewise.
7489 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
7491 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
7492 arm_active_target.profile.
7494 2017-06-28 Richard Biener <rguenther@suse.de>
7497 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
7498 TYPE_OVERFLOW_WRAPS.
7499 * match.pd (negate_expr_p): Likewise.
7500 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
7501 fold_build2, not fold_binary.
7503 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7505 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
7506 Convert memory address to Pmode.
7507 (aarch64_print_operand): Assert MEM operands are always Pmode.
7509 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
7512 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
7513 Remove redundant if.
7514 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
7515 * config/arm/aarch-common-protos.h
7516 (aarch_forward_to_shift_is_not_shifted_re): Remove.
7517 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
7519 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
7522 * multiple_target.c (create_dispatcher_calls): Set the default
7523 clone to be static, not public.
7525 2017-06-28 Richard Biener <rguenther@suse.de>
7527 * tree-vect-loop.c (vectorizable_reduction): Move special
7528 cond reduction IV var creation ...
7529 (vect_create_epilog_for_reduction): ... here. Remove induction_index
7530 parameter. Use STMT_VINFO_VECTYPE.
7531 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
7534 2017-06-28 Martin Liska <mliska@suse.cz>
7537 * ipa-visibility.c (non_local_p): Handle visibility.
7539 2017-06-28 Martin Liska <mliska@suse.cz>
7542 * common.opt: Add IntegerRange to various options.
7543 * opt-functions.awk (integer_range_info): New function.
7544 * optc-gen.awk: Add integer_range_info to cl_options struct.
7545 * opts-common.c (decode_cmdline_option): Handle
7546 CL_ERR_INT_RANGE_ARG.
7547 (cmdline_handle_error): Likewise.
7548 * opts.c (print_filtered_help): Show valid interval in
7549 when --help is provided.
7550 * opts.h (struct cl_option): Add range_min and range_max fields.
7551 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
7553 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
7555 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
7556 (x * C EQ/NE y * C): New transformation.
7558 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
7560 * genmultilib (combination_space): Accept '+' in option names.
7562 2017-06-28 Martin Liska <mliska@suse.cz>
7565 * asan.c (instrument_derefs): Bail out inner references
7566 that are hard register variables.
7568 2017-06-28 Jakub Jelinek <jakub@redhat.com>
7571 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
7572 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
7574 2017-06-28 Richard Biener <rguenther@suse.de>
7576 * tree-vectorizer.h (vect_get_vec_defs): Remove.
7577 (vect_get_slp_defs): Adjust.
7578 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
7580 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
7582 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7583 get_initial_defs_for_reduction instead of vect_get_vec_defs.
7584 (vectorizable_reduction): Adjust.
7585 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
7587 (vect_get_slp_defs): Likewise.
7588 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
7589 (vectorizable_bswap): Adjust.
7590 (vectorizable_call): Likewise.
7591 (vectorizable_conversion): Likewise.
7592 (vectorizable_assignment): Likewise.
7593 (vectorizable_shift): Likewise.
7594 (vectorizable_operation): Likewise.
7595 (vectorizable_store): Likewise.
7596 (vectorizable_condition): Likewise.
7597 (vectorizable_comparison): Likewise.
7599 2017-06-28 Michael Collison <michael.collison@arm.com>
7602 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
7604 (arm_gen_movmemqi): Removed unused variable 'i'.
7605 Convert 'for' loop into 'while' loop.
7606 (arm_expand_prologue): Remove last unnecessary set of insn.
7607 (thumb_pop): Remove unused variable 'pushed_words'.
7608 (thumb_exit): Remove last unnecessary set of regs_to_pop.
7610 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7612 * config/s390/predicates.md: Use s390_rel_address_ok_p.
7613 * config/s390/s390-protos.h: Add prototype of
7614 s390_rel_address_ok_p.
7615 * config/s390/s390.c (s390_got_symbol): New function.
7616 (s390_rel_address_ok_p): New function.
7617 (legitimize_pic_address): Use s390_rel_address_ok_p.
7618 (s390_load_got): Use s390_got_symbol.
7619 (s390_option_override): Issue error if
7620 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
7621 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
7623 * config/s390/s390.opt: New option mpic-data-is-text-relative.
7625 2017-06-27 Andrew Pinski <apinski@cavium.com>
7627 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
7628 (X * copysign (1.0, X)): New pattern.
7629 (X * copysign (1.0, -X)): New pattern.
7630 (copysign (-1.0, CST)): New pattern.
7632 2017-06-27 Joseph Myers <joseph@codesourcery.com>
7634 * genmultilib (combination_space): Remove variable.
7635 Validate reuse rules against regular expression for any sequence
7636 of multilib options in any order.
7638 2017-06-27 Michael Collison <michael.collison@arm.com>
7640 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
7641 call aarch64_split_simd_combine.
7642 * (aarch64_combine_internal<mode>): Delete pattern.
7643 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
7644 Allow register and subreg operands.
7646 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7648 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
7649 specific need, just fallback on defaults.
7650 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
7652 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7653 Olivier Hainque <hainque@adacore.com>
7655 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
7657 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
7658 targets. Pick a default if no particular attempt applied.
7659 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
7662 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7664 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
7665 (x86_64-wrs-vxworks7): Likewise.
7667 2017-06-27 Marek Polacek <polacek@redhat.com>
7670 * ubsan.c (instrument_null): Check get_base_address's result for null.
7672 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
7674 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
7676 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
7678 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
7679 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
7680 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
7681 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
7683 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
7684 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
7685 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
7686 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
7687 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
7688 BUILT_IN_FEUPDATEENV): New builtins.
7689 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
7690 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
7691 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
7692 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
7694 (builtin_structptr_types): Adjust size.
7695 * tree.c (builtin_structptr_types): Add four entries.
7697 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7698 Olivier Hainque <hainque@adacore.com>
7700 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
7701 (TLS_SYM): New local macro, forcing reference to __tls__ on
7702 link command lines for VxWorks 7 RTPs, triggering initialization
7704 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
7705 OS features TLS support, true for RTPs on VxWorks 7.
7706 * config/vxworks.c (vxworks_override_options): Setup emutls
7709 2017-06-27 Jakub Jelinek <jakub@redhat.com>
7711 * predict.c (test_prediction_value_range): Use -1U instead of -1
7712 to avoid narrowing conversion warning.
7713 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
7714 to avoid narrowing conversion warning.
7715 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
7717 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
7719 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
7721 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
7722 64bit configurations.
7723 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
7724 (SIZE_TYPE): Likewise.
7725 * config/vxworks.c (vxworks_emutls_var_fields): Use
7726 long_unsigned_type_node instead of unsigned_type_node as the offset
7727 field type, which is "pointer" mode in emutls.c.
7729 2017-06-27 Jakub Jelinek <jakub@redhat.com>
7732 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
7735 * gimple-fold.c (replace_call_with_call_and_fold): Handle
7736 gimple_vuse copying separately from gimple_vdef copying.
7738 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7740 * value-prof.c (free_hist): Remove call to memset and the enclosing if
7743 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
7744 Olivier Hainque <hainque@adacore.com>
7746 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
7747 for all vxworks7 targets.
7748 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
7749 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
7750 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
7751 variations for VX6/VX7 and 32/64bits later on in ...
7752 (VXWORKS_LIB_SPEC): Leverage new macros.
7753 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
7754 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
7756 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
7758 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
7759 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
7761 2017-06-26 Carl Love <cel@us.ibm.com>
7763 * config/rs6000/rs6000-c.c: Add support for built-in functions
7764 vector bool char vec_reve (vector bool char);
7765 vector signed char vec_reve (vector signed char);
7766 vector unsigned char vec_reve (vector unsigned char);
7767 vector bool int vec_reve (vector bool int);
7768 vector signed int vec_reve (vector signed int);
7769 vector unsigned int vec_reve (vector unsigned int);
7770 vector bool long long vec_reve (vector bool long long);
7771 vector signed long long vec_reve (vector signed long long);
7772 vector unsigned long long vec_reve (vector unsigned long long);
7773 vector bool short vec_reve (vector bool short);
7774 vector signed short vec_reve (vector signed short);
7775 vector double vec_reve (vector double);
7776 vector float vec_reve (vector float);
7777 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
7778 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
7779 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
7780 (altivec_vreve): New pattern.
7781 * config/rs6000/altivec.h (vec_reve): New define.
7782 * doc/extend.texi (vec_rev): Update the built-in documentation file
7783 for the new built-in functions.
7785 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7787 PR tree-optimization/71815
7788 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
7790 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
7792 (slsr_process_phi): Likewise.
7793 (replace_uncond_cands_and_profitable_phis): Don't replace a
7794 multiply candidate with a stride of 1 (copy or cast).
7795 (phi_incr_cost): Call uses_consumed_by_stmt rather than
7797 (lowest_cost_path): Likewise.
7798 (total_savings): Likewise.
7800 2017-06-26 Richard Biener <rguenther@suse.de>
7803 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
7804 Use def_builtin_pure for all gather builtins.
7806 2017-06-26 Richard Biener <rguenther@suse.de>
7808 PR tree-optimization/81203
7809 * tree-tailcall.c (find_tail_calls): Do not move stmts into
7812 2017-06-26 Marek Polacek <polacek@redhat.com>
7815 * doc/invoke.texi: Document -Wmultistatement-macros.
7817 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
7819 * doc/sourcebuild.texi (ARM-specific attributes): Document new
7820 arm_neon_ok_no_float_abi effective target.
7822 2017-06-26 Richard Biener <rguenther@suse.de>
7824 PR tree-optimization/80928
7825 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
7826 (copy_bbs): Set BB_DUPLICATED flag early.
7827 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
7829 (execute_on_shrinking_pred): Likewise.
7830 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
7831 BB_DUPLICATED blocks.
7832 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
7833 iterate over all PHIs considering removal of *gsi.
7835 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
7837 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
7840 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7842 * config/rs6000/rs6000-string.c: (expand_block_clear,
7843 do_load_for_compare, select_block_compare_mode,
7844 compute_current_alignment, expand_block_compare,
7845 expand_strncmp_align_check, expand_strn_compare,
7846 expand_block_move, rs6000_output_load_multiple)
7847 Move functions related to string/block move/compare
7849 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
7850 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
7851 for this function which is now used in two files.
7852 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
7853 * config.gcc: Add rs6000-string.o to extra_objs for
7854 targets powerpc*-*-* and rs6000*-*-*.
7856 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
7859 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
7860 32-bit, since indexed is not valid for DImode.
7861 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
7862 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
7863 (define_peephole2 for Altivec d-form load): Add 32-bit support.
7864 (define_peephole2 for Altivec d-form store): Likewise.
7867 * multiple_target.c (create_dispatcher_calls): Only create the
7868 dispatcher call if the function is the default clone of a
7871 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
7874 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
7875 a call, force the call to not be a tail call.
7877 2017-06-23 Jeff Law <law@redhat.com>
7879 * doc/contrib.texi: Add entry for Steven Pemberton's work on
7882 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
7884 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
7885 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
7886 handling for early expansion of vector shifts (sl,sr,sra,rl).
7887 (builtin_function_type): Add vector shift right instructions
7888 to the unsigned argument list.
7890 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7892 rtl-optimizatoin/79286
7893 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
7894 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
7895 trap. PIC register plus a const unspec without offset can never trap.
7897 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
7899 * tree.h (builtin_structptr_type): New type.
7900 (builtin_structptr_types): Declare new array.
7901 * tree.c (builtin_structptr_types): New array.
7902 (free_lang_data, build_common_tree_nodes): Use it.
7904 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
7907 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
7910 2017-06-22 Matt Turner <mattst88@gmail.com>
7912 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
7913 Lake models to skylake case. Assume skylake for unknown
7914 models with clflushopt.
7916 2017-06-22 Jeff Law <law@redhat.com>
7918 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
7919 frame sizes that do not satisfy aarch64_uimm12_shift.
7921 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
7923 * profile-count.h (apply_probability,
7924 apply_scale, probability_in): Fix checks for zero.
7926 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7928 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
7929 * doc/cppdiropts.texi (-I @var{dir}): Document it.
7931 2016-06-22 Richard Biener <rguenther@suse.de>
7933 * tree-vect-loop.c (vect_model_reduction_cost): Handle
7934 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
7935 REDUC_MAX_EXPR support.
7936 (vectorizable_reduction): Likewise.
7937 (vect_create_epilog_for_reduction): Likewise.
7939 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
7941 * match.pd (A / (1 << B) -> A >> B): New.
7942 * generic-match-head.c: Include optabs-tree.h.
7943 * gimple-match-head.c: Likewise.
7944 * optabs-tree.h (target_supports_op_p): New.
7945 * optabs-tree.c (target_supports_op_p): New.
7947 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7949 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
7950 $gcc_cv_ld --help output.
7951 (gcc_cv_ld_demangle): Likewise.
7952 (gcc_cv_ld_eh_frame_hdr): Likewise.
7953 (gcc_cv_ld_pie): Likewise.
7954 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
7955 (gcc_cv_ld_buildid): Likewise.
7956 (gcc_cv_ld_sysroot): Likewise.
7957 (ld_bndplt_support): Likewise.
7958 (ld_pushpopstate_support): Likewise.
7959 * configure: Regenerate.
7960 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
7962 2017-06-21 Jakub Jelinek <jakub@redhat.com>
7965 * config/i386/sse.md (round<mode>2): Renumber match_dup and
7966 operands indexes to avoid gap between operands and match_dups.
7968 2017-06-21 Andrew Pinski <apinski@cavium.com>
7970 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
7971 Increment Arith_shift and Arith_shift_reg by 1.
7972 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
7974 * config/aarch64/aarch64.c (thunderx_tunings): Enable
7975 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
7976 (aarch64_strip_extend): Add new argument and test for it.
7977 (aarch64_cheap_mult_shift_p): New function.
7978 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
7979 add a cost if it is true.
7980 Update calls to aarch64_strip_extend.
7981 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
7983 2017-06-21 Andrew Pinski <apinski@cavium.com>
7985 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
7987 (thunderxt88): Likewise.
7988 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
7989 (thunderx_prefetch_tune): New variable.
7990 (thunderx2t99_prefetch_tune): Update for the correct values.
7991 (thunderxt88_tunings): New variable.
7992 (thunderx_tunings): Use thunderx_prefetch_tune instead of
7993 generic_prefetch_tune.
7994 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
7996 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7998 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
7999 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8000 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8001 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8002 (aarch64_atomic_cas<mode>, GPI): Likewise.
8004 2017-06-21 Martin Liska <mliska@suse.cz>
8006 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8007 statements on cold and hot labels.
8008 * predict.c (tree_estimate_probability_bb): Remove the
8009 prediction from this place.
8011 2017-06-21 Martin Liska <mliska@suse.cz>
8013 PR tree-optimization/79489
8014 * gimplify.c (maybe_add_early_return_predict_stmt): New
8016 (gimplify_return_expr): Call the function.
8017 * predict.c (tree_estimate_probability_bb): Remove handling
8019 * predict.def: Update comment about early return predictor.
8020 * gimple-predict.h (is_gimple_predict): New function.
8021 * predict.def: Change default value of early return to 66.
8022 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8024 * passes.def: Put pass_strip_predict_hints to the beginning of
8027 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8029 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8030 FUNCTION_DECL declarations.
8031 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8033 (dwaf2out_decl): Likewise.
8034 * godump.c (go_early_global_decl): Skip call to the real debug hook
8035 for FUNCTION_DECL declarations.
8036 * passes.c (rest_of_decl_compilation): Skip call to the
8037 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8038 -fdump-go-spec is passed.
8040 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8042 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8043 Reorder for compactness.
8044 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8045 (def_builtin_pure, def_builtin_pure2): New functions.
8046 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8048 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8050 * match.pd (nop_convert): New predicate.
8051 ((A +- CST1) +- CST2): Allow some NOP conversions.
8053 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8056 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8057 with ctors/dtors if GOVD_SHARED is set.
8059 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8061 * config/aarch64/aarch64.md (movti_aarch64):
8062 Emit mov rather than orr.
8063 (movtf_aarch64): Likewise.
8064 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8065 Emit mov rather than orr.
8067 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8069 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8070 Swap alternatives, make integer dup more expensive.
8072 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8074 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8075 Return true for non-tls symbols.
8077 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8079 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8080 (cortex-a75): Likewise.
8081 (cortex-a75.cortex-a55): Likewise.
8082 * config/aarch64/aarch64-tune.md: Regenerate.
8083 * doc/invoke.texi (-mtune): Document new values for -mtune.
8085 2017-06-21 Tom de Vries <tom@codesourcery.com>
8087 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8089 (Effective-Target Keywords, Other attributes): Suggest using
8090 dg-add-options stack_size feature to get stack limit in stack_size
8091 effective target documentation.
8093 2017-06-21 Julian Brown <julian@codesourcery.com>
8094 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8096 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8097 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8098 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8100 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8101 attribute type list for neon_multiply.
8102 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8103 attribute type list for neon_multiply.
8104 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8105 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8106 attribute type list for neon_multiply.
8107 * config/arm/types.md (crypto_pmull): Add.
8108 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8109 attribute type list.
8111 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8113 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8116 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8118 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8119 to make sure not to dereference a NULL cost_classes_ptr pointer.
8121 2017-06-20 Carl Love <cel@us.ibm.com>
8123 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8124 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8125 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8126 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8127 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8128 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8129 VMULOSW): New enum "unspec" values.
8130 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8131 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8132 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8133 altivec_vmulosw): New patterns.
8134 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8135 VMULOSW): Add definitions.
8137 2017-06-20 Julia Koval <julia.koval@intel.com>
8139 * config/i386/i386.c: Fix rounding expand for new pattern.
8140 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8142 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8144 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8145 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8147 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8149 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8152 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8154 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8155 architecture, then by implementer ID.
8156 * config/aarch64/aarch64-tune.md: Regenerate.
8158 2017-06-20 Richard Biener <rguenther@suse.de>
8161 * fold-const.c (split_tree): Fold to type before negating.
8163 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8165 * diagnostic-show-locus.c
8166 (selftest::test_fixit_deletion_affecting_newline): New function.
8167 (selftest::diagnostic_show_locus_c_tests): Call it.
8169 2017-06-20 Andreas Schwab <schwab@suse.de>
8172 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8175 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8177 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8178 __ARM_FEATURE_COPROC according to support.
8180 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8182 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8183 Rewritten to avoid overflow for > 32-bit pointers.
8186 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8187 by removing enum keyword.
8188 (ubsan_type_descriptor): Likewise. Formatting fix.
8191 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8192 splitter): Require TARGET_SSE2 in the condition.
8194 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8197 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8198 for doing vector set of SFmode on ISA 3.0.
8199 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8200 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8202 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8203 SFmode value into a V4SF variable that was extracted from another
8204 V4SF variable without converting the element to double precision
8205 and back to single precision vector format.
8206 (vsx_insert_extract_v4sf_p9_2): Likewise.
8208 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8210 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8211 in UWHI to avoid undefined overflow.
8214 * ubsan.h (enum ubsan_encode_value_phase): New.
8215 (ubsan_encode_value): Change second argument to
8216 enum ubsan_encode_value_phase with default value of
8217 UBSAN_ENCODE_VALUE_GENERIC.
8218 * ubsan.c (ubsan_encode_value): Change second argument to
8219 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8220 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8221 create_tmp_var_raw instead of create_tmp_var and use a
8223 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8224 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8225 ubsan_encode_value callers.
8228 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8229 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8230 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8232 2017-06-19 Richard Biener <rguenther@suse.de>
8235 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8236 estimates if we changed anything.
8238 2017-06-19 Richard Biener <rguenther@suse.de>
8240 PR tree-optimization/80887
8241 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8242 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8243 simplified lookups, then reset mprts_hook.
8244 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8246 (try_to_simplify): Likewise.
8248 2017-06-19 Martin Liska <mliska@suse.cz>
8251 * gimplify.c (gimplify_switch_expr):
8252 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8254 2017-06-19 Martin Liska <mliska@suse.cz>
8256 * doc/install.texi: Document that PGO runs in 4 stages.
8258 2017-06-19 Martin Liska <mliska@suse.cz>
8261 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8262 to dispatcher function name.
8263 * multiple_target.c (replace_function_decl): New function.
8264 (create_dispatcher_calls): Redirect both edges and references.
8266 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8268 * profile-count.c (profile_count::dump): Dump quality.
8269 (profile_count::differs_from_p): Update for unsigned val.
8270 * profile-count.h (profile_count_quality): New enum.
8271 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8273 2017-06-19 Richard Biener <rguenther@suse.de>
8275 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8276 struct function as arg.
8277 (estimate_numbers_of_iterations): Export overload with loop arg.
8278 (free_numbers_of_iterations_estimates_loop): Use an overload of
8279 free_numbers_of_iterations_estimates instead.
8280 * tree-cfg.c (remove_bb): Adjust.
8281 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8282 * tree-parloops.c (gen_parallel_loop): Likewise.
8283 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8285 (tree_unroll_loops_completely): Likewise.
8286 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8287 Use an overload instead and export.
8288 (estimated_loop_iterations): Adjust.
8289 (max_loop_iterations): Likewise.
8290 (likely_max_loop_iterations): Likewise.
8291 (estimate_numbers_of_iterations): Take struct function as arg
8293 (loop_exits_before_overflow): Adjust.
8294 (free_numbers_of_iterations_estimates_loop): Use an overload.
8295 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8296 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8298 2017-06-19 Richard Biener <rguenther@suse.de>
8301 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8302 RANGE_EXPR conservatively.
8304 2017-06-16 Carl Love <cel@us.ibm.com>
8306 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8307 definitions for vec_float, vec_float2, vec_floato,
8308 vec_floate built-ins.
8309 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8310 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8312 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8313 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8314 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8315 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8316 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8317 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8318 vec_floato): Add builtin defines.
8319 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8320 Update the built-in documentation file for the new built-in
8323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8325 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8326 (mthumb): Mark as the negative of -marm.
8328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8330 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8332 (ARM Options, -mtune): Document that this accepts the same
8333 extension options as -mcpu.
8334 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8338 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8339 permitted extensions.
8341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8343 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8344 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8345 (armv8-m.main): Add option +nodsp.
8346 * config/arm/arm-cpu-cdata.h: Regenerated.
8348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8350 * config/arm/t-fuchsia: New file.
8351 * config.gcc (arm*-*-fuchsia*): Use it.
8353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8355 * config/arm/t-symbian: Rewrite for new option infrastructure.
8357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8359 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
8360 (MULTILIB_REQUIRED): Likewise.
8362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8364 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
8365 (MULTILIB_RESUE): Likewise.
8366 (MULTILIB_MATCHES): Likewise.
8367 (MULTLIB_REQUIRED): Likewise.
8369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8371 * config/arm/t-rtems: Rewrite for new option framework.
8373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8375 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
8376 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
8377 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
8378 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
8379 * config/arm/t-multilib: ... here.
8380 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
8381 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
8382 armv7-a and armv8*-a when A-profile libraries have not been built.
8383 * config/arm/t-rmprofile: Rewrite.
8385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8387 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
8388 with a backslash. Remove the backslash after substituting unescaped
8390 * doc/fragments.texi (MULTILIB_REUSE): Document it.
8392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8394 * config.gcc: (arm*-*-*): When building a-profile libraries, force
8395 the driver to pass through the default setting of -mfloat-abi.
8396 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
8398 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
8399 (all_feat_combs): New rule.
8400 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
8402 * config/arm/t-aprofile: Rewrite.
8404 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8406 * config/arm/arm.h (FPUTYPE_AUTO): Define.
8407 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
8408 fpu is not specified by the user/command-line.
8409 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
8410 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
8411 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
8412 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
8413 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
8414 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
8416 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8418 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
8419 * config/arm/t-arm-elf: Rewritten.
8421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8423 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
8424 have some floating-point instructions.
8425 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
8426 (TARGET_MAYBE_HARD_FLOAT): New macro.
8427 * config/arm/arm-builtins.c (arm_init_builtins): Use
8428 TARGET_MAYBE_HARD_FLOAT.
8429 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
8431 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8433 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
8434 (configargs.h): Include it.
8435 (arm_print_hint_for_fpu_option): New function.
8436 (arm_parse_fpu_option): New function.
8437 (candidate_extension): New class.
8438 (arm_canon_for_multilib): New function.
8439 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
8440 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
8441 (ARCH_CANONICAL_SPECS): New macro.
8442 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
8444 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8446 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
8447 are set after handling multilib fragments. Set target_cpu_default2
8450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8452 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
8454 (arm*-*-*): Set target_cpu_default2 to a quoted string.
8455 * config/arm/parsecpu.awk (check_cpu): Validate any extension
8457 (check_arch): Likewise.
8458 * config/arm/arm.c (arm_configure_build_target): Handle
8459 TARGET_CPU_DEFAULT being a string constant. Scan any feature
8460 options in the default.
8462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8464 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
8465 when an option is an alias of another.
8466 * config/arm/parsecpu.awk (optalias): New parser token.
8467 (gen_comm_data): Mark non-alias options as such. Emit entries
8468 for extension aliases.
8469 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
8470 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
8471 (armv6kz, armv6zk, armv6t2): Likewise.
8472 (armv7): Make vfpv3-d16 an alias.
8473 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
8475 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
8476 Sort in canonical order.
8477 (armv8-a): Sort in canonical order.
8478 (armv8.1-a, armv8.2-a): Likewise.
8479 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
8481 (cortex-a9): Sort in canonical order.
8482 * config/arm/arm.c (selftests.h): Include it.
8483 (arm_test_cpu_arch_data): New function.
8484 (arm_run_self_tests): New function.
8485 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
8486 (targetm): Move declaration to the end of the file.
8487 * arm-cpu-cdata.h: Regenerated.
8489 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8491 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
8492 call to target_mode_check describing the type of option passed.
8493 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
8494 (arm_target_thumb_only): Use arm_parse_arch_option_name or
8495 arm_parse_cpu_option_name to match parameters against list of
8497 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
8498 arm_arch_core_flags data structure.
8499 * config/arm/arm-cpu_cdata.h: Regenerated.
8501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8503 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
8505 (arm_print_hint_for_cpu_option): Likewise.
8506 (arm_print_hint_for_arch_option): Likewise.
8507 (arm_parse_cpu_option_name): Likewise.
8508 (arm_parse_arch_option_name): Likewise.
8509 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
8510 of entries in the all_fpus list.
8511 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
8512 (arm_parse_cpu_option_name): Declare.
8513 (arm_parse_arch_option_name): Declare.
8514 (arm_parse_option_features): Declare.
8515 (arm_intialize_isa): Declare.
8516 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
8518 (gen_comm_data): ... here. Make definitions non-static.
8519 * config/arm/arm-cpu-data.h: Regenerated.
8520 * config/arm/arm-cpu-cdata.h: Regenerated.
8522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8524 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
8525 (cpu_arch_extension): New structure.
8526 (cpu_arch_option, arch_option, cpu_option): New structures.
8527 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
8529 (gen_data): Generate new format data tables.
8530 * config/arm/arm.c (cpu_tune): New structure.
8531 (cpu_option, processors): Delete.
8532 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
8533 (arm_print_hint_for_cpu_option): ... this and ...
8534 (arm_print_hint_for_arch_option): ... this.
8535 (arm_parse_arch_cpu_name): Delete. Replace with ...
8536 (arm_parse_cpu_option_name): ... this and ...
8537 (arm_parse_arch_option_name): ... this.
8538 (arm_unrecognized_feature): Change type of target parameter to
8540 (arm_parse_arch_cpu_features): Delete. Replace with ...
8541 (arm_parse_option_features): ... this.
8542 (arm_configure_build_target): Rework to use new configuration data
8544 (arm_print_tune_info): Rework for new configuration data tables.
8545 * config/arm/arm-cpu-data.h: Regenerated.
8546 * config/arm/arm-cpu.h: Regenerated.
8548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8550 * Makefile.in (OBJS): Move sbitmap.o from here ...
8551 (OBJS-libcommon): ... to here.
8553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8555 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
8556 (ISA_ALL_CRYPTO): New macro.
8557 (ISA_ALL_SIMD): New macro
8558 (ISA_ALL_FP): New macro.
8559 * config/arm/arm.c (fpu_bitlist): Update initializer.
8560 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
8562 (arm9e): Add fpu. Add option for nofp
8563 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
8564 (arm926ej-s, arm1026ej-s): Likewise.
8565 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
8566 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
8567 neon-fp16, neon-vfpv4, nofp and nosimd.
8568 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
8569 (cortex-a8): Add fpu. Add option for nofp.
8570 (cortex-a9): Add fpu. Add options for nosimd and nofp.
8571 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
8572 (cortex-r4f): Add fpu.
8573 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
8574 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
8576 (cortex-r8): Likewise.
8577 (cortex-m4): Add fpu. Add option for nofp.
8578 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
8579 (cortex-a17.cortex-a7): Likewise.
8580 (cortex-a32): Add fpu. Add options for crypto and nofp.
8581 (cortex-a35, cortex-a53): Likewise.
8582 (cortex-a57): Add fpu. Add option for crypto.
8583 (cortex-a72, cortex-a73): Likewise.
8584 (exynos-m1): Likewise.
8585 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
8586 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
8587 (cortex-m33): Add fpu. Add option for nofp.
8588 * config/arm/arm-cpu-cdata.h: Regenerated
8589 * config/arm/arm-cpu-data.h: Regenerated.
8591 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8593 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
8594 (armv5te, armv5tej): Likewise.
8595 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
8596 (armv7): Add options fp and vfpv3-d16.
8597 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
8598 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
8600 (armv7ve): Likewise.
8601 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
8602 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
8603 (armv8-a): Add nocrypto option.
8604 (armv8.1-a, armv8.2-a): Likewise.
8605 (armv8-m.main): add options fp, fp.dp and nofp.
8607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8609 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
8611 (armv8-a+crc): Delete.
8612 (armv8.1-a): Add options simd, crypto and nofp.
8613 (armv8.2-a): Add options fp16, simd, crypto and nofp.
8614 (armv8.2-a+fp16): Delete.
8615 (armv8-m.main): Add option dsp.
8616 (armv8-m.main+dsp): Delete.
8617 (cortex-a8): Add fpu. Add nofp option.
8618 (cortex-a9): Add fpu. Add nofp and nosimd options.
8619 * config/arm/parsecpu.awk (gen_data): Generate option tables and
8620 link to main cpu and architecture data structures.
8621 (gen_comm_data): Only put isa attributes from the main architecture
8623 (option): New statement for architecture and CPU entries.
8624 * arm.c (struct cpu_option): New structure.
8625 (struct processors): Add entry for options.
8626 (arm_unrecognized_feature): New function.
8627 (arm_parse_arch_cpu_name): Ignore any characters after the first
8629 (arm_parse_arch_cpu_feature): New function.
8630 (arm_configure_build_target): Separate out any CPU and architecture
8631 features and parse separately. Don't error out if -mfpu=auto is
8632 used with only an architecture string.
8633 (arm_print_asm_arch_directives): New function.
8634 (arm_file_start): Call it.
8635 * config/arm/arm-cpu-cdata.h: Regenerated.
8636 * config/arm/arm-cpu-data.h: Likewise.
8637 * config/arm/arm-tables.opt: Likewise.
8639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8641 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
8642 assembler when it is not -mfpu=auto.
8644 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8646 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
8647 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
8648 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
8649 (ASM_CPU_SPEC): Rewrite.
8650 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
8651 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
8652 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
8653 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
8654 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
8655 copied string is NUL-terminated. Also strip any characters prefixed
8657 (arm_rewrite_selected_arch): New function.
8658 (arm_rewrite_march): New function.
8660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8662 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
8663 (x_arm_cpu_string, x_arm_tune_string): Likewise.
8664 (march, mcpu, mtune): Convert to string-based options.
8665 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
8666 (arm_parse_arch_cpu_name): New function.
8667 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
8668 identify selected architecture or CPU.
8669 (arm_option_save): New function.
8670 (TARGET_OPTION_SAVE): Redefine.
8671 (arm_option_restore): Restore string options.
8672 (arm_option_print): Print string options.
8674 2017-06-16 Martin Sebor <msebor@redhat.com>
8676 PR tree-optimization/80933
8677 PR tree-optimization/80934
8678 * builtins.c (fold_builtin_3): Do not handle bcmp here.
8679 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
8680 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
8681 (gimple_fold_builtin): Call them.
8683 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8685 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
8686 as unlikely; update profile.
8688 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8690 * predict.c (force_edge_cold): Handle declaring edges impossible
8693 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8695 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
8697 (try_unroll_loop_completely): Fix reporting.
8699 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8701 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
8703 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
8706 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
8707 if given a non-constant argument for an intrinsic which requires a
8710 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8712 * profile.c (compare_freqs): New function.
8713 (branch_prob): Sort edge list.
8714 (find_spanning_tree): Assume that the list is priority sorted.
8716 2017-06-16 Richard Biener <rguenther@suse.de>
8718 PR tree-optimization/81090
8719 * passes.def (pass_record_bounds): Remove.
8720 * tree-pass.h (make_pass_record_bounds): Likewise.
8721 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
8722 make_pass_record_bounds): Likewise.
8723 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
8724 not free niter estimates at the beginning but at the end.
8725 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
8727 2017-06-16 Richard Biener <rguenther@suse.de>
8729 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
8730 initializer to workaround ICE in host GCC 4.8.
8732 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
8734 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
8736 (clone_inlined_nodes): Update.
8738 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8740 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
8741 prefetch settings, and enable prefetching by default at -O3.
8743 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8745 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8746 Set flag_prefetch_loop_arrays according to tuning data.
8748 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
8750 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
8752 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
8753 [Unrelated to main purpose of the patch] Place the pointer field last
8754 to enable type checking errors when tune structure are wrongly merged.
8755 * config/aarch64/aarch64.c (generic_prefetch_tune,)
8756 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
8757 (thunderx2t99_prefetch_tune): New tune constants.
8758 (tune_params *_tunings): Update all tunings (no functional change).
8759 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
8760 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
8761 from tunings structures.
8763 2017-06-16 Jakub Jelinek <jakub@redhat.com>
8766 * ubsan.c (instrument_null): Add T argument, use it instead
8767 of computing it based on IS_LHS.
8768 (instrument_object_size): Likewise.
8769 (pass_ubsan::execute): Adjust instrument_null and
8770 instrument_object_size callers to pass gimple_get_lhs or
8771 gimple_assign_rhs1 result to it. Use instrument_null instead of
8772 calling get_base_address and instrument_mem_ref. Handle
8773 aggregate call arguments for object-size sanitization.
8775 2017-06-16 Yury Gribov <tetra2005@gmail.com>
8777 PR tree-optimization/81089
8778 * tree-vrp.c (is_masked_range_test): Validate operands of
8781 2017-06-15 Martin Sebor <msebor@redhat.com>
8784 * dumpfile.c (dump_register): Avoid calling memset to initialize
8785 a class with a default ctor.
8786 * gcc.c (struct compiler): Remove const qualification.
8787 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
8788 * hash-table.h: Ditto.
8789 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
8791 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
8792 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
8794 * params.h (struct param_info): Make struct members non-const.
8795 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
8796 with default initialization.
8797 * vec.h (vec_copy_construct, vec_default_construct): New helper
8799 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
8800 with vec_copy_construct.
8801 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
8802 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
8803 * doc/invoke.texi (-Wclass-memaccess): Document.
8805 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8807 * emit-rtl.h (is_leaf): Update comment about local
8810 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8813 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
8814 for a variable to have a section before checking if the section has a
8816 Set section to.persistent if persistent attribute is set.
8817 Warn if .persistent attribute is used on an automatic variable.
8819 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
8821 PR rtl-optimization/80474
8822 * reorg.c (update_block): Do not ignore instructions in a delay slot.
8824 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
8826 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
8829 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
8831 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
8832 (casesi): Emit bounds checking as RTL.
8833 (casesi_internal_mips16_<mode>): Remove bounds checking.
8835 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
8837 * config/xtensa/xtensa.c (xtensa_option_override): Append
8838 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
8839 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
8840 xtensa_doloop_hooks): Define unconditionally.
8841 (xtensa_reorg_loops): Only call reorg_loops in the presence of
8843 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
8844 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
8845 for it in xtensa_option_override.
8846 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
8847 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
8849 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
8851 * doc/cppopts.texi: Document '-' special value to -MF.
8853 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
8855 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
8856 (cortex_a53_fconst): Likewise.
8857 (cortex_a53_fpmul): Likewise.
8858 (cortex_a53_f_load_64): Likewise.
8859 (cortex_a53_f_load_many): Likewise.
8860 (cortex_a53_advsimd_alu): Likewise.
8861 (cortex_a53_advsimd_alu_q): Likewise.
8862 (cortex_a53_advsimd_mul): Likewise.
8863 (cortex_a53_advsimd_mul_q): Likewise.
8864 (fpmac bypass): Add new bypass for fpmac-fpmac case.
8865 Add missing fmul, r2f_cvt and fconst cases.
8867 2017-06-14 Richard Biener <rguenther@suse.de>
8870 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
8872 (fold_binary_loc): When associating do not treat pre-existing
8873 TREE_OVERFLOW on literal constants as a reason to allow
8874 TREE_OVERFLOW on associated literal constants.
8876 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
8878 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
8879 (MASK_FEATURES): New macro.
8880 * config/sparc/sparc.c (sparc_option_override): Remove the special
8881 handling of -mfpu and generalize it to all MASK_FEATURES switches.
8883 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
8885 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
8886 a division of 0 if non-call exceptions are enabled.
8888 2017-06-14 Andrew Pinski <apinski@cavium.com>
8889 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8892 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
8893 Improve vector initialization code gen for only variable case.
8895 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
8897 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
8899 2017-06-14 Richard Biener <rguenther@suse.de>
8901 PR tree-optimization/81083
8902 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
8905 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8907 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
8908 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
8909 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
8910 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
8911 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
8912 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
8914 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8916 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
8917 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
8919 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8921 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
8923 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8925 * config/rs6000/t-rtems: Don't handle SPE.
8927 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8929 * config/rs6000/t-linux: Don't handle SPE.
8931 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8933 * config/rs6000/eabispe.h: Delete file.
8935 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8937 * config/rs6000/t-spe: Delete file.
8939 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8941 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
8942 (rs6000_legitimate_offset_address_p): Return false for anything in
8943 V2SImode or V2SFmode.
8945 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
8947 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
8948 except V2SF and V2SI. Rearrange the vector modes, and add comments.
8949 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
8951 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
8952 (rs6000_legitimate_offset_address_p): Ditto.
8953 (rs6000_emit_move): Ditto.
8954 (rs6000_init_builtins): Remove V4HI_type_node.
8956 2017-06-13 Martin Liska <mliska@suse.cz>
8959 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
8960 (gate_asan): Likewise.
8961 * asan.h (asan_no_sanitize_address_p): Remove the function.
8962 (sanitize_flags_p): New function.
8963 * builtins.def: Fix coding style.
8964 * common.opt: Use renamed enum value.
8965 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
8966 * doc/extend.texi: Document no_sanitize attribute.
8967 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
8968 to SANITIZE_UNDEFINED_NONDEFAULT.
8969 * gcc.c (sanitize_spec_function): Use the renamed enum value.
8970 * gimple-fold.c (optimize_atomic_compare_exchange_p):
8971 Use sanitize_flags_p.
8972 * gimplify.c (gimplify_function_tree): Likewise.
8973 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
8974 * opts.c (parse_no_sanitize_attribute): New function.
8975 (common_handle_option): Use renamed enum value.
8976 * opts.h (parse_no_sanitize_attribute): Declare.
8977 * tree.c (sanitize_flags_p): New function.
8978 * tree.h: Declared here.
8979 * tsan.c: Use sanitize_flags_p.
8980 * ubsan.c (ubsan_expand_null_ifn): Likewise.
8981 (instrument_mem_ref): Likewise.
8982 (instrument_bool_enum_load): Likewise.
8983 (do_ubsan_in_current_function): Remove the function.
8984 (pass_ubsan::execute): Use sanitize_flags_p.
8985 * ubsan.h: Remove do_ubsan_in_current_function
8986 * tree-cfg.c (print_no_sanitize_attr_value): New function.
8987 (dump_function_to_file): Use it here.
8989 2017-06-13 Martin Jambor <mjambor@suse.cz>
8991 PR tree-optimization/80803
8992 PR tree-optimization/81063
8993 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
8994 (propagate_subaccesses_across_link): Enqueue subtree whenever
8995 necessary instead of relying on the caller.
8997 2017-06-13 Martin Jambor <mjambor@suse.cz>
8999 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9000 that have a first_link.
9001 (sort_and_splice_var_accesses): Do not check first_link before
9003 (subtree_mark_written_and_enqueue): Likewise.
9004 (propagate_all_subaccesses): Likewise and do not stop at first
9005 parent with a first_link.
9007 2017-06-13 Martin Jambor <mjambor@suse.cz>
9009 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9012 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9014 * match.pd: New pattern.
9016 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9018 * tree-vrp.c (is_masked_range_test): New function.
9019 (register_edge_assert_for): Determine ranges for
9022 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9024 PR tree-optimization/67328
9025 * fold-const.c (maskable_range_p): New function.
9026 (build_range_check): Generate bittests if possible.
9028 2017-06-13 Martin Liska <mliska@suse.cz>
9030 * gimple-pretty-print.c (dump_probability): Add new argument.
9031 (dump_edge_probability): Dump both probability and count.
9032 (dump_gimple_label): Likewise.
9033 (dump_gimple_bb_header): Likewise.
9035 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9038 * config/avr/avr-devices.c: Fix indentation.
9039 * config/avr/gen-avr-mmcu-specs.c: Dito.
9041 2017-06-13 Richard Biener <rguenther@suse.de>
9043 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9044 instead get vector type from stmt_info.
9045 (vectorizable_reduction): Adjust. Remove dead code.
9047 2017-06-13 Richard Biener <rguenther@suse.de>
9050 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9051 case of C * (x * C2 + C3).
9052 (fold_addr_of_array_ref_difference): Properly fold index difference.
9054 2017-06-12 David S. Miller <davem@davemloft.net>
9057 * config/sparc/sparc.md (return expander): Emit frame blockage if
9058 function uses alloca.
9060 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9062 * combine.c (make_field_assignment): Check len rather than the mode
9063 precision when calling force_to_mode.
9065 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9067 Support multilibs and devices that see flash in RAM address range.
9070 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9071 (avr_mcu_t) <flash_pm_offset>: New field.
9072 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9073 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9074 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9075 (AVR_TINY_PM_OFFSET): Remove macro.
9076 * config/avr/avr.opt (-mshort-calls): New option.
9077 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9078 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9079 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9080 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9081 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9082 instead of avr_arch->have_jmp_call.
9083 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9084 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9085 avr_arch->flash_pm_offset to define.
9086 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9087 new field flash_pm_offset. Add entry for avrxmega3.
9088 (avr_texinfo): Add entry for avrxmega3.
9089 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9090 attiny212, attiny214,
9091 attiny412, attiny414, attiny416, attiny417,
9092 attiny814, attiny816, attiny817,
9093 attiny1614, attiny1616, attiny1617,
9094 attiny3214, attiny3216, attiny3217.
9095 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9096 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9097 (avr_print_operand_address) [AVR_TINY]: Same.
9098 (avr_asm_init_sections) <readonly_data_section>: Only patch
9099 callback if avr_arch->flash_pm_offset = 0.
9100 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9101 for rodata if avr_arch->flash_pm_offset != 0.
9102 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9103 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9104 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9105 (m_options): Append opt_rcall.
9106 (m_dirnames): Append dir_rcall.
9107 * config/avr/t-multilib: Regenerate.
9109 * configure.ac [target=avr]: Check whether avrxmega3 default
9110 linker description file works as needed.
9111 * configure: Regenerate.
9112 * doc/avr-mmcu.texi: Regenerate.
9113 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9114 <__AVR_ARCH__>: Document avrxmega3 and 103.
9115 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9116 <__AVR_SHORT_CALLS__>: Document it.
9117 <__AVR_PM_BASE_ADDRESS__>: Document it.
9118 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9119 (AVR Variable Attributes) <progmem>: Document this is
9120 not needed for avrxmega3.
9121 (AVR Named Address Spaces) <__flash>: Dito.
9123 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9125 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9127 2017-06-12 Doug Rupp <rupp@adacore.com>
9129 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9130 Append vxworks-stdint.h to the tm_file list.
9131 * config/vxworks-stdint.h: New file.
9133 2017-06-12 Martin Liska <mliska@suse.cz>
9135 PR tree-optimization/81041
9136 * tree-profile.c (gimple_gen_ic_func_profiler):
9137 Create an extra BB in profile-generate
9138 (gimple_gen_time_profiler): Likewise.
9140 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9142 PR tree-optimization/81003
9143 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9144 (update_range_test): Use it instead of force_gimple_operand_gsi.
9146 2017-06-12 Richard Biener <rguenther@suse.de>
9148 PR tree-optimization/81053
9149 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9150 with backedge value not defined in loop. Simplify def stmt
9153 2017-06-11 Tom de Vries <tom@codesourcery.com>
9156 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9158 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9159 nvptx_cannot_force_const_mem.
9161 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9163 * opts.c (finish_options): Move test for flag_split_stack after
9164 it has been initialized.
9166 2017-06-11 Jason Merrill <jason@redhat.com>
9168 * tree.h (id_equal): New.
9169 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9170 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9171 instead of strcmp of IDENTIFIER_POINTER.
9173 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9175 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9176 (mark_all_inlined_calls_cdtor): Fix formating.
9177 (inline_transform): Rescale profile before inlining.
9179 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9181 * cgraph.h (cgraph_edge::clone): Update prototype.
9182 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9183 (cgraph_node::create_clone): Update.
9184 (cgraph_node::create_version_clone): Update.
9185 * tree-inline.c (copy_bb): Update.
9186 (expand_call_inline): Update.
9188 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9190 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9191 factored out from ...
9192 (rs6000_emit_prologue): ... here.
9194 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9196 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9197 factored out from ...
9198 (rs6000_emit_prologue): ... here.
9200 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9202 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9204 (handle_missing_profiles): Fix computation of tp_first_run.
9205 (counts_to_freqs): Do not touch freqs when count is 0.
9207 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9209 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9212 2017-06-10 Tom de Vries <tom@codesourcery.com>
9214 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9215 attributes): Document signal effective target.
9217 2017-06-10 Tom de Vries <tom@codesourcery.com>
9219 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9220 Document effective target stack_size.
9222 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9224 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9225 to the edit_context if they can be auto-applied.
9227 2017-06-9 Ian Lance Taylor <iant@golang.org>
9229 * opts.c (finish_options): If -fsplit-stack, disable implicit
9230 -forder-blocks-and-partition.
9231 * doc/invoke.texi (Optimize Options): Document that when using
9232 -fsplit-stack -forder-blocks-and-partition is not implicitly
9235 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9237 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9238 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9239 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9240 * builtins.def (abort, trap, unreachable): Declare cold.
9241 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9242 * tree-core.h (ECF_COLD): New.
9243 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9244 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9246 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9248 * predict.c (unlikely_executed_stmt_p): Cleanup.
9250 2017-06-09 Richard Biener <rguenther@suse.de>
9252 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9253 model if the ref is always written to.
9255 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9257 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9259 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9261 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9264 2017-06-09 Tom de Vries <tom@codesourcery.com>
9267 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9268 "target cannot support label values" when encountering LABEL_REF.
9270 2017-06-09 Martin Liska <mliska@suse.cz>
9272 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9273 (gimple_gen_ic_func_profiler): Emit direct comparison
9274 of __gcov_indirect_call_callee with NULL.
9275 (gimple_gen_time_profiler): Change probability from
9276 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9278 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9280 * profile.c (edge_gcov_counts): Turn to pointer.
9281 (compute_branch_probabilities, compute_branch_probabilities): Update.
9282 (branch_prob): Do not clear edge_gcov_count.
9283 * profile.h (edge_gcov_counts): Turn to pointer.
9284 (edge_gcov_count): Update.
9286 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9288 * gimple.h (gimple_check_failed): Mark cold.
9290 2017-06-09 Richard Biener <rguenther@suse.de>
9292 PR tree-optimization/66623
9293 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9294 refactor check_reduction into two parts, properly computing
9295 whether we have to check reduction validity for outer loop
9298 2017-06-09 Richard Biener <rguenther@suse.de>
9300 PR tree-optimization/79483
9301 * graphite-scop-detection.c (order): New global.
9302 (get_order): Compute bb to order mapping that satisfies code
9303 generation constraints.
9304 (cmp_pbbs): New helper.
9305 (build_scops): Start domwalk at entry block, sort generated
9308 2017-06-09 Richard Biener <rguenther@suse.de>
9311 * ipa-polymorphic-call.c
9312 (ipa_polymorphic_call_context::restrict_to_inner_class):
9313 Skip FIELD_DECLs with error_mark_node type.
9314 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9317 2017-06-09 Martin Liska <mliska@suse.cz>
9319 * predict.c (struct branch_predictor): New struct.
9320 (test_prediction_value_range): New test.
9321 (predict_c_tests): New function.
9322 * selftest-run-tests.c (selftest::run_tests): Run the function.
9323 * selftest.h: Declare new tests.
9325 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9328 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9329 gen_add3_insn did not fail.
9330 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9331 r0, construct that number in a temporary reg and add that reg to r0.
9332 If asked to put the result in r0 as well, fail.
9334 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9337 for early expansion of vec_eqv.
9339 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9342 * ubsan.c (instrument_null): Avoid pointless code temporary.
9343 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9345 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9349 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
9352 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
9355 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9357 * predict.c (unlikely_executed_edge_p): Move ahead.
9358 (probably_never_executed_edge_p): Use it.
9360 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9363 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
9364 gimple_call_builtin_p call.
9366 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9368 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
9369 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
9370 rtl_check_failed_type2, rtl_check_failed_code1,
9371 rtl_check_failed_code2, rtl_check_failed_code_mode,
9372 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
9373 rtvec_check_failed_bounds, rtl_check_failed_flag,
9374 _fatal_insn_not_found, _fatal_insn): Likewise.
9375 * tree.h (tree_contains_struct_check_failed,
9376 tree_check_failed, tree_not_check_failed,
9377 tree_class_check_failed, tree_range_check_failed,
9378 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
9379 tree_vec_elt_check_failed, phi_node_elt_check_failed,
9380 tree_operand_check_failed, omp_clause_check_failed,
9381 omp_clause_operand_check_failed, omp_clause_range_check_failed):
9384 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9386 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
9387 flag_branch_probabilities.
9388 * ipa-inline.c (edge_badness): Likewise.
9389 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9390 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
9391 * predict.c (maybe_hot_frequency_p): Likewise.
9392 (probably_never_executed): Likewise.
9393 * sched-ebb.c (schedule_ebbs): Likewise.
9394 * sched-rgn.c (find_single_block_region): Likewise.
9395 * tracer.c (tail_duplicate): Likewise.
9397 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9399 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
9400 longer requires x_flag_profile_use.
9402 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9404 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
9405 instead of flag_reorder_blocks_and_partition.
9406 * dbxout.c (dbxout_function_end): Likewise.
9407 * dwarf2out.c (gen_subprogram_die): Likewise.
9408 * haifa-sched.c (sched_create_recovery_edges): Likewise.
9409 * hw-doloop.c (reorg_loops): Likewise.
9410 * varasm.c (assemble_start_function,
9411 assemble_end_function): Likewise.
9412 (decide_function_section): Do not check for
9413 flag_reorder_blocks_and_partition.
9415 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
9417 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9419 (chkp_get_hard_register_fake_addr_expr): Ditto.
9420 (chkp_build_addr_expr): Add check for hard reg case.
9421 (chkp_parse_array_and_component_ref): Ditto.
9422 (chkp_find_bounds_1): Ditto.
9423 (chkp_process_stmt): Don't generate bounds store for
9426 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9428 * predict.c (maybe_hot_bb_p): Do not check profile status.
9429 (maybe_hot_edge_p): Likewise.
9430 (probably_never_executed): Check for zero counts even if profile
9432 (unlikely_executed_edge_p): New function.
9433 (unlikely_executed_stmt_p): New function.
9434 (unlikely_executed_bb_p): New function.
9435 (set_even_probabilities): Use unlikely predicates.
9436 (combine_predictions_for_bb): Likewise.
9437 (predict_paths_for_bb): Likewise.
9438 (predict_paths_leading_to_edge): Likewise.
9439 (determine_unlikely_bbs): New function.
9440 (estimate_bb_frequencies): Use it.
9441 (compute_function_frequency): Use zero counts even if profile is
9443 * profile-count.h: Fix typo.
9445 2017-08-08 Julia Koval <julia.koval@intel.com>
9447 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
9448 _mm512_mask_cvtsepi16_storeu_epi8,
9449 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
9450 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
9451 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
9452 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
9453 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
9454 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
9455 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
9456 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
9457 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
9458 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
9459 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
9460 __builtin_ia32_pmovuswb256mem_mask,
9461 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
9462 __builtin_ia32_pmovwb512mem_mask): New builtins.
9464 2017-08-08 Julia Koval <julia.koval@intel.com>
9466 PR target/73350,80862
9467 * config/i386/subst.md (round): Fix round pattern.
9468 * config/i386/i386.c (ix86_erase_embedded_rounding):
9469 Fix erasing rounding for the fixed pattern.
9471 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
9473 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
9475 2017-06-08 Martin Liska <mliska@suse.cz>
9477 PR gcov-profile/80911
9478 * gcov.c (block_info::block_info): New constructor.
9480 2017-06-07 Carl Love <cel@us.ibm.com>
9482 * config/rs6000/rs6000-c: The return type of the following
9483 built-in functions was implemented as int not long long. Fix sign
9484 of return value for the unsigned version of vec_mulo and vec_mule.
9485 vector unsigned long long vec_bperm (vector unsigned long long,
9486 vector unsigned char)
9487 vector signed long long vec_mule (vector signed int,
9489 vector unsigned long long vec_mule (vector unsigned int,
9490 vector unsigned int)
9491 vector signed long long vec_mulo (vector signed int,
9493 vector unsigned long long vec_mulo (vector unsigned int,
9494 vector unsigned int)
9495 * doc/extend.texi: Fix the documentation for the built-in
9498 2017-06-07 Carl Love <cel@us.ibm.com>
9501 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
9504 2017-06-07 Carl Love <cel@us.ibm.com>
9506 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
9507 support, Generate doublehv for signed int/float for BE case only.
9509 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
9511 * doc/invoke.texi (mcx16): Rewrite.
9513 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9515 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
9516 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
9517 *mov<mode>_softfloat, and an anonymous splitter): Use
9518 nonimmediate_operand instead of rs6000_nonimmediate_operand.
9520 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9522 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
9524 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9525 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
9526 (rs6000_debug_reg_global): Adjust.
9527 (rs6000_init_hard_regno_mode_ok): Adjust.
9528 (rs6000_dbx_register_number): Adjust.
9529 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
9530 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9531 Remove SPE_ACC and SPEFSCR.
9532 (REG_ALLOC_ORDER): Ditto.
9533 (FRAME_POINTER_REGNUM): Change to 111.
9534 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
9535 (REG_CLASS_NAMES): Ditto.
9536 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
9537 (REGISTER_NAMES): Ditto.
9538 (ADDITIONAL_REG_NAMES): Ditto.
9539 (rs6000_reg_names): Ditto.
9540 * config/rs6000/rs6000.md: Renumber some register number
9543 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9545 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
9547 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
9548 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
9550 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
9551 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
9552 Delete the SPE high registers.
9553 (REG_ALLOC_ORDER): Ditto.
9554 (enum reg_class): Remove SPE_HIGH_REGS.
9555 (REG_CLASS_NAMES): Ditto.
9556 (REG_CLASS_CONTENTS): Delete the SPE high registers.
9557 (REGISTER_NAMES): Ditto.
9558 (rs6000_reg_names): Ditto.
9559 * doc/tm.texi.in: Remove SPE as example.
9560 * doc/tm.texi: Regenerate.
9562 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9564 * config/rs6000/8540.md (ppc8540_brinc): Delete.
9565 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
9566 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
9567 * config/rs6000/rs6000.md (type): Remove "brinc".
9569 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9571 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
9572 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
9573 * config/rs6000/linuxspe.h: Delete file.
9574 * config/rs6000/rs6000.md: Don't include spe.md.
9575 * config/rs6000/spe.h: Delete file.
9576 * config/rs6000/spe.md: Delete file.
9577 * config/rs6000/t-rs6000: Remove spe.md.
9579 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9581 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
9582 (reg_or_none500mem_operand): Delete.
9583 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
9584 instead of reg_or_none500mem_operand.
9586 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9588 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
9589 handling of SPE flags.
9590 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
9592 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9594 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
9596 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
9597 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
9598 paired_divv2sf3): Similar.
9599 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
9600 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
9601 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
9603 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
9604 Rename the paired_* instruction patterns.
9605 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
9607 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
9608 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
9609 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
9610 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
9611 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
9613 (struct machine_function): Delete field spe_insn_chain_scanned_p.
9614 (spe_func_has_64bit_regs_p): Delete.
9615 (spe_expand_predicate_builtin): Delete.
9616 (spe_expand_evsel_builtin): Delete.
9617 (TARGET_DWARF_REGISTER_SPAN): Do not define.
9618 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
9619 (invalid_e500_subreg): Delete.
9620 (rs6000_legitimize_address): Always force_reg op2 as well, for
9621 paired single memory accesses.
9622 (rs6000_member_type_forces_blk): Delete.
9623 (rs6000_spe_function_arg): Delete.
9624 (rs6000_expand_unop_builtin): Delete SPE handling.
9625 (rs6000_expand_binop_builtin): Ditto.
9626 (spe_expand_stv_builtin): Delete.
9627 (bdesc_2arg_spe): Delete.
9628 (spe_expand_builtin): Delete.
9629 (spe_expand_predicate_builtin): Delete.
9630 (spe_expand_evsel_builtin): Delete.
9631 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
9632 (spe_init_builtins): Delete.
9633 (spe_func_has_64bit_regs_p): Delete.
9634 (savres_routine_name): Delete "info" parameter. Adjust callers.
9635 (rs6000_emit_stack_reset): Ditto.
9636 (rs6000_dwarf_register_span): Delete.
9637 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
9638 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
9639 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
9641 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
9643 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
9644 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
9645 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
9646 mulv2sf3, divv2sf3): Delete expanders.
9648 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9650 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
9652 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9654 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
9655 * config/rs6000/rs6000.c: Ditto.
9657 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9659 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
9660 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
9661 comparison_operator.
9663 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9665 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
9666 * config/rs6000/rs6000.opt: Ditto.
9667 * config/rs6000/t-rtems: Ditto.
9669 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9671 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
9672 TARGET_E500_SINGLE by 0, simplify.
9673 * config/rs6000/rs6000.c: Ditto.
9674 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
9675 (spe_build_register_parallel): Delete.
9676 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
9677 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
9678 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
9679 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
9680 (E500_CONVERT): Delete.
9681 * config/rs6000/spe.md: Remove many patterns and all define_constants.
9683 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
9685 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
9686 * config/rs6000/dfp.md: Ditto.
9687 (negdd2, *negdd2_fpr): Merge.
9688 (absdd2, *absdd2_fpr): Merge.
9689 (negtd2, *negtd2_fpr): Merge.
9690 (abstd2, *abstd2_fpr): Merge.
9691 * config/rs6000/e500.h: Delete file.
9692 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
9693 TARGET_FPRS by 1 and simplify.
9694 * config/rs6000/rs6000-c.c: Ditto.
9695 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
9697 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
9699 * config/rs6000/rs6000.md: Ditto.
9700 (floatdidf2, *floatdidf2_fpr): Merge.
9701 (move_from_CR_gt_bit): Delete.
9702 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
9703 (E500_CR_IOR_COMPARE): Delete.
9704 (All patterns that require !TARGET_FPRS): Delete.
9705 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
9707 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9709 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
9711 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9713 * graphds.c (add_edge): Intitialize edge's attached data.
9714 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
9715 pointer parameter. Call pointed function on each edge during
9716 graph traversing. Skip traversing the edge when the function
9718 (graphds_dfs, graphds_scc): Ditto.
9719 (for_each_edge): New parameter. Pass the new parameter to callback
9721 * graphds.h (skip_edge_callback): New function pointer type.
9722 (graphds_dfs, graphds_scc): New function pointer parameter.
9723 (graphds_edge_callback, for_each_edge): New parameter.
9725 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9727 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
9728 out code checking if runtime alias check is possible to below ...
9729 Call the new function.
9730 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
9731 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
9733 2017-06-07 Marek Polacek <polacek@redhat.com>
9736 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
9737 TYPE_OVERFLOW_WRAPS check.
9739 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9741 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
9742 if versioning is required.
9743 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
9744 peeling with the check for versioning.
9746 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9748 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
9749 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
9750 Set true to new parameter if new ssa variable is defined.
9751 (vect_gen_vector_loop_niters): Refactor. Set range information
9752 for the new vector loop bound variable.
9753 (vect_do_peeling): Ditto.
9755 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9757 * tree-affine.c (ssa.h): Include header file.
9758 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
9759 has wrapping overflow behavior.
9761 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9763 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
9765 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9767 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
9768 (tree_to_aff_combination): ... here.
9770 2017-06-07 Bin Cheng <bin.cheng@arm.com>
9772 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
9773 reg_pressure model function.
9774 (ivopts_global_cost_for_size): Delete.
9775 (determine_set_costs, iv_ca_recount_cost): Call new model function
9776 ivopts_estimate_reg_pressure.
9778 2017-06-07 Tamar Christina <tamar.christina@arm.com>
9780 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
9781 expensive than udiv. Remove floating point cases from mod.
9783 2017-06-07 Tamar Christina <tamar.christina@arm.com>
9785 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
9788 2017-06-07 Tamar Christina <tamar.christina@arm.com>
9790 * config/aarch64/aarch64.md
9791 (copysignsf3): Fix mask generation.
9793 2017-06-07 Jakub Jelinek <jakub@redhat.com>
9795 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
9797 (class dump_manager): Add register_dumps method.
9798 * dumpfile.c: Include langhooks.h.
9799 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
9800 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
9801 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
9802 (dump_manager::dump_register): Start with 512 entries instead of 32.
9803 (dump_manager::register_dumps): New method.
9804 * toplev.c (general_init): Instead of invoking register_dumps
9805 langhook, invoke register_dumps method on the dump manager.
9806 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
9809 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
9811 * doc/md.texi: Clarify the restrictions on a define_insn condition.
9812 Say that # requires an associated define_split to exist, and that
9813 the define_split must be suitable for use after register allocation.
9815 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9817 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
9818 (compute_outgoing_frequencies): Also initialize zero counts.
9819 (find_many_sub_basic_blocks): Do not produce uninitialized profile
9820 around loops; preserve more of profile when nothing changes.
9822 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
9824 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
9826 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
9827 * config/arm/arm-cpu-cdata.h: Regenerate.
9828 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
9829 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
9830 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
9831 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
9832 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
9834 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
9836 * config/arm/t-rmprofile: Likewise.
9837 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
9839 2017-06-06 David S. Miller <davem@davemloft.net>
9842 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
9843 blockage if function uses alloca.
9845 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9847 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
9848 New "uid" fields to hold pretty-print IDs of group and ref.
9849 Memory references are now identified as <group_id>:<ref_id>
9850 instead of using [random] addresses.
9851 (dump_mem_details): Simplify, no functional change.
9852 (dump_mem_ref): Simplify and make output more concise.
9853 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
9854 (find_or_create_group): Initialize group uid.
9855 (record_ref): Initialize ref uid. Improve debug output.
9856 (prune_group_by_reuse, should_issue_prefetch_p,)
9857 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
9858 (mark_nontemporal_store, determine_loop_nest_reuse):
9859 Improve debug output.
9861 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9863 * dbgcnt.def (prefetch): New debug counter.
9864 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
9865 (schedule_prefetches): Stop issueing prefetches if debug counter
9868 2017-06-06 Tom de Vries <tom@codesourcery.com>
9870 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
9871 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
9873 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9875 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
9876 Use aarch64_reg_or_zero predicate for operand 4.
9877 (aarch64_compare_and_swap<mode> define_insn_and_split):
9878 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
9879 (aarch64_store_exclusive<mode>): Likewise for operand 2.
9881 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9883 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
9884 (arm_compute_save_core_reg_mask): This.
9885 (thumb1_compute_save_reg_mask): Rename into ...
9886 (thumb1_compute_save_core_reg_mask): This.
9887 (arm_compute_save_reg0_reg12_mask): Adapt comment.
9888 (arm_compute_frame_layout): Likewise.
9890 2017-06-06 Richard Biener <rguenther@suse.de>
9892 PR tree-optimization/80974
9893 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
9894 keep or clear leaders SSA info.
9896 2017-06-06 Tom de Vries <tom@codesourcery.com>
9898 * config/nvptx/nvptx.c (split_mode_p): New function.
9899 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
9901 2017-06-06 Tom de Vries <tom@codesourcery.com>
9903 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
9905 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9908 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
9911 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9913 * shrink-wrap.c (handle_simple_exit): Update profile.
9914 (try_shrink_wrapping): Upate profile.
9916 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9918 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
9919 (tree_guess_outgoing_edge_probabilities): New.
9920 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
9921 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
9923 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9925 * ipa-split.c (split_function): Initialize return bb profile.
9927 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
9929 * profile.c (compute_branch_probabilities): Also initialize
9932 2017-06-06 Richard Biener <rguenther@suse.de>
9934 PR tree-optimization/80928
9935 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
9936 (vect_analyze_loop_operations): Properly guard analysis for
9938 (vect_transform_loop): Likewise.
9939 (vect_analyze_loop_2): Also reset SLP type on PHIs.
9940 (vect_model_induction_cost): Do not cost for pure SLP.
9941 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
9942 of induction in inner loop vectorization.
9943 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
9944 (vect_get_and_check_slp_defs): Handle vect_induction_def.
9945 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
9947 (vect_analyze_slp_cost_1): Cost induction.
9948 (vect_detect_hybrid_slp_stmts): Handle PHIs.
9949 (vect_get_slp_vect_defs): Likewise.
9950 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
9951 (vect_transform_stmt): Handle SLP reductions.
9952 * tree-vectorizer.h (vectorizable_induction): Adjust.
9954 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9956 * config/rs6000/rs6000.c (make_resolver_func): Update
9957 init_lowered_empty_function call.
9959 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
9961 * doc/invoke.texi: Document the -fprofile-abs-path option.
9962 * common.opt (fprofile-abs-path): New option.
9963 * gcov-io.h (gcov_write_filename): Declare.
9964 * gcov-io.c (gcov_write_filename): New function.
9965 * coverage.c (coverage_begin_function): Use gcov_write_filename.
9966 * profile.c (output_location): Likewise.
9968 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
9970 * shring-wrap.c: Revert accidental commit.
9972 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
9974 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
9976 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
9978 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
9980 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
9981 profile in callgraph edge.
9982 * profile-count.h (apply_probability): If THIS is 0, then result is 0
9983 (apply_scale): Likewise.
9984 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
9985 Also scale profile when inlining function with zero profile.
9986 (initialize_cfun): Update exit block profile even when it is zero.
9987 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
9988 when profile is read.
9990 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9992 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
9993 (CLONE_*): New constants to define the processors we can generate
9994 code for with the target_clone attribute.
9995 (rs6000_clone_map): New array to identify which clone processors
9996 the current program is running on.
9997 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
9998 target_clone attribute.
9999 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10000 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10001 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10002 (cpu_expand_builtin): Add support for target_clone attribute.
10003 (rs6000_valid_attribute_p): Allow "default" attribute.
10004 (get_decl_name): New debug function to simplify printing the
10005 current function name in debugging statements.
10006 (rs6000_clone_priority): New functions to support the target_clone
10007 attribute, and be able to generate code to switch between ISA 2.05
10008 through ISA 3.0 (power6 through power9).
10009 (rs6000_compare_version_priority): Likewise.
10010 (rs6000_get_function_versions_dispatcher): Likewise.
10011 (make_resolver_func): Likewise.
10012 (add_condition_to_bb): Likewise.
10013 (dispatch_function_versions): Likewise.
10014 (rs6000_generate_version_dispatcher_body): Likewise.
10015 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10016 (fusion_gpr_load_p): Fix a spacing issue.
10017 * doc/extend.texi (Common Function Attributes): Document that the
10018 PowerPC supports the target_clone attribute.
10020 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10022 * config/arm/arm.h: explain F symbol found in description of ARM
10023 register allocation in its legend.
10025 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10027 * config/mips/frame-header-opt.c: Include profile-count.h.
10028 * config/riscv/riscv.c: Include profile-count.h
10030 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10032 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10034 (sm_set_flag_if_changed): Add bbs field.
10035 (execute_sm_if_changed_flag_set): Pass BBS.
10036 (execute_sm): Update.
10038 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10040 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10043 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10045 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10046 (peephole2): New peephole2 to emit the above.
10047 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10049 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10051 * config/aarch64/aarch64.c (define_peephole2 above
10052 *sub_<shift>_<mode>): New peephole.
10054 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10056 * config/i386/i386.c (make_resolver_func): Update.
10057 * Makefile.in: Add profile-count.h and profile-count.o
10058 * auto-profile.c (afdo_indirect_call): Update to new API.
10059 (afdo_set_bb_count): Update.
10060 (afdo_propagate_edge): Update.
10061 (afdo_propagate_circuit): Update.
10062 (afdo_calculate_branch_prob): Update.
10063 (afdo_annotate_cfg): Update.
10064 * basic-block.h: Include profile-count.h
10065 (struct edge_def): Turn count to profile_count.
10066 (struct basic_block_def): Likewie.
10067 (REG_BR_PROB_BASE): Move to profile-count.h
10068 (RDIV): Move to profile-count.h
10069 * bb-reorder.c (max_entry_count): Turn to profile_count.
10070 (find_traces): Update.
10071 (rotate_loop):Update.
10072 (connect_traces):Update.
10073 (sanitize_hot_paths):Update.
10074 * bt-load.c (migrate_btr_defs): Update.
10075 * cfg.c (RDIV): Remove.
10076 (init_flow): Use alloc_block.
10077 (alloc_block): Uninitialize count.
10078 (unchecked_make_edge): Uninitialize count.
10079 (check_bb_profile): Update.
10080 (dump_edge_info): Update.
10081 (dump_bb_info): Update.
10082 (update_bb_profile_for_threading): Update.
10083 (scale_bbs_frequencies_int): Update.
10084 (scale_bbs_frequencies_gcov_type): Update.
10085 (scale_bbs_frequencies_profile_count): New.
10086 * cfg.h (update_bb_profile_for_threading): Update.
10087 (scale_bbs_frequencies_profile_count): Declare.
10088 * cfgbuild.c (compute_outgoing_frequencies): Update.
10089 (find_many_sub_basic_blocks): Update.
10090 * cfgcleanup.c (try_forward_edges): Update.
10091 (try_crossjump_to_edge): Update.
10092 * cfgexpand.c (expand_gimple_tailcall): Update.
10093 (construct_exit_block): Update.
10094 * cfghooks.c (verify_flow_info): Update.
10095 (dump_bb_for_graph): Update.
10096 (split_edge): Update.
10097 (make_forwarder_block): Update.
10098 (duplicate_block): Update.
10099 (account_profile_record): Update.
10100 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10101 (get_estimated_loop_iterations): Update.
10102 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10103 (single_likely_exit): Update.
10104 * cfgloopmanip.c (scale_loop_profile): Update.
10106 (set_zero_probability): Update.
10107 (lv_adjust_loop_entry_edge): Update.
10108 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10109 (purge_dead_edges): Update.
10110 (rtl_account_profile_record): Update.
10111 * cgraph.c (cgraph_node::create): Uninitialize count.
10112 (symbol_table::create_edge): Uninitialize count.
10113 (cgraph_update_edges_for_call_stmt_node): Update.
10114 (cgraph_edge::dump_edge_flags): Update.
10115 (cgraph_node::dump): Update.
10116 (cgraph_edge::maybe_hot_p): Update.
10117 * cgraph.h: Include profile-count.h
10118 (create_clone), create_edge, create_indirect_edge): Update.
10119 (cgraph_node): Turn count to profile_count.
10120 (cgraph_edge0: Likewise.
10121 (make_speculative, clone): Update.
10122 (create_edge): Update.
10123 (init_lowered_empty_function): Update.
10124 * cgraphclones.c (cgraph_edge::clone): Update.
10125 (duplicate_thunk_for_node): Update.
10126 (cgraph_node::create_clone): Update.
10127 * cgraphunit.c (cgraph_node::analyze): Update.
10128 (cgraph_node::expand_thunk): Update.
10129 * final.c (dump_basic_block_info): Update.
10130 * gimple-streamer-in.c (input_bb): Update.
10131 * gimple-streamer-out.c (output_bb): Update.
10132 * graphite.c (print_global_statistics): Update.
10133 (print_graphite_scop_statistics): Update.
10134 * hsa-brig.c: Include basic-block.h.
10135 * hsa-dump.c: Include basic-block.h.
10136 * hsa-gen.c (T sum_slice): Update.
10137 (convert_switch_statements):Update.
10138 * hsa-regalloc.c: Include basic-block.h.
10139 * ipa-chkp.c (chkp_produce_thunks): Update.
10140 * ipa-cp.c (struct caller_statistics): Update.
10141 (init_caller_stats): Update.
10142 (gather_caller_stats): Update.
10143 (ipcp_cloning_candidate_p): Update.
10144 (good_cloning_opportunity_p): Update.
10145 (get_info_about_necessary_edges): Update.
10146 (dump_profile_updates): Update.
10147 (update_profiling_info): Update.
10148 (update_specialized_profile): Update.
10149 (perhaps_add_new_callers): Update.
10150 (decide_about_value): Update.
10151 (ipa_cp_c_finalize): Update.
10152 * ipa-devirt.c (struct odr_type_warn_count): Update.
10153 (struct decl_warn_count): Update.
10154 (struct final_warning_record): Update.
10155 (possible_polymorphic_call_targets): Update.
10156 (ipa_devirt): Update.
10157 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10158 * ipa-icf.c (sem_function::merge): Update.
10159 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10160 * ipa-inline.c (compute_uninlined_call_time): Update.
10161 (compute_inlined_call_time): Update.
10162 (want_inline_small_function_p): Update.
10163 (want_inline_self_recursive_call_p): Update.
10164 (edge_badness): Update.
10165 (lookup_recursive_calls): Update.
10166 (recursive_inlining): Update.
10167 (inline_small_functions): Update.
10168 (dump_overall_stats): Update.
10169 (dump_inline_stats): Update.
10170 * ipa-profile.c (ipa_profile_generate_summary): Update.
10171 (ipa_propagate_frequency): Update.
10172 (ipa_profile): Update.
10173 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10174 * ipa-utils.c (ipa_merge_profiles): Update.
10175 * loop-doloop.c (doloop_modify): Update.
10176 * loop-unroll.c (report_unroll): Update.
10177 (unroll_loop_runtime_iterations): Update.
10178 * lto-cgraph.c (lto_output_edge): Update.
10179 (lto_output_node): Update.
10180 (input_node): Update.
10181 (input_edge): Update.
10182 (merge_profile_summaries): Update.
10183 * lto-streamer-in.c (input_cfg): Update.
10184 * lto-streamer-out.c (output_cfg): Update.
10185 * mcf.c (create_fixup_graph): Update.
10186 (adjust_cfg_counts): Update.
10187 (sum_edge_counts): Update.
10188 * modulo-sched.c (sms_schedule): Update.
10189 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10190 * predict.c (maybe_hot_count_p): Update.
10191 (probably_never_executed): Update.
10192 (dump_prediction): Update.
10193 (combine_predictions_for_bb): Update.
10194 (propagate_freq): Update.
10195 (handle_missing_profiles): Update.
10196 (counts_to_freqs): Update.
10197 (rebuild_frequencies): Update.
10198 (force_edge_cold): Update.
10199 * predict.h: Include profile-count.h
10200 (maybe_hot_count_p, counts_to_freqs): UPdate.
10201 * print-rtl-function.c: Do not include cfg.h
10202 * print-rtl.c: Include basic-block.h
10203 * profile-count.c: New file.
10204 * profile-count.h: New file.
10205 * profile.c (is_edge_inconsistent): Update.
10206 (correct_negative_edge_counts): Update.
10207 (is_inconsistent): Update.
10208 (set_bb_counts): Update.
10209 (read_profile_edge_counts): Update.
10210 (compute_frequency_overlap): Update.
10211 (compute_branch_probabilities): Update; Initialize and deinitialize
10213 (branch_prob): Update.
10214 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10215 (edge_gcov_count): New.
10216 (bb_gcov_count): New.
10217 * shrink-wrap.c (try_shrink_wrapping): Update.
10218 * tracer.c (better_p): Update.
10219 * trans-mem.c (expand_transaction): Update.
10220 (ipa_tm_insert_irr_call): Update.
10221 (ipa_tm_insert_gettmclone_call): Update.
10222 * tree-call-cdce.c: Update.
10223 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10224 (gimple_duplicate_sese_tail): Update.
10225 (gimple_account_profile_record): Update.
10226 (execute_fixup_cfg): Update.
10227 * tree-inline.c (copy_bb): Update.
10228 (copy_edges_for_bb): Update.
10229 (initialize_cfun): Update.
10230 (freqs_to_counts): Update.
10231 (copy_cfg_body): Update.
10232 (expand_call_inline): Update.
10233 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10234 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10235 (try_unroll_loop_completely): Update.
10236 (try_peel_loop): Update.
10237 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10238 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10239 * tree-ssa-loop-split.c (connect_loops): Update.
10240 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10241 * tree-ssa-reassoc.c (branch_fixup): Update.
10242 * tree-ssa-tail-merge.c (replace_block_by): Update.
10243 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10244 (compute_path_counts): Update.
10245 (update_profile): Update.
10246 (recompute_probabilities): Update.
10247 (update_joiner_offpath_counts): Update.
10248 (estimated_freqs_path): Update.
10249 (freqs_to_counts_path): Update.
10250 (clear_counts_path): Update.
10251 (ssa_fix_duplicate_block_edges): Update.
10252 (duplicate_thread_path): Update.
10253 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10254 (struct switch_conv_info): Update.
10255 * tree-tailcall.c (decrease_profile): Update.
10256 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10257 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10258 * value-prof.c (check_counter): Update.
10259 (gimple_divmod_fixed_value): Update.
10260 (gimple_mod_pow2): Update.
10261 (gimple_mod_subtract): Update.
10262 (gimple_ic_transform): Update.
10263 (gimple_stringop_fixed_value): Update.
10264 * value-prof.h (gimple_ic): Update.
10266 2017-06-02 Carl Love <cel@us.ibm.com>
10268 * config/rs6000/rs6000-c: Add support for built-in functions
10269 vector double vec_doublee (vector signed int);
10270 vector double vec_doublee (vector unsigned int);
10271 vector double vec_doublee (vector float);
10272 vector double vec_doubleh (vector signed int);
10273 vector double vec_doubleh (vector unsigned int);
10274 vector double vec_doubleh (vector float);
10275 vector double vec_doublel (vector signed int);
10276 vector double vec_doublel (vector unsigned int);
10277 vector double vec_doublel (vector float);
10278 vector double vec_doubleo (vector signed int);
10279 vector double vec_doubleo (vector unsigned int);
10280 vector double vec_doubleo (vector float);.
10281 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10282 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10284 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10285 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10286 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10288 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10289 vec_doublel, vec_doubleh.
10290 * doc/extend.texi: Update the built-in documentation file for the
10291 new built-in functions.
10293 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10296 * ipa-inline-analysis.c (free_growth_caches): Set
10297 edge_removal_hook_holder to NULL after removing it.
10299 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10301 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10302 comparision with zero.
10304 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10305 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10306 for early expansion of vec_min and vec_max builtins.
10307 (builtin_function_type): Add min/max unsigned variants to those
10308 identified as having unsigned arguments.
10310 2017-06-02 Olivier Hainque <hainque@adacore.com>
10312 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10314 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10316 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10317 Use VALL_F16 iterator rather than VALL.
10319 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10321 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10322 Emit CBNZ inside loop when doing a strong exchange and comparing
10323 against zero. Generate the CC flags after the loop.
10325 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10327 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10328 (dl_section_ref): New.
10329 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10330 On AIX, append an expression to subtract the size of the
10331 section length to dl_section_ref.
10333 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10336 for early expansion of vector absolute builtins.
10338 2017-06-02 Richard Biener <rguenther@suse.de>
10340 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10341 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10343 2017-06-02 Richard Biener <rguenther@suse.de>
10345 PR tree-optimization/80948
10346 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10347 stmt order as well.
10349 2017-06-02 Richard Biener <rguenther@suse.de>
10351 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
10353 * tree-vect-stmts.c (process_use): Do not mark backedge defs
10354 for inductions as relevant.
10356 2017-06-02 Richard Biener <rguenther@suse.de>
10358 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
10359 (vectorizable_induction): ... this. Remove dead code.
10361 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
10363 * builtins. (expand_builtin_alloca): Remove second parameter and
10364 infer its value from the first parameter instead.
10365 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
10367 2017-06-02 Jakub Jelinek <jakub@redhat.com>
10369 PR rtl-optimization/80903
10370 * loop-doloop.c (add_test): Unshare sequence.
10372 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
10374 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
10376 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
10378 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
10380 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
10381 xlogue_layout::get_instance, logue_layout::xlogue_layout,
10382 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
10383 (xlogue_layout::get_stub_rtx): Make static.
10384 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
10385 (xlogue_layout::compute_stub_managed_regs): Rename to...
10386 (xlogue_layout::count_stub_managed_regs): ...this.
10387 (xlogue_layout::is_stub_managed_reg): New function.
10388 (xlogue_layout::m_stub_names): Rename to...
10389 (xlogue_layout::s_stub_names): ...this, make static.
10390 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
10391 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
10392 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
10393 xlogue_layout::s_stub_names): Instantiate statics.
10394 (stub_managed_regs): Remove.
10395 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
10396 (disable_call_ms2sysv_xlogues): Rename to...
10397 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
10398 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
10400 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
10401 change after reload_completed.
10402 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
10404 (ix86_expand_prologue): Likewise.
10405 (ix86_expand_epilogue): Likewise.
10406 (ix86_expand_split_stack_prologue): Likewise.
10407 (ix86_compute_frame_layout): Remove frame parameter ...
10408 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
10409 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
10411 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
10412 (ix86_frame): Move from here ...
10413 * config/i386/i386.h (ix86_frame): ... to here.
10414 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
10415 complete ix86_frame data structure instead. Remove some_ld_name.
10417 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10419 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
10420 symbols that hold a DECL_VALUE_EXPR.
10422 2017-06-01 Martin Jambor <mjambor@suse.cz>
10424 PR tree-optimization/80898
10425 * tree-sra.c (process_subtree_disqualification): Removed.
10426 (disqualify_candidate): Do not acll
10427 process_subtree_disqualification.
10428 (subtree_mark_written_and_enqueue): New function.
10429 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
10430 RHS has been disqualified and re-queue LHS if necessary. Apart
10431 from that, ignore disqualified RHS.
10433 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10435 * config/s390/s390.c (s390_emit_epilogue): Disable early return
10436 address fetch for z10 or later.
10438 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10440 * config/arc/arc.md (tst_movb): Add guard when splitting.
10442 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10444 * config/arc/arc.c (arc_can_eliminate): Test against
10445 arc_frame_pointer_needed.
10447 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10449 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
10450 to prevent store reordering.
10451 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
10452 (type): Add block type.
10453 (stack_tie): Define special instruction to be used in
10456 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10458 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
10459 constraint. It is not valid for the pattern.
10460 (noncommutative_binary_comparison): Likewise.
10462 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10464 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
10467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10469 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
10470 be used by the reg-alloc.
10472 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10474 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
10475 reg-alloc when having mul64 or mul32x16 instructions.
10476 (mulsidi3): Likewise.
10477 (umulsidi3): Likewise.
10478 (mulsi32x16): New pattern.
10479 (mulsi64): Likewise.
10480 (mulsidi64): Likewise.
10481 (umulsidi64): Likewise.
10482 (MUL32x16_REG): Define.
10483 (mul64_600): Use MUL32x16_REG.
10484 (mac64_600): Likewise.
10485 (umul64_600): Likewise.
10486 (umac64_600): Likewise.
10488 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
10490 * config/arc/arc.md (mulsi3_700): Make it commutative.
10492 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
10494 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
10496 (*sign_extendsidi2_insn): Likewise for movstosw.
10498 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
10500 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
10501 the type of the input discriminant value. Convert the
10502 discriminant value of signedness vary.
10504 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
10506 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
10507 Add to -Wall section.
10509 2017-06-01 Richard Biener <rguenther@suse.de>
10511 PR middle-end/66313
10512 * fold-const.c (fold_plusminus_mult_expr): If the factored
10513 factor may be zero use a wrapping type for the inner operation.
10514 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
10515 and handle moved defs.
10516 (process_assignment): Properly guard the unary op case. Return a
10517 tri-state indicating that moving the stmt before the call may allow
10518 to continue. Pass through to_move.
10519 (find_tail_calls): Handle moving unrelated defs before
10522 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10525 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
10526 splitter result in the canonical way.
10528 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
10530 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
10531 also for 32bit target. Update insn attributes.
10532 (zero-extendsidi2 splitter): Allow all registers for operand 1.
10534 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
10536 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
10537 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
10538 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
10539 (_mm_maskz_min_ss): New intrinsics.
10541 2017-05-31 Martin Liska <mliska@suse.cz>
10543 * tree-vect-loop.c (vect_create_epilog_for_reduction):
10544 Change comment style to one we normally use.
10545 (vectorizable_reduction): Likewise.
10546 (vectorizable_induction): Likewise.
10547 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10548 (vectorizable_call): Likewise.
10549 (vectorizable_simd_clone_call): Likewise.
10550 (vectorizable_conversion): Likewise.
10551 (vectorizable_assignment): Likewise.
10552 (vectorizable_shift): Likewise.
10553 (vectorizable_operation): Likewise.
10554 (vectorizable_store): Likewise.
10555 (vectorizable_load): Likewise.
10556 * tree-vectorizer.h: Likewise.
10558 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
10560 * passes.c (emergency_dump_function): New.
10561 * tree-pass.h (emergency_dump_function): Declare.
10562 * plugin.c (plugins_internal_error_function): Remove.
10563 * plugin.h (plugins_internal_error_function): Remove declaration.
10564 * toplev.c (internal_error_function): New static function. Use it...
10565 (general_init): ...here.
10567 2017-05-31 Graham Markall <graham.markall@embecosm.com>
10569 * config/arc/arc.c (arc_print_operand): Handle constant operands.
10570 (arc_rtx_costs): Add costs for new patterns.
10571 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
10572 * config/arc/predicates.md: Add _1_2_3_operand predicate.
10574 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10576 * tree-ssa-strlen.c (get_next_strinfo): New function.
10577 (get_stridx_plus_constant): Use it.
10578 (zero_length_string): Likewise.
10579 (adjust_related_strinfos): Likewise.
10580 (adjust_last_stmt): Likewise.
10582 2017-05-31 Richard Biener <rguenther@suse.de>
10585 * config/i386/i386.c (ix86_expand_builtin): Remove assert
10586 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
10588 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
10590 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
10591 loop_vinfo argument and use of dependence distance vectors.
10592 Check instead whether the two references differ only in their
10593 initial value and assume that they have the same alignment if the
10594 difference is a multiple of the vector alignment.
10595 (vect_analyze_data_refs_alignment): Update call accordingly.
10597 2017-05-31 Martin Liska <mliska@suse.cz>
10600 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
10602 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10604 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
10605 (create_intersect_range_checks): Move from ...
10606 * tree-data-ref.c (create_intersect_range_checks_index)
10607 (create_intersect_range_checks): ... to here.
10608 (create_runtime_alias_checks): New function factored from ...
10609 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
10610 here. Call above function.
10611 * tree-data-ref.h (create_runtime_alias_checks): New function.
10613 2017-05-31 Bin Cheng <bin.cheng@arm.com>
10615 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
10616 segment length for dr_b and compute it in wide_int.
10618 2017-05-31 Richard Biener <rguenther@suse.de>
10620 PR tree-optimization/80906
10621 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
10622 and pass through iv_map.
10623 (copy_bb_and_scalar_dependences): Adjust.
10624 (translate_pending_phi_nodes): Likewise.
10625 (copy_loop_close_phi_args): Handle code-generating IVs instead
10628 2017-05-30 David Malcolm <dmalcolm@redhat.com>
10630 * diagnostic-color.c (color_dict): Add "type-diff".
10631 (parse_gcc_colors): Update comment.
10632 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10633 -fdiagnostics-show-template-tree and -fno-elide-type.
10634 (GCC_COLORS): Add type-diff to example.
10636 (-fdiagnostics-show-template-tree): New.
10637 (-fno-elide-type): New.
10638 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
10639 the pp_format_decoder callback. Call any m_format_postprocessor's
10641 (pretty_printer::pretty_printer): Initialize
10642 m_format_postprocessor.
10643 (pretty_printer::~pretty_printer): Delete any
10644 m_format_postprocessor.
10645 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
10646 (class format_postprocessor): New class.
10647 (struct pretty_printer::format_decoder): Document the new parameters.
10648 (struct pretty_printer::m_format_postprocessor): New field.
10649 * tree-diagnostic.c (default_tree_printer): Update for new
10650 bool and const char ** params.
10651 * tree-diagnostic.h (default_tree_printer): Likewise.
10653 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
10655 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
10656 (lwa_operand): Delete rs6000_gen_cell_microcode test.
10657 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10658 rs6000_gen_cell_microcode code.
10659 (rs6000_final_prescan_insn): Delete.
10660 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
10661 "warn-cell-microcode" entries.
10662 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
10663 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
10664 throughout. Change cc_reg_not_micro_cr0_operand to
10665 cc_reg_not_cr0_operand throughout.
10666 (*extendhi<mode>2_noload): Delete.
10667 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
10668 (mwarn-cell-microcode): Delete.
10669 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
10670 -mgen-cell-microcode and -mwarn-cell-microcode.
10672 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
10675 * config/i386/constraints.md (Yd): New constraint.
10677 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
10678 and (?Yd, r) alternatives. Update insn attributes.
10679 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
10680 and (?*Yd, r) alternatives. Update insn attributes.
10681 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
10683 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
10685 * gimplify.c (gimplify_modify_expr): Don't create a
10686 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
10689 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10691 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
10693 2017-05-30 Richard Biener <rguenther@suse.de>
10695 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
10696 and reduc_def fields.
10697 (STMT_VINFO_REDUC_TYPE): New define.
10698 (STMT_VINFO_REDUC_DEF): Likewise.
10699 (vect_force_simple_reduction): Adjust prototype.
10700 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
10701 (vect_is_simple_reduction): Remove check_reduction argument.
10702 (vect_force_simple_reduction): Adjust and set
10703 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
10704 (vectorizable_reduction): Do not re-do reduction analysis
10705 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
10706 * tree-parloops.c (gather_scalar_reductions): Adjust.
10708 2017-05-30 Richard Biener <rguenther@suse.de>
10710 PR middle-end/80901
10711 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
10714 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
10716 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
10717 Introduce unknown_misalignment parameter and remove vf.
10718 (vect_peeling_hash_get_lowest_cost):
10719 Pass unknown_misalignment parameter.
10720 (vect_enhance_data_refs_alignment):
10721 Fix unsupportable data ref treatment.
10723 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10725 * tree-vect-data-refs.c (vect_get_data_access_cost):
10726 Workaround for SLP handling.
10727 (vect_enhance_data_refs_alignment):
10728 Compute costs for doing no peeling at all, compare to the best
10729 peeling costs so far and avoid peeling if cheaper.
10731 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10733 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
10734 Return peeling info and set costs to zero for unlimited cost
10736 (vect_enhance_data_refs_alignment): Also inspect all datarefs
10737 with unknown misalignment. Compute and costs for unknown
10738 misalignment, compare them to the costs for known misalignment
10739 and choose the cheapest for peeling.
10741 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10743 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
10744 (vect_get_peeling_costs_all_drs): Create function.
10745 (vect_peeling_hash_get_lowest_cost):
10746 Use vect_get_peeling_costs_all_drs.
10747 (vect_peeling_supportable): Create function.
10748 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
10750 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
10752 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
10753 DR_HAS_NEGATIVE_STEP.
10754 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
10755 (vect_enhance_data_refs_alignment): Use.
10756 (vect_duplicate_ssa_name_ptr_info): Use.
10757 * tree-vectorizer.h (dr_misalignment): Use.
10758 (known_alignment_for_access_p): Use.
10760 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10763 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
10765 (has_section_name): New function.
10767 2017-05-30 Martin Liska <mliska@suse.cz>
10770 * auto-profile.c (get_function_decl_from_block): Fix
10773 2017-05-30 Richard Biener <rguenther@suse.de>
10775 PR middle-end/80876
10776 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
10778 2017-05-30 Martin Liska <mliska@suse.cz>
10780 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
10781 * dumpfile.h (struct dump_file_info): Remove ctors.
10783 2017-05-30 Martin Liska <mliska@suse.cz>
10785 * predict.def: Fix GNU coding style.
10787 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
10789 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
10790 Mark 'to' argument with ATTRIBUTE_UNUSED.
10792 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
10794 * config/xtensa/xtensa.c (xtensa_emit_call): Use
10795 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
10796 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
10799 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
10801 * doc/install.texi (Options specification): Restore entry of
10802 --enable-sjlj-exceptions.
10804 2017-05-27 Michael Eager <eager@eagercon.com>
10807 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
10809 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
10811 * config/microblaze/microblaze.h
10812 (FIXED_REGISTERS): Update in macro.
10813 (CALL_USED_REGISTERS): Update in macro.
10815 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
10817 * doc/install.texi: Add links to macOS binary distributions.
10819 2017-05-27 Jakub Jelinek <jakub@redhat.com>
10823 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10825 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
10827 2017-05-26 Martin Liska <mliska@suse.cz>
10829 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
10831 2017-05-26 Martin Liska <mliska@suse.cz>
10833 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
10834 always leading ';; '.
10835 (dump_bb_info): Likewise.
10836 (brief_dump_cfg): Likewise.
10837 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
10838 * dumpfile.c: Remove usage of TDF_VERBOSE.
10839 * dumpfile.h (enum dump_kind): Likewise.
10840 (dump_gimple_bb_header): Do not use TDF_COMMENT.
10841 * print-tree.c (debug_verbose): Remove.
10842 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
10843 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
10844 * tree-diagnostic.c (default_tree_printer): Replace
10845 TDF_DIAGNOSTIC with TDF_SLIM.
10847 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10849 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
10850 in parameter loop, rather than loop_vinfo.
10851 (create_intersect_range_checks): Ditto.
10852 (vect_create_cond_for_alias_checks): Update call to above functions.
10854 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10856 PR tree-optimization/80815
10857 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
10858 for merging runtime alias checks. Handle negative DR_STEPs.
10860 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10862 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
10864 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
10865 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
10866 out code pruning runtime alias checks.
10867 * tree-data-ref.c (prune_runtime_alias_test_list): New function
10868 factored out from above.
10869 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
10871 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
10873 (prune_runtime_alias_test_list): New decalaration.
10875 2017-05-26 Bin Cheng <bin.cheng@arm.com>
10877 * tree-vect-data-refs.c (compare_tree): Rename and move ...
10878 * tree-data-ref.c (data_ref_compare_tree): ... to here.
10879 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
10880 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
10881 (operator==, comp_dr_with_seg_len_pair): Ditto.
10882 (vect_prune_runtime_alias_test_list): Ditto.
10884 2017-05-26 Martin Liska <mliska@suse.cz>
10887 * params.def: Bound partial-inlining-entry-probability param.
10889 2017-05-26 Marek Polacek <polacek@redhat.com>
10892 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
10895 2017-05-26 Richard Biener <rguenther@suse.de>
10897 PR tree-optimization/80842
10898 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
10901 2017-05-26 Richard Biener <rguenther@suse.de>
10903 PR tree-optimization/80844
10904 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
10906 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
10908 * doc/md.texi (Machine Constraints): Update x86 family
10909 machine constraints section to match 'config/i386/constraints.md'.
10911 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
10913 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
10915 2017-05-25 Nathan Sidwell <nathan@acm.org>
10917 * doc/invoke.texi (--enable-languages): Update documentation.
10919 2017-05-25 Martin Liska <mliska@suse.cz>
10921 * dumpfile.c: Add TDF_FOLDING.
10922 * dumpfile.h (enum dump_kind): Likewise.
10923 * genmatch.c (dt_simplify::gen_1): Use it.
10925 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10927 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
10929 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10931 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
10932 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
10934 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
10936 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
10937 * match.pd (X == C): Rewrite it here.
10938 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
10939 with_certain_nonzero_bits2): New predicates.
10940 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
10942 2017-05-24 Nathan Sidwell <nathan@acm.org>
10944 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
10947 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
10950 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
10952 * config/powerpcspe: New port. Files are copied from the rs6000
10953 port, with "rs6000" in filenames replaced by "powerpcspe".
10955 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
10957 PR rtl-optimization/80754
10958 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
10960 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
10962 * config/sparc/sparc.md (length): Return the correct value for -mflat
10963 sibcalls to match output_sibcall.
10965 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
10969 * config/rs6000/rs6000.c (struct machine_function): Add new field
10971 (rs6000_get_separate_components): Init that field, use it.
10972 (rs6000_components_for_bb): Use the field.
10974 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
10976 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
10978 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
10980 PR middle-end/80823
10981 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
10983 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10986 * config/s390/s390.c (s390_check_qrst_address): Check incoming
10987 address against address_operand predicate.
10988 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
10990 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
10992 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
10993 parameters passed indirectly.
10995 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
10997 * config/i386/i386.md (*movdi_internal): Remove SSE4
10998 alternative 18 (?r, *v). Update insn attributes.
10999 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11000 Update insn attributes.
11001 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11002 Update insn attributes.
11003 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11004 alternative 1 (r, v). Remove isa attribute.
11005 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11006 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11007 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11009 2017-05-23 Tom de Vries <tom@codesourcery.com>
11011 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11014 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11016 * cgraphunit.c (symbol_table::process_new_functions): Update.
11017 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11018 (inline_generate_summary): Rename to ...
11019 (ipa_fn_summary_generate): ... this one.
11020 (inline_read_summary): Rename to ...
11021 (ipa_fn_summary_read): ... this one.
11022 (inline_write_summary): Rename to ...
11023 (ipa_fn_summary_write): ... this one.
11024 (inline_free_summary): Rename to ...
11025 (ipa_free_fn_summary): ... this one.
11026 (pass_data_local_fn_summary, pass_local_fn_summary,
11027 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11028 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11029 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11030 make_pass_ipa_fn_summary): New.
11031 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11032 inline_write_summary, inline_free_summary): Remove.
11033 (ipa_free_fn_summary) : New.
11034 * ipa-inline.c (ipa_inline): Update.
11035 (pass_ipa_inline): Do not generate summaries.
11036 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11038 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11039 and add pass_ipa_fn_summary.
11040 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11042 (make_pass_inline_parameters): Remove.
11044 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11046 * omp-low.c (struct omp_context): Remove "default_kind" member.
11049 * omp-offload.c (execute_oacc_device_lower): Remove the
11050 parallelism dimensions function attributes for unparallelized
11051 OpenACC kernels constructs.
11053 2017-05-23 Martin Liska <mliska@suse.cz>
11055 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11057 (cgraph_edge::make_speculative): Likewise.
11058 (cgraph_edge::resolve_speculation): Likewise.
11059 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11060 (cgraph_node::dump): Likewise.
11061 * cgraph.h: Likewise.
11062 * cgraphunit.c (analyze_functions): Likewise.
11063 (symbol_table::compile): Likewise.
11064 * ipa-cp.c (print_all_lattices): Likewise.
11065 (determine_versionability): Likewise.
11066 (initialize_node_lattices): Likewise.
11067 (ipcp_verify_propagated_values): Likewise.
11068 (estimate_local_effects): Likewise.
11069 (update_profiling_info): Likewise.
11070 (create_specialized_node): Likewise.
11071 (perhaps_add_new_callers): Likewise.
11072 (decide_about_value): Likewise.
11073 (decide_whether_version_node): Likewise.
11074 (identify_dead_nodes): Likewise.
11075 (ipcp_store_bits_results): Likewise.
11076 * ipa-devirt.c (dump_targets): Likewise.
11077 (ipa_devirt): Likewise.
11078 * ipa-icf.c (sem_item::dump): Likewise.
11079 (sem_function::equals): Likewise.
11080 (sem_variable::equals): Likewise.
11081 (sem_item_optimizer::read_section): Likewise.
11082 (sem_item_optimizer::execute): Likewise.
11083 (congruence_class::dump): Likewise.
11084 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11085 (dump_inline_summary): Likewise.
11086 (estimate_node_size_and_time): Likewise.
11087 (inline_analyze_function): Likewise.
11088 * ipa-inline-transform.c (inline_call): Likewise.
11089 * ipa-inline.c (report_inline_failed_reason): Likewise.
11090 (want_early_inline_function_p): Likewise.
11091 (edge_badness): Likewise.
11092 (update_edge_key): Likewise.
11093 (inline_small_functions): Likewise.
11094 * ipa-profile.c (ipa_profile): Likewise.
11095 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11096 (ipa_make_edge_direct_to_target): Likewise.
11097 (remove_described_reference): Likewise.
11098 (ipa_impossible_devirt_target): Likewise.
11099 (propagate_controlled_uses): Likewise.
11100 (ipa_print_node_params): Likewise.
11101 (ipcp_transform_function): Likewise.
11102 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11103 (propagate_pure_const): Likewise.
11104 * ipa-reference.c (generate_summary): Likewise.
11105 (read_write_all_from_decl): Likewise.
11106 (propagate): Likewise.
11107 (ipa_reference_read_optimization_summary): Likewise.
11108 * ipa-utils.c (ipa_merge_profiles): Likewise.
11109 * ipa.c (walk_polymorphic_call_targets): Likewise.
11110 (symbol_table::remove_unreachable_nodes): Likewise.
11111 (ipa_single_use): Likewise.
11112 * passes.c (execute_todo): Likewise.
11113 * predict.c (drop_profile): Likewise.
11114 * symtab.c (symtab_node::get_dump_name): New function.
11115 (symtab_node::dump_name): Likewise.
11116 (symtab_node::dump_asm_name): Likewise.
11117 (symtab_node::dump_references): Likewise.
11118 (symtab_node::dump_referring): Likewise.
11119 (symtab_node::dump_base): Likewise.
11120 (symtab_node::debug_symtab): Likewise.
11121 * tree-sra.c (convert_callers_for_node): Likewise.
11122 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11123 * value-prof.c (init_node_map): Likewise.
11125 2017-05-23 Martin Liska <mliska@suse.cz>
11127 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11128 and symtab_node::debug_symtab to symbol_table::debug.
11129 * cgraphunit.c (analyze_functions): Use the renamed function.
11130 (symbol_table::compile): Likewise.
11131 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11132 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11133 * passes.c (execute_todo): Likewise.
11134 * symtab.c (symbol_table::dump): New function.
11135 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11137 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11139 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11140 that nonconst implies exec.
11142 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11144 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11145 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11146 (inline_edge_summary_vec): Turn into ...
11147 (ipa_call_summaries): ... this one.
11148 (redirect_to_unreachable, edge_set_predicate,
11149 evaluate_properties_for_edge, inline_summary_alloc,
11150 reset_ipa_call_summary, reset_inline_summary,
11151 inline_summary_t::duplicate): Update.
11152 (inline_edge_duplication_hook): Turn to ...
11153 (ipa_call_summary_t::duplicate): ... this one.
11154 (inline_edge_removal_hook): Turn to ...
11155 (ipa_call_summary_t::remove): ... this one.
11156 (dump_inline_edge_summary): Turn to ...
11157 (dump_ipa_call_summary): ... this one.
11158 (estimate_function_body_sizes): Update.
11159 (inline_update_callee_summaries): Update.
11160 (remap_edge_change_prob): Update.
11161 (remap_edge_summaries): Update.
11162 (inline_merge_summary): Update.
11163 (do_estimate_edge_time): Update.
11164 (inline_generate_summary): Update.
11165 (inline_read_section): Update.
11166 (inline_read_summary): Update.
11167 (inline_free_summary): Update.
11168 * ipa-inline.c (can_inline_edge_p): Update.
11169 (compute_inlined_call_time): Update.
11170 (want_inline_small_function_p): Update.
11171 (edge_badness): Update.
11172 (early_inliner): Update.
11173 * ipa-inline.h (inline_edge_summary): Turn to ...
11174 (ipa_call_summary): ... this one.
11175 (ipa_call_summary_t): New class.
11176 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11177 (ipa_call_summaries): New.
11178 (inline_edge_summary): Remove.
11179 (estimate_edge_growth): Update.
11180 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11181 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11182 * ipa-split.c (execute_split_functions): Update.
11183 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11185 2017-05-23 Tom de Vries <tom@codesourcery.com>
11187 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11188 attributes): Document rdrand effective target.
11190 2017-05-23 Tom de Vries <tom@codesourcery.com>
11192 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11193 attributes): Sort alphabetically.
11195 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11197 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11199 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11202 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11203 V2DF/V2DI splat into two separate patterns, one that handles
11204 registers, and the other that only handles memory. Drop support
11205 for splatting from a GPR on ISA 2.07 and then splitting the
11206 splat into direct move and splat.
11207 (vsx_splat_<mode>_reg): Likewise.
11208 (vsx_splat_<mode>_mem): Likewise.
11210 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11212 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11214 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11216 PR middle-end/80809
11217 * omp-low.c (finish_taskreg_remap): New function.
11218 (finish_taskreg_scan): If unit size of ctx->record_type
11219 is non-constant, unshare the size expression and replace
11220 decls in it with possible outer var refs.
11222 PR middle-end/80809
11223 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11224 GOVD_SHARED rather than GOVD_PRIVATE with it.
11225 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11226 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11228 PR middle-end/80853
11229 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11230 as last argument to build_outer_var_ref for pointer bases of array
11231 section reductions.
11233 2017-05-19 Martin Sebor <msebor@redhat.com>
11235 * print-tree.c (print_node): Print DECL_READ_P flag.
11237 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11239 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11240 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11241 * cgraph.c: Likewise.
11242 * cgraphunit.c: Likewise.
11243 * gengtype.c: Likewise.
11244 * ipa-cp.c: Likewise.
11245 * ipa-devirt.c: Likewise.
11246 * ipa-icf.c: Likewise.
11247 * ipa-predicate.c: Likewise.
11248 * ipa-profile.c: Likewise.
11249 * ipa-prop.c: Likewise.
11250 * ipa-split.c: Likewise.
11252 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11253 edge_predicate_pool, dump_inline_hints,
11254 inline_summary::account_size_time, redirect_to_unreachable,
11255 edge_set_predicate, set_hint_predicate,
11256 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11257 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11258 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11259 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11260 ipa_call_summary_t::remove, initialize_growth_caches,
11261 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11262 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11263 mark_modified, unmodified_parm_1, unmodified_parm,
11264 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11265 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11266 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11267 will_be_nonconstant_predicate, record_modified_bb_info,
11268 get_minimal_bb, record_modified, param_change_prob,
11269 phi_result_unknown_predicate, predicate_for_phi_result,
11270 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11271 estimate_function_body_sizes, compute_inline_parameters,
11272 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11273 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11274 inline_update_callee_summaries, remap_edge_change_prob,
11275 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11276 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11277 inline_analyze_function, inline_summary_t::insert,
11278 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11279 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11280 inline_free_summary): Move to ipa-fnsummary.h
11281 (predicate_t): Remove.
11282 * ipa-fnsummary.c: New file.
11283 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11284 (enum inline_hints_vals, inline_hints, agg_position_info,
11285 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11286 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11287 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11288 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11289 inline_read_summary, inline_write_summary, inline_free_summary,
11290 inline_analyze_function, initialize_inline_failed,
11291 inline_merge_summary, inline_update_overall_summary,
11292 compute_inline_parameters): Move to ipa-fnsummary.h
11293 * ipa-fnsummary.h: New file.
11294 * ipa-inline-transform.h: Include ipa-inline.h.
11295 * ipa-inline.c: LIkewise.
11297 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11299 * ipa-inline.c (edge_badness): Use inlined_time instead of
11300 inline_summaries->get.
11302 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11304 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11306 2017-05-22 Nathan Sidwell <nathan@acm.org>
11308 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11309 (fdump-lang): Document 'raw' option.
11310 * dumpfile.h (TDI_tu): Delete.
11311 * dumpfile.c (dump_files): Remove translation-unit.
11312 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11314 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11316 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11317 command option from $(AWK) call.
11318 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11319 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11320 [FORMAT]: Remove handling of variable.
11321 * config/avr/t-multilib: Regenerate.
11323 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11325 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11327 (dump_inline_summary): Do not print self_time.
11328 (estimate_function_body_sizes): Do not set self_time.
11329 (compute_inline_parameters): Likewise.
11330 (inline_read_section, inline_write_summary): Do not stream self_time.
11331 * ipa-inline.h (inline_summary): Drop self_time.
11333 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11335 * ipa-inline-analysis.c (account_size_time): Rename to ...
11336 (inline_summary::account_size_time): ... this one.
11337 (reset_ipa_call_summary): Turn to ...
11338 (ipa_call_summary::reset): ... this one.
11339 (reset_inline_summary): Turn to ...
11340 (inline_summary::reset): ... this one.
11341 (inline_summary_t::remove): Update.
11342 (inline_summary_t::duplicate): Update.
11343 (ipa_call_summary_t::remove): Update.
11344 (dump_inline_summary): Update.
11345 (estimate_function_body_sizes): Update.
11346 (compute_inline_parameters): Update.
11347 (estimate_node_size_and_time): Update.
11348 (inline_merge_summary): Update.
11349 (inline_update_overall_summary): Update.
11350 (inline_read_section): Update.
11351 (inline_write_summary): Update.
11352 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
11353 add account_size_time and reset member functions.
11354 (ipa_call_summary): Add reset function.
11355 * ipa-predicate.h (predicate::operator &): Constify.
11357 2017-05-22 Richard Biener <rguenther@suse.de>
11359 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
11361 2017-05-19 Jason Merrill <jason@redhat.com>
11363 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
11365 2017-05-19 Marek Polacek <polacek@redhat.com>
11368 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
11369 TYPE_OVERFLOW_WRAPS checks.
11371 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
11373 * tree-core.h (enum omp_clause_default_kind): Add
11374 "OMP_CLAUSE_DEFAULT_PRESENT".
11375 * tree-pretty-print.c (dump_omp_clause): Handle it.
11376 * gimplify.c (enum gimplify_omp_var_data): Add
11377 "GOVD_MAP_FORCE_PRESENT".
11378 (gimplify_adjust_omp_clauses_1): Map it to
11379 "GOMP_MAP_FORCE_PRESENT".
11380 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
11382 * gimplify.c (oacc_default_clause): Clarify.
11384 2017-05-19 Nathan Sidwell <nathan@acm.org>
11386 LANG_HOOK_REGISTER_DUMPS
11387 * toplev.c (general_init): Call register dump lang hook.
11388 * doc/invoke.texi: Document -fdump-lang option family.
11389 * dumpfile.c (dump_files): Remove class dump here.
11390 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
11391 * dumpfile.h (tree_dump_index): Remove TDI_class.
11392 * langhooks-def.h (lhd_register_dumps): Declare.
11393 (LANG_HOOKS_REGISTER_DUMPS): Define.
11394 (LANG_HOOKS_INITIALIZER): Add it.
11395 * langhooks.c (lhd_register_dumps): Define.
11396 * langhooks.h (struct lang_hooks): Add register_dumps.
11398 2017-05-19 Nathan Sidwell <nathan@acm.org>
11400 * context.h (context::set_passes): New.
11401 * context.c (context::context): Do not create pass manager.
11402 * toplev.c (general_init): Create pass manager here.
11404 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
11406 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
11407 use this splitter if two add or or instructions would also work for
11408 the constant we want to generate.
11410 2017-05-19 Richard Biener <rguenther@suse.de>
11413 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
11414 predicate evaluation.
11416 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11418 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
11420 * ipa-inline.c (want_inline_small_function_p): Do not cast to
11423 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
11425 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11426 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11427 (inline_edge_summary_vec): Turn into ...
11428 (ipa_call_summaries): ... this one.
11429 (redirect_to_unreachable, edge_set_predicate,
11430 evaluate_properties_for_edge, inline_summary_alloc,
11431 reset_ipa_call_summary, reset_inline_summary,
11432 inline_summary_t::duplicate): Update.
11433 (inline_edge_duplication_hook): Turn to ...
11434 (ipa_call_summary_t::duplicate): ... this one.
11435 (inline_edge_removal_hook): Turn to ...
11436 (ipa_call_summary_t::remove): ... this one.
11437 (dump_inline_edge_summary): Turn to ...
11438 (dump_ipa_call_summary): ... this one.
11439 (estimate_function_body_sizes): Update.
11440 (inline_update_callee_summaries): Update.
11441 (remap_edge_change_prob): Update.
11442 (remap_edge_summaries): Update.
11443 (inline_merge_summary): Update.
11444 (do_estimate_edge_time): Update.
11445 (inline_generate_summary): Update.
11446 (inline_read_section): Update.
11447 (inline_read_summary): Update.
11448 (inline_free_summary): Update.
11449 * ipa-inline.c (can_inline_edge_p): Update.
11450 (compute_inlined_call_time): Update.
11451 (want_inline_small_function_p): Update.
11452 (edge_badness): Update.
11453 (early_inliner): Update.
11454 * ipa-inline.h (inline_edge_summary): Turn to ...
11455 (ipa_call_summary): ... this one.
11456 (ipa_call_summary_t): New class.
11457 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11458 (ipa_call_summaries): New.
11459 (inline_edge_summary): Remove.
11460 (estimate_edge_growth): Update.
11461 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11462 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11463 * ipa-split.c (execute_split_functions): Update.
11464 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11466 2017-05-19 Richard Biener <rguenther@suse.de>
11468 PR middle-end/80764
11469 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
11471 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
11473 * config/rs6000/rs6000.c (struct machine_function): Add field
11474 fpr_is_wrapped_separately.
11475 (rs6000_get_separate_components): Use 64 components. Handle the
11476 new FPR components.
11477 (rs6000_components_for_bb): Handle the FPR components.
11478 (rs6000_emit_prologue_components): Handle the FPR components.
11479 (rs6000_emit_epilogue_components): Handle the FPR components.
11480 (rs6000_set_handled_components): Handle the FPR components.
11481 (rs6000_emit_prologue): Don't output prologue code for those FPRs
11482 that are already separately shrink-wrapped.
11483 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
11484 that are already separately shrink-wrapped.
11486 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11489 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
11492 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
11493 (define_peephole2 for Altivec d-form load): Add peepholes to catch
11494 cases where the register allocator uses a move and an offsettable
11495 memory operation to/from a FPR register on ISA 2.06/2.07.
11496 (define_peephole2 for Altivec d-form store): Likewise.
11498 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
11501 * config/i386/mmx.md (*mov<mode>_internal): Enable
11502 alternatives 11, 12, 13 and 14 also for 32bit targets.
11503 Remove alternatives 15, 16, 17 and 18.
11504 * config/i386/sse.md (vec_concatv2di): Change
11505 alternative (!x, *y) to (x, ?!*Yn).
11507 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
11509 * dumpfile.h (enum dump_kind): Remove stray comma.
11511 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11513 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
11514 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
11515 predicate::num_conditions
11516 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
11517 (CHANGED): turn into predicate::changed.
11518 (agg_position_info): Move to ipa-predicate.h
11519 (add_condition, predicate::add_clause, predicate::operator &=,
11520 predicate::or_with, predicate::evaluate, predicate::probability,
11521 dump_condition, dump_clause, predicate::dump,
11522 predicate::remap_after_duplication, predicate::remap_after_inlining,
11523 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
11524 (evaluate_conditions_for_known_args): Update.
11525 (set_cond_stmt_execution_predicate): Update.
11526 * ipa-inline.h: Include ipa-predicate.h
11527 (condition, inline_param_summary, conditions, agg_position_info,
11528 predicate): Move to ipa-predicate.h
11529 * ipa-predicate.c: New file.
11530 * ipa-predicate.h: New file.
11532 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
11534 * final.c (leaf_function_p): Check we are not in a sequence.
11536 2017-05-18 Martin Liska <mliska@suse.cz>
11538 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
11539 * dumpfile.c (dump_register): Use new enum dump_kind.
11540 (get_dump_file_name): Likewise.
11541 (dump_enable_all): Likewise.
11542 (dump_switch_p_1): Likewise.
11543 (enable_rtl_dump_file): Remove usage of TDF_RTL.
11544 * dumpfile.h (enum dump_kind): New enum type.
11545 (struct dump_file_info): Create constructor and
11546 format fields and comments.
11547 * passes.c (pass_manager::register_one_dump_file):
11549 * statistics.c (statistics_early_init): Likewise.
11550 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
11551 TDF_TREE with TDF_SLIM.
11552 (gather_memory_references_ref): Likewise.
11554 2017-05-18 Martin Liska <mliska@suse.cz>
11556 * vec.h (struct vnull): Use it.
11558 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
11560 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
11561 (true_predicate, false_predicate, true_predicate_p,
11562 false_predicate_p): Remove.
11563 (single_cond_predicate, not_inlined_predicate): Turn to member function
11565 (add_condition): Update.
11566 (add_clause): Turn to...
11567 (predicate::add_clause): ... this one; update; allow passing NULL
11569 (and_predicates): Turn to ...
11570 (predicate::operator &=): ... this one.
11571 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
11572 (or_predicates): Turn to ...
11573 (predicate::or_with): ... this one.
11574 (evaluate_predicate): Turn to ...
11575 (predicate::evaluate): ... this one.
11576 (predicate_probability): Turn to ...
11577 (predicate::probability): ... this one.
11578 (dump_condition): Update.
11579 (dump_predicate): Turn to ...
11580 (predicate::dump): ... this one.
11581 (account_size_time): Update.
11582 (edge_set_predicate): Update.
11583 (set_hint_predicate): UPdate.
11584 (evaluate_conditions_for_known_args): Update.
11585 (evaluate_properties_for_edge): Update.
11586 (remap_predicate_after_duplication): Turn to...
11587 (predicate::remap_after_duplication): ... this one.
11588 (remap_hint_predicate_after_duplication): Update.
11589 (inline_summary_t::duplicate): UPdate.
11590 (dump_inline_edge_summary): Update.
11591 (dump_inline_summary): Update.
11592 (set_cond_stmt_execution_predicate): Update.
11593 (set_switch_stmt_execution_predicate): Update.
11594 (compute_bb_predicates): Update.
11595 (will_be_nonconstant_expr_predicate): Update.
11596 (will_be_nonconstant_predicate): Update.
11597 (phi_result_unknown_predicate): Update.
11598 (predicate_for_phi_result): Update.
11599 (array_index_predicate): Update.
11600 (estimate_function_body_sizes): Update.
11601 (estimate_node_size_and_time): Update.
11602 (estimate_ipcp_clone_size_and_time): Update.
11603 (remap_predicate): Rename to ...
11604 (predicate::remap_after_inlining): ... this one.
11605 (remap_hint_predicate): Update.
11606 (inline_merge_summary): Update.
11607 (inline_update_overall_summary): Update.
11608 (estimate_size_after_inlining): Update.
11609 (read_predicate): Rename to ...
11610 (predicate::stream_in): ... this one.
11611 (read_inline_edge_summary): Update.
11612 (write_predicate): Rename to ...
11613 (predicate::stream_out): ... this one.
11614 (write_inline_edge_summary): Update.
11615 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
11616 (clause_t): Turn to uint32_t
11617 (predicate): Turn to class; implement constructor and operators
11619 (size_time_entry): Update.
11620 (inline_summary): Update.
11621 (inline_edge_summary): Update.
11623 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
11625 * fold-const.c (fold_binary_loc): Move transformation...
11626 * match.pd (C - X CMP X): ... here.
11628 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
11630 * config/sparc/sparc.c (sparc_option_override): Set function
11631 alignment for -mcpu=niagara7 to 64 to match the I$ line.
11632 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
11634 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
11636 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
11638 2017-05-18 Marek Polacek <polacek@redhat.com>
11641 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
11642 (pass_ubsan::execute): Call gimple_assign_single_p instead of
11643 gimple_assign_load_p.
11645 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
11647 PR middle-end/80692
11648 * real.c (do_compare): Give decimal_do_compare preference over
11649 comparing just the signs.
11651 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
11653 * doc/md.texi (Canonicalization of Instructions): Describe the
11654 canonical form of instructions that inherently set a condition
11657 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
11659 PR middle-end/80775
11660 * tree-cfg.c: Move deletion of unreachable case statements to after
11661 the merging of consecutive case labels.
11663 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
11665 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
11666 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
11667 restoring of callee-saved registers.
11669 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
11671 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
11672 * config/visium/visium.c (single_set_and_flags): Likewise.
11673 * config/visium/visium.md (Substitutions): Likewise.
11675 2017-05-17 Martin Liska <mliska@suse.cz>
11677 * cfg.c: Introduce dump_flags_t type and
11678 use it instead of int type.
11680 * cfghooks.c: Likewise.
11681 * cfghooks.h (struct cfg_hooks): Likewise.
11682 * cfgrtl.c: Likewise.
11683 * cfgrtl.h: Likewise.
11684 * cgraph.c (cgraph_node::get_body): Likewise.
11685 * coretypes.h: Likewise.
11686 * domwalk.c: Likewise.
11687 * domwalk.h: Likewise.
11688 * dumpfile.c (struct dump_option_value_info): Likewise.
11689 (dump_enable_all): Likewise.
11690 (dump_switch_p_1): Likewise.
11691 (opt_info_switch_p): Likewise.
11692 * dumpfile.h (enum tree_dump_index): Likewise.
11693 (struct dump_file_info): Likewise.
11694 * genemit.c: Likewise.
11695 * generic-match-head.c: Likewise.
11696 * gengtype.c (open_base_files): Likewise.
11697 * gimple-pretty-print.c: Likewise.
11698 * gimple-pretty-print.h: Likewise.
11699 * graph.c (print_graph_cfg): Likewise.
11700 * graphite-scop-detection.c (dot_all_sese): Likewise.
11701 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
11702 * loop-unroll.c (report_unroll): Likewise.
11703 * passes.c (pass_manager::register_one_dump_file): Likewise.
11704 * print-tree.c: Likewise.
11705 * statistics.c: Likewise.
11706 * tree-cfg.c: Likewise.
11707 * tree-cfg.h: Likewise.
11708 * tree-dfa.c: Likewise.
11709 * tree-dfa.h: Likewise.
11710 * tree-dump.c (dump_function): Likewise.
11711 * tree-dump.h (struct dump_info): Likewise.
11712 * tree-pretty-print.c: Likewise.
11713 * tree-pretty-print.h: Likewise.
11714 * tree-ssa-live.c: Likewise.
11715 * tree-ssa-live.h: Likewise.
11716 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
11717 * tree-vect-loop.c: Likewise.
11718 * tree-vect-slp.c: Likewise.
11720 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
11721 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11723 PR tree-optimization/80457
11724 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
11725 of all arguments to a statement as scalar_to_vec operations.
11726 (vectorizable_call): Adjust call to vect_model_simple_cost for
11728 (vectorizable_conversion): Likewise.
11729 (vectorizable_assignment): Likewise.
11730 (vectorizable_shift): Likewise.
11731 (vectorizable_operation): Likewise.
11732 (vectorizable_comparison): Likewise.
11733 (vect_is_simple_cond): Record the def types for operands.
11734 (vectorizable_condition): Likewise, call vect_model_simple_cost.
11735 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
11736 for statement argument count.
11738 2017-05-16 Carl Love <cel@us.ibm.com>
11740 * config/rs6000/rs6000-c: Add support for built-in functions
11741 vector unsigned long long vec_bperm (vector unsigned long long,
11742 vector unsigned char)
11743 vector signed long long vec_mule (vector signed int,
11745 vector unsigned long long vec_mule (vector unsigned int,
11746 vector unsigned int)
11747 vector signed long long vec_mulo (vector signed int,
11749 vector unsigned long long vec_mulo (vector unsigned int,
11750 vector unsigned int)
11751 vector signed char vec_sldw (vector signed char,
11752 vector signed char,
11754 vector unsigned char vec_sldw (vector unsigned char,
11755 vector unsigned char,
11757 vector signed short vec_sldw (vector signed short,
11758 vector signed short,
11760 vector unsigned short vec_sldw (vector unsigned short,
11761 vector unsigned short,
11763 vector signed int vec_sldw (vector signed int,
11766 vector unsigned int vec_sldw (vector unsigned int,
11767 vector unsigned int,
11769 vector signed long long vec_sldw (vector signed long long,
11770 vector signed long long,
11772 vector unsigned long long vec_sldw (vector unsigned long long,
11773 vector unsigned long long,
11775 * config/rs6000/rs6000-c: Add support for built-in functions
11776 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
11777 * config/rs6000/altivec.h: Add defintion for vec_sldw.
11778 * doc/extend.texi: Update the built-in documentation for the
11779 new built-in functions.
11781 2017-05-16 Marek Polacek <polacek@redhat.com>
11785 * tree.c (save_expr): Don't fold the expression.
11787 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
11789 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
11790 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
11791 and (?*y,m). Update insn attributes.
11793 2017-05-16 Martin Liska <mliska@suse.cz>
11795 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
11796 flags argument of print_gimple_stmt, print_gimple_expr,
11797 print_generic_stmt and print_generic_expr.
11798 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
11799 * coretypes.h: Likewise.
11800 * except.c (dump_eh_tree): Likewise.
11801 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11802 * gimple-pretty-print.h: Likewise.
11803 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
11804 (backprop::push_to_worklist): Likewise.
11805 (backprop::pop_from_worklist): Likewise.
11806 (backprop::process_use): Likewise.
11807 (backprop::intersect_uses): Likewise.
11808 (note_replacement): Likewise.
11809 * gimple-ssa-store-merging.c
11810 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
11811 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
11812 (pass_store_merging::execute): Likewise.
11813 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
11814 (ssa_base_cand_dump_callback): Likewise.
11815 (dump_incr_vec): Likewise.
11816 (replace_refs): Likewise.
11817 (replace_mult_candidate): Likewise.
11818 (create_add_on_incoming_edge): Likewise.
11819 (create_phi_basis): Likewise.
11820 (insert_initializers): Likewise.
11821 (all_phi_incrs_profitable): Likewise.
11822 (introduce_cast_before_cand): Likewise.
11823 (replace_one_candidate): Likewise.
11824 * gimplify.c (gimplify_expr): Likewise.
11825 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
11826 (set_rename): Likewise.
11827 (rename_uses): Likewise.
11828 (copy_loop_phi_nodes): Likewise.
11829 (add_close_phis_to_merge_points): Likewise.
11830 (copy_loop_close_phi_args): Likewise.
11831 (copy_cond_phi_args): Likewise.
11832 (graphite_copy_stmts_from_block): Likewise.
11833 (translate_pending_phi_nodes): Likewise.
11834 * graphite-poly.c (print_pdr): Likewise.
11835 (dump_gbb_cases): Likewise.
11836 (dump_gbb_conditions): Likewise.
11837 (print_scop_params): Likewise.
11838 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
11839 (build_cross_bb_scalars_use): Likewise.
11840 (gather_bbs::before_dom_children): Likewise.
11841 * hsa-dump.c (dump_hsa_immed): Likewise.
11842 * ipa-cp.c (print_ipcp_constant_value): Likewise.
11843 (get_replacement_map): Likewise.
11844 * ipa-inline-analysis.c (dump_condition): Likewise.
11845 (estimate_function_body_sizes): Likewise.
11846 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
11847 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
11848 * ipa-prop.c (ipa_dump_param): Likewise.
11849 (ipa_print_node_jump_functions_for_edge): Likewise.
11850 (ipa_modify_call_arguments): Likewise.
11851 (ipa_modify_expr): Likewise.
11852 (ipa_dump_param_adjustments): Likewise.
11853 (ipa_dump_agg_replacement_values): Likewise.
11854 (ipcp_modif_dom_walker::before_dom_children): Likewise.
11855 * ipa-pure-const.c (check_stmt): Likewise.
11856 (pass_nothrow::execute): Likewise.
11857 * ipa-split.c (execute_split_functions): Likewise.
11858 * omp-offload.c (dump_oacc_loop_part): Likewise.
11859 (dump_oacc_loop): Likewise.
11860 * trans-mem.c (tm_log_emit): Likewise.
11861 (tm_memopt_accumulate_memops): Likewise.
11862 (dump_tm_memopt_set): Likewise.
11863 (dump_tm_memopt_transform): Likewise.
11864 * tree-cfg.c (gimple_verify_flow_info): Likewise.
11865 (print_loop): Likewise.
11866 * tree-chkp-opt.c (chkp_print_addr): Likewise.
11867 (chkp_gather_checks_info): Likewise.
11868 (chkp_get_check_result): Likewise.
11869 (chkp_remove_check_if_pass): Likewise.
11870 (chkp_use_outer_bounds_if_possible): Likewise.
11871 (chkp_reduce_bounds_lifetime): Likewise.
11872 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
11873 (chkp_mark_completed_bounds): Likewise.
11874 (chkp_register_incomplete_bounds): Likewise.
11875 (chkp_mark_invalid_bounds): Likewise.
11876 (chkp_maybe_copy_and_register_bounds): Likewise.
11877 (chkp_build_returned_bound): Likewise.
11878 (chkp_get_bound_for_parm): Likewise.
11879 (chkp_build_bndldx): Likewise.
11880 (chkp_get_bounds_by_definition): Likewise.
11881 (chkp_generate_extern_var_bounds): Likewise.
11882 (chkp_get_bounds_for_decl_addr): Likewise.
11883 * tree-chrec.c (chrec_apply): Likewise.
11884 * tree-data-ref.c (dump_data_reference): Likewise.
11885 (dump_subscript): Likewise.
11886 (dump_data_dependence_relation): Likewise.
11887 (analyze_overlapping_iterations): Likewise.
11888 * tree-inline.c (expand_call_inline): Likewise.
11889 (tree_function_versioning): Likewise.
11890 * tree-into-ssa.c (dump_defs_stack): Likewise.
11891 (dump_currdefs): Likewise.
11892 (dump_names_replaced_by): Likewise.
11893 (dump_update_ssa): Likewise.
11894 (update_ssa): Likewise.
11895 * tree-object-size.c (pass_object_sizes::execute): Likewise.
11896 * tree-parloops.c (build_new_reduction): Likewise.
11897 (try_create_reduction_list): Likewise.
11898 (ref_conflicts_with_region): Likewise.
11899 (oacc_entry_exit_ok_1): Likewise.
11900 (oacc_entry_exit_single_gang): Likewise.
11901 * tree-pretty-print.h: Likewise.
11902 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
11903 (get_scalar_evolution): Likewise.
11904 (add_to_evolution): Likewise.
11905 (get_loop_exit_condition): Likewise.
11906 (analyze_evolution_in_loop): Likewise.
11907 (analyze_initial_condition): Likewise.
11908 (analyze_scalar_evolution): Likewise.
11909 (instantiate_scev): Likewise.
11910 (number_of_latch_executions): Likewise.
11911 (gather_chrec_stats): Likewise.
11912 (final_value_replacement_loop): Likewise.
11913 (scev_const_prop): Likewise.
11914 * tree-sra.c (dump_access): Likewise.
11915 (disqualify_candidate): Likewise.
11916 (create_access): Likewise.
11917 (reject): Likewise.
11918 (maybe_add_sra_candidate): Likewise.
11919 (create_access_replacement): Likewise.
11920 (analyze_access_subtree): Likewise.
11921 (analyze_all_variable_accesses): Likewise.
11922 (sra_modify_assign): Likewise.
11923 (initialize_constant_pool_replacements): Likewise.
11924 (find_param_candidates): Likewise.
11925 (decide_one_param_reduction): Likewise.
11926 (replace_removed_params_ssa_names): Likewise.
11927 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
11928 * tree-ssa-copy.c (dump_copy_of): Likewise.
11929 (copy_prop_visit_cond_stmt): Likewise.
11930 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
11931 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
11932 (record_equivalences_from_stmt): Likewise.
11933 * tree-ssa-dse.c (compute_trims): Likewise.
11934 (delete_dead_call): Likewise.
11935 (delete_dead_assignment): Likewise.
11936 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
11937 (forward_propagate_into_cond): Likewise.
11938 (pass_forwprop::execute): Likewise.
11939 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11940 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
11942 (move_computations_worker): Likewise.
11943 (execute_sm): Likewise.
11944 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
11945 (remove_exits_and_undefined_stmts): Likewise.
11946 (remove_redundant_iv_tests): Likewise.
11947 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
11948 (adjust_iv_update_pos): Likewise.
11949 * tree-ssa-math-opts.c (bswap_replace): Likewise.
11950 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
11951 (value_replacement): Likewise.
11952 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
11953 * tree-ssa-pre.c (print_pre_expr): Likewise.
11954 (get_representative_for): Likewise.
11955 (create_expression_by_pieces): Likewise.
11956 (insert_into_preds_of_block): Likewise.
11957 (eliminate_insert): Likewise.
11958 (eliminate_dom_walker::before_dom_children): Likewise.
11959 (eliminate): Likewise.
11960 (remove_dead_inserted_code): Likewise.
11961 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
11962 * tree-ssa-reassoc.c (get_rank): Likewise.
11963 (eliminate_duplicate_pair): Likewise.
11964 (eliminate_plus_minus_pair): Likewise.
11965 (eliminate_not_pairs): Likewise.
11966 (undistribute_ops_list): Likewise.
11967 (eliminate_redundant_comparison): Likewise.
11968 (update_range_test): Likewise.
11969 (optimize_range_tests_var_bound): Likewise.
11970 (optimize_vec_cond_expr): Likewise.
11971 (rewrite_expr_tree): Likewise.
11972 (rewrite_expr_tree_parallel): Likewise.
11973 (linearize_expr): Likewise.
11974 (break_up_subtract): Likewise.
11975 (linearize_expr_tree): Likewise.
11976 (attempt_builtin_powi): Likewise.
11977 (attempt_builtin_copysign): Likewise.
11978 (transform_stmt_to_copy): Likewise.
11979 (transform_stmt_to_multiply): Likewise.
11980 (dump_ops_vector): Likewise.
11981 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
11982 (print_scc): Likewise.
11983 (set_ssa_val_to): Likewise.
11984 (visit_reference_op_store): Likewise.
11985 (visit_use): Likewise.
11986 (sccvn_dom_walker::before_dom_children): Likewise.
11987 (run_scc_vn): Likewise.
11988 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11990 (expr_hash_elt::print): Likewise.
11991 (const_and_copies::pop_to_marker): Likewise.
11992 (const_and_copies::record_const_or_copy_raw): Likewise.
11993 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
11994 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
11995 (dump_predicates): Likewise.
11996 (find_uninit_use): Likewise.
11997 (warn_uninitialized_phi): Likewise.
11998 (pass_late_warn_uninitialized::execute): Likewise.
11999 * tree-ssa.c (verify_vssa): Likewise.
12000 (verify_ssa): Likewise.
12001 (maybe_optimize_var): Likewise.
12002 * tree-vrp.c (dump_value_range): Likewise.
12003 (dump_all_value_ranges): Likewise.
12004 (dump_asserts_for): Likewise.
12005 (register_edge_assert_for_2): Likewise.
12006 (vrp_visit_cond_stmt): Likewise.
12007 (vrp_visit_switch_stmt): Likewise.
12008 (vrp_visit_stmt): Likewise.
12009 (vrp_visit_phi_node): Likewise.
12010 (simplify_cond_using_ranges_1): Likewise.
12011 (fold_predicate_in): Likewise.
12012 (evrp_dom_walker::before_dom_children): Likewise.
12013 (evrp_dom_walker::push_value_range): Likewise.
12014 (evrp_dom_walker::pop_value_range): Likewise.
12015 (execute_early_vrp): Likewise.
12017 2017-05-16 Richard Biener <rguenther@suse.de>
12019 * dwarf2out.c (loc_list_from_tree_1): Do not create
12020 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12022 2017-05-16 Richard Biener <rguenther@suse.de>
12024 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12026 (note_variable_value_in_expr): If we resolved the decl ref
12027 do not push to the stack.
12029 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12031 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12032 operations in fast-math mode.
12033 (vaddq_f16): Likewise.
12034 (vmul_f16): Likewise.
12035 (vmulq_f16): Likewise.
12036 (vsub_f16): Likewise.
12037 (vsubq_f16): Likewise.
12038 * config/arm/neon.md (add<mode>3): New.
12040 (fma:<VH:mode>3): New. Also remove outdated comment.
12043 2017-05-16 Martin Liska <mliska@suse.cz>
12047 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12048 (odr_types_equivalent_p): Likewise.
12050 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12052 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12054 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12057 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12058 non-interunit SSE move alternatives with '?'.
12059 (zero-extendsidi peephole2): New peephole to skip intermediate
12060 general register in SSE zero-extend sequence.
12062 2017-05-15 Jeff Law <law@redhat.com>
12064 * reorg.c (relax_delay_slots): Create a new variable to hold
12065 the temporary target rather than clobbering TARGET_LABEL.
12067 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12068 missing argument to extract_bit_field call.
12069 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12071 2017-05-15 Martin Liska <mliska@suse.cz>
12074 * gcc.c (process_command): Do not allow empty argument of -o option.
12076 2017-05-15 Renlin Li <renlin.li@arm.com>
12078 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12079 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12080 * config/aarch64/constraints.md (Usf): Add long call check.
12081 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12082 (call_value): Likewise.
12083 (sibcall): Likewise.
12084 (sibcall_value): Likewise.
12086 (call_value_insn): New.
12087 (sibcall_insn): Update rtx pattern.
12088 (sibcall_value_insn): Likewise.
12089 (call_internal): Remove.
12090 (call_value_internal): Likewise.
12091 (sibcall_internal): Likewise.
12092 (sibcall_value_internal): Likewise.
12093 (call_reg): Likewise.
12094 (call_symbol): Likewise.
12095 (call_value_reg): Likewise.
12096 (call_value_symbol): Likewise.
12098 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12101 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12103 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12105 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12106 compatible with CCGOCmode and with CCZmode.
12108 2017-05-14 Martin Sebor <msebor@redhat.com>
12110 PR middle-end/77671
12111 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12112 (gimple_fold_builtin_snprintf): Same.
12113 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12114 (gimple_fold_builtin_snprintf): Same.
12115 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12116 of character types.
12117 (is_call_safe): New function.
12118 (try_substitute_return_value): Call it.
12119 (try_simplify_call): New function.
12120 (pass_sprintf_length::handle_gimple_call): Call it.
12122 2017-05-14 Martin Sebor <msebor@redhat.com>
12124 PR middle-end/80669
12125 * builtins.c (expand_builtin_stpncpy): Simplify.
12127 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12129 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12130 * config/i386/i386.h
12131 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12132 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12133 (struct machine_function): Add new members call_ms2sysv,
12134 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12135 (struct machine_frame_state): New fields sp_realigned and
12136 sp_realigned_offset.
12137 * config/i386/i386.c
12138 (enum xlogue_stub): New enum.
12139 (enum xlogue_stub_sets): New enum.
12140 (class xlogue_layout): New class.
12141 (struct ix86_frame): New fields stack_realign_allocate_offset,
12142 stack_realign_offset and outlined_save_offset. Modify comments to
12143 detail stack layout when using out-of-line stubs.
12144 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12145 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12146 -mcall-ms2sysv-xlogues.
12147 (stub_managed_regs): New static variable.
12148 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12149 registers managed by out-of-line stub.
12150 (disable_call_ms2sysv_xlogues): New function.
12151 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12152 m->call_ms2sysv when appropriate and compute frame layout for
12154 (sp_valid_at, fp_valid_at): New inline functions.
12155 (choose_basereg): New function.
12156 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12158 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12159 Use align parameter of choose_baseaddr to generated aligned SSE movs
12161 (pro_epilogue_adjust_stack): Modify to track
12162 machine_frame_state::sp_realigned.
12163 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12164 (ix86_nsaved_sseregs): Likewise.
12165 (ix86_emit_save_regs): Likewise.
12166 (ix86_emit_save_regs_using_mov): Likewise.
12167 (ix86_emit_save_sse_regs_using_mov): Likewise.
12168 (get_scratch_register_on_entry): Likewise.
12169 (gen_frame_set): New function.
12170 (gen_frame_load): Likewise.
12171 (gen_frame_store): Likewise.
12172 (emit_outlined_ms2sysv_save): Likewise.
12173 (emit_outlined_ms2sysv_restore): Likewise.
12174 (ix86_expand_prologue): Modify stack re-alignment code and call
12175 emit_outlined_ms2sysv_save when appropriate.
12176 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12177 parameter rtx_insn *insn, which allows the function to be used to only
12178 generate the notes.
12179 (ix86_expand_epilogue): Modify validity checks of frame and stack
12180 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12181 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12182 * config/i386/predicates.md
12183 (save_multiple): New predicate.
12184 (restore_multiple): Likewise.
12185 * config/i386/sse.md
12186 (save_multiple<mode>): New pattern.
12187 (save_multiple_realign<mode>): Likewise.
12188 (restore_multiple<mode>): Likewise.
12189 (restore_multiple_and_return<mode>): Likewise.
12190 (restore_multiple_leave_return<mode>): Likewise.
12191 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12193 2017-05-14 Julia Koval <julia.koval@intel.com>
12195 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12196 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12197 (__builtin_ia32_xsetbv): New builtins.
12198 * config/i386/i386.c (ix86_expand_special_args_builtin):
12200 (ix86_expand_builtin): Special expand for new intrinsics.
12201 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12202 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12203 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12205 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12207 * cfganal.c (inverted_post_order_compute): Change argument type
12209 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12210 * df-core.c (rest_of_handle_df_initialize): Adjust.
12211 (rest_of_handle_df_finish): Likewise.
12212 (df_analyze_1): Likewise.
12213 (df_analyze): Likewise.
12214 (loop_inverted_post_order_compute): Change argument to be a vec *.
12215 (df_analyze_loop): Adjust.
12216 (df_get_n_blocks): Likewise.
12217 (df_get_postorder): Likewise.
12218 * df.h (struct df_d): Change field to be a vec.
12219 * lcm.c (compute_laterin): Adjust.
12220 (compute_available): Likewise.
12221 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12222 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12223 * tree-ssa-pre.c (compute_antic): Likewise.
12225 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12227 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12228 (depth_first_search::depth_first_search): Change structure init
12229 function to this constructor.
12230 (depth_first_search::add_bb): Rename function to this member.
12231 (depth_first_search::execute): Likewise.
12232 (flow_dfs_compute_reverse_finish): Adjust.
12234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12236 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12237 (longest_simple_path): Likewise.
12238 * shrink-wrap.c (spread_components): Likewise.
12239 (disqualify_problematic_components): Likewise.
12240 (emit_common_heads_for_components): Likewise.
12241 (emit_common_tails_for_components): Likewise.
12242 (insert_prologue_epilogue_for_components): Likewise.
12244 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12246 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12249 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12251 * df-core.c (df_set_blocks): Start using auto_bitmap.
12252 (df_compact_blocks): Likewise.
12253 * df-problems.c (df_rd_confluence_n): Likewise.
12254 * df-scan.c (df_insn_rescan_all): Likewise.
12255 (df_process_deferred_rescans): Likewise.
12256 (df_update_entry_block_defs): Likewise.
12257 (df_update_exit_block_uses): Likewise.
12258 (df_entry_block_bitmap_verify): Likewise.
12259 (df_exit_block_bitmap_verify): Likewise.
12260 (df_scan_verify): Likewise.
12261 * lra-constraints.c (lra_constraints): Likewise.
12262 (undo_optional_reloads): Likewise.
12263 (lra_undo_inheritance): Likewise.
12264 * lra-remat.c (calculate_gen_cands): Likewise.
12265 (do_remat): Likewise.
12266 * lra-spills.c (assign_spill_hard_regs): Likewise.
12267 (spill_pseudos): Likewise.
12268 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12269 (bitmap_set_subtract_values): Likewise.
12271 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12273 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12274 management with auto_bitmap.
12275 (fix_inter_tick): Likewise.
12276 (fix_recovery_deps): Likewise.
12277 * ira.c (add_store_equivs): Likewise.
12278 (find_moveable_pseudos): Likewise.
12279 (split_live_ranges_for_shrink_wrap): Likewise.
12280 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12281 (rtx_reuse_manager::seen_def_p): Likewise.
12282 (rtx_reuse_manager::set_seen_def): Likewise.
12283 * print-rtl.h (class rtx_reuse_manager): Likewise.
12285 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12287 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12289 (migrate_btr_def): Likewise.
12290 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12291 * df-core.c (loop_post_order_compute): Likewise.
12292 (loop_inverted_post_order_compute): Likewise.
12293 * hsa-common.h: Likewise.
12294 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12295 * init-regs.c (initialize_uninitialized_regs): Likewise.
12296 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12297 (inline_small_functions): Likewise.
12298 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12299 * ira.c (combine_and_move_insns): Likewise.
12300 (build_insn_chain): Likewise.
12301 * loop-invariant.c (find_invariants): Likewise.
12302 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12303 * predict.c (tree_predict_by_opcode): Likewise.
12304 (predict_paths_leading_to): Likewise.
12305 (predict_paths_leading_to_edge): Likewise.
12306 (estimate_loops_at_level): Likewise.
12307 (estimate_loops): Likewise.
12308 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12309 (spread_components): Likewise.
12310 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12311 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12312 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12313 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12314 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12315 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12316 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12317 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12318 (mark_threaded_blocks): Likewise.
12319 (thread_through_all_blocks): Likewise.
12320 * tree-ssa.c (verify_ssa): Likewise.
12321 (execute_update_addresses_taken): Likewise.
12322 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12324 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12326 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12328 (post_order_compute): Likewise.
12329 (inverted_post_order_compute): Likewise.
12330 (pre_and_rev_post_order_compute_fn): Likewise.
12332 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12334 * genrecog.c (int_set::int_set): Explicitly construct our
12335 auto_vec base class.
12336 * vec.h (auto_vec::auto_vec): New constructor.
12338 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12340 * bitmap.h (class auto_bitmap): New constructor taking
12341 bitmap_obstack * argument.
12343 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12345 * bitmap.h (class auto_bitmap): Change type of m_bits to
12346 bitmap_head, and adjust ctor / dtor and member operators.
12348 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12350 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12351 when returned register mode doesn't match original mode.
12353 2017-05-12 Jeff Law <law@redhat.com>
12354 Jakub Jelinek <jakub@redhat.com>
12356 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
12357 we look for cc setter after the compare-elim changes.
12358 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
12359 within the vector to match what compare-elim now expects.
12360 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
12361 (xorsi3_flags, one_cmplsi2_flags): Likewise.
12363 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
12364 after the compare-elim changes.
12365 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
12366 the vector to match what compare-elim now expects.
12367 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
12368 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
12369 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
12370 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
12371 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
12373 * config/visium/visium.c (single_set_and_flags): Fix where
12374 we look for cc setter after the compare-elim changes.
12375 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
12376 with the vector to match what compare-elim now expects.
12377 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
12378 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
12379 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
12380 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
12381 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
12382 (neg<mode>2_insn_set_overflow): Likewise.
12384 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
12386 PR middle-end/79794
12387 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
12388 maybe_expand_insn call, set ops[0].target. If still set after call,
12389 set alt_rtl. Add extra arg to recursive calls.
12390 (extract_bit_field): Add alt_rtl argument. Pass to
12392 * expmed.h (extract_bit_field): Fix prototype.
12393 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
12394 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
12395 to extract_bit_field_calls.
12396 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
12397 Pass alt_rtl to extract_bit_field calls.
12398 * calls.c (store_unaligned_arguments_into_psuedos)
12399 load_register_parameters): Pass extra NULL to extract_bit_field calls.
12400 * optabs.c (maybe_legitimize_operand): Clear op->target when call
12402 * optabs.h (struct expand_operand): Add target bitfield.
12404 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12406 * compare-elim.c (try_eliminate_compare): Canonicalize
12407 operation with embedded compare to
12408 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
12409 (set (reg) (operation)].
12411 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
12413 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
12416 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
12417 cost of adding a carry flag for ADC instruction.
12418 [case MINUS]: Ignore the cost of subtracting a carry flag
12419 for SBB instruction.
12421 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
12423 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
12425 * config/rs6000/bmiintrin.h: New file.
12426 * config/rs6000/bmi2intrin.h: New file.
12427 * config/rs6000/x86intrin.h: New file.
12429 2017-05-12 Jeff Law <law@redhat.com>
12431 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
12434 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
12436 PR middle-end/80707
12437 * tree-cfg.c: Remove cfg edges of unreachable case statements.
12439 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12441 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12442 early expansion of vector divide builtins.
12443 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
12444 builtins identified as having unsigned arguments.
12446 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12448 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
12449 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
12450 expansion of vector logical operations (and, andc, or, xor,
12453 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12455 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
12456 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
12458 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12460 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
12461 early GIMPLE expansion of vector multiplies.
12463 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12465 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
12466 TARGET_HAVE_MOVT conditional.
12467 (movt splitter): Likewise.
12469 2017-05-12 Richard Biener <rguenther@suse.de>
12471 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
12472 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12473 Fold all stmts not inplace.
12475 2017-05-12 Richard Biener <rguenther@suse.de>
12477 PR tree-optimization/80713
12478 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
12479 inserted_exprs bit for not removed stmts.
12481 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
12483 PR middle-end/69921
12484 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
12485 parallelized" attribute for parallelized OpenACC kernels.
12486 * omp-offload.c (execute_oacc_device_lower): Use it.
12488 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
12489 Set "oacc kernels" attribute.
12490 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
12491 parameter. Adjust all users.
12492 (oacc_fn_attrib_kernels_p): Remove function.
12493 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
12494 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
12495 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
12496 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
12497 assert "oacc kernels" attribute is set.
12499 2017-05-11 Carl Love <cel@us.ibm.com>
12501 * config/rs6000/rs6000-c: Add support for built-in functions
12502 vector unsigned char vec_popcnt (vector signed char)
12503 vector unsigned char vec_popcnt (vector unsigned char)
12504 vector unsigned short vec_popcnt (vector signed short)
12505 vector unsigned short vec_popcnt (vector unsigned short)
12506 vector unsigned int vec_popcnt (vector signed int)
12507 vector unsigned int vec_popcnt (vector unsigned int)
12508 vector unsigned long long vec_popcnt (vector signed long long)
12509 vector unsigned long long vec_popcnt (vector unsigned long long)
12510 vector signed long long vec_slo (vector signed long long,
12511 vector signed char)
12512 vector signed long long vec_slo (vector signed long long,
12513 vector unsigned char)
12514 vector unsigned long long vec_slo (vector unsigned long long,
12515 vector signed char)
12516 vector unsigned long long vec_slo (vector unsigned long long,
12517 vector unsigned char)
12518 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
12519 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
12520 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
12521 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
12522 * doc/extend.texi: Update the built-in documentation file for the
12523 new built-in functions.
12525 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12527 * attribs.h (sorted_attr_string): Move machine independent
12528 functions for target clone support from the i386 port to common
12529 code. Rename ix86_function_versions to common_function_versions.
12530 Rename make_name to make_unique_name.
12531 (common_function_versions): Likewise.
12532 (make_unique_name): Likewise.
12533 (make_dispatcher_decl): Likewise.
12534 (is_function_default_version): Likewise.
12535 * attribs.c (attr_strcmp): Likewise.
12536 (sorted_attr_string): Likewise.
12537 (common_function_versions): Likewise.
12538 (make_unique_name): Likewise.
12539 (make_dispatcher_decl): Likewise.
12540 (is_function_default_version): Likewise.
12541 * config/i386/i386.c (attr_strcmp): Likewise.
12542 (sorted_attr_string): Likewise.
12543 (ix86_function_versions): Likewise.
12544 (make_name): Likewise.
12545 (make_dispatcher_decl): Likewise.
12546 (is_function_default_version): Likewise.
12547 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
12549 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12552 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12553 Account for direct move costs for vec_construct of integer
12556 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
12559 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
12560 (UNSPEC_STX_ATOMIC): Ditto.
12561 (loaddi_via_sse): New insn.
12562 (storedi_via_sse): Ditto.
12563 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
12564 Update corresponding peephole2 patterns.
12565 (atomic_storedi_fpu): Ditto.
12567 2017-05-11 Julia Koval <julia.koval@intel.com>
12569 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
12570 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
12572 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
12573 (__builtin_ia32_rsqrt14ss_mask): New builtins.
12574 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
12576 2017-05-11 Nathan Sidwell <nathan@acm.org>
12578 * graphite-poly.c: Include dumpfile.h.
12580 * dumpfle.h (dump_function): Declare here ...
12581 * tree-dump.h (dump_function): ... not here.
12582 * dumpfile.c: #include tree-cfg.h.
12583 (dump_function): Move here from ...
12584 * tree-dump.c (dump_function): ... here.
12585 * gimplify.c: #include splay-tree.h, not tree-dump.h.
12586 * graphite-poly.c: Don't include tree-dump.h.
12587 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
12588 * print-tree.c: Likewise.
12589 * stor-layout.c: Likewise.
12590 * tree-nested.c: Likewise.
12592 * dumpfile.c (dump_start): Use TDF_FLAGS.
12593 (dump_enable_all): Fix TDF_KIND check thinko.
12595 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12597 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12598 array entries to represent two legal parameterizations of the
12599 overloaded __builtin_cmpb function, as represented by the
12600 P6_OV_BUILTIN_CMPB constant.
12601 (altivec_resolve_overloaded_builtin): Add special case handling
12602 for the __builtin_cmpb function, as represented by the
12603 P6_OV_BUILTIN_CMPB constant.
12604 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
12605 (BU_P6_64BIT_2): New macro.
12606 (BU_P6_OVERLOAD_2): New macro
12607 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
12608 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
12609 (CMPB): Add overload support to represent both 32-bit and 64-bit
12610 compare-bytes function.
12611 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12612 support for TARGET_CMPB.
12613 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
12614 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12615 documentation of the __builtin_cmpb overloaded built-in function.
12617 2017-05-11 Richard Biener <rguenther@suse.de>
12619 PR tree-optimization/80705
12620 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
12621 bases are not vectorizable.
12623 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12625 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
12626 when counting register pressure.
12628 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12630 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
12631 (iv_ca_more_deps): Renamed to ...
12632 (iv_ca_compare_deps): ... this.
12633 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
12635 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12637 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
12639 (determine_group_iv_costs): ... here.
12640 (find_inv_vars_cb): Record inv var if it's not recorded before.
12642 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12644 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
12645 (get_shiftadd_cost): Ditto.
12647 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12649 * tree-ssa-address.c: Include header file.
12650 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
12652 (add_to_parts): Refactor.
12653 (addr_to_parts): New parameter. Update use of move_hint_to_base.
12654 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
12657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12659 PR tree-optimization/53090
12660 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
12662 (extract_cond_operands): Detect condition with IV on both sides
12663 and return COMP_IV_EXPR_2.
12664 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
12665 (rewrite_use_compare): Simplify by removing call to function
12666 extract_cond_operands.
12668 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12670 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
12671 (extract_cond_operands): Detect condition comparing against non-
12672 invariant bound and return appropriate enum value.
12673 (find_interesting_uses_cond): Update use of extract_cond_operands.
12674 Handle its return value accordingly.
12675 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
12677 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12679 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
12680 nonlinear iv_use computation in loop invariant sensitive way.
12682 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12684 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
12685 (find_iv_candidates): Call relate_compare_use_with_all_cands.
12687 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12689 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
12690 (dump_cand): Support iv_cand.inv_exprs.
12691 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
12693 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
12696 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12698 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
12700 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
12701 as local function. Include necessary header files.
12702 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
12704 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12706 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
12708 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12710 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
12711 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
12712 RSHIFT_EXPR and BIT_NOT_EXPR.
12714 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12716 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
12717 (adjust_setup_cost): New parameter supporting round up adjustment.
12718 (struct address_cost_data): Delete.
12719 (force_expr_to_var_cost): Don't bound cost with spill_cost.
12720 (split_address_cost, ptr_difference_cost): Delete.
12721 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
12722 (struct ainc_cost_data): New struct.
12723 (get_address_cost_ainc): New function.
12724 (get_address_cost, get_computation_cost): Reimplement.
12725 (determine_group_iv_cost_address): Record inv_expr for all uses of
12727 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
12728 (iv_ca_has_deps): Reimplemented to ...
12729 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
12731 (iv_ca_extend): Call iv_ca_more_deps.
12733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12735 * tree-ssa-address.c (struct mem_address): Move to header file.
12736 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
12737 * tree-ssa-address.h (struct mem_address): Move from C file.
12738 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
12740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12742 * tree-affine.h (aff_combination_type): New interface.
12743 (aff_combination_zero_p): Remove static.
12744 (aff_combination_const_p): New interface.
12745 (aff_combination_singleton_var_p): New interfaces.
12747 2017-05-11 Richard Biener <rguenther@suse.de>
12749 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12750 Skip unreachable blocks and destinations.
12751 (eliminate): Move stmt removal and fixup ...
12752 (fini_eliminate): ... here. Skip inserted exprs.
12753 (pass_pre::execute): Move fini_pre after fini_eliminate.
12754 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
12755 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
12756 PRE to get rid of dead code that has invalid SSA form and
12757 split critical edges again.
12759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
12761 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
12763 2017-05-11 Richard Biener <rguenther@suse.de>
12765 * passes.c (execute_function_todo): Verify loops if they are
12766 said to be up-to-date.
12767 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
12768 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
12770 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
12773 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
12774 handle calling assemble_external ourself.
12777 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
12778 modes with zero size. Enhance comment.
12780 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12782 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
12783 built-ins for vec_xl and vec_xst with short and char pointer
12786 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
12788 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
12789 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
12790 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
12791 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
12792 (_mm_maskz_min_round_ss): New intrinsics.
12793 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
12794 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
12795 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
12796 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
12797 (__builtin_ia32_minss_mask_round): New builtins.
12798 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
12799 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
12800 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
12802 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
12803 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
12805 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
12808 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
12810 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
12811 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
12812 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
12813 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
12814 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
12815 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
12816 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
12817 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
12818 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
12819 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
12820 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
12821 (__builtin_ia32_mulss_mask_round): New builtins.
12822 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
12823 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
12824 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
12826 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
12827 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
12829 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
12832 2017-05-10 Julia Koval <julia.koval@intel.com>
12834 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
12835 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
12836 (_mm256_setr_m128i): New intrinsics.
12838 2017-05-10 Julia Koval <julia.koval@intel.com>
12840 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
12841 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
12842 (_mm_maskz_rcp14_ss): New intrinsics.
12843 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
12844 (__builtin_ia32_rcp14ss_mask): New builtins.
12845 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
12847 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
12849 PR tree-optimization/51513
12850 * tree-cfg.c (gimple_seq_unreachable_p): New function.
12851 (assert_unreachable_fallthru_edge_p): Use it.
12852 (group_case_labels_stmt): Likewise.
12853 * tree-cfg.h: Prototype it.
12854 * stmt.c: Include cfghooks.h and tree-cfg.h.
12855 (emit_case_dispatch_table) <gap_label>: New local variable.
12856 Use it to fill dispatch table gaps.
12857 Test for default_label before updating probabilities.
12858 (expand_case) <default_label>: Remove unneeded initialization.
12859 Test for unreachable default case statement and remove its edge.
12860 Set default_label accordingly.
12861 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
12863 2017-05-10 Carl Love <cel@us.ibm.com>
12865 * config/rs6000/rs6000-c: Add support for built-in functions
12866 vector signed char vec_neg (vector signed char)
12867 vector signed short int vec_neg (vector short int)
12868 vector signed int vec_neg (vector signed int)
12869 vector signed long long vec_neg (vector signed long long)
12870 vector float vec_neg (vector float)
12871 vector double vec_neg (vector double)
12872 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
12874 * config/rs6000/altivec.h: Add define for vec_neg
12875 * doc/extend.texi: Update the built-in documentation for the
12876 new built-in functions.
12878 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12880 PR tree-optimization/77644
12881 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
12883 2017-05-10 Nathan Sidwell <nathan@acm.org>
12885 * dumpfile.h (TDI_lang_all): New.
12886 (TDF_KIND): New. Renumber others
12887 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
12889 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
12891 (get_dump_file_name): Adjust suffix generation.
12892 (dump_enable_all): Use TDF_KIND.
12893 * doc/invoke.texi (-fdump-lang-all): Document.
12895 * dumpfile.h: Tabify.
12897 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
12900 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
12901 Move member access before delete.
12903 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
12905 * tree-inline.c (expand_call_inline): Split block at stmt
12908 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12911 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
12912 are now unused after splitting mov{sf,sd}_hardfloat.
12913 (f32_lr2): Likewise.
12914 (f32_lm): Likewise.
12915 (f32_lm2): Likewise.
12916 (f32_li): Likewise.
12917 (f32_li2): Likewise.
12918 (f32_lv): Likewise.
12919 (f32_sr): Likewise.
12920 (f32_sr2): Likewise.
12921 (f32_sm): Likewise.
12922 (f32_sm2): Likewise.
12923 (f32_si): Likewise.
12924 (f32_si2): Likewise.
12925 (f32_sv): Likewise.
12926 (f32_dm): Likewise.
12927 (f32_vsx): Likewise.
12928 (f32_av): Likewise.
12929 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
12930 For movsf, order stores so the VSX stores occur before the GPR
12931 store which encourages the register allocator to use a traditional
12932 FPR instead of a GPR. For movsd, order the stores so that the GPR
12933 store comes before the VSX stores to allow the power6 to work.
12934 This is due to the power6 not having a 32-bit integer store
12935 instruction from a FPR.
12936 (movsf_hardfloat): Likewise.
12937 (movsd_hardfloat): Likewise.
12939 2017-05-09 Martin Sebor <msebor@redhat.com>
12941 PR translation/80280
12942 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
12945 PR translation/80280
12946 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
12947 data member added in r247778.
12948 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
12950 2017-05-09 Nathan Sidwell <nathan@acm.org>
12952 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
12954 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
12957 2017-05-09 Marek Polacek <polacek@redhat.com>
12959 * doc/invoke.texi: Fix typo.
12961 2017-05-09 Richard Biener <rguenther@suse.de>
12963 * tree-vrp.c (vrp_val_is_max): Adjust comment.
12964 (vrp_val_is_min): Likewise.
12965 (set_value_range_to_value): Likewise.
12966 (set_value_range_to_nonnegative): Likewise.
12967 (gimple_assign_nonzero_p): Likewise.
12968 (gimple_stmt_nonzero_p): Likewise.
12969 (vrp_int_const_binop): Likewise. Remove unreachable case.
12970 (adjust_range_with_scev): Adjust comments.
12971 (compare_range_with_value): Likewise.
12972 (extract_range_from_phi_node): Likewise.
12973 (test_for_singularity): Likewise.
12975 2017-05-09 Richard Biener <rguenther@suse.de>
12977 * tree-vrp.c (get_single_symbol): Add assert that we don't
12978 get overflowed constants as invariant part.
12979 (compare_values_warnv): Add comment before the TREE_NO_WARNING
12980 checks. Use wi::cmp instead of recursing for integer constants.
12981 (compare_values): Just ignore whether we assumed undefined
12982 overflow instead of failing the compare.
12983 (extract_range_for_var_from_comparison_expr): Add comment before the
12984 TREE_NO_WARNING sets.
12985 (test_for_singularity): Likewise.
12986 (extract_range_from_comparison): Do not disable optimization
12987 when we assumed undefined overflow.
12988 (extract_range_basic): Remove init of unused var.
12990 2017-05-09 Richard Biener <rguenther@suse.de>
12992 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
12993 (extract_range_from_multiplicative_op_1): Adjust.
12994 (extract_range_from_binary_expr_1): Use int_const_binop.
12996 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
12999 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13000 rs6000_store_data_bypass_p in seven define_bypass directives and
13001 in several comments.
13002 * config/rs6000/rs6000-protos.h: Add prototype for
13003 rs6000_store_data_bypass_p function.
13004 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13005 function implements slightly different (rs6000-specific) semantics
13006 than store_data_bypass_p, returning false rather than aborting
13007 with assertion error when arguments do not satisfy the
13008 requirements of store data bypass.
13009 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13010 rs6000_store_data_bypass_p.
13012 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13014 * config/xtensa/xtensa-protos.h
13015 (xtensa_initial_elimination_offset): New declaration.
13016 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13017 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13018 macro definition, add case for FRAME_POINTER_REGNUM when
13019 FRAME_GROWS_DOWNWARD.
13020 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13021 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13022 xtensa_initial_elimination_offset.
13024 2017-05-08 Nathan Sidwell <nathan@acm.org>
13026 * doc/invoke.texi: Alphabetize -fdump options.
13028 2017-05-08 Martin Sebor <msebor@redhat.com>
13030 PR translation/80280
13031 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13033 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13035 * target.def (compute_frame_layout): New optional target hook.
13036 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13037 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13038 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13040 * reload1.c (verify_initial_elim_offsets): Likewise.
13041 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13042 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13043 (arm_get_frame_offsets): Split up into this ...
13044 (arm_compute_frame_layout): ... and this function.
13046 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13048 * config/aarch64/constraints.md (Usa): New constraint.
13049 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13051 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13053 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13054 with_multilib_list after it has been checked.
13056 2017-05-08 Richard Biener <rguenther@suse.de>
13058 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13059 (bitmap_set_subtract_values): Likewise.
13061 2017-05-08 Richard Biener <rguenther@suse.de>
13063 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13064 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13066 (gimple_stmt_nonzero_warnv_p): Rename to ...
13067 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13069 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13070 (extract_range_basic): Adjust, do not disable propagation on
13071 strict overflow sensitive simplification.
13072 (vrp_visit_cond_stmt): Likewise.
13074 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13076 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13077 body size unconditionally.
13079 2017-05-07 Jeff Law <law@redhat.com>
13082 2017-05-06 Jeff Law <law@redhat.com>
13083 PR tree-optimization/78496
13084 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13087 PR tree-optimization/78496
13088 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13089 (simplify_stmt_using_ranges): Call it.
13090 (vrp_dom_walker::before_dom_children): Extract equivalences
13091 from an ASSERT_EXPR with an equality comparison against a
13094 2017-05-06 Jeff Law <law@redhat.com>
13096 PR tree-optimization/78496
13097 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13100 PR tree-optimization/78496
13101 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13102 (simplify_stmt_using_ranges): Call it.
13103 (vrp_dom_walker::before_dom_children): Extract equivalences
13104 from an ASSERT_EXPR with an equality comparison against a
13107 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13109 * lra-constraints.c (lra_copy_reg_equiv): New function.
13110 (split_reg): Use it to copy equivalence information from the
13111 original register to the spill register.
13113 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13115 PR rtl-optimization/75964
13116 * simplify-rtx.c (simplify_const_relational_operation): Remove
13117 invalid handling of comparisons of integer ABS.
13119 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13121 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13122 initialize to zero.
13123 (init_regs): Remove declaration.
13124 (function_arg_advance_32): Initialize error_p as boolean variable.
13126 2017-05-05 Nathan Sidwell <nathan@acm.org>
13128 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13129 lines. Use for (;;).
13131 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13133 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13134 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13135 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13136 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13137 VF=2 that require versioning.
13139 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13141 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13144 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13146 * diagnostic.h (diagnostic_override_option_index): Convert from
13147 macro to inline function.
13149 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13151 * diagnostic.c (last_module_changed_p): New function.
13152 (set_last_module): New function.
13153 (diagnostic_report_current_module): Convert macro usage to
13154 the above functions.
13155 * diagnostic.h (diagnostic_context::last_module): Strengthen
13156 from const line_map * to const line_map_ordinary *.
13157 (diagnostic_last_module_changed): Delete macro.
13158 (diagnostic_set_last_module): Delete macro.
13160 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13162 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13163 with diagnostic_report_diagnostic.
13164 (diagnostic_n_impl_richloc): Likewise.
13165 * diagnostic.h (report_diagnostic): Delete macro.
13166 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13167 with diagnostic_report_diagnostic.
13168 * substring-locations.c (format_warning_va): Likewise.
13170 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13172 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13173 save/restor of format_spec. Move option-printing code to...
13174 (print_option_information): ...this new function, and
13175 reimplement by simply printing to the pretty_printer,
13176 rather than appending to the format string.
13178 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13180 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13181 handling logic into...
13182 (update_effective_level_from_pragmas): ...this new function.
13184 2017-05-04 Andrew Waterman <andrew@sifive.com>
13186 * config/riscv/riscv.opt (mstrict-align): New option.
13187 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13188 (SLOW_UNALIGNED_ACCESS): Define.
13189 (riscv_slow_unaligned_access): Declare.
13190 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13192 (riscv_slow_unaligned_access): New variable.
13193 (rocket_tune_info): Set slow_unaligned_access to true.
13194 (optimize_size_tune_info): Set slow_unaligned_access to false.
13195 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13196 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13197 (riscv_option_override): Set riscv_slow_unaligned_access.
13198 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13200 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13202 * config/riscv/riscv.md: Unify indentation.
13204 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13209 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13211 (extendsi<mode>2): Add support for doing sign extension via
13212 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13213 don't have ISA 3.0 instructions.
13214 (extendsi<mode>2 splitter): Likewise.
13215 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13216 generate the normal insns since SImode can now go in vector
13217 registers. Disallow the special UNSPECs needed for previous
13218 machines to hide SImode being used. Add new insns
13219 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13220 (fix_trunc<mode>si2_stfiwx): Likewise.
13221 (fix_trunc<mode>si2_internal): Likewise.
13222 (fixuns_trunc<mode>si2): Likewise.
13223 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13224 (fctiw<u>z_<mode>_smallint): Likewise.
13225 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13226 of floating point to 32-bit integer from doing a direct move to
13227 the GPR registers to do a store.
13228 (fctiwz_<mode>): Break long line.
13230 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13232 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13233 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13234 (addr_list, addr_offset_valid_p): New.
13235 (split_address_groups): Check offset validity with above function.
13236 (gt-tree-ssa-loop-ivopts.h): Include header file.
13238 2017-05-05 Nathan Sidwell <nathan@acm.org>
13240 * config.gcc (arm*-*-*): Add missing 'fi'.
13242 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13244 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13245 included in -fopt-info does not matter.
13246 * doc/optinfo.texi (-fopt-info): Fix description of default
13247 behavour. Explicitly say order of options included in -fopt-info
13250 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13252 * config.gcc: Allow combinations of aprofile and rmprofile values for
13253 --with-multilib-list.
13254 * config/arm/t-multilib: New file.
13255 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13256 variables. Remove setting of ISA and floating-point ABI in
13257 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13258 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13259 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13260 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13262 * config/arm/t-rmprofile: Likewise except for the matches changes.
13263 * doc/install.texi (--with-multilib-list): Document the combination of
13264 aprofile and rmprofile values and warn about pitfalls in doing that.
13266 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13268 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13269 (movdi_aarch64): Likewise.
13271 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13273 PR tree-optimization/80632
13274 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13276 (build_arrays): Initialize it for virtual phis.
13277 (fix_phi_nodes): Use it for virtual phis.
13279 PR tree-optimization/80558
13280 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13281 [x, y] op z into [x op, y op z] for op & or | if conditions
13284 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13285 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13288 * config/arm/arm.md (use_literal_pool): Remove.
13289 (64-bit immediate split): No longer takes cost into consideration
13290 if arm_disable_literal_pool is enabled.
13291 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13292 used when arm_disable_literal_pool is enabled.
13293 (arm_max_const_double_inline_cost): Remove use of
13294 arm_disable_literal_pool.
13295 (push_minipool_fix): Add assert.
13296 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13297 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13298 (no_literal_pool_sf_immediate): New.
13300 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13302 PR tree-optimization/80613
13303 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13304 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13306 2017-05-05 Richard Biener <rguenther@suse.de>
13308 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13310 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13312 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13313 of this flag from insn conditions due to removal from r247495.
13315 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13317 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13319 (arm_early_store_addr_dep_ptr): Likewise.
13320 * config/arm/aarch-common-protos.h
13321 (arm_early_load_addr_dep_ptr): Add prototype.
13322 (arm_early_store_addr_dep_ptr): Likewise.
13323 * config/arm/cortex-a53.md: Add new bypasses.
13325 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13327 * tree.c (next_type_uid): Change type to unsigned.
13328 (type_hash_canon): Decrement back next_type_uid if
13329 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13330 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13333 2017-05-04 Martin Sebor <msebor@redhat.com>
13335 * builtins.c: Fix a trivial typo in a comment.
13337 PR middle-end/79234
13338 * builtins.c (check_sizes): Adjust to handle reading past the end.
13339 Avoid printing excessive upper bound of ranges. Use %E to print
13340 tree nodes instead of converting them to %wu.
13341 (expand_builtin_memchr): New function.
13342 (compute_dest_size): Rename...
13343 (compute_objsize): ...to this.
13344 (expand_builtin_memcpy): Adjust.
13345 (expand_builtin_mempcpy): Adjust.
13346 (expand_builtin_strcat): Adjust.
13347 (expand_builtin_strcpy): Adjust.
13348 (check_strncat_sizes): Adjust.
13349 (expand_builtin_strncat): Adjust.
13350 (expand_builtin_strncpy): Adjust and simplify.
13351 (expand_builtin_memset): Adjust.
13352 (expand_builtin_bzero): Adjust.
13353 (expand_builtin_memcmp): Adjust.
13354 (expand_builtin): Handle memcmp.
13355 (maybe_emit_chk_warning): Check strncat just once.
13357 2017-05-04 Martin Sebor <msebor@redhat.com>
13359 PR preprocessor/79214
13360 PR middle-end/79222
13361 PR middle-end/79223
13362 * builtins.c (check_sizes): Add inlining context and issue
13363 warnings even when -Wno-system-headers is set.
13364 (check_strncat_sizes): Same.
13365 (expand_builtin_strncat): Same.
13366 (expand_builtin_memmove): New function.
13367 (expand_builtin_stpncpy): Same.
13368 (expand_builtin): Handle memmove and stpncpy.
13370 2017-05-04 Bin Cheng <bin.cheng@arm.com>
13372 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
13373 which is not used any more.
13375 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13377 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
13379 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13381 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
13382 (cortexa53_tunings): Likewise.
13383 (cortexa57_tunings): Likewise.
13384 (cortexa72_tunings): Likewise.
13385 (cortexa73_tunings): Likewise.
13387 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13389 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
13390 Set loop alignment to 8.
13392 2017-05-04 Martin Sebor <msebor@redhat.com>
13394 PR translation/80280
13395 * builtins.c (expand_builtin_object_size): Add missing quoting to
13396 %D and like directives.
13397 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
13398 (hsa_type_for_tree_type): Same.
13399 (verify_function_arguments): Same.
13400 * symtab.c (symbol_table::change_decl_assembler_name): Same.
13401 * varasm.c (get_section): Same.
13404 2017-05-04 Martin Sebor <msebor@redhat.com>
13406 PR translation/80280
13407 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
13409 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
13411 * config/aarch64/aarch64.c (generic_addrcost_table):
13412 Change HI/TI mode setting.
13414 2017-05-04 Martin Jambor <mjambor@suse.cz>
13416 PR tree-optimization/80622
13417 * tree-sra.c (comes_initialized_p): New function.
13418 (build_accesses_from_assign): Only set write lazily when
13419 comes_initialized_p is false.
13420 (analyze_access_subtree): Use comes_initialized_p.
13421 (propagate_subaccesses_across_link): Assert !comes_initialized_p
13422 instead of testing for PARM_DECL.
13424 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13426 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
13427 constraint on operand 0 to allow more general addressing modes.
13428 Adjust output template.
13429 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
13431 * config/aarch64/aarch64-protos.h
13432 (aarch64_address_valid_for_prefetch_p): Declare prototype.
13433 * config/aarch64/constraints.md (Dp): New address constraint.
13434 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
13437 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
13439 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
13440 update use of estimate_ipcp_clone_size_and_time.
13441 (estimate_local_effects): Update use of
13442 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
13443 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
13444 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
13445 Return nonspecialized time.
13447 2017-05-04 Richard Biener <rguenther@suse.de>
13449 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
13450 for the last VUSE which def dominates the PHI. Directly call
13452 (get_continuation_for_phi_1): Remove.
13454 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
13456 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
13457 to explain the use of truncating division. Cap the number of
13458 iterations to the maximum given by nb_iterations_upper_bound,
13461 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13463 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
13464 * configure: Regenerate.
13465 * config.in: Regenerate.
13466 * config/i386/driver-mingw32.c: new file.
13467 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
13468 * config.host: Link driver-mingw32.o on MinGW host.
13469 * doc/install.texi: Document new --enable-mingw-wildcard configure
13472 2017-05-04 Marek Polacek <polacek@redhat.com>
13474 PR tree-optimization/80612
13475 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
13477 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13478 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
13480 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
13481 (movt splitter): Likewise.
13482 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
13483 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
13484 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
13485 block for Thumb-1 with MOVT.
13486 (thumb2_legitimate_address_p): Move code block ...
13487 (can_avoid_literal_pool_for_label_p): ... into this new function.
13488 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
13490 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
13491 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
13492 "M-profile targets with the MOVT instruction".
13494 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13496 * config/arm/arm-builtins.c (arm_init_builtins): Rename
13497 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
13498 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
13500 2017-05-04 Martin Liska <mliska@suse.cz>
13502 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
13503 variable cond_code.
13505 2017-05-04 Richard Biener <rguenther@suse.de>
13507 * tree.c (array_at_struct_end_p): Handle arrays at struct
13508 end with flexarrays more conservatively. Refactor and treat
13509 arrays of arrays or aggregates more strict. Fix
13510 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
13511 * tree.c (array_at_struct_end_p): Adjust prototype.
13512 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
13513 * gimple-fold.c (get_range_strlen): Likewise.
13514 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
13516 2017-05-04 Richard Biener <rguenther@suse.de>
13518 PR tree-optimization/31130
13519 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
13521 (supports_overflow_infinity): Likewise.
13522 (is_negative_overflow_infinity): Likewise.
13523 (is_positive_overflow_infinity): Likewise.
13524 (is_overflow_infinity): Likewise.
13525 (stmt_overflow_infinity): Likewise.
13526 (overflow_infinity_range_p): Likewise.
13527 (usable_range_p): Remove as always returning true.
13528 (make_overflow_infinity): Remove.
13529 (negative_overflow_infinity): Likewise.
13530 (positive_overflow_infinity): Likewise.
13531 (avoid_overflow_infinity): Likewise.
13532 (set_value_range): Adjust accordingly.
13533 (set_value_range_to_nonnegative): Likewise, remove now unused
13534 overflow_infinity arg.
13535 (vrp_operand_equal_p): Adjust.
13536 (update_value_range): Likewise.
13537 (range_int_cst_singleton_p): Likewise.
13538 (operand_less_p): Likewise.
13539 (compare_values_warnv): Likewise.
13540 (extract_range_for_var_from_comparison_expr): Likewise.
13541 (vrp_int_const_binop): Likewise.
13542 (zero_nonzero_bits_from_vr): Likewise.
13543 (extract_range_from_multiplicative_op_1): Likewise.
13544 (extract_range_from_binary_expr_1): Likewise.
13545 (extract_range_from_unary_expr): Likewise.
13546 (extract_range_from_comparison): Likewise.
13547 (extract_range_basic): Likewise.
13548 (adjust_range_with_scev): Likewise.
13549 (compare_ranges): Likewise.
13550 (compare_range_with_value): Likewise.
13551 (dump_value_range): Likewise.
13552 (test_for_singularity): Likewise, remove strict_overflow_p parameter
13554 (simplify_cond_using_ranges): Adjust.
13556 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13558 * brig-builtins.def: Added a builtin for class_f64.
13559 * builtin-types.def: Added a builtin type needed by class_f64.
13561 2017-05-03 Jason Merrill <jason@redhat.com>
13563 * timevar.def: Add TV_CONSTEXPR.
13565 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13567 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
13569 2017-05-03 Martin Jambor <mjambor@suse.cz>
13571 * ipa-prop.c (ipa_update_after_lto_read): Removed.
13572 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
13573 * ipa-cp.c (ipcp_propagate_stage): Do not call
13574 ipa_update_after_lto_read.
13575 * ipa-inline.c (ipa_inline): Likewise.
13577 2017-05-03 Martin Jambor <mjambor@suse.cz>
13579 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
13580 tag. Added a default constructor and a destructor.
13581 (ipa_edge_args_sum_t): New class;
13582 (ipa_edge_args_sum): Declare.
13583 (ipa_edge_args_vector): Remove declaration.
13584 (IPA_EDGE_REF): Use ipa_edge_args_sum.
13585 (ipa_free_edge_args_substructures): Remove declaration.
13586 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
13587 (ipa_edge_args_info_available_for_edge_p): Likewise.
13588 * ipa-prop.c (ipa_edge_args_vector): Removed.
13589 (edge_removal_hook_holder): Likewise.
13590 (edge_duplication_hook_holder): Likewise.
13591 (ipa_edge_args_sum): New variable.
13592 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
13593 ipa_edge_args_vector.
13594 (ipa_free_edge_args_substructures): Likewise.
13595 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
13596 ipa_edge_args_vector.
13597 (ipa_edge_removal_hook): Turned into method
13598 ipa_edge_args_sum_t::remove.
13599 (ipa_edge_duplication_hook): Turned into method
13600 ipa_edge_args_sum_t::duplicate.
13601 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
13602 registering edge hooks.
13603 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
13604 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
13605 ipa_edge_args_sum instead of ipa_edge_args_vector.
13606 * ipa-profile.c (ipa_profile): Likewise.
13608 2017-05-03 Martin Jambor <mjambor@suse.cz>
13610 * symbol-summary.h (function_summary): New method exists.
13611 (function_summary::symtab_removal): Deallocate through release.
13612 (call_summary): New class.
13613 (gt_ggc_mx): New overload.
13614 (gt_pch_nx): Likewise.
13615 (gt_pch_nx): Likewise.
13617 2017-05-03 Jeff Law <law@redhat.com>
13619 PR tree-optimization/78496
13620 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
13621 from simplify_cond_using_ranges. Split off code to walk
13622 backwards through casts into ...
13623 (simplify_cond_using_ranges_2): New function.
13624 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
13625 (execute_vrp): After identifying jump threads, call
13626 simplify_cond_using_ranges_2.
13628 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
13631 * ipa-inline.h (inline_summary): Add ctor.
13632 (create_ggc): Do not use ggc_cleared_alloc.
13634 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
13635 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13637 * gcc.c (handle_braces): Support escaping in switch matching
13639 * doc/invoke.texi (Spec Files): Document it.
13640 Remove superfluous @code markup in items.
13642 2017-05-03 David Malcolm <dmalcolm@redhat.com>
13644 * diagnostic-show-locus.c (struct column_range): New struct.
13645 (get_affected_columns): New function.
13646 (get_printed_columns): New function.
13647 (struct correction): New struct.
13648 (correction::ensure_capacity): New function.
13649 (correction::ensure_terminated): New function.
13650 (struct line_corrections): New struct.
13651 (line_corrections::~line_corrections): New dtor.
13652 (line_corrections::add_hint): New function.
13653 (layout::print_trailing_fixits): Reimplement in terms of the new
13655 (selftest::test_overlapped_fixit_printing): New function.
13656 (selftest::diagnostic_show_locus_c_tests): Call it.
13658 2017-05-03 Nathan Sidwell <nathan@acm.org>
13660 Canonicalize canonical type hashing
13661 * tree.h (type_hash_canon_hash): Declare.
13662 * tree.c (type_hash_list, attribute_hash_list): Move into
13663 type_hash_canon_hash.
13664 (build_type_attribute_qual_variant): Break out hash code calc into
13665 type_hash_canon_hash.
13666 (type_hash_canon_hash): New. Generic type hash computation.
13667 (build_range_type_1, build_array_type_1, build_function_type,
13668 build_method_type_directly, build_offset_type, build_complex_type,
13669 make_vector_type): Call it.
13671 2017-05-03 Richard Biener <rguenther@suse.de>
13673 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
13674 When all DRs have unknown misaligned do not always peel
13675 when there is a store but apply the same costing model as if
13676 there were only loads.
13678 2017-05-03 Richard Biener <rguenther@suse.de>
13681 PR tree-optimization/80492
13682 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
13683 compare_base_decls returning dont-know properly.
13685 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
13687 * config/arm/iterators.md (CCSI): New mode iterator.
13688 (arch): New mode attribute.
13689 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
13690 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
13691 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
13692 code iterator for success result mode.
13693 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
13694 the corresponding new insn generators.
13696 2017-05-03 Bin Cheng <bin.cheng@arm.com>
13699 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13700 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13702 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
13704 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
13705 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
13706 (DDR_A): Wrap DDR argument in brackets.
13707 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
13708 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
13709 (DDR_REVERSED_P): Likewise.
13711 2017-05-03 Jakub Jelinek <jakub@redhat.com>
13713 PR tree-optimization/79472
13714 * tree-switch-conversion.c (struct switch_conv_info): Add
13715 contiguous_range and default_case_nonstandard fields.
13716 (collect_switch_conv_info): Compute contiguous_range and
13717 default_case_nonstandard fields, don't clear final_bb if
13718 contiguous_range and only the default case doesn't have the required
13720 (check_all_empty_except_final): Set default_case_nonstandard instead
13721 of failing if contiguous_range and the default case doesn't have empty
13723 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
13724 and only the default case doesn't have the required constants. Skip
13726 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
13727 if default_case_nonstandard.
13728 (build_constructors): Build constant 1 just once. Assert that default
13729 values aren't inserted in between cases if contiguous_range. Skip
13731 (build_arrays): Skip virtual phis.
13732 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
13733 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
13734 Handle virtual phis.
13735 (gen_inbound_check): Handle default_case_nonstandard case.
13736 (process_switch): Adjust check_final_bb caller. Call
13737 gather_default_values with the first non-default case instead of
13738 default case if default_case_nonstandard.
13740 2017-05-02 Nathan Sidwell <nathan@acm.org>
13742 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
13743 check. Fix formatting.
13745 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
13747 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
13748 errors when comparing specialized and unspecialized times.
13750 2017-05-02 David Malcolm <dmalcolm@redhat.com>
13752 * diagnostic-show-locus.c
13753 (layout::should_print_annotation_line_p): Make private.
13754 (layout::print_annotation_line): Make private.
13755 (layout::annotation_line_showed_range_p): Make private.
13756 (layout::show_ruler): Make private.
13757 (layout::print_source_line): Make private. Pass in line and
13758 line_width, rather than calling location_get_source_line. Drop
13760 (layout::print_leading_fixits): New method.
13761 (layout::print_any_fixits): Rename to...
13762 (layout::print_trailing_fixits): ...this, and make private.
13763 Don't print newline fixits.
13764 (diagnostic_show_locus): Move logic for printing one row into...
13765 (layout::print_line): ...this new function. Move the
13766 location_get_source_line call and error-handling from
13767 print_source_line to here. Call print_leading_fixits, and rename
13768 print_any_fixits to print_trailing_fixits.
13769 (selftest::test_fixit_insert_containing_newline): Update now that
13770 newlines are partially supported.
13771 (selftest::test_fixit_insert_containing_newline_2): New test.
13772 (selftest::test_fixit_replace_containing_newline): Update comments.
13773 (selftest::diagnostic_show_locus_c_tests): Call the new test.
13774 * edit-context.c (class added_line): New class.
13775 (class edited_line): Describe newline handling in comment.
13776 (edited_line::actually_edited_p): New method.
13777 (edited_line::print_content): Delete redundant decl.
13778 (edited_line::m_predecessors): New field.
13779 (edited_file::print_content): Call edited_line::print_content.
13780 (edited_file::print_diff): Update to support newlines.
13781 (edited_file::print_diff_hunk): Likewise.
13782 (edited_file::print_run_of_changed_lines): New function.
13783 (edited_file::print_diff_line): Convert to...
13784 (print_diff_line): ...this.
13785 (edited_file::get_effective_line_count): New function.
13786 (edited_line::edited_line): Initialize new field m_predecessors.
13787 (edited_line::~edited_line): Clean up m_predecessors.
13788 (edited_line::apply_fixit): Handle newlines.
13789 (edited_line::get_effective_line_count): New function.
13790 (edited_line::print_content): New function.
13791 (edited_line::print_diff_lines): New function.
13792 (selftest::test_applying_fixits_insert_containing_newline): New
13794 (selftest::test_applying_fixits_replace_containing_newline): New
13796 (selftest::insert_line): New function.
13797 (selftest::test_applying_fixits_multiple_lines): Add example of
13799 (selftest::edit_context_c_tests): Call the new tests.
13801 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13803 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
13804 parameter cand. Update dump information.
13805 (get_computation_cost): Update uses.
13807 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13809 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
13810 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
13811 (get_computation_at, rewrite_use_address): Update use of
13812 get_computation_aff.
13814 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13816 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
13817 (get_computation): Delete.
13818 (get_computation_cost): Implement like get_computation_cost_at.
13819 Use get_computation_at.
13820 (get_computation_cost_at): Delete.
13821 (rewrite_use_nonlinear_expr): Use get_computation_at.
13822 (rewrite_use_compare, remove_unused_ivs): Ditto.
13824 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13826 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
13828 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13830 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
13831 (ivopts_global_cost_for_size): Rename parameter and update uses.
13832 (iv_ca_recount_cost): Update uses.
13833 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
13834 candidates seperately in n_invs and n_cands.
13835 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
13837 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13839 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
13840 (find_inv_vars_cb): New.
13841 (find_depends): Renamed to ...
13842 (find_inv_vars): ... this.
13843 (add_candidate_1, force_var_cost): Call find_inv_vars.
13844 (split_address_cost, determine_group_iv_cost_cond): Ditto.
13846 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13848 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
13849 inv_vars. Add inv_exprs.
13850 (struct iv_cand): Rename depends_on to inv_vars.
13851 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
13852 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
13853 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
13854 (dump_cand): Dump inv_vars.
13855 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
13856 (record_invariant, find_depends, add_candidate_1): Ditto.
13857 (set_group_iv_cost, force_var_cost): Ditto.
13858 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
13859 (get_computation_cost_at, get_computation_cost): Ditto.
13860 (determine_group_iv_cost_generic): Ditto.
13861 (determine_group_iv_cost_address): Ditto.
13862 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
13863 (determine_group_iv_costs): Ditto.
13864 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
13865 (iv_ca_set_remove_invariants): Renamed to ...
13866 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
13867 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
13868 (iv_ca_set_add_invariants): Renamed to ...
13869 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
13870 (iv_ca_set_cp): Use iv_ca_set_add_invs.
13871 (iv_ca_has_deps): Support inv_vars and inv_exprs.
13872 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
13873 (create_new_ivs): Remove useless dump.
13875 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13877 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
13879 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
13880 (iv_ca_set_no_cp, create_new_iv): Ditto.
13882 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13884 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13886 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
13888 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
13889 function tree_check2.
13891 2017-05-02 Martin Liska <mliska@suse.cz>
13893 * doc/gcov.texi: Add missing preposition.
13894 * gcov.c (function_info::function_info): Properly fill up
13895 all member variables.
13897 2017-05-02 Tamar Christina <tamar.christina@arm.com>
13899 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
13901 2017-05-02 Tamar Christina <tamar.christina@arm.com>
13903 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
13905 2017-05-02 Martin Liska <mliska@suse.cz>
13908 * lto-streamer-in.c (lto_read_tree_1): Remove
13909 LTO_STREAMER_DEBUG.
13910 * lto-streamer.c (struct tree_hash_entry): Likewise.
13911 (struct tree_entry_hasher): Likewise.
13912 (tree_entry_hasher::hash): Likewise.
13913 (tree_entry_hasher::equal): Likewise.
13914 (lto_streamer_init): Likewise.
13915 (lto_orig_address_map): Likewise.
13916 (lto_orig_address_get): Likewise.
13917 (lto_orig_address_remove): Likewise.
13918 * lto-streamer.h: Likewise.
13919 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
13920 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
13922 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
13924 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
13925 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
13926 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
13927 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
13928 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
13929 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
13930 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
13931 (mm_maskz_sub_ss): New intrinsics.
13932 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13933 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13934 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
13935 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
13936 (__builtin_ia32_subss_mask_round): New builtins.
13937 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13938 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13939 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
13941 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
13942 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13944 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13947 2017-05-02 Martin Jambor <mjambor@suse.cz>
13949 PR tree-optimization/78687
13950 * tree-sra.c (access): New field parent.
13951 (process_subtree_disqualification): New function.
13952 (disqualify_candidate): Call it.
13953 (build_accesses_from_assign): Reset write flag if creating an
13955 (build_access_subtree): Fill in parent field and also prpagate
13956 down grp_write flag.
13957 (create_artificial_child_access): New parameter set_grp_write, set
13958 grp_write to its value.
13959 (propagate_subaccesses_across_link): Also propagate grp_write flag
13961 (propagate_all_subaccesses): Push the closest parent back to work
13962 queue if add_access_to_work_queue returned true.
13964 2017-05-02 Richard Biener <rguenther@suse.de>
13966 * common.opt (fstrict-overflow): Alias negative to fwrapv.
13967 * doc/invoke.texi (fstrict-overflow): Remove all traces of
13968 -fstrict-overflow documentation.
13969 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
13970 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
13971 flag_strict_overflow.
13972 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
13973 * lto-opts.c (lto_write_options): Do not stream it.
13974 * lto-wrapper.c (merge_and_complain): Do not handle it.
13975 * opts.c (default_options_table): Do not set -fstrict-overflow.
13976 (finish_options): Likewise do not clear it when sanitizing.
13977 * simplify-rtx.c (simplify_const_relational_operation): Do not
13978 test flag_strict_overflow.
13980 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
13982 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
13983 using enabled attribute.
13984 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
13985 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
13986 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
13987 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
13988 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
13989 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
13990 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
13991 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
13992 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
13993 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
13995 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
13997 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
13999 2017-05-02 Richard Biener <rguenther@suse.de>
14001 PR tree-optimization/80591
14003 2017-04-10 Richard Biener <rguenther@suse.de>
14005 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14008 2017-05-02 Richard Biener <rguenther@suse.de>
14010 PR tree-optimization/80549
14011 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14012 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14013 headers if they do not have a preheader.
14015 2017-05-02 Martin Liska <mliska@suse.cz>
14018 * common.opt: Fix typo.
14019 * doc/invoke.texi: Likewise.
14021 2017-05-01 Jan Beulich <jbeulich@suse.com>
14023 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14024 swapping, add (x,x,m,x,n) alternative.
14026 2017-05-01 Nathan Sidwell <nathan@acm.org>
14028 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14029 unnecessary unadjusted_alignment check.
14031 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14034 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14036 * gimplify.c (gimplify_cilk_detach): New function.
14037 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14038 * tree-core.h: Document EXPR_CILK_SPAWN.
14039 * tree.h (EXPR_CILK_SPAWN): Define.
14041 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14043 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14044 to use new fixit_hint representation, using the "replace" logic.
14045 (get_line_span_for_fixit_hint): Likewise.
14046 (layout::print_any_fixits): Likewise.
14047 (selftest::test_one_liner_many_fixits): Rename to...
14048 (selftest::test_one_liner_many_fixits_1): ...this, and update
14049 comment and expected output to reflect that the multiple fix-it
14050 hints are now consolidated into one insertion.
14051 (selftest::test_one_liner_many_fixits_2): New test.
14052 (selftest::test_diagnostic_show_locus_one_liner): Update for
14054 (selftest::test_fixit_consolidation): Update for fix-it API
14056 * diagnostic.c (print_parseable_fixits): Likewise.
14057 * edit-context.c (edited_line::m_line_events): Convert from
14058 auto_vec <line_event *> to auto_vec <line_event>.
14059 (class line_event): Convert from abstract base class to a concrete
14060 class, taking over the role of replace_event.
14061 (class insert_event): Delete.
14062 (class replace_event): Rename to class line_event. Convert to
14064 (edit_context::add_fixits): Reimplement.
14065 (edit_context::apply_insert): Delete.
14066 (edit_context::apply_replace): Rename to...
14067 (edit_context::apply_fixit): ...this. Convert to half-open range.
14068 (edited_file::apply_insert): Delete.
14069 (edited_file::apply_replace): Rename to...
14070 (edited_file::apply_fixit): ...this.
14071 (edited_line::~edited_line): Drop deletion of events.
14072 (edited_line::apply_insert): Delete.
14073 (edited_line::apply_replace): Rename to...
14074 (edited_line::apply_fixit): ...this. Convert to half-open range.
14075 Update for change to type of m_line_events.
14076 * edit-context.h (edit_context::apply_insert): Delete.
14077 (edit_context::apply_replace): Rename to...
14078 (edit_context::apply_fixit): ...this.
14080 2017-05-01 Martin Sebor <msebor@redhat.com>
14082 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14085 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14088 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14089 __get_cpuid_max returns 0.
14090 (__get_cpuid_count): Ditto.
14092 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14094 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14095 replacement expression is another instance of one of its arguments.
14097 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14100 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14101 check for stack push/pop autoinc.
14102 * config/i386/i386.c (ix86_agi_dependent): Return false
14103 if the only reason why modified_in_p returned true is that
14104 addr is SP based and set_insn is a push or pop.
14106 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14108 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14111 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14114 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14115 (account_size_time): Use two predicates - exec_pred and
14117 (evaluate_conditions_for_known_args): Compute both clause and
14119 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14120 (inline_summary_t::duplicate): Update.
14121 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14123 (compute_inline_parameters): Likewise.
14124 (estimate_edge_size_and_time): Update caluclation of time.
14125 (estimate_node_size_and_time): Compute both time and nonspecialized
14127 (estimate_ipcp_clone_size_and_time): Update.
14128 (inline_merge_summary): Update.
14129 (do_estimate_edge_time): Update.
14130 (do_estimate_edge_size): Update.
14131 (do_estimate_edge_hints): Update.
14132 (inline_read_section, inline_write_summary): Stream both new predicates.
14133 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14135 (compute_inlined_call_time): Cleanup.
14136 (big_speedup_p): Update.
14137 (edge_badness): Update.
14138 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14139 (size_time_entry): Replace predicate by exec_predicate and
14140 nonconst_predicate.
14141 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14142 (estimate_edge_time): Return also nonspec_time.
14143 (reset_edge_growth_cache): Update.
14145 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14147 PR rtl-optimization/80491
14148 * ifcvt.c (noce_process_if_block): When looking for x setter
14149 with missing else_bb, don't check only the insn right before
14150 cond_earliest, but look for the last insn that x is modified in
14151 within the same bb.
14153 PR rtl-optimization/80491
14154 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14156 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14158 PR tree-optimization/80487
14159 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14161 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14163 PR tree-optimization/79697
14164 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14165 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14166 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14168 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14169 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14171 2017-04-28 Martin Sebor <msebor@redhat.com>
14173 PR tree-optimization/80523
14174 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14175 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14177 (maybe_warn, format_directive, parse_directive): Use new functions.
14178 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14180 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14182 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14184 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14186 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14187 target_header_dir): Set correctly.
14188 * configure: Regenerated.
14189 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14190 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14191 instead of SYSTEM_HEADER_DIR.
14193 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14195 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14196 (estimate_local_effects): Likewise.
14197 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14198 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14199 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14200 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14201 do_estimate_edge_time, estimate_edge_time): Likewise.
14202 * ipa-inline-analysis.c (estimate_node_size_and_time,
14203 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14204 (estimate_time_after_inlining): Remove.
14206 2017-04-28 Martin Liska <mliska@suse.cz>
14208 * doc/gcov.texi: Enhance documentation of gcov.
14210 2017-04-28 Martin Liska <mliska@suse.cz>
14212 * doc/gcov.texi: Sort options in alphabetic order.
14213 * doc/gcov-dump.texi: Likewise.
14214 * doc/gcov-tool.texi: Likewise.
14215 * gcov.c (print_usage): Likewise.
14216 * gcov-dump.c (print_usage): Likewise.
14217 * gcov-tool.c (print_merge_usage_message): Likewise.
14218 (print_rewrite_usage_message): Likewise.
14219 (print_overlap_usage_message): Likewise.
14221 2017-04-28 Martin Liska <mliska@suse.cz>
14223 PR gcov-profile/53915
14224 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14226 2017-04-28 Martin Liska <mliska@suse.cz>
14228 PR gcov-profile/79891
14229 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14230 is marked by compiler as living on a line.
14231 (get_cycles_count): Remove usage of the union.
14232 (output_intermediate_file): Likewise.
14233 (find_source): Fix GNU coding style.
14234 (accumulate_line_counts): Remove old non-all block mode.
14235 (output_lines): Remove usage of the union.
14236 * profile.c (output_location): Include all BBs, even if
14237 belonging to a same line (and file) as a previous BB.
14239 2017-04-28 Martin Liska <mliska@suse.cz>
14241 * gcov.c (process_args): Handle new argument 'w'.
14242 (read_graph_file): Assign ID to BBs.
14243 (output_branch_count): Display BB # if verbose flag is set.
14244 (output_lines): Likewise for arcs.
14245 (print_usage): Add '--verbose' option help.
14246 * doc/gcov.texi: Document --verbose (-w) option.
14248 2017-04-28 Martin Liska <mliska@suse.cz>
14250 * gcov.c (struct block_location_info): New struct.
14251 (process_file): Fill up the new structure.
14252 (read_graph_file): Replace usage of encoding by the newly added
14254 (add_line_counts): Likewise.
14255 (accumulate_line_counts): Remove usage of the union.
14256 (function_info::function_info): New function.
14257 (function_info::~function_info): Likewise.
14258 (process_file): Call delete instead of release_function.
14259 (release_function): Release the function.
14260 (release_structures): Call delete instead of release_function.
14261 (solve_flow_graph): Replace usage of num_blocks.
14262 (find_exception_blocks): Likewise.
14263 (output_lines): Fix GNU coding style.
14265 2017-04-28 Martin Liska <mliska@suse.cz>
14268 * coverage.c (coverage_remove_note_file): New function.
14269 * coverage.h: Declare the function.
14270 * toplev.c (finalize): Clean if an error has been seen.
14272 2017-04-28 Martin Liska <mliska@suse.cz>
14274 PR gcov-profile/80031
14275 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14276 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14277 * gcov.c (read_graph_file): Read just number of blocks.
14278 * profile.c (branch_prob): Do not stream 0 flags per a basic
14281 2017-04-28 Martin Liska <mliska@suse.cz>
14283 * gcov-dump.c (tag_*): Add new argument to declarations.
14284 (dump_gcov_file): Likewise.
14285 (tag_blocks): Add and use new argument depth.
14286 (tag_arcs): Likewise.
14287 (tag_lines): Likewise.
14288 (tag_counters): Likewise.
14289 (tag_summary): Likewise.
14290 (dump_working_sets): Use depth to do a proper indentation.
14292 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14295 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14296 * symtab.c (symtab_node::debug_symtab): Move definition here.
14298 2017-04-28 Richard Biener <rguenther@suse.de>
14300 * lto-streamer.h (LTO_major_version): Bump to 7.
14302 2017-04-28 Richard Biener <rguenther@suse.de>
14304 * tree-vrp.c (assert_info): New struct.
14305 (add_assert_info): New helper.
14306 (register_edge_assert_for_2): Refactor to add asserts to a vector
14308 (register_edge_assert_for_1): Likewise.
14309 (register_edge_assert_for): Likewise.
14310 (finish_register_edge_assert_for): New helper actually registering
14311 asserts where live on edge.
14312 (find_conditional_asserts): Adjust.
14313 (find_switch_asserts): Likewise.
14314 (evrp_dom_walker::try_find_new_range): Generalize.
14315 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14317 2017-04-27 Marek Polacek <polacek@redhat.com>
14320 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14321 arg10 and arg11 to itype.
14323 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14325 * doc/extend.texi (Object Size Checking): Improve grammar.
14327 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14330 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14331 that the logic for permitting reciprocal estimates matches that
14334 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14337 * tree.c (type_cache_hasher::equal): Only compare
14338 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14339 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14340 non-aggregate element types.
14341 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14342 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14344 2017-04-27 Richard Biener <rguenther@suse.de>
14346 PR middle-end/80533
14347 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14348 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14349 keeping a reference to a trailing array.
14351 2017-04-27 Richard Biener <rguenther@suse.de>
14353 PR middle-end/80539
14354 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
14355 being in loop-closed SSA form conservatively.
14356 (chrec_fold_multiply_poly_poly): Likewise.
14358 2017-04-27 Tamar Christina <tamar.christina@arm.com>
14360 PR middle-end/79665
14361 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
14362 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
14364 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14367 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
14368 (aarch64_function_arg_alignment): Return unsigned int again, but still
14369 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
14370 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14371 Don't emit -Wpsabi note.
14372 (aarch64_function_arg_boundary): Likewise.
14373 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14376 2017-04-26 Nathan Sidwell <nathan@acm.org>
14378 * tree.h (crc32_unsigned_n): Declare.
14379 (crc32_unsigned, crc32_unsigned): Make inline.
14380 * tree.c (crc32_unsigned_bits): Replace with ...
14381 (crc32_unsigned_n): ... this.
14382 (crc32_unsigned, crc32_byte): Remove.
14383 (crc32_string): Remove unnecessary braces.
14385 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14387 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
14388 * ipa-inline-analysis.c (MAX_TIME): Remove.
14389 (account_size_time): Use sreal for time.
14390 (dump_inline_summary): Update.
14391 (estimate_function_body_sizes): Update.
14392 (estimate_edge_size_and_time): Update.
14393 (estimate_calls_size_and_time): Update.
14394 (estimate_node_size_and_time): Update.
14395 (inline_merge_summary): Update.
14396 (inline_update_overall_summary): Update.
14397 (estimate_time_after_inlining): Update.
14398 (inline_read_section): Update.
14399 (inline_write_summary): Update.
14400 * ipa-inline.c (compute_uninlined_call_time): Update.
14401 (compute_inlined_call_time): Update.
14402 (recursive_inlining): Update.
14403 (inline_small_functions): Update.
14404 (dump_overall_stats): Update.
14405 * ipa-inline.h: Include sreal.h.
14406 (size_time_entry): Turn time to sreal.
14407 (inline_summary): Turn self_time nad time to sreal.
14409 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
14411 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
14413 (sreal::stream_out, sreal::stream_in): New.
14414 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
14416 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14418 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
14421 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
14424 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
14426 (dimode_scalar_chain::compute_convert_gain): Ditto.
14427 (dimode_scalar_chain::make_vector_copies): Ditto.
14428 (dimode_scalar_chain::convert_reg): Ditto.
14429 (dimode_scalar_chain::convert_insn): Ditto.
14430 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
14431 (VI248_AVX512BW_1): New mode iterator.
14432 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
14433 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
14436 2017-04-25 Martin Sebor <msebor@redhat.com>
14438 PR tree-optimization/80497
14439 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
14440 constants are representable in HOST_WIDE_INT.
14441 (parse_directive): Ditto.
14443 2017-04-25 Martin Sebor <msebor@redhat.com>
14446 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
14447 (new_zero_array): Adjust signature.
14448 (dom_info::dom_init): Used unsigned rather that size_t.
14449 (dom_info::dom_info): Same.
14451 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14452 Jakub Jelinek <jakub@redhat.com>
14455 * config/arm/arm.c: Include gimple.h.
14456 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14457 returns negative, increment ncrn only if it returned positive.
14458 (arm_needs_doubleword_align): Return int instead of bool,
14459 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
14460 members, but if there is any such non-FIELD_DECL
14461 > PARM_BOUNDARY aligned decl, return -1 instead of false.
14462 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
14463 returns negative, increment nregs only if it returned positive.
14464 (arm_setup_incoming_varargs): Likewise.
14465 (arm_function_arg_boundary): Emit -Wpsabi note if
14466 arm_needs_doubleword_align returns negative, return
14467 DOUBLEWORD_ALIGNMENT only if it returned positive.
14469 2017-04-25 Marek Polacek <polacek@redhat.com>
14472 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
14473 first argument to type.
14475 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
14478 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
14479 type checks to test for compatibility instead of equality.
14481 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14482 Jakub Jelinek <jakub@redhat.com>
14485 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
14487 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
14488 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
14489 the alignment computation, but return their maximum in warn_alignment.
14490 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
14491 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
14493 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
14494 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
14497 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14499 * config/arc/simdext.md (dmpyh): Fix typo.
14501 2017-04-25 Richard Biener <rguenther@suse.de>
14503 PR tree-optimization/80492
14504 * alias.c (compare_base_decls): Handle registers with asm
14505 specification conservatively.
14506 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14507 compare_base_decls returning dont-know properly.
14509 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14511 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
14512 (legitimate_offset_address_p): New function.
14513 (arc_legitimate_address_p): Use above function.
14515 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14517 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
14519 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14521 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
14522 ACCH registers whenever they are available.
14524 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14526 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
14527 double regs fix when not used.
14529 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14531 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
14533 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
14534 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
14536 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14538 * config/arc/arc.c (arc_output_addsi): Check for h-register class
14539 when emitting short ADD instructions.
14541 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
14543 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
14545 (cmpsi_cc_c_insn): Likewise.
14546 (cbranchsi4_scratch): Compute proper instruction length using
14547 compact_hreg_operand.
14548 * config/arc/predicates.md (compact_hreg_operand): New predicate.
14550 2017-04-25 Richard Biener <rguenther@suse.de>
14552 PR middle-end/80509
14553 * passes.c (pass_manager::pass_manager): Initialize
14554 m_name_to_pass_map.
14556 2017-04-25 Richard Biener <rguenther@suse.de>
14558 PR tree-optimization/79201
14559 * tree-ssa-sink.c (statement_sink_location): Handle calls.
14561 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14564 * config/s390/vector.md: Split MEM->GPR vector moves for
14565 non-s_operand addresses.
14567 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14570 * config/s390/predicates.md (reload_const_wide_int_operand): New
14572 * config/s390/s390.md ("movti"): Remove d/P alternative.
14573 ("movti_bigconst"): New pattern definition.
14575 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14578 * s390-protos.h (s390_expand_cs_hqi): Removed.
14579 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
14580 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
14581 modes as well as CCZ1mode and CCZmode.
14582 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
14583 signature of s390_emit_compare_and_swap.
14584 (s390_expand_cs_hqi): Likewise, make static.
14585 (s390_expand_cs_tdsi): Generate an explicit compare before trying
14586 compare-and-swap, in some cases.
14587 (s390_expand_cs): Wrapper function.
14588 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
14590 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
14591 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
14592 patterns for small and large integers. Forbid symref memory operands.
14593 Move expander to s390.c. Require cc register.
14594 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
14595 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
14596 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
14597 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
14598 symref memory operands. Remove CC mode and call s390_match_ccmode
14600 ("atomic_exchange<mode>"): Allow and implement all integer modes.
14602 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14604 * config/s390/s390.md (define_peephole2): New peephole to help
14605 combining the load-and-test pattern with volatile memory.
14607 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
14609 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
14610 with CCZmode for TARGET_Z196.
14612 2017-04-25 Jakub Jelinek <jakub@redhat.com>
14614 PR rtl-optimization/80501
14615 * combine.c (make_compound_operation_int): Set subreg_code to SET
14616 even for AND with mask of the sign bit of mode.
14618 PR rtl-optimization/80500
14619 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
14620 sum's initial value.
14622 2017-04-25 Julian Brown <julian@codesourcery.com>
14623 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14625 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
14627 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
14629 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
14631 2017-04-25 Julian Brown <julian@codesourcery.com>
14632 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14634 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
14635 (thunderx2t99_sha): New Reservation.
14637 2017-04-25 Julian Brown <julian@codesourcery.com>
14638 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14640 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
14641 type for 1-element load.
14643 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
14645 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
14647 2017-04-24 Martin Jambor <mjambor@suse.cz>
14649 PR tree-optimization/80293
14650 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
14651 char arrays not totally scalarizable if it is false.
14652 (analyze_all_variable_accesses): Pass correct value in the new
14653 parameter. Add a statistics counter.
14655 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
14657 PR middle-end/79931
14658 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
14660 2017-04-24 Richard Biener <rguenther@suse.de>
14662 PR tree-optimization/80494
14663 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
14664 out for complex types.
14666 2017-04-24 Richard Biener <rguenther@suse.de>
14668 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
14669 * tree-ssa-sccvn.c (print_scc): Print SCC size.
14670 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
14671 (DFS): Adjust and never fail.
14672 (sccvn_dom_walker::fail): Remove.
14673 (sccvn_dom_walker::before_dom_children): Adjust.
14674 (run_scc_vn): Likewise and never fail.
14675 * tree-ssa-pre.c (pass_pre::execute): Adjust.
14676 (pass_fre::execute): Likewise.
14678 2017-04-24 Richard Biener <rguenther@suse.de>
14680 PR tree-optimization/79725
14681 * tree-ssa-sink.c (statement_sink_location): Return whether
14682 failure reason was zero uses. Move that check later.
14683 (sink_code_in_bb): Deal with zero uses by removing the stmt
14686 2017-04-24 Richard Biener <rguenther@suse.de>
14689 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
14690 pointer-based references.
14692 2017-04-24 Richard Biener <rguenther@suse.de>
14695 * pass_manager.h (pass_manager::operator new): Remove.
14696 (pass_manager::operator delete): Likewise.
14697 * passes.c (pass_manager::operator new): Remove.
14698 (pass_manager::operator delete): Likewise.
14699 (pass_manager::pass_manager): Zero individual pass members.
14701 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
14704 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
14705 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
14706 Check "XEXP (src, 1)" operand here.
14707 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
14708 Check "XEXP (src, 1)" operand here.
14709 (dimode_scalar_chain::make_vector_copies): Detect count register
14710 of a shift instruction. Zero extend count register from QImode
14711 to DImode to satisfy vector shift pattern count operand predicate.
14712 Substitute vector shift count operand with a DImode copy.
14713 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
14716 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
14718 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
14719 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
14720 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
14721 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
14722 (UNSPEC_NOREX_MEM): Remove definition.
14724 2017-04-21 Richard Biener <rguenther@suse.de>
14726 PR tree-optimization/79547
14727 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14728 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
14729 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
14730 without any constraints.
14732 2017-04-21 Richard Biener <rguenther@suse.de>
14734 PR tree-optimization/78847
14735 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
14737 2017-04-21 Richard Biener <rguenther@suse.de>
14739 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
14740 (build_distinct_type_copy): Likewise.
14741 (build_variant_type_copy): Likewise.
14742 * tree.c (build_qualified_type): Pass down mem-stat info.
14743 (build_distinct_type_copy): Likewise.
14744 (build_variant_type_copy): Likewise.
14746 2017-04-21 Richard Biener <rguenther@suse.de>
14748 PR tree-optimization/80237
14749 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
14751 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
14752 for a simplified result.
14754 2016-04-21 Richard Biener <rguenther@suse.de>
14756 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
14757 sth as strict as a simple_iv but a chrec without symbols and an
14758 operand defined in the loop we are peeling (and not some subloop).
14759 (propagate_constants_for_unrolling): Propagate all constants.
14761 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
14764 * config/i386/i386.c (print_reg): Remove assert for disalowed
14765 regno values, call output_operand_lossage instead.
14767 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
14770 * config/i386/constraints.md (Yc): New register constraint.
14771 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
14772 Use Yc constraint for alternative 2 of operand 0. Remove
14773 preferred_for_speed attribute.
14775 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
14777 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
14778 lastprivate clauses in SIMT case.
14780 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
14782 * doc/invoke.texi (-Wextra-semi): Document new warning option.
14784 2017-04-20 Richard Biener <rguenther@suse.de>
14786 PR tree-optimization/57796
14787 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
14788 as N scalar stores.
14789 (vect_model_load_cost): Cost gathers as N scalar loads.
14791 2017-04-20 Richard Biener <rguenther@suse.de>
14793 * ggc-page.c (ggc_allocated_p): Rename to ...
14794 (safe_lookup_page_table_entry): ... this and return the lookup
14796 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
14798 2017-04-20 Richard Biener <rguenther@suse.de>
14800 PR tree-optimization/80453
14801 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
14802 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
14803 from the conditions.
14804 (vn_phi_eq): Pass them down.
14805 (vn_phi_lookup): Record them.
14806 (vn_phi_insert): Likewise.
14808 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
14810 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
14811 uninitialized variable warning to avoid buffer overrun.
14813 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
14816 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
14817 is suppressed for '{ 0 }' in C.
14819 2017-04-20 Jakub Jelinek <jakub@redhat.com>
14821 * BASE-VER: Set to 8.0.0.
14823 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
14825 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
14826 priority .init_array and .fini_array section with SECTION_NOTYPE
14829 2017-04-20 Jakub Jelinek <jakub@redhat.com>
14831 PR middle-end/80423
14832 * tree.h (build_array_type): Add typeless_storage default argument.
14833 * tree.c (type_cache_hasher::equal): Also compare
14834 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
14835 (build_array_type): Add typeless_storage argument, set
14836 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
14838 (build_nonshared_array_type): Adjust build_array_type_1 caller.
14839 (build_array_type): Likewise. Add typeless_storage argument.
14841 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
14842 Jakub Jelinek <jakub@redhat.com>
14844 PR tree-optimization/80426
14845 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
14846 operation on symbolic operands, also compute the overflow for the
14847 invariant part when the operation degenerates into a negation.
14849 2017-04-19 Jakub Jelinek <jakub@redhat.com>
14852 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
14853 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
14856 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
14858 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
14861 * config/avr/avr.c (tree.h): Include it.
14862 (cgraph.h): Include it.
14863 (avr_encode_section_info): Don't warn for uninitialized progmem
14864 variable if it's just an alias.
14866 2017-04-19 Richard Biener <rguenther@suse.de>
14869 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
14870 when needed by AutoPGO.
14872 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
14875 * doc/lto.texi: Remove an extra 'that'.
14877 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
14879 PR rtl-optimization/80429
14880 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
14881 are only used in debug insns.
14883 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
14884 Vladimir Makarov <vmakarov@redhat.com>
14886 * config/sparc/predicates.md (input_operand): Add comment. Return
14887 true for any memory operand when LRA is in progress.
14888 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
14890 2017-04-18 Jeff Law <law@redhat.com>
14893 * mips.md ({return,simple_return}_internal): Do not overwrite
14896 2017-04-18 Jakub Jelinek <jakub@redhat.com>
14898 PR tree-optimization/80443
14899 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
14900 instead of adding 1, subtract -1 and similarly instead of subtracting
14903 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
14905 PR rtl-optimization/80357
14906 * haifa-sched.c (tmp_bitmap): New variable.
14907 (model_recompute): Handle duplicate use records.
14908 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
14909 (free_global_sched_pressure_data): Free it.
14911 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
14914 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
14915 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14916 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14917 instead of SYSTEM_HEADER_DIR.
14919 2017-04-18 Jeff Law <law@redhat.com>
14921 PR middle-end/80422
14922 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
14923 predecessors after walking up the insn chain.
14925 2017-04-18 Jakub Jelinek <jakub@redhat.com>
14928 * dwarf2out.c (modified_type_die): Try harder not to emit internal
14929 sizetype type into debug info.
14931 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14934 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
14935 unneeded test for TARGET_UPPER_REGS_SF.
14936 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
14938 2017-04-18 Jakub Jelinek <jakub@redhat.com>
14941 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
14942 instead of gsi_after_labels.
14944 2017-04-18 Jeff Law <law@redhat.com>
14946 * regcprop.c (maybe_mode_change): Avoid creating copies of the
14950 2017-04-13 Jeff Law <law@redhat.com>
14951 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
14952 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
14954 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
14957 * config/avr/avr.c (intl.h): Include it.
14958 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
14960 2017-04-18 Martin Liska <mliska@suse.cz>
14962 PR gcov-profile/78783
14963 * gcov-tool.c (gcov_output_files): Validate that destination
14964 file is either removed by the tool or by a user.
14966 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
14967 Guy Benyei <guybe@mellanox.com>
14969 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
14970 block, and do not negate it, the stored id is already negative.
14972 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
14974 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
14976 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14979 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
14980 masks of options that should be turned off if the VSX vector
14981 options are turned off.
14982 (OTHER_P8_VECTOR_MASKS): Likewise.
14983 (OTHER_VSX_VECTOR_MASKS): Likewise.
14984 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
14985 rs6000_disable_incompatible_switches to validate no type switches
14987 (rs6000_incompatible_switch): New function to disallow turning on
14988 other vector options if -mno-vsx, -mno-power8-vector, or
14989 -mno-power9-vector are specified.
14991 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
14993 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
14995 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
14997 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
14998 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
14999 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15000 (ARG_POINTER_CFA_OFFSET): Likewise.
15002 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15004 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15005 conditions to take advantage of various optimizations.
15007 2017-04-13 Jeff Law <law@redhat.com>
15009 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15010 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15011 (zero_extendsidi2_dext): Likewise.
15013 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15016 * fold-const.c (fold_ternary_loc): Revert
15017 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15020 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15022 PR rtl-optimization/80343
15023 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15024 new scratch pseudo.
15026 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15029 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15030 to ubsan_encode_value.
15032 2017-04-13 Jeff Law <law@redhat.com>
15034 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15035 appearing in DEBUG_INSNs.
15037 2017-04-13 Martin Liska <mliska@suse.cz>
15039 PR gcov-profile/80413
15040 * gcov-io.c (gcov_write_string): Copy to buffer just when
15041 allocated size is greater than zero.
15043 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15046 * dwarf2out.c (decls_for_scope): Ignore declarations of
15047 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15049 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15052 * ipa-visibility.c (non_local_p): Fix typos.
15053 (localize_node): When localizing symbol in same comdat group,
15054 dissolve the group only when we know external symbols are going
15056 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15058 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15060 PR tree-optimization/79390
15061 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15062 order does not result in usable sequence, retry with reversed operand
15068 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15069 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15070 op0 instead of fold_convert_loc (loc, type, arg0).
15072 2017-04-12 Jeff Law <law@redhat.com>
15074 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15075 has a delay slot in the generated code.
15077 * config/cris/cris.md (cris_preferred_reload_class): Return
15078 GENNONACR_REGS rather than GENERAL_REGS.
15080 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15083 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15084 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15085 signedness of the result type.
15087 2017-04-12 Richard Biener <rguenther@suse.de>
15088 Jeff Law <law@redhat.com>
15090 PR tree-optimization/80359
15091 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15092 trim stores to TARGET_MEM_REFs.
15094 2017-04-12 Richard Biener <rguenther@suse.de>
15096 PR tree-optimization/79390
15097 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15098 threading case even more.
15100 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15103 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15104 for quad_address_p for TImode, instead of just not indexed_address.
15106 2017-04-12 Richard Biener <rguenther@suse.de>
15107 Bernd Edlinger <bernd.edlinger@hotmail.de>
15109 PR middle-end/79671
15110 * alias.c (component_uses_parent_alias_set_from): Handle
15111 TYPE_TYPELESS_STORAGE.
15112 (get_alias_set): Likewise.
15113 * tree-core.h (tree_type_common): Add typeless_storage flag.
15114 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15115 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15116 for types containing members with TYPE_TYPELESS_STORAGE.
15117 (place_field): Likewise.
15118 (layout_type): Likewise for ARRAY_TYPE.
15119 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15120 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15121 TYPE_TYPELESS_STORAGE.
15122 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15124 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15127 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15128 first argument to type.
15130 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15134 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15135 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15136 (rs6000_expand_binop_builtin): Likewise.
15137 (rs6000_expand_ternop_builtin): Likewise; also add missing
15138 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15139 vshasigma built-ins.
15140 * doc/extend.texi: Document that vec_xxpermdi's third argument
15141 must be a constant.
15143 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15145 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15146 Use shift_const cost parameter when calculating gain of STV shifts.
15148 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15150 PR rtl-optimization/70478
15151 * lra-constraints.c (process_alt_operands): Check memory for
15152 disfavoring memory insn operand.
15154 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15156 PR middle-end/80100
15157 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15158 left shift in unsigned HOST_WIDE_INT type.
15160 PR rtl-optimization/80385
15161 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15162 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15165 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15166 if they have any depend clauses.
15168 2017-04-11 Martin Liska <mliska@suse.cz>
15171 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15172 * ipa-split.c (split_function): Create a local comdat symbol
15173 if caller is in a comdat group.
15175 2017-04-11 Martin Liska <mliska@suse.cz>
15178 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15181 2017-04-11 Martin Sebor <msebor@redhat.com>
15183 PR middle-end/80364
15184 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15185 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15187 (directive::set_width, directive::set_precision, format_character):
15189 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15192 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15195 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15196 conflict, set target->arch_name instead of target->cpu_name.
15198 2017-04-11 Richard Biener <rguenther@suse.de>
15200 PR tree-optimization/80374
15201 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15202 build_zero_cst, remove fold_convertible_p check again.
15204 2017-04-11 Martin Liska <mliska@suse.cz>
15207 * ubsan.c (instrument_object_size): Do not instrument register
15210 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15213 * config/i386/i386-builtin-types.def
15214 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15215 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15216 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15217 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15218 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15219 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15220 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15221 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15222 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15223 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15224 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15225 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15226 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15227 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15228 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15229 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15230 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15231 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15232 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15233 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15234 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15235 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15236 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15237 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15238 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15239 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15240 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15241 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15242 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15243 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15244 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15245 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15246 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15247 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15248 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15249 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15250 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15251 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15252 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15253 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15254 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15255 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15256 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15257 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15258 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15259 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15261 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15262 flag to second_arg_count, handle 4 argument function type _COUNT
15263 aliases, handle second_arg_count on second argument rather than last.
15265 2017-04-10 Jeff Law <law@redhat.com>
15267 PR tree-optimization/80374
15268 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15269 record anything if we can not convert integer_zero_node to the
15272 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15275 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15276 Enhance special handling given to the TARGET_P9_MINMAX option in
15277 relation to certain other options.
15279 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15281 PR tree-optimization/80153
15282 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15283 remove POINTER_PLUS_EXPR's base part directly, rather than through
15286 2017-04-10 Richard Biener <rguenther@suse.de>
15287 Bin Cheng <bin.cheng@arm.com>
15289 PR tree-optimization/80153
15290 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15291 the first element of pointer type aff_tree. Build result expr in
15293 (add_elt_to_tree): Convert to type unconditionally. Remove other
15294 fold_convert calls.
15295 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15296 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15298 2017-04-10 Richard Biener <rguenther@suse.de>
15300 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15303 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15305 PR rtl-optimization/70478
15306 * lra-constraints.c (curr_small_class_check): New.
15307 (update_and_check_small_class_inputs): New.
15308 (process_alt_operands): Update curr_small_class_check. Disfavor
15309 alternative insn memory operands. Check available regs for small
15312 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15315 * config/mips/mips.opt (-mvirt): Update description.
15316 * doc/invoke.texi (-mvirt): Likewise.
15318 2017-04-10 Richard Biener <rguenther@suse.de>
15320 PR middle-end/80362
15321 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15322 looking for NEGATE_EXPR in -A / -B to A / B folding.
15324 2017-04-10 Martin Liska <mliska@suse.cz>
15326 PR gcov-profile/80224
15327 * gcov.c (print_usage): Fix usage string.
15328 (get_gcov_intermediate_filename): Remove.
15329 (output_gcov_file): Use both for normal and intermediate format.
15330 (generate_results): Do not initialize special file for
15331 intermediate format.
15333 2017-04-10 Richard Biener <rguenther@suse.de>
15335 PR tree-optimization/80304
15336 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15339 2017-04-10 Nathan Sidwell <nathan@acm.org>
15342 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15343 (rs6000_init_builtins): Use it.
15345 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15347 * config/arm/arm.md (<mrc>): Add mode to SET source.
15348 (<mrrc>): Likewise.
15350 2017-04-10 Richard Biener <rguenther@suse.de>
15352 PR middle-end/80344
15353 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
15355 2017-04-10 Jakub Jelinek <jakub@redhat.com>
15358 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
15359 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
15360 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
15361 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
15362 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
15363 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
15364 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
15365 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
15366 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
15367 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
15368 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
15369 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
15370 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
15371 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
15372 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
15373 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
15374 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
15375 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
15376 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
15377 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
15378 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
15379 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
15380 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
15382 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
15384 PR rtl-optimization/70478
15385 * lra-constraints.c: Reverse the last patch.
15387 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
15389 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
15390 Add comment for WCHAR_T.
15392 2017-04-08 Martin Liska <mliska@suse.cz>
15395 2017-04-07 Martin Liska <mliska@suse.cz>
15398 * ipa-split.c (split_function): Add function part to a same comdat
15401 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15404 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
15406 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
15408 * rs6000/rs6000.c (vec_load_pendulum): Rename...
15409 (vec_pairing): ...to this.
15410 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
15411 (rs6000_sched_init): Adjust for name change.
15412 (struct rs6000_sched_context): Likewise.
15413 (rs6000_init_sched_context): Likewise.
15414 (rs6000_set_sched_context): Likewise.
15416 2017-04-07 Jakub Jelinek <jakub@redhat.com>
15422 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
15424 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
15425 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
15426 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
15428 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
15430 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
15432 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15434 PR rtl-optimization/70703
15435 * ira-color.c (update_conflict_hard_regno_costs): Use
15436 int64_t instead of HOST_WIDE_INT.
15438 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
15440 PR rtl-optimization/70478
15441 * lra-constraints.c (process_alt_operands): Disfavor alternative
15442 insn memory operands.
15444 2017-04-07 Jeff Law <law@redhat.com>
15446 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
15447 CALL and NOTE_INSN_CALL_ARG_LOCATION.
15449 2017-04-07 Martin Liska <mliska@suse.cz>
15452 * config/aarch64/aarch64.c (aarch64_process_target_attr):
15453 Show error message instead of an ICE.
15455 2017-04-07 Martin Liska <mliska@suse.cz>
15458 * ipa-split.c (split_function): Add function part to a same comdat
15461 2017-04-07 Richard Biener <rguenther@suse.de>
15463 PR middle-end/80341
15464 * tree.c (get_unwidened): Also handle ! for_type case for
15466 * convert.c (do_narrow): Split out from ...
15467 (convert_to_integer_1): ... here. Do not pass final truncation
15468 type to get_unwidened for TRUNC_DIV_EXPR.
15470 2017-04-07 Richard Biener <rguenther@suse.de>
15472 * tree-affine.c (wide_int_ext_for_comb): Take type rather
15474 (aff_combination_const): Adjust.
15475 (aff_combination_scale): Likewise.
15476 (aff_combination_add_elt): Likewise.
15477 (aff_combination_add_cst): Likewise.
15478 (aff_combination_convert): Likewise.
15479 (add_elt_to_tree): Likewise. Remove unused argument.
15480 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
15482 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
15484 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
15486 * config/arm/arm.c (arm_default_short_enums): Use
15487 ARM_DEFAULT_SHORT_ENUMS.
15488 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
15490 2017-04-06 Jakub Jelinek <jakub@redhat.com>
15493 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
15494 members with redundant out-of-class redeclaration.
15496 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15499 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
15500 * config/i386/i386.md (*zero_extendsidi2):
15501 Add (?*x,*x) and (?*v,*v) alternatives.
15503 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
15506 * config/i386/i386.c (ix86_expand_builtin)
15507 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
15508 mode from insn data. Convert operands to insn operand mode.
15509 Copy operands that don't satisfy insn predicate to a register.
15511 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
15513 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
15516 2017-04-06 Richard Biener <rguenther@suse.de>
15518 PR tree-optimization/80334
15519 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
15520 preserve alignment of accesses.
15522 2017-04-06 Richard Biener <rguenther@suse.de>
15524 PR tree-optimization/80262
15525 * tree-sra.c (build_ref_for_offset): Preserve address-space
15527 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
15528 Drop useless address-space information on MEM_REF offsets.
15530 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
15532 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
15534 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15536 PR rtl-optimization/70703
15537 * ira-color.c (update_conflict_hard_regno_costs): Use
15538 HOST_WIDE_INT instead of long.
15540 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
15543 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
15544 not defined for x86_64 target. Add -mmmx target option when __SSE2__
15546 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
15547 for x86_64 target. Handle -m3dnowa option.
15549 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
15551 PR rtl-optimization/70703
15552 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
15553 (update_conflict_hard_regno_costs): Use long instead of unsigned
15554 arithmetic for cost calculation.
15556 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15557 Bernd Edlinger <bernd.edlinger@hotmail.de>
15560 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
15563 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
15566 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
15567 ptr_mode with Pmode throughout.
15568 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
15569 into probe_stack_range and use DImode.
15571 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
15574 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
15575 call_eh_return is true.
15577 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15579 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
15580 Initialize last_match_fntype_index.
15582 2017-04-05 Jakub Jelinek <jakub@redhat.com>
15585 * tree-nvr.c: Include internal-fn.h.
15586 (pass_return_slot::execute): Ignore internal calls without
15589 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15590 Richard Biener <rguenther@suse.de>
15593 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
15594 captures used multiple times, except for the last use.
15595 * generic-match-head.c: Include gimplify.h.
15597 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15599 PR tree-optimization/79390
15600 * target.h (struct noce_if_info): Declare.
15601 * targhooks.h (default_noce_conversion_profitable_p): Declare.
15602 * target.def (noce_conversion_profitable_p): New target hook.
15603 * ifcvt.h (struct noce_if_info): New type, moved from ...
15604 * ifcvt.c (struct noce_if_info): ... here.
15605 (noce_conversion_profitable_p): Renamed to ...
15606 (default_noce_conversion_profitable_p): ... this. No longer
15608 (noce_try_store_flag_constants, noce_try_addcc,
15609 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
15610 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
15611 instead of noce_conversion_profitable_p.
15612 * config/i386/i386.c: Include ifcvt.h.
15613 (ix86_option_override_internal): Don't override
15614 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
15615 (ix86_noce_conversion_profitable_p): New function.
15616 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
15617 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
15618 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
15619 * doc/tm.texi: Regenerated.
15621 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15623 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
15626 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15629 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
15630 instructions for small multiply cores.
15632 2017-04-04 Jeff Law <law@redhat.com>
15634 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
15636 (mips_expand_vec_perm_const): Initialize elements in orig_perm
15637 that are not set by the loop over the elements.
15639 2017-04-04 Jakub Jelinek <jakub@redhat.com>
15642 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
15643 int mode, convert_modes it to mode as unsigned, otherwise use
15644 lowpart_subreg to mode rather than SImode.
15645 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
15646 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
15647 Use DImode instead of SImode for the shift count operand.
15648 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
15651 2017-04-04 Richard Biener <rguenther@suse.de>
15653 PR middle-end/80281
15654 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
15655 arithmetic done for the negate or the plus. Simplify.
15656 (A - (-B) -> A + B): Likewise.
15657 * fold-const.c (split_tree): Make sure to not negate pointers.
15659 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
15661 PR rtl-optimization/60818
15662 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
15663 a compare of comparisons with the thing compared if this results
15664 in a different machine mode.
15666 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
15668 * alias.c (base_alias_check): Fix typo in comment.
15669 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
15670 * cgraphunit.c (symbol_table::compile): Likewise.
15671 * collect2.c (maybe_run_lto_and_relink): Likewise.
15672 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
15673 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
15674 * config/avr/avr.c (avr_map_op_t): Likewise.
15675 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
15676 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
15677 * config/epiphany/epiphany.md (movcc): Likewise.
15678 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
15679 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
15681 * config/mips/mips.c (mips_save_restore_reg): Likewise.
15682 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
15683 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
15684 * config/sh/sh.c (sh_rtx_costs): Likewise.
15685 * fold-const.c (fold_truth_andor): Likewise.
15686 * genautomata.c (collapse_flag): Likewise.
15687 * gengtype.h (struct type::u::s): Likewise.
15688 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
15689 * input.c (FORMAT_AMOUNT): Likewise.
15690 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
15691 (known_aggs_to_agg_replacement_list): Likewise.
15692 * ipa-inline-analysis.c: Likewise.
15693 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
15694 * ipa-polymorphic-call.c
15695 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
15696 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
15697 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
15699 * modulo-sched.c (apply_reg_moves): Likewise.
15700 * omp-expand.c (build_omp_regions_1): Likewise.
15701 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
15702 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
15703 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
15704 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
15705 * value-prof.c: Likewise.
15706 * var-tracking.c (val_reset): Likewise.
15708 2017-04-03 Richard Biener <rguenther@suse.de>
15710 PR tree-optimization/80275
15711 * fold-const.c (split_address_to_core_and_offset): Handle
15714 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
15716 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
15717 descriptors is at least equal to that of functions.
15719 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
15721 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
15723 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
15726 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
15727 (mov<IMOD4:mode>): New expander.
15728 (*mov<IMOD4:mode>_internal): New insn and split pattern.
15730 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15732 PR rtl-optimization/79405
15733 * fwprop.c (propagations_left): New variable.
15734 (forward_propagate_into): Decrement it.
15735 (fwprop_init): Initialize it.
15736 (fw_prop): If the variable has reached zero, stop propagating.
15737 (fwprop_addr): Ditto.
15739 2017-03-31 Jakub Jelinek <jakub@redhat.com>
15742 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
15743 a FUNCTION_DECL, pass it as decl instead of origin to
15746 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
15748 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
15751 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
15754 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
15755 TARGET_VSX_SMALL_INTEGER.
15757 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15759 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15760 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
15762 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15764 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
15765 extraction from odd-numbered MSA register.
15767 2017-03-31 Jakub Jelinek <jakub@redhat.com>
15769 PR middle-end/80173
15770 * expmed.c (store_bit_field_1): Don't attempt to create
15771 a word subreg out of hard registers wider than word if they
15772 have HARD_REGNO_NREGS of 1 for their mode.
15774 PR middle-end/80163
15775 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
15776 conversions to integer types wider than word and pointer.
15779 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
15780 (rtx_equal_for_cselib_p): Pass 0 to it.
15781 * cselib.c (cselib_hasher::equal): Likewise.
15782 (rtx_equal_for_cselib_1): Add depth argument. If depth
15783 is 128, don't look up VALUE locs and punt. Increment
15784 depth in recursive calls when walking VALUE locs.
15786 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
15788 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
15789 (make_gcov_file_name): Use the canonical path name for generating
15791 (read_line): Fix handling of files with ascii null bytes.
15793 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
15795 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
15796 to initialise a vector register instead
15797 of using a const_int.
15799 2017-03-30 Jakub Jelinek <jakub@redhat.com>
15801 PR translation/80189
15802 * gimplify.c (omp_default_clause): Use %qs instead of %s in
15803 diagnostic messages.
15805 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
15808 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
15809 (dfp_diex_<mode>): Update mode of operand 1.
15810 * doc/extend.texi (dxex, dxexq): Document change to return type.
15811 (diex, diexq): Document change to argument type.
15813 2017-03-30 Martin Jambor <mjambor@suse.cz>
15816 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
15817 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
15818 it reflects the signature changes performed at the callee side.
15819 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
15820 to cgraph_build_function_type_skip_args.
15821 (build_function_decl_skip_args): Adjust call to the above function.
15823 2017-03-30 Jakub Jelinek <jakub@redhat.com>
15826 * config/i386/sse.md
15827 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
15828 register as dest whenever it is a MEM not rtx_equal_p to the
15829 corresponding dup operand, and when forcing into reg move the
15830 reg into the memory afterwards.
15831 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
15832 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
15833 for the force_reg mode.
15834 (avx512vl_vextractf128<mode>): Use register as dest either
15835 always when a MEM, or when it is a MEM not rtx_equal_p to the
15836 corresponding dup operand, or even not when it is a CONST_VECTOR
15837 depending on the mode and lo vs. hi.
15838 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
15840 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
15841 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
15842 Likewise. Require that operands[2] is even.
15843 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
15844 Remove extraneous parens. Require that operands[2] is a multiple
15846 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
15847 operands[0] is a MEM if <mask_applied>, the predicates/constraints
15848 disallow memory then.
15850 2017-03-30 Richard Biener <rguenther@suse.de>
15852 PR tree-optimization/77498
15853 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
15854 to non-constants over backedges.
15856 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
15858 PR rtl-optimization/80233
15859 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
15860 as last_combined_insn. Do not test for BARRIER_P separately.
15862 2017-03-29 Andreas Schwab <schwab@suse.de>
15865 * calls.c (prepare_call_address): Convert funexp to Pmode before
15866 copying to temp reg.
15868 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15870 PR tree-optimization/80158
15871 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
15872 Handle possible future case of more than one alternate
15874 (replace_rhs_if_not_dup): Likewise.
15875 (replace_one_candidate): Likewise.
15877 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
15879 PR rtl-optimization/80193
15880 * ira.c (ira): Do not check allocation for LRA.
15882 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
15884 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
15885 (nvptx_output_simt_exit): Declare.
15886 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
15887 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
15888 (init_softstack_frame): Move initialization of crtl->is_leaf to...
15889 (nvptx_declare_function_name): ...here. Emit declaration of local
15890 memory space buffer for omp_simt_enter insn.
15891 (nvptx_output_unisimt_switch): New.
15892 (nvptx_output_softstack_switch): New.
15893 (nvptx_output_simt_enter): New.
15894 (nvptx_output_simt_exit): New.
15895 * config/nvptx/nvptx.h (struct machine_function): New fields
15896 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
15897 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
15898 (UNSPECV_SIMT_EXIT): Ditto.
15899 (omp_simt_enter_insn): New insn.
15900 (omp_simt_enter): New expansion.
15901 (omp_simt_exit): New insn.
15902 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
15904 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
15905 (expand_GOMP_SIMT_ENTER_ALLOC): New.
15906 (expand_GOMP_SIMT_EXIT): New.
15907 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
15908 (GOMP_SIMT_ENTER_ALLOC): Ditto.
15909 (GOMP_SIMT_EXIT): Ditto.
15910 * target-insns.def (omp_simt_enter): New insn.
15911 (omp_simt_exit): Ditto.
15912 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
15914 (lower_rec_simd_input_clauses): Implement SIMT privatization.
15915 (lower_rec_input_clauses): Likewise.
15916 (lower_lastprivate_clauses): Handle SIMT privatization.
15918 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
15919 (ompdevlow_adjust_simt_enter): New.
15920 (find_simtpriv_var_op): New.
15921 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
15922 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
15924 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
15925 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
15926 (copy_decl_for_dup_finish): Ditto.
15928 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
15930 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
15933 * config/i386/i386.c (ix86_option_override_internal): Always
15934 allow -mpreferred-stack-boundary=3 for 64-bit targets.
15936 2017-03-28 Bin Cheng <bin.cheng@arm.com>
15938 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
15940 2017-03-28 Bin Cheng <bin.cheng@arm.com>
15942 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
15943 mark new edge's irreducible flag accordign to it.
15944 (vect_do_peeling): Check loop preheader edge's irreducible flag
15945 and pass it to function slpeel_add_loop_guard.
15947 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
15949 PR tree-optimization/80218
15950 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
15951 Update block frequencies and counts.
15953 2017-03-28 Richard Biener <rguenther@suse.de>
15955 PR tree-optimization/78644
15956 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
15957 of a simplification result we may not use it at all.
15959 2017-03-28 Richard Biener <rguenther@suse.de>
15962 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
15963 without arguments, generate default definition of a SSA name.
15965 2017-03-28 Richard Biener <rguenther@suse.de>
15967 PR middle-end/80222
15968 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
15969 TYPE_REF_CAN_ALIAS_ALL references.
15970 * fold-const.c (fold_indirect_ref_1): Likewise.
15972 2017-03-28 Martin Liska <mliska@suse.cz>
15975 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
15976 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
15978 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
15979 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
15981 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
15982 (EXTRA_SPECS): Define.
15983 (SUBTARGET_EXTRA_SPECS): Likewise.
15984 (SUBTARGET_CPP_SPEC): Likewise.
15985 * config/arc/elf.h (EXTRA_SPECS): Renamed to
15986 SUBTARGET_EXTRA_SPECS.
15987 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
15989 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
15991 * config/arc/simdext.md (vst64_insn): Update pattern.
15992 (vld32wh_insn): Likewise.
15993 (vld32wl_insn): Likewise.
15994 (vld64_insn): Likewise.
15995 (vld32_insn): Likewise.
15997 2017-03-28 Marek Polacek <polacek@redhat.com>
16000 * fold-const.c (fold_comparison): Use protected_set_expr_location
16001 instead of SET_EXPR_LOCATION.
16003 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16005 * tree.c (add_expr): Avoid name lookup warning.
16007 2017-03-27 Jeff Law <law@redhat.com>
16009 PR tree-optimization/80216
16010 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16011 function name. Limit recursion depth.
16012 (record_temporary_equivalences): Corresponding changes.
16014 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16016 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16019 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16022 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16024 * cfgcleanup.c (reg_note_cfa_p): New array.
16025 (insns_have_identical_cfa_notes): New function.
16026 (old_insns_match_p): Don't cross-jump in between /f
16027 and non-/f instructions. If both i1 and i2 are frame related,
16028 verify all CFA notes, their order and content.
16030 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16033 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16034 HImode and SImode with zero extend to DImode to one insn.
16035 (bswap<mode>2_extenddi): Likewise.
16036 (bswapsi2_extenddi): Likewise.
16037 (bswaphi2_extendsi): Likewise.
16038 (bswaphi2): Combine bswap HImode and SImode into one insn.
16039 Separate memory insns from swapping register.
16040 (bswapsi2): Likewise.
16041 (bswap<mode>2): Likewise.
16042 (bswaphi2_internal): Delete, no longer used.
16043 (bswapsi2_internal): Likewise.
16044 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16045 store, and gpr<-gpr swap insns.
16046 (bswap<mode>2_store): Likewise.
16047 (bswaphi2_reg): Register only splitter, combine with the splitter.
16048 (bswaphi2 splitter): Likewise.
16049 (bswapsi2_reg): Likewise.
16050 (bswapsi2 splitter): Likewise.
16051 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16052 the insns into load, store, and register/register insns.
16053 (bswapdi2_ldbrx): Likewise.
16054 (bswapdi2_load): Likewise.
16055 (bswapdi2_store): Likewise.
16056 (bswapdi2_reg): Likewise.
16058 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16060 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16061 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16063 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16066 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16068 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16069 special handling for target option conflicts between dform
16070 options (-mpower9-dform, -mpower9-dform-vector,
16071 -mpower9-dform-scalar) and -mno-direct-move.
16073 2017-03-27 Richard Biener <rguenther@suse.de>
16075 PR tree-optimization/80181
16076 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16078 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16080 * config/arc/predicates.md (move_double_src_operand): Replace the
16081 call to move_double_src_operand with a call to address_operand.
16083 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16085 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16086 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16087 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16089 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16091 * config/arc/predicates.md (long_immediate_loadstore_operand):
16092 Consider scaled addresses cases.
16094 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16096 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16097 restored when in interrupt.
16098 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16099 doesn't have delay slot.
16101 2017-03-27 Richard Biener <rguenther@suse.de>
16104 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16105 inlined thunk clones.
16107 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16110 * asan.c (instrument_derefs): Copy over last operand from
16111 original COMPONENT_REF to the new COMPONENT_REF with
16112 DECL_BIT_FIELD_REPRESENTATIVE.
16113 * ubsan.c (instrument_object_size): Likewise.
16115 2017-03-27 Richard Biener <rguenther@suse.de>
16117 PR tree-optimization/80170
16118 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16119 sure DR/SCEV didnt fold in constants we do not see when looking
16120 at the reference base alignment.
16122 2017-03-27 Richard Biener <rguenther@suse.de>
16124 PR middle-end/80171
16125 * gimple-fold.c (fold_ctor_reference): Properly guard against
16126 NULL return value from canonicalize_constructor_val.
16128 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16131 * config/i386/i386.c (ix86_expand_builtin)
16132 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16133 flags reg setting and flags reg using instructions.
16134 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16135 clobbering instructions to zero extend op2.
16137 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16139 * doc/install.texi (Configuration) <--with-aix-soname>:
16140 Update link to AIX ld.
16142 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16144 PR rtl-optimization/80160
16145 PR rtl-optimization/80159
16146 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16147 reg_alternate_class into account.
16149 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16152 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16153 to consider in curr_insn_transform.
16155 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16157 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16158 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16159 and emit_mode_inner.
16161 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16163 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16164 argument to the overloaded builtin variants. Use the new flag to
16165 deprecate certain builtin variants.
16166 * config/s390/s390-builtin-types.def: Add new builtin types.
16167 * config/s390/s390-builtins.h: Support new flags field for
16168 overloaded builtins.
16169 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16170 (s390_macro_to_expand): Enable vector float data type.
16171 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16172 builtins by incrementing the __VEC__ version number.
16173 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16175 (s390_resolve_overloaded_builtin): Emit error messages depending
16176 on the builtin flags.
16177 * config/s390/s390.c (s390_expand_builtin): Support additional
16178 flags argument. Change error message to match the messages
16179 emitted in s390-c.c.
16180 * config/s390/s390.md: New UNSPEC_* constants.
16181 (op_type): Add new instruction types.
16182 * config/s390/vecintrin.h: Add new builtins and test data class
16184 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16185 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16186 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16187 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16188 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16189 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16191 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16192 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16193 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16194 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16196 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16197 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16198 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16199 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16201 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16202 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16203 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16204 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16205 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16206 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16207 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16209 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16210 ("vec_scatter_element<V_HW_4:mode>_DI")
16211 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16212 ("vec_fpint<mode>", "vflls")
16213 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16214 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16215 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16216 ("*vec_cmphe<mode>_cc"): ... these.
16218 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16219 mode constant instead of magic value.
16221 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16223 * config/s390/s390.c (s390_expand_vec_compare): Support other
16224 vector floating point modes than just V2DF.
16225 (s390_expand_vcond): Likewise.
16226 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16227 (s390_cannot_change_mode_class): Prevent mode changes between TF
16228 and V1TF in vector registers.
16229 * config/s390/s390.md (DF, SF): New mode attributes.
16230 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16231 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16232 SFmode support for VRs.
16233 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16235 (VFT, VF_HW): New mode iterators.
16236 (vw, sdx): New mode attributes.
16237 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16238 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16239 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16240 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16241 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16242 also the new vector floating point modes. Renaming to ...
16244 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16245 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16246 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16247 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16248 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16249 ("vec_unordered<mode>"): ... these.
16251 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16252 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16253 ("*vec_extendv2df"): New insn definitions.
16255 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16257 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16258 ("mulditi3_2", "*muldi3_sign"): New patterns.
16259 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16260 rename the pattern definition.
16262 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16264 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16266 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16268 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16270 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16271 instruction if possible.
16272 * config/s390/vector.md (vec_halfnumelts): New mode
16274 ("*vec_vllezlf<mode>"): New pattern.
16276 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16278 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16279 ("popcountv4si2", "popcountv2di2"): Rename to ...
16280 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16281 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16283 ("popcount<mode>2_vxe"): New pattern.
16285 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16287 * common/config/s390/s390-common.c (processor_flags_table): Add
16289 * config.gcc: Add arch12.
16290 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16291 Default to arch12 for unknown CPU model numbers.
16292 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16293 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16294 PROCESSOR_max sanity check.
16295 * config/s390/s390-opts.h (enum processor_type): Add
16297 * config/s390/s390.c (processor_table): Add arch12.
16298 (s390_expand_builtin): Add check for B_VXE flag.
16299 (s390_issue_rate): Add PROCESSOR_ARCH12.
16300 (s390_get_sched_attrmask): Likewise.
16301 (s390_get_unit_mask): Likewise.
16302 (s390_sched_score): Enable z13 scheduling for arch12.
16303 (s390_sched_reorder): Likewise.
16304 (s390_sched_variable_issue): Likewise.
16305 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16307 (s390_tune_attr): Use z13 scheduling also for arch12.
16308 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16309 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16310 (TARGET_VXE_P): New macros.
16311 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16312 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16313 * config/s390/s390.opt: Add arch12 as processor_type.
16315 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16317 * config/s390/s390.md
16318 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16319 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16320 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16322 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16323 Rename expanders to ...
16325 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16326 ("fixuns_truncdddi2_emu"): ... these.
16328 ("fixuns_trunc<mode>si2_emu"): New expander.
16330 ("*fixuns_truncdfdi2_z13"): Rename to ...
16331 ("*fixuns_truncdfdi2_vx"): ... this.
16333 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16335 * config/s390/2964.md: Remove the single element vector compare
16336 instructions which are no longer used.
16337 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16339 (s390_canonicalize_comparison): Remove handling of DFmode
16341 (s390_expand_vec_compare_scalar): Remove function.
16342 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16343 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16345 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16347 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16349 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16351 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
16352 will anyway by matched by mov<mode>_64dfp.
16354 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16356 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
16357 vlef/vstef. Add missing operand to vleif.
16359 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16361 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
16362 pair for all vector types with 64 bit elements.
16363 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
16364 * config/s390/vector.md (V_HW_64): ... here.
16365 (V_128_NOSINGLE): New mode iterator.
16366 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
16367 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
16368 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
16369 ("*vec_load_pairv2di"): Change to ...
16370 ("*vec_load_pair<mode>"): ... this one.
16372 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16374 * config/s390/constraints.md: Add comments.
16375 (jKK): Reject element sizes > 8 bytes.
16376 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
16378 * config/s390/s390.md: Add the s_operand checks formerly in
16379 s390_split_ok_p to various splitters where they are still
16381 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
16382 for 128 bit vectors. Plus two splitters.
16384 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16386 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
16389 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16392 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
16393 error if the boundary argument is not constant.
16395 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16397 PR rtl-optimization/80112
16398 * loop-doloop.c (doloop_condition_get): Don't check condition
16399 if cmp isn't SET with IF_THEN_ELSE src.
16401 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16403 PR tree-optimization/80158
16404 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
16405 replacing a candidate statement, also replace it for the
16406 candidate's alternate interpretation.
16407 (replace_rhs_if_not_dup): Likewise.
16408 (replace_one_candidate): Likewise.
16410 2017-03-24 Richard Biener <rguenther@suse.de>
16412 PR tree-optimization/80167
16413 * graphite-isl-ast-to-gimple.c
16414 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
16416 (translate_isl_ast_to_gimple::get_rename): Likewise.
16418 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
16420 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
16421 handling of certain combinations of target options, including the
16422 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
16423 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
16425 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16428 * config/arm/arm.md (*load_multiple): Add reload_completed to
16429 matching condition.
16431 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16432 Richard Biener <rguenth@suse.de>
16434 PR tree-optimization/79908
16435 PR tree-optimization/80136
16436 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16437 been cast away, gimplify_and_add suffices.
16439 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
16441 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
16443 2017-03-23 Richard Biener <rguenther@suse.de>
16445 PR tree-optimization/80032
16446 * gimplify.c (gimple_push_cleanup): Forced unconditional
16447 cleanups still have to go to the conditional_cleanups
16450 2017-03-22 Jakub Jelinek <jakub@redhat.com>
16452 PR tree-optimization/80072
16453 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
16455 (next_operand_entry_id): Change type to unsigned int.
16456 (sort_by_operand_rank): Make sure to return the right return value
16457 even if unsigned fields are bigger than INT_MAX.
16458 (struct oecount): Change cnt and id type to unsigned int.
16459 (oecount_hasher::equal): Formatting fix.
16460 (oecount_cmp): Make sure to return the right return value
16461 even if unsigned fields are bigger than INT_MAX.
16462 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
16465 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
16466 TREE_READONLY on result if writing it more than once.
16469 * doc/invoke.texi (-fsanitize=thread): Document that with
16470 -fnon-call-exceptions atomics are not able to throw
16474 * tsan.c: Include tree-eh.h.
16475 (instrument_builtin_call): Call maybe_clean_eh_stmt or
16476 maybe_clean_or_replace_eh_stmt where needed.
16477 (instrument_memory_accesses): Add cfg_changed argument.
16478 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
16479 if it returned true.
16480 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
16482 PR rtl-optimization/63191
16483 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
16484 wrapper function, moved the whole old content into ...
16485 (ix86_delegitimize_address_1): ... this. New inline function.
16486 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
16487 true as last argument instead of ix86_delegitimize_address.
16489 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16491 * config/aarch64/aarch64.c (generic_branch_cost): Copy
16492 cortexa57_branch_cost.
16494 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
16496 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
16498 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16501 * doc/md.texi (Constraints): Document wA constraint.
16502 * config/rs6000/constraints.md (wA): New.
16503 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
16504 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
16505 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
16506 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
16508 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
16511 * gimplify.c (is_oacc_declared): New function.
16512 (oacc_default_clause): Use it to set default flags for acc declared
16513 variables inside parallel regions.
16514 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
16515 declared variables.
16516 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
16517 declare attribute to any decl as necessary.
16519 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16522 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
16523 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
16524 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
16525 (arm_arch_lpae): This.
16526 * config/arm/arm.c (arm_arch7ve): Rename into ...
16527 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
16528 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
16531 2017-03-22 Martin Liska <mliska@suse.cz>
16534 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
16535 error message instead of an ICE.
16537 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16539 * doc/extend.texi (6.11 Additional Floating Types): Revise.
16541 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
16543 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
16545 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16548 2017-03-21 Martin Sebor <msebor@redhat.com>
16550 * doc/extend.texi: Use "cannot" instead of "can't."
16551 * doc/hostconfig.texi: Same.
16552 * doc/install.texi: Same.
16553 * doc/invoke.texi: Same.
16554 * doc/loop.texi: Same.
16555 * doc/md.texi: Same.
16556 * doc/objc.texi: Same.
16557 * doc/rtl.texi: Same.
16558 * doc/tm.texi: Same.
16559 * doc/tm.texi.in: Same.
16560 * doc/trouble.texi: Same.
16562 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
16565 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
16566 (collect_checksum_attributes): Set it.
16567 (die_checksum_ordered): Use it.
16569 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16571 PR tree-optimization/79908
16572 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
16573 change: For a VA_ARG whose LHS has been cast away, use
16574 force_gimple_operand to construct the side effects.
16576 2017-03-21 David Malcolm <dmalcolm@redhat.com>
16578 PR translation/80001
16579 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
16580 more amenable to translation.
16581 (oacc_loop_auto_partitions): Likewise.
16583 2017-03-21 Marek Polacek <polacek@redhat.com>
16584 Martin Sebor <msebor@redhat.com>
16586 PR tree-optimization/80109
16587 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
16588 on INTEGRAL_TYPE_P.
16590 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16591 Segher Boessenkool <segher@kernel.crashing.org>
16594 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
16595 check reg_used_between_p between insn and one of succ or succ2
16596 depending on if succ is artificial insn not inserted into insn
16599 2017-03-21 Martin Liska <mliska@suse.cz>
16601 PR gcov-profile/80081
16602 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
16603 * doc/gcc.texi: Include gcov-dump stuff.
16604 * doc/gcov-dump.texi: New file.
16606 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
16608 PR rtl-optimization/79150
16609 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
16610 conditional jump, if the jump is the last insn of the loop.
16612 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16613 Richard Biener <rguenth@suse.de>
16615 PR tree-optimization/79908
16616 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
16617 been cast away, use force_gimple_operand to construct the side
16620 2017-03-21 Martin Liska <mliska@suse.cz>
16622 PR libfortran/79956
16623 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
16626 2017-03-21 Brad Spengler <spender@grsecurity.net>
16629 * plugin.c (htab_hash_plugin): New function.
16630 (add_new_plugin): Use it and adjust.
16631 (parse_plugin_arg_opt): Adjust.
16632 (init_one_plugin): Likewise.
16634 2017-03-21 Richard Biener <rguenther@suse.de>
16636 PR tree-optimization/80032
16637 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
16638 if set force the cleanup to happen unconditionally.
16639 (gimplify_target_expr): Push inserted clobbers with force_uncond
16640 to avoid them being removed by control-dependent DCE.
16642 2017-03-21 Richard Biener <rguenther@suse.de>
16644 PR tree-optimization/80122
16645 * tree-inline.c (copy_bb): Do not expans va-arg packs or
16646 va_arg_pack_len when the inlined call stmt requires pack
16648 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
16650 2017-03-21 Jakub Jelinek <jakub@redhat.com>
16653 * tsan.c (instrument_builtin_call): If the memory model argument
16654 is not a constant, assume it is valid.
16657 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
16660 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
16662 PR rtl-optimization/79910
16663 * combine.c (can_combine_p): Do not allow combining an I0 or I1
16664 if its dest is used by an insn before I2 (other than the combined
16665 insns themselves, which are properly handled already).
16667 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
16670 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
16672 * combine.c (record_used_regs): New static function.
16673 (try_combine): Handle situations where there is an additional
16674 instruction between I2 and I3 which needs to have a LOG_LINK
16678 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
16680 * combine.c (try_combine): Delete redundant i1 test. Call
16681 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
16683 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16686 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
16687 alternatives 13/14.
16689 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16691 PR tree-optimization/80054
16692 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
16693 the optimization if a PHI or any of its arguments is not dominated
16694 by the candidate's basis. Use gphi* rather than gimple* as
16696 (replace_profitable_candidates): Clean up a gimple* variable that
16697 should be a gphi* variable.
16699 2017-03-20 Martin Sebor <msebor@redhat.com>
16702 * doc/extend.texi (attribute constructor): Document present limitation.
16704 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
16707 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
16708 __POWER9_VECTOR__ #ifdef control, change template definition to
16709 use Power9-specific built-in function.
16710 (vec_any_eq): Likewise.
16711 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
16712 to control outcomes from this test.
16713 (vector_ae_<mode>p): For VEC_F modes, likewise.
16715 2017-03-20 Ian Lance Taylor <iant@google.com>
16717 * config/i386/i386.c (ix86_function_regparm): Save an extra
16718 register for -fsplit-stack with DECL_STATIC_CHAIN.
16720 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
16723 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
16724 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
16726 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
16728 * config/riscv/riscv.c (riscv_print_operand): Use "fence
16730 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
16733 2017-03-20 Marek Polacek <polacek@redhat.com>
16736 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
16738 2017-03-20 Richard Biener <rguenther@suse.de>
16740 PR tree-optimization/80113
16741 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
16742 allocate extra SSA name for PHI def.
16743 (add_close_phis_to_outer_loops): Likewise.
16744 (add_close_phis_to_merge_points): Likewise.
16745 (copy_loop_close_phi_args): Likewise.
16746 (copy_cond_phi_nodes): Likewise.
16748 2017-03-20 Martin Liska <mliska@suse.cz>
16750 PR middle-end/79753
16751 * tree-chkp.c (chkp_build_returned_bound): Do not build
16752 returned bounds for a LHS that's not a BOUNDED_P type.
16754 2017-03-20 Martin Liska <mliska@suse.cz>
16758 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
16759 COMPLEX_CST and VECTOR_CST.
16761 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16764 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
16765 target operand. A new splitter adds the clobber statement in case
16766 the target operand is dead anyway.
16768 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
16770 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
16771 to age-old versions of binutils and glibc.
16773 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
16775 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
16777 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
16779 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
16781 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
16783 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
16784 requirement for binutils 2.13.
16786 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
16788 * combine.c (try_combine): Delete redundant i1 test. Call
16789 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
16791 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
16793 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
16794 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
16796 <riscv64-*-elf>: Re-arrange section
16797 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
16798 <riscv32-*-linux>: Likewise.
16799 <riscv64-*-elf>: Likewise
16800 <riscv64-*-linux>: Likewise.
16802 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
16805 * aarch64.opt(verbose-cost-dump): Fix typo.
16807 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
16810 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
16811 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
16813 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
16815 * reload.c (find_reloads): When reloading a nonoffsettable address,
16816 use RELOAD_OTHER for it and its address reloads.
16818 PR rtl-optimization/79910
16819 * combine.c (record_used_regs): New static function.
16820 (try_combine): Handle situations where there is an additional
16821 instruction between I2 and I3 which needs to have a LOG_LINK
16824 2017-03-17 Jeff Law <law@redhat.com>
16826 PR tree-optimization/71437
16827 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
16828 conditional in the hash table first.
16829 (vrp_dom_walker::before_dom_children): Extract condition from
16830 ASSERT_EXPR. Record condition, its inverion and any implied
16831 conditions as well.
16833 2017-03-17 Marek Polacek <polacek@redhat.com>
16834 Markus Trippelsdorf <markus@trippelsdorf.de>
16836 PR tree-optimization/80079
16837 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
16840 2017-03-17 Richard Biener <rguenther@suse.de>
16842 PR middle-end/80075
16843 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
16844 Properly verify the LHS before the RHS possibly claims to be
16846 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
16849 2017-03-17 Martin Jambor <mjambor@suse.cz>
16851 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
16852 (List of -O2 options): Likewise.
16853 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
16856 2017-03-17 Tom de Vries <tom@codesourcery.com>
16858 * gcov-dump.c (print_usage): Print bug_report_url.
16860 2017-03-17 Richard Biener <rguenther@suse.de>
16862 PR middle-end/80050
16863 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
16864 (parser::peek): Likewise.
16866 2017-03-17 Richard Biener <rguenther@suse.de>
16868 PR tree-optimization/80048
16869 * sese.c (free_sese_info): Properly release rename_map and
16870 copied_bb_map elements.
16872 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
16874 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
16875 Add linked-list forward and backlinks. Insert on
16876 construction, remove on destruction.
16877 (class pass_store_merging): Add m_stores_head field.
16878 (pass_store_merging::terminate_and_process_all_chains):
16879 Iterate over m_stores_head list.
16880 (pass_store_merging::terminate_all_aliasing_chains):
16882 (pass_store_merging::execute): Check for debug stmts first.
16883 Push new chains onto the m_stores_head stack.
16885 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
16888 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
16889 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
16890 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
16892 2017-03-16 Jeff Law <law@redhat.com>
16894 PR tree-optimization/71437
16895 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
16896 member function. Implementation moved into after_dom_children
16897 member function and into the threader's thread_outgoing_edges
16899 (dom_opt_dom_walker::after_dom_children): Simplify by moving
16900 some code into new thread_outgoing_edges.
16901 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
16902 definition. Simplify marker handling (do it here). Assume we always
16903 have the available expression and the const/copies tables.
16904 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
16906 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
16907 * tree-vrp.c (equiv_stack): No longer file scoped.
16908 (vrp_dom_walker): New class.
16909 (vrp_dom_walker::before_dom_children): New member function.
16910 (vrp_dom_walker::after_dom_children): Likewise.
16911 (identify_jump_threads): Setup domwalker. Use it rather than
16912 walking edges in a random order by hand. Simplify setup/finalization.
16913 (finalize_jump_threads): Remove.
16914 (vrp_finalize): Do not call identify_jump_threads here.
16915 (execute_vrp): Do it here instead and call thread_through_all_blocks
16918 PR tree-optimization/71437
16919 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
16921 (simplify_stmt_for_jump_threading): Add basic_block argument. All
16923 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
16924 (dom_opt_dom_walker::thread_across_edge): Remove
16925 handle_dominating_asserts argument. All callers changed.
16926 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
16927 changes. Remove calls to lhs_of_dominating_assert. Other
16928 uses of handle_dominating_asserts turn into unconditional code
16929 (simplify_control_stmt_condition_1): Likewise.
16930 (simplify_control_stmt_condition): Likewise.
16931 (thread_through_normal_block, thread_across_edge): Likewise.
16932 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
16933 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
16934 object if it is not an SSA_NAME.
16935 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
16936 before calling into the VRP specific simplifiers.
16937 (identify_jump_threads): Remove handle_dominating_asserts
16940 2017-03-16 Jakub Jelinek <jakub@redhat.com>
16943 * tree-diagnostic.c (default_tree_printer): No longer static.
16944 * tree-diagnostic.h (default_tree_printer): New prototype.
16946 2017-03-16 Tamar Christina <tamar.christina@arm.com>
16948 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
16949 Change ins into fmov.
16951 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16953 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
16954 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
16955 Use h_con constraint for operand 1.
16956 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
16957 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
16959 2017-03-15 Jeff Law <law@redhat.com>
16961 PR tree-optimization/71437
16962 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
16963 (record_temporary_equivalences): Use it.
16965 PR tree-optimization/71437
16966 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
16967 tree-ssa-scopedtables.
16968 (lookup_avail_expr, build_and_record_new_cond): Likewise.
16969 (record_conditions, record_cond, vuse_eq): Likewise.
16970 (record_edge_info): Adjust to API tweak of record_conditions.
16971 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
16972 (record_temporary_equivalences, optimize_stmt): Likewise.
16973 (eliminate_redundant_computations): Likewise.
16974 (record_equivalences_from_stmt): Likewise.
16975 * tree-ssa-scopedtables.c: Include options.h and params.h.
16976 (vuse_eq): New function, moved from tree-ssa-dom.c
16977 (build_and_record_new_cond): Likewise.
16978 (record_conditions): Likewise. Accept vector of conditions rather
16979 than edge_equivalence structure for first argument.
16980 for the first argument.
16981 (avail_exprs_stack::lookup_avail_expr): New member function, moved
16982 from tree-ssa-dom.c.
16983 (avail_exprs_stack::record_cond): Likewise.
16984 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
16985 from tree-ssa-dom.c.
16986 (avail_exprs_stack): Add new member functions lookup_avail_expr
16988 (record_conditions): Declare.
16990 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
16993 * lra-constraints.c (process_alt_operands): Increase reject for
16994 reloading an input/output operand.
16996 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
16999 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17000 insns to convert from signed/unsigned char/short to IEEE 128-bit
17002 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17004 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17007 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17008 subreg of inner mode for values already in registers.
17010 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17012 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17013 iteration reg is used after the loop.
17015 2017-03-14 Martin Sebor <msebor@redhat.com>
17017 PR tree-optimization/79800
17018 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17019 precision in negative-positive range.
17020 (format_floating): Call non-const overload with adjusted precision.
17022 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17025 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17028 2017-03-14 Martin Sebor <msebor@redhat.com>
17030 PR middle-end/80020
17031 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17032 * builtins.def (aligned_alloc): Use it.
17035 * Makefile.in (GTFILES): Add calls.c.
17036 * calls.c: Include "gt-calls.h".
17038 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17040 PR rtl-optimization/79728
17041 * regs.h (struct target_regs): New field
17042 x_contains_allocatable_regs_of_mode.
17043 (contains_allocatable_regs_of_mode): New macro.
17044 * reginfo.c (init_reg_sets_1): Initialize it, and change
17045 contains_reg_of_mode so it includes global regs as well.
17046 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17047 rather than contains_regs_of_mode.
17049 2017-03-14 Martin Liska <mliska@suse.cz>
17051 * doc/invoke.texi: Document options that can't be combined with
17052 -fcheck-pointer-bounds.
17054 2017-03-14 Martin Liska <mliska@suse.cz>
17056 PR middle-end/79831
17057 * doc/invoke.texi (-Wchkp): Document the option.
17059 2017-03-14 Martin Liska <mliska@suse.cz>
17061 * Makefile.in: Install gcov-dump.
17063 2017-03-14 Martin Liska <mliska@suse.cz>
17065 * multiple_target.c (expand_target_clones): Bail out for
17066 an invalid attribute.
17068 2017-03-14 Richard Biener <rguenther@suse.de>
17070 * alias.c (struct alias_set_entry): Pack properly.
17071 * cfgloop.h (struct loop): Likewise.
17072 * cse.c (struct set): Likewise.
17073 * ipa-utils.c (struct searchc_env): Likewise.
17074 * loop-invariant.c (struct invariant): Likewise.
17075 * lra-remat.c (struct cand): Likewise.
17076 * recog.c (struct change_t): Likewise.
17077 * rtl.h (struct address_info): Likewise.
17078 * symbol-summary.h (function_summary): Likewise.
17079 * tree-loop-distribution.c (struct partition): Likewise.
17080 * tree-object-size.c (struct object_size_info): Likewise.
17081 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17082 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17083 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17084 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17085 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17086 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17087 (struct _stmt_vec_info): Likewise.
17089 2017-03-14 Martin Liska <mliska@suse.cz>
17092 * multiple_target.c (create_dispatcher_calls): Check that
17093 a target can create a function dispatcher.
17095 2017-03-14 Martin Liska <mliska@suse.cz>
17098 * multiple_target.c (expand_target_clones): Drop local.local
17099 flag for default implementation.
17101 2017-03-14 Richard Biener <rguenther@suse.de>
17103 PR tree-optimization/80030
17104 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17106 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17108 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17109 gcc_fallthrough() instead of __attribute__((fallthrough));
17111 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17113 * doc/gcc.texi: Remove "up" link to (DIR).
17114 * doc/gccint.texi: Ditto.
17116 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17118 * doc/install.texi (Specific) <avr>: Remove reference to
17121 2017-03-13 Jeff Law <law@redhat.com>
17123 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17124 attribute rather than comments.
17126 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17127 match_scratch operand is highest.
17129 2017-03-13 Martin Liska <mliska@suse.cz>
17131 PR middle-end/78339
17132 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17133 is a CHKP clone, use original declaration.
17135 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17137 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17138 (arc_conditional_register_usage): Use a different allocation order
17139 when optimizing for size.
17140 * common/config/arc/arc-common.c (arc_option_optimization_table):
17141 Section anchors default on when optimizing for size.
17143 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17145 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17147 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17149 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17150 * config/arc/arc.md (cpu_facility): Add cd variant.
17151 (*movqi_insn): Add code density variant.
17152 (*movhi_insn): Likewise.
17153 (*movqi_insn): Likewise.
17154 (*addsi3_mixed): Likewise.
17155 (subsi3_insn): Likewise.
17157 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17159 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17161 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17163 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17164 expressions with MINUS and UNARY ops.
17166 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17169 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17171 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17172 between vec_select and vector argument.
17173 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17174 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17175 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17176 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17177 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17178 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17180 2017-03-13 Richard Biener <rguenther@suse.de>
17183 * params.def (vect-max-peeling-for-alignment): Fix typo.
17185 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17187 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17188 issue that only occurred with binutils below 2.18.
17190 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17192 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17193 refer to binutils 2.11/2.12 minimum.
17195 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17197 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17198 ftp.kernel.org and simplify binutils requirement.
17200 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17202 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17204 (Optimize Options): Ditto. Also remove redundancy.
17206 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17208 PR translation/79848
17209 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17211 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17212 to G_ to avoid double translation.
17214 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17216 PR translation/79923
17217 * auto-profile.c (get_combined_location): Convert leading
17218 character of diagnostics to lower case and remove trailing period.
17219 (read_profile): Likewise for various diagnostics.
17220 * config/arm/arm.c (arm_option_override): Remove trailing period
17221 from various diagnostics.
17222 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17223 (msp430_expand_delay_cycles): Likewise.
17225 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17228 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17229 full command-line argument, rather than just "str".
17230 (aarch64_validate_march): Likewise.
17231 (aarch64_validate_mtune): Likewise.
17233 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17235 PR rtl-optimization/78911
17236 * lra-assigns.c (must_not_spill_p): New function.
17237 (spill_for): Use it.
17239 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17241 PR tree-optimization/79981
17242 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17243 ATOMIC_COMPARE_EXCHANGE ifn result.
17244 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17245 IFN_ATOMIC_COMPARE_EXCHANGE.
17247 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17250 * opts.c (parse_sanitizer_options): Add missing question mark to
17251 "did you mean" message.
17253 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17255 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17257 (VMULEUH_UNS): Likewise.
17258 (VMULOUB_UNS): Likewise.
17259 (VMULOUH_UNS): Likewise.
17260 * config/rs6000/rs6000.c (builtin_function_type): Remove
17261 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17263 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17266 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17267 x with result of extra_parsing_for_operand_code_0.
17268 (function_reader::extra_parsing_for_operand_code_0): Convert
17269 return type from void to rtx, returning x. When reading
17270 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17271 larger size containing struct block_symbol.
17273 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17275 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17276 -mfloat128-hardware without -m64.
17278 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17281 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17282 entries to the case statement that marks unsigned arguments to
17283 overloaded functions.
17285 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17288 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17290 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17293 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17294 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17296 2017-03-10 Martin Liska <mliska@suse.cz>
17300 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17301 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17304 2017-03-10 Olivier Hainque <hainque@adacore.com>
17306 * tree-switch-conversion (array_value_type): Start by resetting
17307 candidate type to it's main variant.
17309 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17311 PR rtl-optimization/79909
17312 * combine.c (try_combine): Use simplify_replace_rtx on individual
17313 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17314 of the whole CALL_INSN_FUNCTION_USAGE.
17316 PR tree-optimization/79972
17317 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17318 get_range_info on SSA_NAMEs. Formatting fixes.
17320 2017-03-10 Richard Biener <rguenther@suse.de>
17321 Jakub Jelinek <jakub@redhat.com>
17323 PR tree-optimization/77975
17324 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17325 edge to be constant.
17326 (get_val_for): For constant x return it. Formatting fix.
17327 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17328 would use the same bases as the current one.
17330 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17332 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17333 instead of vec_select for V1TImode.
17334 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17336 (VSX_LE_128): Add V1TI to this mode iterator.
17337 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17338 (*vsx_le_perm_store_<mode>): Likewise.
17339 (pre-reload splitter for VSX stores): Likewise.
17340 (post-reload splitter for VSX stores): Likewise.
17341 (*vsx_xxpermdi2_le_<mode>): Likewise.
17342 (*vsx_lxvd2x2_le_<mode>): Likewise.
17343 (*vsx_stxvd2x2_le_<mode>): Likewise.
17345 2017-03-09 Michael Eager <eager@eagercon.com>
17347 Correct failures with --enable-checking=yes,rtl.
17349 * config/microblaze/microblaze.c (microblaze_expand_shift):
17350 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17351 test for const0_rtx.
17352 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
17353 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
17355 2017-03-09 Richard Biener <rguenther@suse.de>
17357 PR tree-optimization/79977
17358 * graphite-scop-detection.c (scop_detection::merge_sese):
17359 Handle the case of extra exits to blocks dominating the entry.
17361 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
17363 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17366 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
17368 PR rtl-optimization/79949
17369 * lra-constraints.c (process_alt_operands): Check memory when
17370 trying to predict a cycle. Print about the overall increase.
17372 2017-03-09 Richard Biener <rguenther@suse.de>
17374 PR middle-end/79971
17375 * gimple-expr.c (useless_type_conversion_p): Preserve
17376 TYPE_SATURATING for fixed-point types.
17378 2017-03-09 Richard Biener <rguenther@suse.de>
17381 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
17382 alignment of BLKmode params.
17384 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17387 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
17388 (VALL_NO_V2Q): Likewise.
17390 * config/aarch64/aarch64-simd.md
17391 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
17393 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
17394 VALL_NO_V2Q mode iterator.
17395 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
17397 2017-03-09 Martin Liska <mliska@suse.cz>
17399 PR tree-optimization/79631
17400 * tree-chkp-opt.c (chkp_is_constant_addr): Call
17401 tree_int_cst_sign_bit just for INTEGER constants.
17403 2017-03-09 Martin Liska <mliska@suse.cz>
17407 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
17410 2017-03-09 Marek Polacek <polacek@redhat.com>
17413 * tree.c (inchash::add_expr): Handle TREE_VEC.
17415 2017-03-09 Martin Liska <mliska@suse.cz>
17418 (chkp_narrow_size_and_offset): New function.
17419 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
17420 (void chkp_parse_bit_field_ref): New function.
17421 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
17422 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
17424 2017-03-09 Martin Liska <mliska@suse.cz>
17427 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
17428 (chkp_find_bounds_1): Remove gcc_unreachable.
17430 2017-03-09 Jakub Jelinek <jakub@redhat.com>
17433 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
17434 BUILT_IN_SYNC*, determine the access type from the size suffix and
17435 always build a MEM_REF with that type. Handle forgotten
17436 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
17439 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
17440 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
17441 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
17442 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
17443 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
17444 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
17445 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
17446 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
17447 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
17448 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
17449 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
17450 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
17451 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
17452 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
17453 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
17454 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
17455 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
17456 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
17457 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
17458 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
17459 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
17460 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
17461 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
17462 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
17463 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
17464 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
17465 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
17466 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
17467 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
17468 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
17469 definitions outside of __OPTIMIZE__ guarded section.
17472 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
17473 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
17474 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
17475 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
17478 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17480 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
17481 ("vfenez<mode>"): Add missing constraints.
17483 2017-03-08 Martin Sebor <msebor@redhat.com>
17486 * config/nds32/nds32.c (nds32_option_override):
17487 Fix misspelled diagnostic.
17489 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17492 * gimplify.c (gimplify_omp_for): Replace index var in outer
17493 taskloop statement with an artificial variable and add
17494 OMP_CLAUSE_PRIVATE clause for it.
17496 2017-03-08 Richard Biener <rguenther@suse.de>
17498 PR tree-optimization/79955
17499 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
17500 for accesses that are completely outside of the variable.
17502 2017-03-08 Andrew Haley <aph@redhat.com>
17504 PR tree-optimization/79943
17505 * tree-ssa-loop-split.c (compute_new_first_bound): When
17506 calculating the new upper bound, (END-BEG) should be added, not
17509 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17511 * config/avr/avr.md (setmemhi): Make sure match_dup
17512 operand number comes before match_scratch.
17514 2017-03-08 Richard Biener <rguenther@suse.de>
17516 PR tree-optimization/79920
17517 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
17518 with ncopies == 1 to ...
17519 (vect_transform_slp_perm_load): ... here. Properly compute
17520 all element loads by iterating VF times over the group. Do
17521 not handle ncopies (computed in a broken way) in
17522 vect_create_mask_and_perm.
17524 2017-03-08 Jakub Jelinek <jakub@redhat.com>
17527 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
17528 is a uniform vector, use uniform_vector_p return value instead of
17529 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
17531 2017-03-07 Marek Polacek <polacek@redhat.com>
17533 PR middle-end/79809
17534 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
17535 (alloca_call_type): Likewise.
17537 2017-03-07 Martin Liska <mliska@suse.cz>
17539 * gcov.c (process_args): Put comment to correct location.
17541 2017-03-07 Martin Liska <mliska@suse.cz>
17543 PR middle-end/68270
17544 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
17545 Use array_at_struct_end_p instead of DECL_CHAIN (field).
17546 (chkp_narrow_bounds_for_field): Likewise.
17547 (chkp_parse_array_and_component_ref): Pass one more argument to
17550 2017-03-07 Richard Biener <rguenther@suse.de>
17552 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
17555 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17557 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
17558 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
17560 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17563 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
17564 to end of description.
17565 (PARAM_MAX_STORES_TO_MERGE): Likewise.
17567 2017-03-07 Jakub Jelinek <jakub@redhat.com>
17569 PR rtl-optimization/79901
17570 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
17572 (*avx512f_<code><mode>3<mask_name>): ... this.
17573 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
17574 iterator instead of VI8_AVX2_AVX512BW.
17576 PR rtl-optimization/79901
17577 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
17578 min/max expander, expand it using expand_vec_cond_expr.
17581 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
17584 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17587 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
17588 to void * for PCH reasons.
17589 * dwarf2out.c (output_loc_operands, output_die): Cast
17590 v.val_vec.array to unsigned char *.
17592 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
17595 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
17598 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
17600 PR rtl-optimization/79571
17601 * lra-constraints.c (process_alt_operands): Calculate static
17602 reject and subtract it from overall when only addresses will be
17605 2017-03-06 Julia Koval <julia.koval@intel.com>
17608 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
17609 incoming stack boundary to 128 for 64-bit targets.
17611 2017-03-06 Richard Biener <rguenther@suse.de>
17613 PR tree-optimization/79894
17614 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
17615 to NULL after folding it.
17617 2017-03-06 Richard Biener <rguenther@suse.de>
17619 PR tree-optimization/79824
17620 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
17621 check disabling peeling for gaps.
17623 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
17625 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17626 attributes): Document gettimeofday.
17628 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17630 * config/s390/s390.c (s390_option_override_internal): Set
17631 PARAM_MIN_VECT_LOOP_BOUND
17633 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
17635 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
17636 * config/s390/s390.md: Likewise.
17638 2017-03-06 Jakub Jelinek <jakub@redhat.com>
17641 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
17642 (<avx2_avx512>_perm<mode>): Rename to ...
17643 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
17645 (<avx512>_perm<mode>_mask): Rename to ...
17646 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
17648 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
17649 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
17650 instead of VI8F_256_512.
17651 (avx512f_perm<mode>): New define_expand.
17652 (avx512f_perm<mode>_mask): Likewise.
17653 (avx512f_perm<mode>_1<mask_name>): New define_insn.
17654 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
17656 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17658 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
17659 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
17661 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
17663 2017-03-06 Martin Liska <mliska@suse.cz>
17666 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
17667 when having a SSA NAME w/o VAR_DECL assigned to it.
17669 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17671 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
17672 msa_dpsub_<su>_d): Fix MODE for vec_select.
17674 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
17676 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
17678 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
17680 2017-03-06 Richard Biener <rguenther@suse.de>
17682 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
17683 * plugin.c (register_plugin_info): Likewise.
17684 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
17686 2017-03-05 Jakub Jelinek <jakub@redhat.com>
17688 * config/i386/sse.md (sse_storehps, sse_storelps,
17689 avx_<castmode><avxsizesuffix>_<castmode>,
17690 avx512f_<castmode><avxsizesuffix>_<castmode>,
17691 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
17692 in condition that at least one operand is not a MEM.
17694 2017-03-03 Jakub Jelinek <jakub@redhat.com>
17696 PR middle-end/79805
17697 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
17698 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
17700 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
17701 gimple_call_nothrow_p flag based on whether original builtin can throw.
17702 If it can, emit following stmts on the fallthrough edge.
17703 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
17704 don't create new bb if inserting just debug stmts on the edge, try to
17705 insert them on the fallthru bb or just reset debug stmts.
17707 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
17710 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
17711 restore recog_data (including the operand rtxes inside it) around
17712 the call to get_insn_template.
17714 2017-03-03 Martin Sebor <msebor@redhat.com>
17716 PR tree-optimization/79699
17717 * context.c (context::~context): Free MPFR caches to avoid
17718 a memory leak on program exit.
17720 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17722 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
17723 Use wide_int::ulow () instead of .elt (0).
17725 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
17727 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
17728 (*pushxf): Limit oF constraint to 32bit targets and add oC
17729 constraint for 64bit targets.
17730 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
17731 (*pushdf): Change rmF constraint to rmC.
17733 2017-03-03 Martin Liska <mliska@suse.cz>
17735 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
17736 Remove unused variable.
17738 2017-03-03 Jakub Jelinek <jakub@redhat.com>
17741 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
17742 is a memory operand, increase num_memory.
17743 (ix86_expand_args_builtin): Likewise.
17745 2017-03-03 Jan Hubicka <jh@suse.cz>
17748 * ipa-devirt.c (maybe_record_node): Properly handle
17749 __cxa_pure_virtual visibility.
17751 2017-03-03 Martin Liska <mliska@suse.cz>
17753 PR tree-optimization/79803
17754 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
17756 (pass_loop_prefetch::execute): Disabled optimization if an
17757 assumption about L1 cache size is not met.
17759 2017-03-03 Martin Liska <mliska@suse.cz>
17761 PR rtl-optimization/79574
17762 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
17763 (hash_scan_set): Likewise.
17764 (dump_hash_table): Likewise.
17765 (hoist_code): Likewise.
17767 2017-03-03 Richard Biener <rguenther@suse.de>
17769 * fixed-value.c (fixed_from_string): Restore use of elt (1)
17770 in place of uhigh ().
17771 (fixed_convert_from_real): Likewise.
17773 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
17776 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
17778 2017-03-03 Richard Biener <rguenther@suse.de>
17780 PR middle-end/79818
17781 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
17782 TYPE_OVERFLOW_UNDEFINED check.
17784 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17786 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
17788 (vector_ae_<mode>_p): Likewise.
17789 (vector_nez_<mode>_p): Likewise.
17790 (vector_ne_v2di_p): Likewise.
17791 (vector_ae_v2di_p): Likewise.
17792 (vector_ne_<mode>_p): Likewise.
17793 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
17795 (vsx_tsqrt<mode>2_fe): Likewise.
17797 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
17800 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
17802 2017-03-02 Jakub Jelinek <jakub@redhat.com>
17804 PR rtl-optimization/79780
17805 * cprop.c (one_cprop_pass): When second and further conditional trap
17806 in a single basic block is turned into an unconditional trap, turn it
17807 into a deleted note to avoid RTL verification failures.
17809 2017-03-02 Richard Biener <rguenther@suse.de>
17811 * fold-const.c (const_binop): Use ulow () instead of elt (0).
17813 2017-03-02 Richard Biener <rguenther@suse.de>
17815 PR tree-optimization/79345
17817 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
17818 param and abort the walk, returning -1 if it is hit.
17819 (walk_aliased_vdefs): Take a limit param and pass it on.
17820 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
17821 defaulting to 0 and return a signed int.
17822 * tree-ssa-uninit.c (struct check_defs_data): New struct.
17823 (check_defs): New helper.
17824 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
17825 about uninitialized memory.
17826 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
17827 bogus uninitialized warning.
17828 (fixed_convert_from_real): Likewise.
17830 2017-03-02 Bin Cheng <bin.cheng@arm.com>
17832 PR tree-optimization/66768
17833 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
17834 iv_use if base object can't be determined.
17836 2017-03-02 Jakub Jelinek <jakub@redhat.com>
17838 PR tree-optimization/79345
17839 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
17840 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
17841 (get_pattern_stats): Initialize it.
17842 * genemit.c (gen_expand): Verify match_scratch numbers come after
17843 match_operand/match_dup numbers.
17844 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
17845 match_scratch numbers.
17846 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
17848 * config/s390/s390.md (trunctdsd2): Likewise.
17850 2017-03-02 Richard Biener <rguenther@suse.de>
17852 * wide-int.h (wide_int_storage::operator=): Implement in terms
17855 2017-03-02 Richard Biener <rguenther@suse.de>
17857 PR tree-optimization/79777
17858 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
17859 the to insert expression to sth existing.
17861 2017-03-01 Martin Sebor <msebor@redhat.com>
17863 PR middle-end/79692
17864 * gimple-ssa-sprintf.c
17865 (directive::known_width_and_precision): New function.
17866 (format_integer): Use it.
17867 (get_mpfr_format_length): Consider the full range of precision
17868 when computing %g output with the # flag. Set the likely byte
17869 count to 3 rather than 1 when precision is indeterminate.
17870 (format_floating): Correct the lower bound of precision.
17872 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17874 * doc/invoke.texi: Document default code model for 64-bit Linux.
17876 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17879 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
17880 udiv rather than div since input pattern is unsigned.
17882 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
17884 * config/i386/i386.c (print_reg): Warn for values of
17885 unsupported size in integer register.
17887 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
17890 * config/rs6000/predicates.md (current_file_function_operand): Do
17891 not allow self calls to be local if the function is replaceable.
17893 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
17896 * config/rs6000/altivec.h (vec_ctz and others): Change the
17897 preprocessor macro that controls conditional compilation from
17898 _ARCH_PWR9 to __POWER9_VECTOR__.
17899 (vec_all_ne): Change parameterization of __altivec_scalar_pred
17900 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
17901 control (instead of _ARCH_PWR9 control) so that template
17902 definition uses power9-specific function.
17903 (vec_any_eq): Likewise.
17904 (vec_all_ne): Change macro definition to use a power9-specific
17905 expansion under #ifdef __POWER9_VECTOR__ control (instead of
17906 _ARCH_PWR9 control).
17907 (vec_any_eq) Likewise.
17908 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
17909 expansion for CMPNEF to remove support for xvcmpnesp instruction.
17910 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
17911 support for xvcmpnedp instruction.
17912 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
17913 macro expansion so that Power9 implementation of vec_all_ne does
17914 not use the AltiVec predicate framework.
17915 (VCMPNEH_P): Likewise.
17916 (VCMPNEW_P): Likewise.
17917 (VCMPNED_P): Likewise.
17918 (VCMPNEFP_P): Likewise.
17919 (VCMPNEDP_P): Likewise.
17920 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
17921 implementation of vec_any_eq to not use AltiVec predicate
17923 (VCMPAEH_P): Likewise.
17924 (VCMPAEW_P): Likewise.
17925 (VCMPAED_P): Likewise.
17926 (VCMPAEFP_P): Likewise.
17927 (VCMPAEDP_P): Likewise.
17928 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
17929 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
17930 not use the AltiVec predicate framework.
17931 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
17932 of vec_any_eq to not use AltiVec predicate framework.
17933 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17934 support for predefined __POWER9_VECTOR__ macro to indicate that
17935 Power9 instruction selection is enabled.
17936 (altivec_overloaded_builtins): Remove extraneous
17937 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
17938 function argument types RS6000_BTI_bool_V16QI and
17939 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
17940 entry for overloaded function argument types RS6000_BTI_bool_V4SI
17941 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
17942 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
17943 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
17944 Power9 for implementations of vec_cmpne. Change the signature for
17945 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
17946 (representing vec_all_ne) to remove the previously described first
17947 argument of type RS6000_BTI_INTSI, as this was an artifact of
17948 reliance on the AltiVec predicate framework, which is no longer
17949 used in the implementation of these functions. Add
17950 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
17951 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
17952 since, unlike the AltiVec predicate framework implementation, we
17953 do not share function descriptors between vec_alle and vec_anyeq.
17954 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
17955 set of modes that receive special treatment even when
17956 TARGET_P9_VECTOR is true. The special treatment emits code that
17957 does not depend on Power9 instructions.
17958 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
17959 define_expand to not rely on AltiVec predicate framework.
17960 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
17962 (vector_ne_v2di_p): Change this define_expand to not rely on
17963 AltiVec predicate framework.
17964 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
17966 (vector_ne_<mode>_p): Change this define_expand to not rely on
17967 AltiVec predicate framework.
17968 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
17970 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
17971 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
17972 define_insn pattern.
17973 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
17974 define_insn pattern because the xvcmpne<VSs>. instruction is not
17976 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
17977 instruction is not supported.
17979 2017-03-01 Jakub Jelinek <jakub@redhat.com>
17981 * config/nvptx/nvptx.c: Include intl.h.
17983 2017-03-01 Martin Jambor <mjambor@suse.cz>
17986 * ipa-prop.h (ipa_bits): Removed field known.
17987 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
17988 to pointers. Adjusted their comments to warn about their sharing.
17989 (ipcp_transformation_summary): Change bits to a vector of pointers.
17990 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
17991 (ipa_get_ipa_bits_for_value): Declare.
17992 * tree-vrp.h (value_range): Mark as GTY((for_user)).
17993 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
17994 (ipa_bits_hash_table): Likewise.
17995 (ipa_vr_ggc_hash_traits): Likewise.
17996 (ipa_vr_hash_table): Likewise.
17997 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
17998 being pointers and vr_known being removed.
17999 (ipa_set_jf_unknown): Likewise.
18000 (ipa_get_ipa_bits_for_value): New function.
18001 (ipa_set_jfunc_bits): Likewise.
18002 (ipa_get_value_range): New overloaded functions.
18003 (ipa_set_jfunc_vr): Likewise.
18004 (ipa_compute_jump_functions_for_edge): Use the above functions to
18005 construct bits and vr parts of jump functions.
18006 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18007 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18009 (ipcp_grow_transformations_if_necessary): Also allocate
18010 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18012 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18013 them. Fix too long lines.
18014 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18015 vr_known being removed.
18016 (ipa_read_jump_function): Use new setter functions to construct bits
18017 and vr parts of jump functions or set them to NULL.
18018 (write_ipcp_transformation_info): Adjust for bits being pointers.
18019 (read_ipcp_transformation_info): Likewise.
18020 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18022 Include gt-ipa-prop.h.
18023 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18025 (ipcp_store_bits_results): Likewise.
18026 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18027 Do not write to existing jump functions but use a temporary instead.
18029 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18032 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18033 attempt to use its first operand as BIT_FIELD_REF base.
18035 2017-03-01 Richard Biener <rguenther@suse.de>
18037 PR middle-end/79721
18038 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18039 interpolating formula in wrapping arithmetic.
18040 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18042 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18044 PR tree-optimization/79734
18045 * tree-vect-generic.c (expand_vector_condition): Optimize
18046 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18047 Handle VEC_COND_EXPR where comparison has different inner width from
18048 type's inner width.
18050 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18052 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18053 markup, and similar issues. Remove @opindex entries for things
18054 that aren't options. Add missing -mmpy-option entries.
18056 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18058 PR tree-optimization/79737
18059 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18060 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18061 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18062 instead of byte_size. Formatting fix.
18063 (shift_bytes_in_array_right): Formatting fix.
18065 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18068 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18069 condition on optimize for the leaf function test.
18071 2017-02-28 Martin Liska <mliska@suse.cz>
18074 * read-rtl-function.c (function_reader::handle_unknown_directive):
18075 Bail out when one uses -flto.
18077 2017-02-28 Martin Liska <mliska@suse.cz>
18079 * common.opt: Replace space with tabular for options of <number>
18081 * config/i386/i386.opt: Show <number> value for
18082 -mlarge-data-threshold.
18083 * opts.c (print_filtered_help): Do not display number in hexadecimal
18086 2017-02-28 Martin Liska <mliska@suse.cz>
18088 * common.opt: Fix --help=option -Q for options which are of
18091 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18093 * config/i386/i386.c (print_reg): Error out for values
18094 of 8-bit size in invalid integer register.
18096 2017-02-28 Martin Sebor <msebor@redhat.com>
18098 PR tree-optimization/79691
18099 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18101 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18104 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18105 gcc_unreachable with output_operand_lossage.
18107 2017-02-28 Richard Biener <rguenther@suse.de>
18109 PR tree-optimization/79740
18110 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18112 (visit_nary_op): Insert the nary into the hashtable if we
18113 pattern-matched sth.
18114 * tree-ssa-pre.c (eliminate_insert): Robustify.
18116 2017-02-28 Richard Biener <rguenther@suse.de>
18118 PR middle-end/79731
18119 * fold-const.c (decode_field_reference): Reject out-of-bound
18122 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18124 * config/i386/i386.c: Include intl.h.
18125 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18126 instead of just cond ? "..." : "...".
18127 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18128 * coverage.c (read_counts_file): Likewise.
18129 * omp-offload.c: Include intl.h.
18130 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18131 of just cond ? "..." : "...".
18132 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18133 of just cond ? "..." : "...".
18135 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18138 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18140 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18141 'tune for' CPU name.
18142 * config/arm/arm-cpu-data.h: Regenerated.
18144 2017-02-28 Richard Biener <rguenther@suse.de>
18146 PR tree-optimization/79732
18147 * tree-inline.c (expand_call_inline): Do not shadow var.
18149 2017-02-28 Richard Biener <rguenther@suse.de>
18151 PR tree-optimization/79723
18152 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18153 address-space properly.
18155 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18157 * doc/optinfo.texi (Optimization groups): Fix option used for
18159 * doc/invoke.texi (-fopt-info): Document "omp".
18160 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18161 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18162 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18163 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18164 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18166 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18168 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18169 OPTGROUP_OMP with "omp".
18171 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18174 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18175 for arithmetic shift of unsigned V2DI.
18177 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18179 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18180 arc/linux.h headers.
18181 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18182 (LINK_SPEC): Likewise.
18183 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18184 (EXTRA_SPECS): Likewise.
18185 (STARTFILE_SPEC): Likewise.
18186 (ENDFILE_SPEC): Likewise.
18187 (LIB_SPEC): Likewise.
18188 (TARGET_SDATA_DEFAULT): Likewise.
18189 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18190 (MULTILIB_DEFAULTS): Likewise.
18191 (DWARF2_UNWIND_INFO): Likewise.
18192 * config/arc/big.h: New file.
18193 * config/arc/elf.h: Likewise.
18194 * config/arc/linux.h: Likewise.
18195 * config/arc/t-uClibc: Remove.
18197 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18199 PR tree-optimization/77536
18200 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18201 (tree_transform_and_unroll_loop): Use above function to compute the
18202 estimated niter of unrolled loop and use it when scaling profile.
18203 Also use count info rather than frequency if it's non-zero.
18204 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18205 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18206 (vect_transform_loop): Call above function.
18208 2017-02-27 Richard Biener <rguenther@suse.de>
18210 PR tree-optimization/45397
18211 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18212 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18213 (visit_nary_op): Add pattern matching for CSEing sign-changed
18214 or truncated operations with wider ones.
18216 2017-02-27 Richard Biener <rguenther@suse.de>
18218 PR tree-optimization/79690
18219 * tree-vect-stmts.c (vectorizable_store): Use vector type
18220 built from the DR with address-space.
18222 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18224 * doc/invoke.texi (Optimize Options): Refine the description
18225 of asan-use-after-return.
18227 2017-02-25 Alan Modra <amodra@gmail.com>
18229 PR rtl-optimization/79584
18230 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18231 base, not ad->base_term, the reg within base. Remove assertion
18232 that ad->base == ad->base_term. Replace gen_int_mode using
18233 bogus mode with const0_rtx.
18235 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18237 PR middle-end/79396
18238 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18239 FMA_EXPR like tcc_binary or tcc_unary.
18241 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18244 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18246 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18247 (output_loc_operands): Handle DW_OP_call_ref and
18248 DW_OP_GNU_variable_value.
18249 (struct variable_value_struct): New type.
18250 (struct variable_value_hasher): Likewise.
18251 (variable_value_hash): New variable.
18252 (string_types): Remove.
18253 (copy_loc_descr): New function.
18254 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18255 (prepend_loc_descr_to_each): New function.
18256 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18257 instead of add_loc_descr_to_each if the first argument is single
18258 location list and the second has multiple.
18259 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18260 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18261 when looking for variable value which doesn't have other location info.
18262 (loc_list_from_tree): Formatting fix.
18263 (gen_array_type_die): Simplify DW_AT_string_length handling.
18264 (adjust_string_types): Remove.
18265 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18266 string_types. Call resolve_variable_values.
18267 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18268 (resolve_addr_in_expr): Likewise. Add A argument.
18269 (copy_deref_exprloc): Remove deref argument. Adjust for the
18270 original expression being DW_OP_GNU_variable_value with optionally
18271 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18272 optionally after it.
18273 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18274 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18275 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18276 (variable_value_hasher::hash, variable_value_hasher::equal): New
18278 (resolve_variable_value_in_expr, resolve_variable_value,
18279 resolve_variable_values, note_variable_value_in_expr,
18280 note_variable_value): New functions.
18281 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18284 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18287 * opts.h (handle_generated_option): Add GENERATED_P argument.
18288 * opts-common.c (handle_option): Adjust function comment.
18289 (handle_generated_option): Add GENERATED_P argument, pass it to
18291 (control_warning_option): Pass false to handle_generated_option
18293 * opts.c (maybe_default_option): Pass true to handle_generated_option
18295 * optc-gen.awk: Likewise.
18297 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18299 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18300 a REG, look at the REG it is a SUBREG of.
18301 (splitter for cmpeqsi_t): Ditto.
18303 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18305 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18306 the special USEs with the pattern of the insn, not the insn itself.
18308 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18311 * doc/invoke.texi: Document -mload-store-pairs.
18313 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18314 Sandra Loosemore <sandra@codesourcery.com>
18316 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18317 argument isn't a CONST_INT.
18318 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18319 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18320 (nios2_validate_compare): Bypass alternate compare logic if *op2
18321 is not a CONST_INT.
18322 (ldstwm_operation_p): Return false if first_base is not a REG or
18323 if first_offset is not a CONST_INT.
18325 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18327 * config/cris/cris.md: Use correct operand in a define_peephole2.
18329 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18331 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18333 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18335 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18336 this_insn if it is an INSN or JUMP_INSN.
18337 (force_offsettable): Look at base, not at addr.
18338 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18339 on things that aren't necessarily CONST_INTs.
18341 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18343 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18344 -mfpmath=sse is the default also for x86-32 targets with SSE2
18345 instruction set when @option{-ffast-math} is enabled
18347 2017-02-24 Jeff Law <law@redhat.com>
18349 PR rtl-optimizatoin/79286
18350 * ira.c (update_equiv_regs): Drop may_trap_p exception to
18353 2017-02-24 Richard Biener <rguenther@suse.de>
18355 PR tree-optimization/79389
18356 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
18359 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
18361 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
18362 function comment to reflect reality.
18363 (loop_exits_before_overflow): Fix typo in function description.
18365 2017-02-24 Richard Biener <rguenther@suse.de>
18367 PR tree-optimization/79389
18368 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
18369 properly that a threading opportunity exists. Detect conditional
18370 copy/constant propagation opportunities.
18372 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
18374 * config/visium/visium.md (type): Add trap.
18375 (b): New mode attribute.
18376 (*btst): Rename into...
18377 (*btst<mode>): ...this and adjust.
18378 (*cbranchsi4_btst_insn): Rename into...
18379 (*cbranch<mode>4_btst_insn): ...this and adjust.
18380 (trap): New define_insn.
18382 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18384 PR tree-optimization/79389
18385 * ifcvt.c (struct noce_if_info): Add rev_cond field.
18386 (noce_reversed_cond_code): New function.
18387 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
18388 reversed_comparison_code. Formatting fix.
18389 (noce_try_store_flag): Test rev_cond != NULL in addition to
18390 reversed_comparison_code.
18391 (noce_try_store_flag_constants): Likewise.
18392 (noce_try_store_flag_mask): Likewise.
18393 (noce_try_addcc): Use rev_cond if non-NULL instead of
18394 reversed_comparison_code.
18395 (noce_try_cmove_arith): Likewise. Formatting fixes.
18396 (noce_try_minmax, noce_try_abs): Clear rev_cond.
18397 (noce_find_if_block): Initialize rev_cond.
18398 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
18399 instead of false as last argument never attempt to reverse it
18402 2017-02-23 Bin Cheng <bin.cheng@arm.com>
18404 PR tree-optimization/79663
18405 * tree-predcom.c (combine_chains): Process refs in reverse order
18406 only for ZERO length chains, and add explaining comment.
18408 2017-02-23 Jeff Law <law@redhat.com>
18410 PR tree-optimization/79578
18411 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
18412 in call to operand_equal_p.
18414 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
18417 * config/i386/cpuid.h: Fix another undefined behavior.
18419 2017-02-23 Richard Biener <rguenther@suse.de>
18421 PR tree-optimization/79683
18422 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
18423 vector types for data-refs.
18425 2017-02-23 Martin Liska <mliska@suse.cz>
18427 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
18429 2017-02-23 Jakub Jelinek <jakub@redhat.com>
18431 PR middle-end/79665
18432 * internal-fn.c (get_range_pos_neg): Moved to ...
18433 * tree.c (get_range_pos_neg): ... here. No longer static.
18434 * tree.h (get_range_pos_neg): New prototype.
18435 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
18436 are known to be in between 0 and signed maximum inclusive, try to
18437 expand both unsigned and signed divmod and use the cheaper one from
18440 2017-02-22 Jeff Law <law@redhat.com>
18442 PR tree-optimization/79578
18443 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
18444 to compare base operands.
18446 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
18449 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
18450 gpc_reg_operand instead of fpr_reg_operand.
18452 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
18454 * config/mips/mips.c (mips_return_in_memory): Force FP
18455 vector types to be returned in memory for o32 ABI.
18457 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18459 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
18460 instead of DW_TAG_member for static data member declarations and don't
18461 set no_linkage_name for static inline data members.
18462 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
18465 2017-02-22 Martin Liska <mliska@suse.cz>
18467 * doc/invoke.texi: Replace inequality signs with square brackets
18470 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18472 PR tree-optimization/68644
18473 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
18475 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
18478 * lra-constraints.c (simplify_operand_subreg): Handle
18479 WORD_REGISTER_OPERATIONS targets.
18481 2017-02-22 Jakub Jelinek <jakub@redhat.com>
18484 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
18485 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
18486 elimination by swapping fld*.
18488 2017-02-22 Richard Biener <rguenther@suse.de>
18490 PR tree-optimization/79673
18491 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
18492 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
18493 irrelevant address-space qualifiers and avoiding a
18494 ADDR_SPACE_CONVERT_EXPR from fold_convert.
18496 2017-02-22 Richard Biener <rguenther@suse.de>
18498 PR tree-optimization/79666
18499 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
18500 to not symbolically negate if that may introduce undefined
18503 2017-02-22 Martin Liska <mliska@suse.cz>
18506 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
18507 * data-streamer-out.c (streamer_write_gcov_count_stream):
18509 * value-prof.c (stream_out_histogram_value): Make assert more
18510 precise based on type of counter.
18512 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
18515 * config/i386/i386.md (standard_x87sse_constant_load splitter):
18516 Use nonimmediate_operand instead of memory_operand for operand 1.
18517 (float-extend standard_x87sse_constant_load splitter): Ditto.
18519 2017-02-21 Jeff Law <law@redhat.com>
18521 PR tree-optimization/79621
18522 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
18523 blocks with edges to themselves.
18525 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18528 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
18529 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
18530 Use gimple_call_builtin_p.
18533 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
18534 on temporarily removed DEBUG_INSNs.
18536 PR tree-optimization/79649
18537 * tree-loop-distribution.c (classify_partition): Give up on
18538 non-generic address space loads/stores.
18540 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
18542 * doc/loop.texi (Loop manipulation): Remove nonexistent
18543 tree_ssa_loop_version from the documentation.
18544 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
18546 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18549 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
18550 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
18551 * config/rs6000/rs6000.c: Include except.h.
18552 (rs6000_expand_split_stack_prologue): Call
18553 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
18555 2017-02-21 Martin Jambor <mjambor@suse.cz>
18558 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
18559 have been analyzed.
18561 2017-02-21 Martin Jambor <mjambor@suse.cz>
18563 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
18564 for backward compatibility only.
18565 * doc/invoke.texi (Option Summary): Remove all references to
18566 -fipa-cp-alignment.
18568 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
18572 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18574 * lra-constraints.c (curr_insn_transform): Handle
18575 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18577 2017-02-21 Martin Liska <mliska@suse.cz>
18579 * config/i386/i386.opt: Replace -masm-dialect with -masm.
18581 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
18583 PR translation/79638
18584 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
18586 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
18589 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
18590 (arm_function_ok_for_sibcall): Return false for an indirect call by
18591 descriptor if all the argument registers are used.
18592 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
18593 alignment of the function.
18595 2017-02-21 Jakub Jelinek <jakub@redhat.com>
18597 PR tree-optimization/61441
18598 * simplify-rtx.c (simplify_const_unary_operation): For
18599 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
18600 the sNaN unmodified.
18602 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18604 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18605 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18606 instead of SYSTEM_HEADER_DIR.
18608 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
18609 Martin Liška <mliska@suse.cz>
18611 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
18612 Fix typos and grammar, use active voice, and clarify.
18614 2017-02-20 Marek Polacek <polacek@redhat.com>
18616 PR middle-end/79537
18617 * gimplify.c (gimplify_expr): Handle unused *&&L;.
18620 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
18622 2017-02-20 Jakub Jelinek <jakub@redhat.com>
18625 * config/i386/i386.c (ix86_expand_builtin): Handle
18626 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
18627 ix86_builtins_isa[fcode].isa as a requirement of those
18628 flags and any other flag in the bitmask.
18629 (ix86_init_mmx_sse_builtins): Use 0 instead of
18630 ~OPTION_MASK_ISA_64BIT as mask.
18631 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
18632 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
18633 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
18634 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
18636 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18639 * lra-constraints.c (split_reg): Check requested split mode
18640 is supported by the register.
18642 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18644 * lra-constraints.c (simplify_operand_subreg): Remove early
18647 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18650 * lra-constraints.c (curr_insn_transform): Tighten condition
18651 for converting SUBREG reloads from OP_OUT to OP_INOUT.
18653 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
18656 * lra-constraints.c (curr_insn_transform): Handle
18657 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
18659 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
18662 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
18664 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
18666 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
18669 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
18672 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18674 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
18677 2017-02-18 Jakub Jelinek <jakub@redhat.com>
18679 * final.c (last_columnnum, override_columnnum): New variables.
18680 (final_start_function): Set last_columnnum, pass it to begin_prologue
18681 hook and pass 0 to dwarf2out_begin_prologue.
18682 (final_scan_insn): Update override_columnnum. Pass last_columnnum
18683 to source_line debug hook.
18684 (notice_source_line): Compute last_columnnum and for debug_column_info
18685 return true on column changes.
18686 * debug.h (struct gcc_debug_hooks): Add column argument to
18687 source_line and begin_prologue hooks.
18688 (debug_nothing_int_charstar_int_bool): Remove prototype.
18689 (debug_nothing_int_int_charstar,
18690 debug_nothing_int_int_charstar_int_bool): New prototypes.
18691 (dwarf2out_begin_prologue): Add column argument.
18692 * debug.c (do_nothing_debug_hooks): Adjust source_line and
18693 begin_prologue hooks.
18694 (debug_nothing_int_charstar_int_bool): Remove.
18695 (debug_nothing_int_int_charstar,
18696 debug_nothing_int_int_charstar_int_bool): New functions.
18697 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
18698 through to dwarf2out_source_line.
18699 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
18700 (dwarf2out_source_line): Add column argument, emit it if requested.
18701 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
18703 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
18704 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
18705 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
18706 through to dwarf2out_begin_prologue.
18707 (vmsdbgout_source_line): Add column argument, pass it through to
18708 dwarf2out_source_line.
18709 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
18710 dbxout_source_line caller.
18711 (dbxout_source_line): Add column argument.
18713 * common.opt (gno-column-info, gcolumn-info): New options.
18714 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
18715 (check_die): Also test for multiple DW_AT_decl_column attributes.
18716 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
18717 DW_AT_decl_column if requested.
18718 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
18720 (gen_variable_die): Likewise.
18721 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
18722 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
18725 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
18726 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
18727 (ix86_handle_option): Handle OPT_m3dnowa.
18728 * doc/invoke.texi (-m3dnowa): Document.
18729 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
18730 -m3dnowa instead of -m3dnow -march=athlon.
18733 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
18734 instead of gcc_assert for K, r and R code checks. Formatting fixes.
18736 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18739 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
18740 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
18741 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
18742 generator for vsx_xxpermdi_<mode>_be.
18743 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
18744 force big-endian semantics.
18745 (vsx_xxpermdi_<mode>_be): New define_expand with same
18746 implementation as previous version of vsx_xxpermdi_<mode>.
18748 2017-02-17 Jakub Jelinek <jakub@redhat.com>
18750 PR tree-optimization/79327
18751 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
18752 variable, its initialization and use.
18754 2017-02-17 Julia Koval <julia.koval@intel.com>
18756 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
18757 (OPTION_MASK_ISA_PKU_UNSET): New.
18758 (ix86_handle_option): Handle -mrdpid.
18759 * config/i386/cpuid.h (bit_RDPID): New.
18760 * config/i386/driver-i386.c (host_detect_local_cpu):
18761 Detect RDPID feature.
18762 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
18763 * config/i386/i386-c.c (ix86_target_macros_internal):
18765 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
18766 (ix86_valid_target_attribute_inner_p): Add "rdpid".
18767 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
18768 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
18769 * config/i386/i386.md (define_insn "rdpid"): New.
18770 * config/i386/i386.opt Add -mrdpid.
18771 * config/i386/immintrin.h (_rdpid_u32): New.
18773 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
18775 PR rtl-optimization/79541
18776 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
18777 instead of transforming it into USE.
18779 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
18781 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
18782 If HONOR_SNANS (SFmode) force the input to a register.
18783 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
18784 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
18785 an frsp or similar insn.
18787 2017-02-17 Martin Liska <mliska@suse.cz>
18789 PR rtl-optimization/79577
18790 * params.def (selsched-max-sched-times): Increase minimum to 1.
18792 2017-02-17 Martin Liska <mliska@suse.cz>
18794 PR rtl-optimization/79574
18795 * gcse.c (want_to_gcse_p): Prevent integer overflow.
18797 2017-02-17 Martin Liska <mliska@suse.cz>
18799 PR tree-optimization/79529
18800 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
18801 ssa_defined_default_def_p to handle cases which are implicitly
18803 * tree-ssa.c (ssa_defined_default_def_p): New function.
18804 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
18805 which are implicitly defined.
18806 * tree-ssa.h (ssa_defined_default_def_p): Declare.
18808 2017-02-17 Richard Biener <rguenther@suse.de>
18810 PR middle-end/79576
18811 * params.def (max-ssa-name-query-depth): Limit to 10.
18813 2017-02-17 Richard Biener <rguenther@suse.de>
18815 PR tree-optimization/79552
18816 * tree-ssa-structalias.c (visit_loadstore): Properly verify
18819 2017-02-17 Richard Biener <rguenther@suse.de>
18822 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
18824 2017-02-17 Marek Polacek <polacek@redhat.com>
18826 PR middle-end/79536
18827 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
18828 (fold_negate_expr): New wrapper.
18830 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
18832 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
18833 Correct terminology and de-emphasize pre-standard behavior.
18835 2017-02-16 Alan Modra <amodra@gmail.com>
18837 PR rtl-optimization/79286
18838 * ira.c (def_dominates_uses): New function.
18839 (update_equiv_regs): Don't create an equivalence for insns that
18840 may trap where the register def does not dominate the use.
18842 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
18844 PR rtl-optimization/78127
18845 * lra.c (lra): Call lra_eliminate before finish the loop after
18848 2017-02-16 Richard Biener <rguenther@suse.de>
18850 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
18852 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
18853 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
18854 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
18855 (isl_val_int_from_wi): New function.
18856 (extract_affine_gmp): Rename to ...
18857 (extract_affine_wi): ... this, take a widest_int.
18858 (extract_affine_int): Just wrap extract_affine_wi.
18859 (add_param_constraints): Use isl_val_int_from_wi.
18860 (add_loop_constraints): Likewise, and extract_affine_wi.
18862 2017-02-15 Jeff Law <law@redhat.com>
18864 PR middle-end/79521
18865 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
18866 ira_init_register_move_cost_if_necessary.
18868 2017-02-15 Martin Sebor <msebor@redhat.com>
18870 PR middle-end/32003
18871 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
18872 removed in a prior commit.
18874 2017-02-15 Bin Cheng <bin.cheng@arm.com>
18876 PR tree-optimization/79347
18877 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
18878 counters during peeling.
18880 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
18882 * Makefile.in (site.exp): Remove "set ISLVER".
18884 2017-02-15 Jakub Jelinek <jakub@redhat.com>
18887 * real.c (real_from_integer): Call real_convert even for decimal.
18889 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
18892 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
18894 2017-02-14 Andrew Pinski <apinski@cavium.com>
18896 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
18897 cores and change the partno/implementer to be correct.
18898 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
18899 the 'B" as the implementer.
18900 * config/aarch64/aarch64-tune.md: Regenerate.
18902 2017-02-14 Carl Love <cel@us.ibm.com>
18904 * config/rs6000/rs6000.c: Add case statement entry to make the
18905 xvcvuxdsp built-in argument unsigned.
18906 * config/rs6000/vsx.md: Fix the source and return operand types so they
18907 match the instruction definitions from the ISA document. Fix typo
18908 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
18911 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
18914 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
18915 member early_clobber_alts.
18916 * lra-lives.c (reg_early_clobber_p): New.
18917 (process_bb_lives): Use it.
18918 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
18919 (debug_operand_data): Initialize early_clobber_alts.
18920 (setup_operand_alternative): Set up early_clobber_alts.
18921 (collect_non_operand_hard_regs): Ditto. Pass early clobber
18922 alternatives to new_insn_reg.
18923 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
18925 (lra_update_insn_regno_info): Pass the new arg.
18927 2017-02-14 Jakub Jelinek <jakub@redhat.com>
18929 PR middle-end/79505
18930 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
18931 (new_oacc_loop_raw): Don't clear already cleared fields.
18934 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
18935 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
18936 _mm512_prefetch_i64gather_ps): New inline functions and macros.
18938 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
18941 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
18943 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
18946 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
18947 the extra instruction to the right place to store 128-bit constant
18950 2017-02-14 Martin Sebor <msebor@redhat.com>
18952 PR middle-end/79448
18953 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
18954 warning for strings of unknown length.
18956 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
18958 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
18960 2017-02-14 Jeff Law <law@redhat.com>
18963 * ira-costs.c (scan_one_insn): Initialize register move costs
18964 for pseudos seen in USE/CLOBBER insns.
18966 PR tree-optimization/79095
18967 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
18968 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
18969 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
18970 if the operands are known to be not equal, then the resulting range
18972 (intersect_ranges): If the new range is ~[0,0] and the old range is
18973 wide, then prefer ~[0,0].
18974 * tree-vrp.c (overflow_comparison_p_1): New function.
18975 (overflow_comparison_p): New function.
18976 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
18977 if NAME is used in an overflow test.
18978 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
18979 overflow check that can be expressed as an equality test, then adjust
18980 ops to be that equality test.
18982 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18984 * config/s390/s390-builtin-types.def: Remove flags argument.
18985 * config/s390/s390.c (s390_init_builtins): Likewise.
18987 2017-02-14 Martin Liska <mliska@suse.cz>
18989 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
18990 vector. Fix trailing white spaces.
18992 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
18994 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
18997 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18999 PR rtl-optimization/68664
19000 * config/arm/arm.c (arm_sched_can_speculate_insn):
19001 New function. Declare prototype.
19002 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19004 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19006 PR rtl-optimization/68664
19007 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19009 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19011 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19013 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19014 max skip bytes for function, loop and jump.
19016 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19018 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19019 ABS_EXPR for gimple dump.
19021 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19024 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19026 PR tree-optimization/79408
19027 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19028 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19029 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19030 also if rhs1 is INTEGER_CST.
19032 2017-02-14 Richard Biener <rguenther@suse.de>
19034 PR middle-end/79432
19035 * tree-into-ssa.c (insert_phi_nodes): When the function can
19036 have abnormal edges rewrite SSA names with broken use-def
19037 dominance out of SSA and register them for PHI insertion.
19039 2017-02-13 Martin Sebor <msebor@redhat.com>
19041 PR middle-end/79496
19042 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19043 clearing info.nowrite flag when snprintf size argument is a range.
19045 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19047 * cprop.c (cprop_jump): Add missing space in string literal.
19048 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19049 (get_constraint_for_component_ref): Likewise.
19050 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19051 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19052 * lra-constraints.c (process_alt_operands): Likewise.
19053 * ipa-inline.c (inline_small_functions): Likewise.
19054 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19055 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19056 * trans-mem.c (diagnose_tm_1_op): Likewise.
19057 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19058 (grid_parallel_clauses_gridifiable): Likewise.
19060 * config/nvptx/mkoffload.c (process): Add space in between
19063 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19064 "MOD4_SSE_REGS" and "ALL_REGS".
19066 * spellcheck.c (test_data): Add , in between "foo" and "food".
19068 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19071 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19072 boundary crossing check and subsequent code generation agree.
19074 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19076 * config/aarch64/aarch64.c (has_memory_op): Delete.
19077 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19080 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19082 PR rtl-optimization/79388
19083 PR rtl-optimization/79450
19084 * combine.c (distribute_notes): When removing TEM_INSN for which
19085 corresponding dest has last value recorded, invalidate that last
19088 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19090 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19091 of explicit '@'. Add missing assembly comment marker on branch costs
19094 2017-02-13 Nathan Sidwell <nathan@acm.org>
19096 * gengtype-lex.l (<in_struct>): Add '/'.
19098 2017-02-13 Martin Liska <mliska@suse.cz>
19101 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19103 2017-02-13 Richard Biener <rguenther@suse.de>
19105 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19107 * configure: Re-generate.
19108 * config.in: Likewise.
19109 * graphite-dependences.c: Simplify as if
19110 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19111 * graphite-isl-ast-to-gimple.c: Likewise.
19112 * graphite-optimize-isl.c: Likewise.
19113 * graphite-poly.c: Likewise.
19114 * graphite-sese-to-poly.c: Likewise.
19115 * graphite.h: Likewise.
19116 * toplev.c: Include isl/version.h and use isl_version () for
19117 printing the ISL version.
19118 * doc/install.texi: Update ISL requirement.
19120 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19122 * doc/standards.texi (Standards): Update reference to
19125 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19127 * doc/extend.texi (Named Address Spaces): sourceware.org now
19129 * doc/install.texi (Binaries): Ditto.
19132 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19134 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19135 standard terminology "stringize"/"stringizing" throughout.
19136 * doc/cppinternals.texi: Likewise.
19138 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19140 * doc/extend.texi: Fix some spelling mistakes and typos.
19141 * doc/invoke.texi: Likewise.
19143 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19146 * params.def (inline-min-speedup) Change from 10 to 8.
19148 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19150 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19153 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19156 * ipa-inline-analysis.c (get_minimal_bb): New function.
19157 (record_modified): Use it.
19158 (remap_edge_change_prob): Handle also ancestor functions.
19160 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19162 * doc/contrib.texi (Contributors): Remove broken link into
19163 the Mauve CVS repository.
19165 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19167 PR middle-end/79454
19168 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19169 result computation whenever lhs doesn't have vector mode, not
19170 just when it has BLKmode.
19172 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19174 * doc/makefile.texi (profiledbootstrap): Refer to the
19175 installation instructions only in textual form.
19177 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19180 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19182 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19184 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19185 (Specific): Update mingw-w64 reference.
19187 (Specific): Remove broken link to Renesas RX processor.
19189 2017-02-10 Richard Biener <rguenther@suse.de>
19191 * toplev.c (process_options): Do not mention obsolete graphite
19192 options when printing sorry message about missing graphite support.
19193 Mention -floop-nest-optimize.
19195 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19197 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19198 (vtst_p16): Likewise.
19199 (vtstq_p8): Likewise.
19200 (vtstq_p16): Likewise.
19202 (vtstq_p64): Likewise.
19203 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19204 (vset_lane_p64): New.
19205 (vsetq_lane_p64): New.
19207 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19209 PR tree-optimization/79411
19210 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19211 stmt operands are SSA_NAMEs used in abnormal phis.
19212 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19215 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19218 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19221 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19223 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19225 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19227 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19231 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19232 not arbitrary TREE_CONSTANT.
19235 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19236 "omp declare target link" attribute unless is_global_var.
19237 * omp-offload.c (find_link_var_op): Likewise.
19239 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19240 Chung-Lin Tang <cltang@codesourcery.com>
19242 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19244 (gimplify_adjust_omp_clauses): Don't delete TILE.
19245 (gimplify_omp_for): Deal with TILE.
19246 * internal-fn.c (expand_GOACC_TILE): New function.
19247 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19249 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19250 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19252 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19253 avoid DIV for outermost collapse var.
19254 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19255 Remove out of date comments, fix whitespace.
19256 * omp-general.c (omp_extract_for_data): Deal with tiling.
19257 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19258 adjust OLF_DIM_BASE value.
19259 (struct omp_for_data): Add tiling field.
19260 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19261 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19262 for auto loops. Remove default auto determining, moved to
19263 oacc_loop_fixed_partitions.
19264 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19265 stmts, add e_mask field.
19266 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19267 (oacc_thread_numbers): Use oacc_dim_call.
19268 (oacc_xform_tile): New.
19269 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19270 (finish_oacc_loop): Adjust for ifns vector.
19271 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19272 add case for GOACC_TILE fns.
19273 (oacc_loop_xform_loop): Delete.
19274 (oacc_loop_process): Iterate over call list directly, and add
19275 handling for GOACC_TILE fns.
19276 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19278 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19279 vector partitioning to outer loops. Assign 2 partitions to loops
19280 when available. Add TILE handling.
19281 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19282 (execite_oacc_device_lower): Process GOACC_TILE fns,
19283 ignore unknown specs.
19284 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19285 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19286 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19288 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19290 * configure.ac (ACX_BUGURL): Update.
19291 * configure: Regenerate.
19293 2017-02-09 Richard Biener <rguenther@suse.de>
19295 PR tree-optimization/69823
19296 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19297 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19299 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19301 * config/arc/arc-c.def: Add __NPS400__ definition.
19302 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19303 (TARGET_NPS400): Define.
19305 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19307 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19309 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19310 pointer, arch_info.
19311 (arc_cpu_types): Fill the arch_info field with a pointer into the
19312 arc_arch_types table.
19313 (arc_selected_cpu): Declare.
19314 * config/arc/arc.c (arc_selected_cpu): Make global.
19315 (arc_selected_arch): Delete.
19316 (arc_base_cpu): Delete.
19317 (arc_override_options): Remove references to deleted variables,
19318 update access to arch information.
19319 (ARC_OPT): Update access to arch information.
19320 (ARC_OPTX): Likewise.
19321 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19322 (TARGET_ARC600): Update access to arch information.
19323 (TARGET_ARC601): Likewise.
19324 (TARGET_ARC700): Likewise.
19325 (TARGET_EM): Likewise.
19326 (TARGET_HS): Likewise.
19327 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19330 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19333 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19334 condition/operands for integer GE/LE/GEU/LEU operations.
19336 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19338 PR translation/79397
19339 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19342 2017-02-08 Martin Jambor <mjambor@suse.cz>
19345 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19346 whether allocation happened.
19347 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19348 nothing was allocated.
19350 2017-02-08 Jakub Jelinek <jakub@redhat.com>
19352 PR tree-optimization/79408
19353 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
19354 constant, but SSA_NAME with a known integer range, use the minimum
19355 of that range instead of op1 to determine if modulo can be replaced
19356 with its first operand.
19358 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19360 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
19362 2017-02-08 Richard Biener <rguenther@suse.de>
19364 PR tree-optimization/71824
19365 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19366 Check all loops contained in the merged region.
19368 2017-02-07 Andrew Pinski <apinski@cavium.com>
19370 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
19372 2017-02-07 Andrew Pinski <apinski@cavium.com>
19374 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
19375 (thunderxt88): Likewise.
19376 (thunderxt81): Disable LSE and change v8.1 to v8.
19377 (thunderxt83): Likewise.
19379 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19380 Richard Biener <rguenther@suse.de>
19382 PR middle-end/79399
19383 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
19384 type from int to size_t.
19385 * ira-costs.c (struct_costs_size): Change type from int to size_t.
19387 2017-02-07 Jakub Jelinek <jakub@redhat.com>
19389 PR rtl-optimization/79386
19390 * cprop.c (bypass_conditional_jumps): Initialize
19391 bypass_last_basic_block already before splitting bbs after
19392 unconditional traps...
19393 (bypass_conditional_jumps): ... rather than here.
19396 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
19397 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
19398 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
19399 fix -masm=intel patterns.
19401 2017-02-07 Richard Biener <rguenther@suse.de>
19403 PR tree-optimization/79256
19404 PR middle-end/79278
19405 * builtins.c (get_object_alignment_2): Use min_align_of_type
19406 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
19407 and ADJUST_FIELD_ALIGN.
19409 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
19411 * doc/tm.texi: Regenerate.
19412 * stor-layout.c (layout_decl): Adjust.
19413 (update_alignment_for_field): Likewise.
19414 (place_field): Likewise.
19415 (min_align_of_type): Likewise.
19416 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
19417 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
19418 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
19419 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
19420 * config/frv/frv.c (frv_adjust_field_align): Likewise.
19421 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
19422 * config/i386/i386.c (x86_field_alignment): Likewise.
19423 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
19424 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
19425 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
19426 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
19427 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
19428 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
19432 2017-01-30 Richard Biener <rguenther@suse.de>
19434 PR tree-optimization/79256
19435 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
19436 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
19439 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
19441 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
19442 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
19443 builtins to SImode and emit a zero-extend, if necessary.
19445 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19447 * docs/invoke.texi (RISC-V Options): Alphabetize.
19449 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19451 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
19454 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
19456 * config/riscv/riscv.c: New file.
19457 * common/config/riscv/riscv-common.c: Likewise.
19458 * config.gcc: Likewise.
19459 * config/riscv/constraints.md: Likewise.
19460 * config/riscv/elf.h: Likewise.
19461 * config/riscv/generic.md: Likewise.
19462 * config/riscv/linux.h: Likewise.
19463 * config/riscv/multilib-generator: Likewise.
19464 * config/riscv/peephole.md: Likewise.
19465 * config/riscv/pic.md: Likewise.
19466 * config/riscv/predicates.md: Likewise.
19467 * config/riscv/riscv-builtins.c: Likewise.
19468 * config/riscv/riscv-c.c: Likewise.
19469 * config/riscv/riscv-ftypes.def: Likewise.
19470 * config/riscv/riscv-modes.def: Likewise.
19471 * config/riscv/riscv-opts.h: Likewise.
19472 * config/riscv/riscv-protos.h: Likewise.
19473 * config/riscv/riscv.h: Likewise.
19474 * config/riscv/riscv.md: Likewise.
19475 * config/riscv/riscv.opt: Likewise.
19476 * config/riscv/sync.md: Likewise.
19477 * config/riscv/t-elf-multilib: Likewise.
19478 * config/riscv/t-linux: Likewise.
19479 * config/riscv/t-linux-multilib: Likewise.
19480 * config/riscv/t-riscv: Likewise.
19481 * configure.ac: Likewise.
19482 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
19483 Waterman as RISC-V maintainers.
19484 * doc/install.texi: Add RISC-V entries.
19485 * doc/invoke.texi: Add RISC-V options section.
19486 * doc/md.texi: Add RISC-V constraints section.
19487 * configure: Regenerated.
19489 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19492 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
19493 false values to be constant vectors with all 0 or all 1 bits set.
19494 (vcondu<mode><mode>): Likewise.
19495 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
19497 (fpmask_comparison_operator): Update comment.
19498 (vecint_comparison_operator): New predicate.
19499 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
19500 vector conditionals when the true and false values are constant
19501 vectors with all 0 bits or all 1 bits set.
19503 2017-02-06 Martin Sebor <msebor@redhat.com>
19505 PR tree-optimization/79376
19506 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
19508 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
19510 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
19511 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
19512 to simplify split condition.
19514 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19516 * omp-expand.c (oxpand_omp_atomic_fetch_op,
19517 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
19520 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
19522 PR rtl-optimization/68664
19523 * target.def (can_speculate_insn): New hook.
19524 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
19525 * doc/tm.texi: Regenerate.
19526 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
19527 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
19528 (rs6000_sched_can_speculate_insn): New function.
19530 2017-02-06 Jakub Jelinek <jakub@redhat.com>
19532 PR tree-optimization/79284
19533 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
19534 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
19535 vectorizable_mask_load_store, vectorizable_operation,
19536 vect_is_simple_cond, get_same_sized_vectype): Use it instead
19537 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
19538 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
19539 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
19540 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
19541 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19542 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
19543 is_gimple_assign (stmt). Replace another such test with
19544 is_gimple_assign (stmt).
19546 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19549 * config/avr/avr.c (rtl-iter.h): Include it.
19550 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
19551 (avr_legitimate_combined_insn): ...and implementation.
19553 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
19555 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
19556 * config/s390/s390.c (s390_const_operand_ok)
19557 (s390_canonicalize_comparison, s390_extract_part)
19558 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
19559 (s390_contiguous_bitmask_p, s390_rtx_costs)
19560 (legitimize_pic_address): Likewise.
19561 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
19562 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
19563 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
19564 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
19565 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
19567 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
19569 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
19570 REGNO($0) == REGNO($1).
19572 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19574 * config/s390/linux.h(SIZE_TYPE): Add comment.
19576 2017-02-06 Julian Brown <julian@codesourcery.com>
19577 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19578 Virendra Pathak <virendra.pathak@broadcom.com>
19580 * config/aarch64/aarch64-cores.def: Change the scheduler
19582 * config/aarch64/aarch64.md: Include thunderx2t99.md.
19583 * config/aarch64/thunderx2t99.md: New file.
19585 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19587 * doc/standards.texi (Go Language): Update link to language
19590 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19592 * tree-eh.c (lower_resx): Sanitize profile.
19593 (cleanup_empty_eh_move_lp): Likewise.
19595 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
19598 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
19600 * cfgloopmanip.h (loop_version): Update prototype.
19601 * modulo-sched.c (sms_schedule): Update call of loop_version.
19602 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
19603 * tree-parloops.c (gen_parallel_loop): Likewise.
19604 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
19605 * tree-ssa-loop-split.c (split_loop): Likewise.
19606 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
19607 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19609 2017-02-05 Martin Liska <mliska@suse.cz>
19612 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
19614 (print_operand_address): Initialize a struct to zero.
19616 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19618 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
19619 garbage collector only in textual form.
19621 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
19623 * doc/extend.texi (x86 specific memory model extensions for
19624 transactional memory): Simplify a phrase.
19626 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
19629 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
19630 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
19631 (atomic_storedi_1): Likewise.
19633 2017-02-04 Jakub Jelinek <jakub@redhat.com>
19635 PR tree-optimization/79338
19636 * tree-parloops.c (gather_scalar_reductions): Don't call
19637 vect_analyze_loop_form for loop->inner before destroying loop's
19640 2017-02-03 Martin Sebor <msebor@redhat.com>
19642 PR tree-optimization/79327
19643 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
19644 when precision has resulted in leading zeros.
19645 (format_integer): Adjust the likely counter to assume an unknown
19646 argument that may be zero is non-zero.
19648 2017-02-03 Jason Merrill <jason@redhat.com>
19651 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
19652 avoid copying non-taken branch.
19654 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19656 PR tree-optimization/79340
19657 * tree-vect-loop.c (vectorizable_reduction): Release
19658 vec_defs elements after safe_splicing them into other vectors.
19661 PR tree-optimization/79327
19662 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
19663 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
19665 (format_integer): Use wide_int_to_tree instead of build_int_cst
19666 + to_?hwi. If argmin is NULL, just set argmin and argmax to
19667 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
19668 of shortest and longest sequence.
19670 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
19672 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
19673 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
19675 2017-02-03 Walter Lee <walt@tilera.com>
19678 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
19679 after initial stackframe link reg save.
19680 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
19682 2017-02-03 Jakub Jelinek <jakub@redhat.com>
19685 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
19686 wu for stxssp alternative.
19688 2017-02-03 Martin Sebor <msebor@redhat.com>
19690 PR tree-optimization/79352
19691 * gimple-fold.c (get_range_strlen): Add argument.
19692 (get_range_strlen): Change return type to bool.
19693 (get_maxval_strlen): Pass in a dummy argument.
19694 * gimple-fold.h (get_range_strlen): Change return type to bool.
19695 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
19696 * tree.h (array_at_struct_end_p): Add argument.
19697 * tree.c (array_at_struct_end_p): Handle it.
19699 2017-02-03 Martin Liska <mliska@suse.cz>
19702 * multiple_target.c (create_dispatcher_calls): Redirect edge
19703 from a caller of a dispatcher.
19704 (expand_target_clones): Make the clones local.
19705 (ipa_target_clone): Do both target clones and resolvers.
19706 (ipa_dispatcher_calls): Remove the pass.
19707 (pass_dispatcher_calls::gate): Likewise.
19708 (make_pass_dispatcher_calls): Likewise.
19709 * passes.def (pass_target_clone): Put as very first IPA early
19712 2017-02-03 Martin Liska <mliska@suse.cz>
19714 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
19715 in case of a function with ifunc attribute.
19717 2017-02-03 Martin Liska <mliska@suse.cz>
19719 * cgraph.c (cgraph_node::dump): Dump function version info.
19720 * symtab.c (symtab_node::dump_base): Add missing new line.
19722 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
19724 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
19725 (ifcombine_ifandif): Use it.
19727 2017-02-03 Martin Liska <mliska@suse.cz>
19729 * doc/invoke.texi: Document default value for
19730 use-after-scope-direct-emission-threshold.
19732 2017-02-03 Martin Liska <mliska@suse.cz>
19734 PR tree-optimization/79339
19735 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
19736 (format_floating): Likewise.
19738 2017-02-03 Martin Liska <mliska@suse.cz>
19741 * ipa-prop.c (ipa_node_params_t::insert): Remove current
19743 (ipa_node_params_t::remove): Likewise.
19744 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
19745 initialization from removed ipa_node_params_t::insert.
19746 (ipa_node_params::~ipa_node_params): Move from removed
19747 ipa_node_params_t::release.
19748 * symbol-summary.h (symbol_summary::m_released): New member.
19749 Do not release a summary twice. Do not allow to call finalizer
19750 for types of a summary that live in GGC memory.
19752 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19754 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
19757 2017-02-02 Martin Sebor <msebor@redhat.com>
19759 PR middle-end/79275
19760 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
19761 (format_string): Tighten up the range of output for non-constant
19762 strings and correct the expected range for wide non-constant strings.
19764 2017-02-02 Martin Sebor <msebor@redhat.com>
19766 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
19768 PR middle-end/32003
19769 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
19771 (-fdump-tree-@var): Add to index and document how to come up
19772 with pass-specific option and dump file names.
19773 (-fdump-passes): Clarify where to look for output.
19775 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
19777 PR middle-end/77445
19778 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
19779 statistics of the analyzed path; allow threading for speed when
19780 any of BBs along the path are optimized for speed.
19782 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
19784 PR middle-end/78468
19785 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
19786 settings of the virtual registers.
19789 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
19791 * explow.c (get_dynamic_stack_size): Take known alignment of stack
19792 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
19795 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19797 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
19798 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
19800 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19802 * config/s390/s390.md: Add missing comments with the expanded
19804 * config/s390/vector.md: Likewise.
19805 * config/s390/vx-builtins.md: Likewise.
19807 2017-02-02 Jakub Jelinek <jakub@redhat.com>
19810 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
19811 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
19812 conditions on a single line.
19814 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19816 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
19817 __S390_VX__ to __VX__.
19819 2017-02-01 Andrew Pinski <apinski@cavium.com>
19821 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
19822 stmt_info to record_stmt_cost.
19823 (vect_get_known_peeling_cost): Pass stmt_info if known to
19825 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
19826 cpu_vector_cost field into
19827 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
19828 field into vec_int_stmt_cost and vec_fp_stmt_cost.
19829 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
19830 splitting of scalar_stmt_cost and vec_stmt_cost.
19831 (thunderx_vector_cost): Likewise.
19832 (cortexa57_vector_cost): LIkewise.
19833 (exynosm1_vector_cost): Likewise.
19834 (xgene1_vector_cost): Likewise.
19835 (thunderx2t99_vector_cost): Improve after the splitting of the two
19837 (aarch64_builtin_vectorization_cost): Update for the splitting of
19838 scalar_stmt_cost and vec_stmt_cost.
19840 2017-02-01 Torvald Riegel <triegel@redhat.com>
19841 Richard Henderson <rth@redhat.com>
19843 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
19844 conditional on existance of a fast atomic load.
19845 * optabs-query.c (can_atomic_load_p): New function.
19846 * optabs-query.h (can_atomic_load_p): Declare it.
19847 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
19848 no fast atomic load is available for the particular size of access.
19849 (expand_atomic_compare_and_swap): Likewise.
19850 (expand_atomic_load): Likewise.
19851 (expand_atomic_store): Likewise.
19852 (expand_atomic_fetch_op): Likewise.
19853 * testsuite/lib/target-supports.exp
19854 (check_effective_target_sync_int_128): Remove x86 because it provides
19855 no fast atomic load.
19856 (check_effective_target_sync_int_128_runtime): Likewise.
19858 2017-02-01 Richard Biener <rguenther@suse.de>
19860 * graphite.c: Include tree-vectorizer.h for find_loop_location.
19861 (graphite_transform_loops): Provide opt-info for optimized nests.
19862 * tree-parloop.c (parallelize_loops): Provide opt-info for
19863 parallelized loops.
19865 2017-02-01 Richard Biener <rguenther@suse.de>
19867 PR middle-end/79315
19868 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
19869 was not set before.
19871 2017-02-01 Richard Biener <rguenther@suse.de>
19873 PR tree-optimization/71824
19874 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
19875 Verify the loops are valid in the merged SESE region.
19876 (scop_detection::can_represent_loop_1): Check analyzing the
19877 evolution of the number of iterations in the region succeeds.
19879 2017-01-31 Ian Lance Taylor <iant@golang.org>
19881 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
19882 REG_ARGS_SIZE note to 32-bit push insns and call insn.
19884 2017-01-31 David Malcolm <dmalcolm@redhat.com>
19886 PR preprocessor/79210
19887 * input.c (get_substring_ranges_for_loc): Replace line_width
19888 assertion with error-handling.
19890 2017-01-31 Richard Biener <rguenther@suse.de>
19892 PR tree-optimization/77318
19893 * graphite-sese-to-poly.c (extract_affine): Fix assert.
19894 (create_pw_aff_from_tree): Take loop parameter.
19895 (add_condition_to_pbb): Pass loop of the condition to
19896 create_pw_aff_from_tree.
19898 2017-01-31 Jakub Jelinek <jakub@redhat.com>
19900 * config/s390/s390.c (s390_asan_shadow_offset): New function.
19901 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
19903 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
19907 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
19909 (convert_int_to_float128): Likewise.
19910 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
19911 (convert_int_to_float128): Likewise.
19912 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
19913 (UNSPEC_IEEE128_CONVERT): Likewise.
19914 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
19915 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
19916 Use local variables for IBM extended format.
19917 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
19918 (fix_trunc<mode>si2_fprs): Likewise.
19919 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
19920 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
19921 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
19922 to know that we can now have integers of all sizes in vector
19924 (fix<uns>_<mode>di2_hw): Likewise.
19925 (float<uns>_<mode>si2_hw): Likewise.
19926 (fix_<mode>si2_hw): Likewise.
19927 (fixuns_<mode>si2_hw): Likewise.
19928 (float<uns>_<mode>di2_hw): Likewise.
19929 (float_<mode>di2_hw): Likewise.
19930 (float_<mode>si2_hw): Likewise.
19931 (floatuns_<mode>di2_hw): Likewise.
19932 (floatuns_<mode>si2_hw): Likewise.
19933 (xscvqp<su>wz_<mode>): Delete, no longer used.
19934 (xscvqp<su>dz_<mode>): Likewise.
19935 (xscv<su>dqp_<mode>): Likewise.
19936 (ieee128_mfvsrd_64bit): Likewise.
19937 (ieee128_mfvsrd_32bit): Likewise.
19938 (ieee128_mfvsrwz): Likewise.
19939 (ieee128_mtvsrw): Likewise.
19940 (ieee128_mtvsrd_64bit): Likewise.
19941 (ieee128_mtvsrd_32bit): Likewise.
19943 2017-01-31 Martin Liska <mliska@suse.cz>
19946 * ipa-prop.c (ipa_free_all_node_params): Call release method
19947 instead of ~sumbol_summary to not to trigger double times
19950 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
19952 PR tree-optimization/71691
19953 * bitmap.h (class auto_bitmap): New.
19954 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
19955 is_maybe_undefined instead of ssa_undefined_value_p.
19957 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19959 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
19960 __S390_ARCH_LEVEL__ to __ARCH__.
19962 2017-01-31 Jakub Jelinek <jakub@redhat.com>
19964 PR tree-optimization/79267
19965 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
19966 if should_remove_lhs_p is true.
19968 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
19971 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
19972 (add_alignment_attribute): New.
19973 (base_type_die): Add alignment attribute.
19974 (subrange_type_die): Likewise.
19975 (modified_type_die): Likewise.
19976 (gen_array_type_die): Likewise.
19977 (gen_descr_array_type_die: Likewise.
19978 (gen_enumeration_type_die): Likewise.
19979 (gen_subprogram_die): Likewise.
19980 (gen_variable_die): Likewise.
19981 (gen_field_die): Likewise.
19982 (gen_ptr_to_mbr_type_die): Likewise.
19983 (gen_struct_or_union_type_die): Likewise.
19984 (gen_subroutine_type_die): Likewise.
19985 (gen_typedef_die): Likewise.
19986 (base_type_cmp): Compare alignment attribute.
19988 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19991 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
19992 (setb_unsigned) New pattern for setb with CCUNS.
19993 * config/rs6000/rs6000.c (expand_block_compare): Use a different
19994 subfc./subfe sequence to avoid overflow problems. Generate a
19995 shorter sequence with cmpld/setb for power9.
19996 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
19997 for generating subfc. instruction.
19998 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
19999 now uses this instruction.
20001 2017-01-30 Ian Lance Taylor <iant@google.com>
20004 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20005 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20007 2017-01-30 Martin Sebor <msebor@redhat.com>
20009 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20010 Move constant to the right of a relational operator.
20011 (get_mpfr_format_length, format_character, format_string): Ditto.
20012 (should_warn_p, maybe_warn): Same.
20014 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20016 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20019 * asan.c (get_translation_unit_decl): Remove function.
20020 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20022 2017-01-30 Martin Liska <mliska@suse.cz>
20024 PR gcov-profile/79259
20025 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20026 -fprofile-generate.
20028 2017-01-30 Martin Liska <mliska@suse.cz>
20031 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20032 Initialize variables with NULL value.
20034 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20037 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20039 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20041 2017-01-30 Richard Biener <rguenther@suse.de>
20043 PR tree-optimization/79276
20044 * tree-vrp.c (process_assert_insertions): Properly adjust common
20045 when removing a duplicate.
20047 * gcc.dg/torture/pr79276.c: New testcase.
20049 2017-01-30 Richard Biener <rguenther@suse.de>
20051 PR tree-optimization/79256
20052 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20053 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20055 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20057 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20060 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20061 ("*r<noxa>sbg_<mode>_sll_bitmask")
20062 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20063 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20064 Use contiguous_bitmask_nowrap_operand.
20066 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20069 * config/rs6000/altivec.h (vec_xl): Revise #define.
20070 (vec_xst): Likewise.
20072 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20074 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20076 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20078 PR rtl-optimization/79194
20079 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20080 traps before call to bypass_conditional_jumps.
20082 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20084 PR tree-optimization/71374
20085 * lra-constraints.c (check_conflict_input_operands): New.
20086 (match_reload): Use it.
20088 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20091 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20092 account to calculate conflict_set.
20094 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20096 PR rtl-optimization/78559
20097 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20098 other_insn in combine.
20100 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20102 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20103 uint16_type_node for BT_UINT16.
20105 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20107 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20108 "RTL Tests" to menu.
20109 (GIMPLE Tests): New node.
20110 (RTL Tests): New node.
20112 2017-01-27 Richard Biener <rguenther@suse.de>
20114 PR tree-optimization/79245
20115 * tree-loop-distribution.c (distribute_loop): Apply cost
20116 modeling also to detected patterns.
20118 2017-01-27 Richard Biener <rguenther@suse.de>
20120 PR tree-optimization/71433
20121 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20122 (compare_assert_loc): New function.
20123 (process_assert_insertions): Sort and optimize assert locations
20124 to remove duplicates and push down identical assertions on
20125 edges to their destination block.
20127 2017-01-27 Richard Biener <rguenther@suse.de>
20129 PR tree-optimization/79244
20130 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20131 out SSA names even if abnormal.
20133 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20135 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20136 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20137 instead of MPFR_RNDN.
20139 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20142 * arm.c (arm_option_override): Don't call build_target_option_node
20143 until after doing all option overrides.
20144 (arm_valid_target_attribute_tree): Likewise.
20146 2017-01-27 Martin Liska <mliska@suse.cz>
20148 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20149 for {cd}tors and C++ {cd}tors.
20151 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20153 * config/s390/s390.md ("*setmem_long_and")
20154 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20156 2017-01-26 Martin Sebor <msebor@redhat.com>
20158 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20161 2017-01-26 Martin Sebor <msebor@redhat.com>
20163 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20164 HAVE_DFmode before using XFmode or DFmode.
20165 (parse_directive): Avoid using the z length modifier to avoid
20166 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20168 PR middle-end/78703
20169 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20170 to accept adjustment as an array.
20171 (get_int_range): New function.
20172 (struct directive): Make width and prec arrays.
20173 (directive::set_width, directive::set_precision): Call get_int_range.
20174 (format_integer, format_floating): Handle width and precision ranges.
20175 (format_string, parse_directive): Same.
20177 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20180 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20181 comdat_type_p set, just clone them, but keep the children in the
20185 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20186 which have direct callers with -fvar-tracking-assignments enabled
20188 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20189 inside of type units.
20191 2017-01-26 Martin Sebor <msebor@redhat.com>
20193 PR middle-end/78703
20194 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20196 (struct format_result): Replace number_chars, number_chars_min,
20197 and number_chars_max with a single member of struct result_range.
20199 (format_result::operator+=): Adjust.
20200 (struct fmtresult): Remove bounded. Handle likely and unlikely
20202 (fmtresult::adjust_for_width_or_precision): New function.
20203 (fmtresult:type_max_digits): New function.
20204 (bytes_remaining): Handle likely and unlikely counters.
20205 (min_bytes_remaining): Remove.
20206 (format_percent): Simplify.
20207 (format_integer, format_floating): Set likely and unlikely counters.
20208 (get_string_length, format_character, format_string): Same.
20209 (format_plain, should_warn_p): New function.
20210 (maybe_warn): Call should_warn_p. Update diagnostic messages
20211 and handle those for all directives, including plain strings.
20212 (format_directive): Handle likely and unlikely counters.
20213 Remove unnecessary quoting from diagnostics. Add an informational
20215 (add_bytes): Remove.
20216 (pass_sprintf_length::compute_format_length): Simplify.
20217 (try_substitute_return_value): Handle likely and unlikely counters.
20219 2017-01-26 Carl Love <cel@us.ibm.com>
20221 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20222 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20224 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20227 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20228 endianess for subregs into account.
20229 * lra-constraints.c (lra_constraints): Do risky transformations
20230 always on the first iteration.
20231 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20232 last_call_used_reg_set.
20233 (process_bb_lives): Define and use last_call_used_reg_set.
20234 * lra.c (lra): Always continue after lra_constraints on the first
20237 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20239 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20241 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20243 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20245 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20246 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20247 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20248 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20249 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20250 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20251 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20252 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20253 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20255 2017-01-26 Marek Polacek <polacek@redhat.com>
20258 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20259 for the third operand.
20261 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20263 PR middle-end/79236
20264 * omp-low.c (struct omp_context): Add simt_stmt field.
20265 (scan_omp_for): Return omp_context *.
20266 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20267 context to the _simt_ SIMD stmt.
20268 (lower_omp_for): For combined SIMD with sibling _simt_
20269 SIMD, make sure to use the same decls in _looptemp_
20270 clauses as in the sibling.
20272 2017-01-26 David Sherwood <david.sherwood@arm.com>
20274 PR middle-end/79212
20275 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20278 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20281 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20282 emit fld b; fld a; if possible.
20284 * brig-builtins.def: Update copyright years.
20285 * config/arm/arm_acle_builtins.def: Update copyright years.
20287 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20290 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20291 constraint instead of o for the stxsd instruction.
20293 2017-01-25 Carl Love <cel@us.ibm.com>
20295 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20296 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20298 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20300 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20302 2017-01-25 Richard Biener <rguenther@suse.de>
20304 PR tree-optimization/69264
20305 * target.def (vector_alignment_reachable): Improve documentation.
20306 * doc/tm.texi: Regenerate.
20307 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20309 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20310 earlier changes with respect to TYPE_USER_ALIGN.
20311 (vector_alignment_reachable_p): Likewise. Improve dumping.
20313 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20316 * config/arm/arm.md (xordi3): Force constant operand into a register
20319 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20321 * doc/invoke.texi (-fstore-merging): Correct default optimization
20322 levels at which it is enabled.
20323 (-O): Move -fstore-merging from list to...
20326 2017-01-25 Richard Biener <rguenther@suse.de>
20329 * omp-expand.c: Include debug.h.
20330 (expand_omp_taskreg): Make sure to generate early debug before
20331 outlining anything from a function.
20332 (expand_omp_target): Likewise.
20333 (grid_expand_target_grid_body): Likewise.
20335 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20338 * asan.c (get_translation_unit_decl): New function.
20339 (asan_add_global): Extract modules file name from globals
20340 TRANSLATION_UNIT_DECL name.
20342 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20345 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20346 for long calls with APCS frame and VFP.
20348 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20350 * cfg.c (original_copy_tables_initialized_p): New function.
20351 * cfg.h (original_copy_tables_initialized_p): New decl.
20352 * cfgrtl.c (relink_block_chain): Guard the call to
20353 free_original_copy_tables with a call to
20354 original_copy_tables_initialized_p.
20355 * cgraph.h (symtab_node::native_rtl_p): New decl.
20356 * cgraphunit.c (symtab_node::native_rtl_p): New function.
20357 (symtab_node::needed_p): Don't assert for early assembly output
20358 for __RTL functions.
20359 (cgraph_node::finalize_function): Set "force_output" for __RTL
20361 (cgraph_node::analyze): Bail out early for __RTL functions.
20362 (analyze_functions): Update assertion to support __RTL functions.
20363 (cgraph_node::expand): Bail out early for __RTL functions.
20364 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
20366 * function.h (struct function): Update comment for field
20368 * gimple-expr.c: Include "tree-pass.h".
20369 (gimple_has_body_p): Return false for __RTL functions.
20370 * Makefile.in (OBJS): Add run-rtl-passes.o.
20371 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
20373 (gcc::pass_manager::get_clean_slate): New accessor.
20374 * passes.c: Include "insn-addr.h".
20375 (should_skip_pass_p): Add logging. Update logic for running
20376 "expand" to be compatible with both __GIMPLE and __RTL. Guard
20377 property-provider override so it is only done for gimple passes.
20379 (skip_pass): New function.
20380 (execute_one_pass): Call skip_pass when skipping passes.
20381 * read-md.c (md_reader::read_char): Support filtering
20382 the input to a subset of line numbers.
20383 (md_reader::md_reader): Initialize fields
20384 m_first_line and m_last_line.
20385 (md_reader::read_file_fragment): New function.
20386 * read-md.h (md_reader::read_file_fragment): New decl.
20387 (md_reader::m_first_line): New field.
20388 (md_reader::m_last_line): New field.
20389 * read-rtl-function.c (function_reader::create_function): Only
20390 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
20391 curr_properties. Set DECL_INITIAL to a dummy block.
20392 (read_rtl_function_body_from_file_range): New function.
20393 * read-rtl-function.h (read_rtl_function_body_from_file_range):
20395 * run-rtl-passes.c: New file.
20396 * run-rtl-passes.h: New file.
20398 2017-01-24 Jeff Law <law@redhat.com>
20400 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
20403 2017-01-24 Bin Cheng <bin.cheng@arm.com>
20405 PR tree-optimization/79159
20406 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
20407 (record_nonwrapping_iv): Improve boundary using above function if no
20408 value range information.
20410 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
20411 Martin Jambor <mjambor@suse.cz>
20413 * brig-builtins.def: New file.
20414 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
20415 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
20416 (DEF_HSAIL_SAT_BUILTIN): Likewise.
20417 (DEF_HSAIL_INTR_BUILTIN): Likewise.
20418 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
20419 * builtin-types.def (BT_INT8): New.
20420 (BT_INT16): Likewise.
20421 (BT_UINT8): Likewise.
20422 (BT_UINT16): Likewise.
20423 (BT_FN_ULONG): Likewise.
20424 (BT_FN_UINT_INT): Likewise.
20425 (BT_FN_UINT_ULONG): Likewise.
20426 (BT_FN_UINT_LONG): Likewise.
20427 (BT_FN_UINT_PTR): Likewise.
20428 (BT_FN_ULONG_PTR): Likewise.
20429 (BT_FN_INT8_FLOAT): Likewise.
20430 (BT_FN_INT16_FLOAT): Likewise.
20431 (BT_FN_UINT32_FLOAT): Likewise.
20432 (BT_FN_UINT16_FLOAT): Likewise.
20433 (BT_FN_UINT8_FLOAT): Likewise.
20434 (BT_FN_UINT64_FLOAT): Likewise.
20435 (BT_FN_UINT16_UINT32): Likewise.
20436 (BT_FN_UINT32_UINT16): Likewise.
20437 (BT_FN_UINT16_UINT16_UINT16): Likewise.
20438 (BT_FN_INT_PTR_INT): Likewise.
20439 (BT_FN_UINT_PTR_UINT): Likewise.
20440 (BT_FN_LONG_PTR_LONG): Likewise.
20441 (BT_FN_ULONG_PTR_ULONG): Likewise.
20442 (BT_FN_VOID_UINT64_UINT64): Likewise.
20443 (BT_FN_UINT8_UINT8_UINT8): Likewise.
20444 (BT_FN_INT8_INT8_INT8): Likewise.
20445 (BT_FN_INT16_INT16_INT16): Likewise.
20446 (BT_FN_INT_INT_INT): Likewise.
20447 (BT_FN_UINT_FLOAT_UINT): Likewise.
20448 (BT_FN_FLOAT_UINT_UINT): Likewise.
20449 (BT_FN_ULONG_UINT_UINT): Likewise.
20450 (BT_FN_ULONG_UINT_PTR): Likewise.
20451 (BT_FN_ULONG_ULONG_ULONG): Likewise.
20452 (BT_FN_UINT_UINT_UINT): Likewise.
20453 (BT_FN_VOID_UINT_PTR): Likewise.
20454 (BT_FN_UINT_UINT_PTR: Likewise.
20455 (BT_FN_UINT32_UINT64_PTR): Likewise.
20456 (BT_FN_INT_INT_UINT_UINT): Likewise.
20457 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
20458 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
20459 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
20460 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
20461 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
20462 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
20463 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
20464 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
20465 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
20466 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
20467 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
20468 * doc/frontends.texi: List BRIG FE.
20469 * doc/install.texi (Testing): Add BRIG tesring requirements.
20470 * doc/invoke.texi (Overall Options): Mention BRIG.
20471 * doc/standards.texi (Standards): Doucment BRIG HSA version.
20473 2017-01-24 Richard Biener <rguenther@suse.de>
20475 PR translation/79208
20476 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
20478 2017-01-24 Martin Jambor <mjambor@suse.cz>
20481 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
20482 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
20483 and known_contexts.
20485 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
20487 PR middle-end/79123
20488 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
20489 casts from signed to unsigned really don't have a range.
20491 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
20493 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
20494 GMP_RNDx for compatiblity.
20496 2017-01-24 Martin Liska <mliska@suse.cz>
20499 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
20500 that would prevent us to call alloca with -1 as argument.
20502 2017-01-24 Jakub Jelinek <jakub@redhat.com>
20504 * dwarf2out.c (output_compilation_unit_header, output_file_names):
20505 Avoid -Wformat-security warning.
20507 2017-01-23 Andrew Pinski <apinski@cavium.com>
20509 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
20512 2017-01-23 Martin Sebor <msebor@redhat.com>
20514 PR middle-end/78703
20515 * gimple-ssa-sprintf.c (warn_level): New global.
20516 (format_integer): Use it here and throughout the rest of the file.
20517 Use the same switch to compute sign as base.
20518 (maybe_warn): New function.
20519 (format_directive): Factor out warnings into maybe_warn.
20520 Add debugging output. Use warn_level.
20521 (add_bytes): Use warn_level.
20522 (pass_sprintf_length::compute_format_length): Add debugging output.
20523 (try_substitute_return_value): Same.
20524 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
20526 PR middle-end/78703
20527 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
20528 (struct fmtresult, format_integer, format_floating): Adjust.
20529 (fmtresult::fmtresult): Set max correctly in two argument ctor.
20530 (get_string_length, format_string,format_directive): Same.
20531 (pass_sprintf_length::compute_format_length): Same.
20532 (try_substitute_return_value): Simplify slightly.
20534 PR middle-end/78703
20535 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
20536 (fmtresult::operator+=): Outlined.
20537 (struct fmtresult): Add ctors.
20538 (struct conversion_spec): Rename...
20539 (struct directive): ...to this. Add and remove data members.
20540 (directive::set_width, directive::set_precision): New functions.
20541 (format_percent): Use fmtresult ctor.
20542 (get_width_and_precision): Remove.
20543 (format_integer): Make naming changes. Avoid computing width and
20545 (format_floating): Same. Adjust indentation.
20546 (format_character, format_none): New functions.
20547 (format_string): Moved character handling to format_character.
20548 (format_directive): Remove arguments, change return type.
20549 (parse_directive): New function.
20550 (pass_sprintf_length::compute_format_length): Move directive
20551 parsing to parse_directive.
20553 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20555 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
20556 (assign_assembler_name_if_needed): ... this.
20557 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
20558 (assign_assembler_name_if_needed): ... this.
20559 (free_lang_data_in_cgraph): Adjust callers.
20560 * cgraphunit.c (cgraph_node::analyze): Likewise.
20561 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
20564 2017-01-23 Richard Biener <rguenther@suse.de>
20566 PR tree-optimization/79088
20567 PR tree-optimization/79188
20568 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
20569 resetting loop bounds after last path deletion. Reset loop
20570 bounds of the target loop, make code match the comments.
20571 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
20572 Make sure loops need no fixups.
20574 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20576 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
20577 exponent support with double type for first argument.
20578 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
20579 type returned by __builtin_vec_extract_sig,
20580 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
20581 functions from "vector int" to "vector unsigned int" or from
20582 "vector long long int" to "vector unsigned long long int".
20583 Changed type returned by __builtin_vec_extract_exp,
20584 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
20585 functions from "vector int" to "vector unsigned int" or from
20586 "vector long long int" to "vector unsigned long long int".
20587 Changed return type of __builtin_vec_test_data_class,
20588 __builtin_vec_test_data_class_sp, and
20589 __builtin_vec_test_data_class_dp from "vector int" to
20590 "vector bool int" or from "vector long long int" to "vector bool
20591 long long int" and changed second argument type from "unsigned
20592 int" to "int". Added new overloaded function forms "vector float
20593 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
20594 "vector float __builtin_vec_insert_exp_sp (vector float, vector
20595 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
20596 double, vector unsigned long long int)" and "vector double
20597 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
20598 long int)". Changed return type of
20599 __builtin_scalar_test_data_class and
20600 __builtin_scalar_test_data_class_sp and
20601 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
20602 int" and changed second argument from "unsigned int" to "int".
20603 Changed type returned by __builtin_scalar_test_neg,
20604 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
20605 from "int" to "bool int". Added new overloaded function form
20606 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
20607 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
20608 exponent double-precision with floating point first argument.
20609 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
20610 documentation of scalar_test_data_class, scalar_test_neg,
20611 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
20612 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
20613 vec_test_data_class built-in functions to reflect refinements in
20614 their type signatures.
20616 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
20618 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
20620 (aarch64_elf_asm_destructor): Likewise.
20622 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
20624 PR rtl-optimization/78634
20625 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
20626 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
20627 * ifcvt.c (noce_try_cmove): Add missing cost check.
20629 PR rtl-optimization/71724
20630 * combine.c (if_then_else_cond): Look for situations where it is
20631 beneficial to undo the work of one of the recursive calls.
20633 2017-01-23 Bin Cheng <bin.cheng@arm.com>
20635 PR tree-optimization/70754
20636 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
20637 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
20638 combined stmt before it if not NULL.
20639 (combine_chains): Process refs reversely and compute dominance point
20642 2017-01-23 Martin Liska <mliska@suse.cz>
20644 PR tree-optimization/79196
20645 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
20646 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
20648 (strlen_optimize_stmt): Call the renamed function.
20650 2017-01-23 Michael Matz <matz@suse.de>
20652 PR tree-optimization/78384
20653 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
20655 2017-01-23 Richard Biener <rguenther@suse.de>
20657 PR tree-optimization/79186
20658 * tree-vrp.c (register_new_assert_for): Make sure we've seen
20659 both incoming edges before moving an assert.
20661 2017-01-23 Martin Jambor <mjambor@suse.cz>
20663 * ipa-prop.c (load_from_param_1): Removed.
20664 (load_from_unmodified_param): Bits from load_from_param_1 put back
20666 (load_from_param): Removed.
20667 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
20668 with stmt. Reverted back to use of load_from_unmodified_param.
20670 2017-01-23 Martin Jambor <mjambor@suse.cz>
20673 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
20674 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
20675 field a pointer to garbage collected vector, mark lattices and
20676 ipcp_orig_node with GTY((skip)).
20677 (ipa_get_param_count): Adjust to descriptors being a pointer.
20678 (ipa_get_param): Likewise.
20679 (ipa_get_type): Likewise.
20680 (ipa_get_param_move_cost): Likewise.
20681 (ipa_set_param_used): Likewise.
20682 (ipa_get_controlled_uses): Likewise.
20683 (ipa_set_controlled_uses): Likewise.
20684 (ipa_is_param_used): Likewise.
20685 (ipa_node_params_t): Move into garbage collector. New methods insert
20687 (ipa_node_params_sum): Annotate wth GTY(()).
20688 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
20690 (ipa_load_from_parm_agg): Adjust declaration.
20691 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
20692 * ipa-profile.c (ipa_profile): Likewise.
20693 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
20694 (ipa_populate_param_decls): Make descriptors parameter garbage
20696 (ipa_dump_param): Adjust to descriptors being a pointer.
20697 (ipa_alloc_node_params): Likewise.
20698 (ipa_initialize_node_params): Likewise.
20699 (load_from_param_1): Make descriptors parameter garbage collected.
20700 (load_from_unmodified_param): Likewise.
20701 (load_from_param): Likewise.
20702 (ipa_load_from_parm_agg): Likewise.
20703 (ipa_node_params::~ipa_node_params): Removed.
20704 (ipa_free_all_node_params): Remove call to delete operator.
20705 (ipa_node_params_t::insert): New.
20706 (ipa_node_params_t::remove): Likewise.
20707 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
20708 copy known_csts and known_contexts vectors.
20709 (ipa_read_node_info): Adjust to descriptors being a pointer.
20710 (ipcp_modif_dom_walker): Make m_descriptors field garbage
20712 (ipcp_transform_function): Make descriptors variable garbage
20715 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
20717 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
20718 * config/i386/avx512dqintrin.h: Ditto.
20719 * config/i386/avx512fintrin.h: Ditto.
20720 * config/i386/i386.c: Handle new builtins.
20721 * config/i386/i386-builtin.def: Add new builtins.
20722 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
20723 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
20725 2017-01-23 Jakub Jelinek <jakub@redhat.com>
20726 Martin Liska <mliska@suse.cz>
20728 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
20729 * asan.c (asan_expand_poison_ifn): Support stores and use
20730 appropriate ASAN report function.
20731 * internal-fn.c (expand_ASAN_POISON_USE): New function.
20732 * internal-fn.def (ASAN_POISON_USE): Declare.
20733 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
20734 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
20735 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
20736 ASAN_POISON calls w/o LHS.
20737 * tree-ssa.c (execute_update_addresses_taken): Create clobber
20738 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
20739 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
20740 * gimplify.c (asan_poison_variables): Add attribute
20741 use_after_scope_memory to variables that really needs to live
20743 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
20744 having the attribute.
20746 2017-01-23 Martin Liska <mliska@suse.cz>
20748 * asan.c (create_asan_shadow_var): New function.
20749 (asan_expand_poison_ifn): Likewise.
20750 * asan.h (asan_expand_poison_ifn): New declaration.
20751 * internal-fn.c (expand_ASAN_POISON): Likewise.
20752 * internal-fn.def (ASAN_POISON): New builtin.
20753 * sanopt.c (pass_sanopt::execute): Expand
20754 asan_expand_poison_ifn.
20755 * tree-inline.c (copy_decl_for_dup_finish): Make function
20757 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
20758 * tree-ssa.c (is_asan_mark_p): New function.
20759 (execute_update_addresses_taken): Rewrite local variables
20760 (identified just by use-after-scope as addressable) into SSA.
20762 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
20764 * doc/install.texi (Specific): opensource.apple.com uses https
20765 now. Remove trailing slash.
20767 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
20769 * README.Portability: Remove note on an Irix compatibility issue.
20771 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
20773 * gcov.c (INCLUDE_ALGORITHM): Define.
20774 (INCLUDE_VECTOR): Define.
20775 No longer include <vector> and <algorithm> directly.
20777 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
20779 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
20781 * doc/invoke.texi (Code Gen Options): Ditto.
20783 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
20786 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
20788 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
20790 rtl-optimization/79125
20791 * cprop.c (local_cprop_pass): Handle cases where we make an
20792 unconditional trap.
20794 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
20798 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
20799 read from, for big endian.
20801 2017-01-20 Jiong Wang <jiong.wang@arm.com>
20803 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
20804 register pauth builtins for LP64 only.
20806 2017-01-20 Marek Polacek <polacek@redhat.com>
20809 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
20812 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
20814 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
20816 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
20817 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
20818 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
20820 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20823 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
20824 in big-endian mode when they are not a single duplicated value.
20826 2017-01-20 Richard Biener <rguenther@suse.de>
20828 * BASE-VER: Bump to 7.0.1.
20830 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
20832 * omp-low.c (omplow_simd_context): New struct. Use it...
20833 (lower_rec_simd_input_clauses): ...here and...
20834 (lower_rec_input_clauses): ...here to hold common data. Adjust all
20835 references to idx, lane, max_vf, is_simt.
20837 2017-01-20 Graham Markall <graham.markall@embecosm.com>
20839 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
20842 2017-01-20 Martin Jambor <mjambor@suse.cz>
20844 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
20845 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
20847 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
20848 (GTFILES): Rename hsa.c to hsa-common.c.
20849 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
20850 * hsa-dump.c: Likewise.
20851 * hsa-gen.c: Likewise.
20852 * hsa-regalloc.c: Likewise.
20853 * ipa-hsa.c: Likewise.
20854 * omp-expand.c: Likewise.
20855 * omp-low.c: Likewise.
20856 * toplev.c: Likewise.
20858 2017-01-20 Marek Polacek <polacek@redhat.com>
20861 * doc/invoke.texi: Document -Wduplicated-branches.
20862 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
20863 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
20864 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
20865 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
20866 return 0 only when not OEP_LEXICOGRAPHIC.
20867 (fold_build_cleanup_point_expr): Use the expression
20868 location when building CLEANUP_POINT_EXPR.
20869 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
20870 * tree.c (add_expr): Handle error_mark_node.
20872 2017-01-20 Martin Liska <mliska@suse.cz>
20875 * tree-profile.c (init_ic_make_global_vars): Do not call
20877 (gimple_init_gcov_profiler): Likewise.
20879 2017-01-20 Martin Liska <mliska@suse.cz>
20882 * cgraph.h (maybe_create_reference): Remove argument and
20884 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
20886 * ipa-cp.c (create_specialized_node): Likewise.
20887 * symtab.c (symtab_node::maybe_create_reference): Handle
20888 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
20890 2017-01-20 Martin Liska <mliska@suse.cz>
20892 * read-rtl-function.c (function_reader::create_function): Use
20893 build_decl instread of build_decl_stat.
20895 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
20897 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
20898 * config/i386/avx512dqintrin.h: Ditto.
20899 * config/i386/avx512fintrin.h: Ditto.
20900 * config/i386/i386-builtin-types.def: Add new types.
20901 * config/i386/i386.c: Handle new types.
20902 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
20903 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
20904 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
20905 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
20906 (__builtin_ia32_kshiftridi): New.
20907 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
20909 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
20913 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
20914 define to rs6000_init_stack_protect_guard.
20915 (rs6000_init_stack_protect_guard): New function.
20917 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
20918 Yunqiang Su <yunqiang.su@imgtec.com>
20920 * config.gcc (supported_defaults): Add madd4.
20921 (with_madd4): Add validation.
20922 (all_defaults): Add madd4.
20923 * config/mips/mips.opt (mmadd4): New option.
20924 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
20926 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
20928 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
20929 (ISA_HAS_FUSED_MADD4): Likewise.
20930 * doc/invoke.texi (-mmadd4): Document the new option.
20931 * doc/install.texi (--with-madd4): Document the new option.
20933 2017-01-19 Jiong Wang <jiong.wang@arm.com>
20935 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
20936 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
20937 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
20938 (aarch64_init_pauth_hint_builtins): New.
20939 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
20940 (aarch64_expand_builtin): Expand new builtins.
20942 2017-01-19 Jiong Wang <jiong.wang@arm.com>
20944 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
20945 * combine-stack-adj.c (no_unhandled_cfa): Handle
20946 REG_CFA_TOGGLE_RA_MANGLE.
20947 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
20948 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
20949 info for return address signing.
20950 (aarch64_expand_epilogue): Likewise.
20952 2017-01-19 Jiong Wang <jiong.wang@arm.com>
20954 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
20955 * config/aarch64/aarch64-protos.h
20956 (aarch64_return_address_signing_enabled): New declaration.
20957 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
20959 (aarch64_expand_prologue): Sign return address before it's pushed onto
20961 (aarch64_expand_epilogue): Authenticate return address fetched from
20963 (aarch64_override_options): Sanity check for ILP32 and ISA level.
20964 (aarch64_attributes): New function attributes for "sign-return-address".
20965 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
20966 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
20967 ("*do_return"): Generate combined instructions according to key index.
20968 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
20969 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
20971 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
20972 * config/aarch64/aarch64.opt (msign-return-address=): New.
20973 * doc/extend.texi (AArch64 Function Attributes): Documents
20974 "sign-return-address=".
20975 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
20977 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
20979 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
20980 overall option summary.
20982 2017-01-19 Jiong Wang <jiong.wang@arm.com>
20984 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
20985 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
20986 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
20987 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
20989 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
20991 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
20992 -mpower9-minmax by default for -mcpu=power9.
20993 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
20994 128-bit floating point.
20996 2017-01-20 Alan Modra <amodra@gmail.com>
20998 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
20999 optimizing for size.
21001 2017-01-20 Alan Modra <amodra@gmail.com>
21004 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21005 for strcmp and strncmp from corresponding builtin decl.
21007 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21009 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21010 instead of i386/rtems-64.h.
21011 * config/i386/rtems-64.h: Remove.
21013 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21017 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21019 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21021 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21023 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21024 Change int to HOST_WIDE_INT.
21025 * config/aarch64/aarch64-protos.h
21026 (aarch64_simd_gen_const_vector_dup): Likewise.
21027 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21029 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21031 * langhooks-def.h (lhd_type_for_size): New decl.
21032 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21033 * langhooks.c (lhd_type_for_size): New function, taken from
21036 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21038 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21039 define_bypass for CR latency.
21040 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21041 (power9-alu2): Add define_bypass for CR latency.
21043 (power9-mul): Update insn latency.
21044 (power9-mul-compare): Update insn latency, bypass latency and remove
21047 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21049 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21051 * config/aarch64/aarch64.md
21052 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21053 aarch64_nopcrelative_literal_loads.
21054 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21056 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21058 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21059 TARGET_LOONGSON_3A.
21060 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21062 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21065 * config.gcc (supported_defaults): Add lxc1-sxc1.
21066 (with_lxc1_sxc1): Add validation.
21067 (all_defaults): Add lxc1-sxc1.
21068 * config/mips/mips.opt (mlxc1-sxc1): New option.
21069 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21071 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21072 __mips_no_lxc1_sxc1.
21073 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21074 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21075 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21077 2017-01-19 Richard Biener <rguenther@suse.de>
21079 PR tree-optimization/72488
21080 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21081 sure to restore SSA info.
21082 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21084 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21086 PR rtl-optimization/79121
21087 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21088 of the inner type when shifting an extended value.
21090 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21093 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21094 interposable aliases.
21096 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21099 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21100 Use the evmergelohi instruction.
21101 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21102 (mov_sitf_e500_subreg8_2_be): Likewise.
21103 (mov_sitf_e500_subreg12_2_le): Likewise.
21104 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21105 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21106 (mov_sitf_e500_subreg8_2_le): Likewise.
21107 (mov_sitf_e500_subreg12_2_be): Likewise.
21109 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21111 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21112 attribute from vecsimple to vecperm.
21113 (altivec_vbpermq2): Likewise.
21115 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21118 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21120 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21121 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21122 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21123 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21124 case where N arg is SIZE_MAX.
21125 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21126 (cmpstrsi): Add pattern.
21128 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21130 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21131 __builtin_vec_revb builtins.
21132 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21133 built-in functions to support generation of the ISA 3.0 XXBR<x>
21134 vector byte reverse instructions.
21135 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21136 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21137 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21138 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21139 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21140 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21141 (P9V_BUILTIN_VEC_REVB): Likewise.
21142 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21143 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21144 (p9_xxbrq_v16qi): Likewise.
21145 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21146 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21147 (p9_xxbrh_v8hi): Likewise.
21148 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21149 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21150 vec_revb built-in functions.
21152 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21154 PR rtl-optimization/78952
21155 * config/i386/i386.md (any_extract): New code iterator.
21156 (*insvqi_2): Use any_extract for source operand.
21157 (*insvqi_3): Use any_shiftrt for source operand.
21159 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21161 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21163 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21165 2017-01-18 Matthias Klose <doko@ubuntu.com>
21167 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21169 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21171 * config/rs6000/altivec.h (vec_bperm): Change #define.
21172 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21173 (altivec_vbpermq2): New define_insn.
21174 (altivec_vbpermd): Likewise.
21175 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21176 function interface.
21177 (VBPERMD): Likewise.
21178 (VBPERM): New polymorphic function interface.
21179 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21180 Add entries for P9V_BUILTIN_VEC_VBPERM.
21181 * doc/extend.texi: Add interfaces for vec_bperm.
21183 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21185 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21186 first letter of error messages.
21187 (s390_resolve_overloaded_builtin): Likewise.
21188 * config/s390/s390.c (s390_expand_builtin): Likewise.
21189 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21190 (s390_valid_target_attribute_inner_p): Likewise.
21191 * config/s390/s390.md ("tabort"): Likewise.
21193 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21195 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21196 (ISA_AVOID_DIV_HILO): New macro.
21197 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21198 (ISA_HAS_DDIV): Likewise.
21200 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21202 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21204 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21206 * doc/invoke.texi (fabi-version): Spelling fix.
21208 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21211 * doc/invoke.texi (fabi-version): Mention mangling fix for
21214 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21217 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21219 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21222 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21225 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21227 (stack_protect_test): Ditto.
21228 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21229 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21231 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21232 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21233 -mstack-protector-guard-offset=.
21234 (RS/6000 and PowerPC Options): Ditto.
21236 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21238 * config/i386/i386.h (MASK_CLASS_P): New define.
21239 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21240 there are no registers from different register sets also when
21241 mask registers are used. Update function comment.
21242 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21243 to (*k/*r) and (*k/*km) alternatives.
21245 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21247 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21248 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21249 (EH_RETURN_HANDLER_RTX): New define.
21250 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21251 Force frame pointer in EH return functions.
21252 (aarch64_expand_epilogue): Add barrier for eh_return.
21253 (aarch64_final_eh_return_addr): Remove.
21254 (aarch64_eh_return_handler_rtx): New function.
21255 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21257 (aarch64_eh_return_handler_rtx): New prototype.
21259 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21261 * config/rs6000/altivec.h (vec_rlmi): New #define.
21262 (vec_vrlnm): Likewise.
21263 (vec_rlnm): Likewise.
21264 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21265 (UNSPEC_VRLNM): Likewise.
21266 (VIlong): New mode iterator.
21267 (altivec_vrl<VI_char>mi): New define_insn.
21268 (altivec_vrl<VI_char>nm): Likewise.
21269 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21271 (VRLDNM): Likewise.
21272 (RLNM): New polymorphic function entry.
21273 (VRLWMI): New monomorphic function entry.
21274 (VRLDMI): Likewise.
21275 (RLMI): New polymorphic function entry.
21276 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21277 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21278 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21281 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21284 * dwarf2out.c (field_byte_offset): Restore the
21285 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21286 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21287 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21290 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21293 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21295 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21298 * dwarf2out.c (add_data_member_location_attribute): For constant
21299 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21300 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21301 DW_AT_byte_size attributes.
21303 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21305 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21306 after forcing to constant memory when the code model is medium.
21308 2017-01-17 Julia Koval <julia.koval@intel.com>
21311 * config/i386/avx512fintrin.h
21312 (_mm512_i32gather_ps): Change __addr type to void const*.
21313 (_mm512_mask_i32gather_ps): Ditto.
21314 (_mm512_i32gather_pd): Ditto.
21315 (_mm512_mask_i32gather_pd): Ditto.
21316 (_mm512_i64gather_ps): Ditto.
21317 (_mm512_mask_i64gather_ps): Ditto.
21318 (_mm512_i64gather_pd): Ditto.
21319 (_mm512_mask_i64gather_pd): Ditto.
21320 (_mm512_i32gather_epi32): Ditto.
21321 (_mm512_mask_i32gather_epi32): Ditto.
21322 (_mm512_i32gather_epi64): Ditto.
21323 (_mm512_mask_i32gather_epi64): Ditto.
21324 (_mm512_i64gather_epi32): Ditto.
21325 (_mm512_mask_i64gather_epi32): Ditto.
21326 (_mm512_i64gather_epi64): Ditto.
21327 (_mm512_mask_i64gather_epi64): Ditto.
21328 (_mm512_i32scatter_ps): Change __addr type to void*.
21329 (_mm512_mask_i32scatter_ps): Ditto.
21330 (_mm512_i32scatter_pd): Ditto.
21331 (_mm512_mask_i32scatter_pd): Ditto.
21332 (_mm512_i64scatter_ps): Ditto.
21333 (_mm512_mask_i64scatter_ps): Ditto.
21334 (_mm512_i64scatter_pd): Ditto.
21335 (_mm512_mask_i64scatter_pd): Ditto.
21336 (_mm512_i32scatter_epi32): Ditto.
21337 (_mm512_mask_i32scatter_epi32): Ditto.
21338 (_mm512_i32scatter_epi64): Ditto.
21339 (_mm512_mask_i32scatter_epi64): Ditto.
21340 (_mm512_i64scatter_epi32): Ditto.
21341 (_mm512_mask_i64scatter_epi32): Ditto.
21342 (_mm512_i64scatter_epi64): Ditto.
21343 (_mm512_mask_i64scatter_epi64): Ditto.
21344 * config/i386/avx512pfintrin.h
21345 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21346 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21347 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21348 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21349 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21350 (_mm512_prefetch_i32scatter_ps): Ditto.
21351 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
21352 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
21353 (_mm512_prefetch_i64scatter_pd): Ditto.
21354 (_mm512_prefetch_i64scatter_ps): Ditto.
21355 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
21356 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
21357 * config/i386/avx512vlintrin.h
21358 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
21359 (_mm_mmask_i32gather_ps): Ditto.
21360 (_mm256_mmask_i32gather_pd): Ditto.
21361 (_mm_mmask_i32gather_pd): Ditto.
21362 (_mm256_mmask_i64gather_ps): Ditto.
21363 (_mm_mmask_i64gather_ps): Ditto.
21364 (_mm256_mmask_i64gather_pd): Ditto.
21365 (_mm_mmask_i64gather_pd): Ditto.
21366 (_mm256_mmask_i32gather_epi32): Ditto.
21367 (_mm_mmask_i32gather_epi32): Ditto.
21368 (_mm256_mmask_i32gather_epi64): Ditto.
21369 (_mm_mmask_i32gather_epi64): Ditto.
21370 (_mm256_mmask_i64gather_epi32): Ditto.
21371 (_mm_mmask_i64gather_epi32): Ditto.
21372 (_mm256_mmask_i64gather_epi64): Ditto.
21373 (_mm_mmask_i64gather_epi64): Ditto.
21374 (_mm256_i32scatter_ps): Change __addr type to void*.
21375 (_mm256_mask_i32scatter_ps): Ditto.
21376 (_mm_i32scatter_ps): Ditto.
21377 (_mm_mask_i32scatter_ps): Ditto.
21378 (_mm256_i32scatter_pd): Ditto.
21379 (_mm256_mask_i32scatter_pd): Ditto.
21380 (_mm_i32scatter_pd): Ditto.
21381 (_mm_mask_i32scatter_pd): Ditto.
21382 (_mm256_i64scatter_ps): Ditto.
21383 (_mm256_mask_i64scatter_ps): Ditto.
21384 (_mm_i64scatter_ps): Ditto.
21385 (_mm_mask_i64scatter_ps): Ditto.
21386 (_mm256_i64scatter_pd): Ditto.
21387 (_mm256_mask_i64scatter_pd): Ditto.
21388 (_mm_i64scatter_pd): Ditto.
21389 (_mm_mask_i64scatter_pd): Ditto.
21390 (_mm256_i32scatter_epi32): Ditto.
21391 (_mm256_mask_i32scatter_epi32): Ditto.
21392 (_mm_i32scatter_epi32): Ditto.
21393 (_mm_mask_i32scatter_epi32): Ditto.
21394 (_mm256_i32scatter_epi64): Ditto.
21395 (_mm256_mask_i32scatter_epi64): Ditto.
21396 (_mm_i32scatter_epi64): Ditto.
21397 (_mm_mask_i32scatter_epi64): Ditto.
21398 (_mm256_i64scatter_epi32): Ditto.
21399 (_mm256_mask_i64scatter_epi32): Ditto.
21400 (_mm_i64scatter_epi32): Ditto.
21401 (_mm_mask_i64scatter_epi32): Ditto.
21402 (_mm256_i64scatter_epi64): Ditto.
21403 (_mm256_mask_i64scatter_epi64): Ditto.
21404 (_mm_i64scatter_epi64): Ditto.
21405 (_mm_mask_i64scatter_epi64): Ditto.
21406 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
21407 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
21408 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
21409 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
21410 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
21411 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
21412 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
21413 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
21414 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
21415 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
21416 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
21417 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
21418 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
21419 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
21420 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
21421 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
21422 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
21423 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
21424 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
21425 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
21426 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
21427 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
21428 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
21429 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
21430 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
21431 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
21432 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
21433 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
21434 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
21435 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
21436 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
21437 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
21438 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
21439 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
21440 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
21441 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
21442 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
21443 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
21444 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
21445 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
21446 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
21447 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
21448 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
21449 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
21450 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
21451 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
21452 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
21453 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
21454 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
21455 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
21456 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
21457 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
21458 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
21459 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
21460 definitions accordingly.
21462 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
21463 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21466 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
21469 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
21472 * ira-conflicts.c (ira_build_conflicts): Update total conflict
21473 hard regs for inner regno.
21475 2017-01-17 Martin Liska <mliska@suse.cz>
21478 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
21479 assumption and add comment.
21481 2017-01-17 Nathan Sidwell <nathan@acm.org>
21483 * ipa-visibility.c (localize_node): New function, broken out of ...
21484 (function_and_variable_visibility): ... here. Call it.
21486 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21488 PR middle-end/77445
21489 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
21490 correctly set frequency of oudgoing edge.
21491 (duplicate_thread_path): Fix profile updating.
21493 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21496 * configure.ac: Add GCC_BASE_VER.
21497 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
21498 version from BASE-VER file.
21499 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
21500 (gcc.o): Depend on $(BASEVER).
21501 * common.opt (dumpfullversion): New option.
21502 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
21503 * doc/invoke.texi: Document -dumpfullversion.
21504 * doc/install.texi: Document --with-gcc-major-version-only.
21505 * configure: Regenerated.
21507 2017-01-17 Richard Biener <rguenther@suse.de>
21509 PR tree-optimization/71433
21510 * tree-vrp.c (register_new_assert_for): Merge same asserts
21511 on all incoming edges.
21512 (process_assert_insertions_for): Handle insertions at the
21515 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
21517 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
21518 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
21520 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
21523 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
21526 2017-01-17 Alan Modra <amodra@gmail.com>
21529 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
21530 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
21531 symbolic stack limit when pic.
21533 2017-01-16 Martin Sebor <msebor@redhat.com>
21535 PR tree-optimization/78608
21536 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
21538 2017-01-16 Jeff Law <law@redhat.com>
21541 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
21542 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
21543 for several include directories that may be relative to sysroot.
21544 * config/i386/x-mingw32 (gplus_includedir): Define.
21545 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
21546 (native_system_includedir): Likewise.
21547 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
21548 override if TARGET_SYSTEM_ROOT is defined.
21549 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
21551 PR tree-optimization/79090
21552 PR tree-optimization/33562
21553 PR tree-optimization/61912
21554 PR tree-optimization/77485
21555 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
21556 and computed trims into the dump file.
21558 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21560 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
21562 2017-01-16 Jakub Jelinek <jakub@redhat.com>
21565 * gimplify.c (gimplify_init_constructor): If want_value and
21566 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
21570 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
21571 sequence. Formatting fixes.
21572 (doloop_optimize): Formatting fixes.
21575 * gcc.c (debug_level_greater_than_spec_func): New function.
21576 (static_spec_functions): Add debug-level-gt spec function.
21577 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
21579 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21580 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
21581 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
21582 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
21583 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
21584 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
21586 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
21588 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
21589 QImode fixups to general and mask registers only.
21591 2017-01-16 Carl Love <cel@us.ibm.com>
21593 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
21594 for built-in functions
21595 vector signed char vec_nabs (vector signed char)
21596 vector signed short vec_nabs (vector signed short)
21597 vector signed int vec_nabs (vector signed int)
21598 vector signed long long vec_nabs (vector signed long long)
21599 vector float vec_nabs (vector float)
21600 vector double vec_nabs (vector double)
21601 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
21603 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
21604 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
21605 * doc/extend.texi: Update the documentation file for the new built-in
21608 2017-01-16 Martin Sebor <msebor@redhat.com>
21610 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
21613 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21615 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
21616 UNSPEC_VSX__XXSPLTD to require special splat handling.
21618 2017-01-16 David Malcolm <dmalcolm@redhat.com>
21621 * system.h: Poison strndup.
21623 2017-01-16 Alan Modra <amodra@gmail.com>
21626 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
21629 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
21631 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
21633 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
21635 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
21636 call recog here. Assert that INSN_CODE (insn) is non-negative.
21638 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21641 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
21643 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
21644 in the currently scheduled RTL fragment.
21646 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21648 PR rtl-optimization/78751
21649 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
21652 2017-01-14 Jeff Law <law@redhat.com>
21654 PR tree-optimization/79090
21655 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
21656 variable length stores.
21657 (compute_trims): Delete dead assignment to *trim_tail.
21658 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
21661 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
21663 PR rtl-optimization/78626
21664 PR rtl-optimization/78727
21665 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
21666 of a block, and split such blocks after everything else is finished.
21668 2017-01-14 Alan Modra <amodra@gmail.com>
21671 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
21672 target legitimate_combined_insn.
21673 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
21674 (rs6000_legitimate_combined_insn): New function.
21675 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
21677 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
21678 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
21679 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
21681 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
21683 * doc/frontends.texi (G++ and GCC): Remove references to Java.
21685 2017-01-13 Jeff Law <law@redhat.com>
21687 PR tree-optimization/33562
21688 PR tree-optimization/61912
21689 PR tree-optimization/77485
21690 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
21692 (delete_dead_assignment): Likewise.
21693 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
21694 statement to delete_dead_call and delete_dead_assignment.
21696 2017-01-13 David Malcolm <dmalcolm@redhat.com>
21699 * substring-locations.c (format_warning_va): Strengthen case 1 so
21700 that both endpoints of the substring must be within the format
21701 range for just the substring to be printed.
21703 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
21705 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
21706 * config/i386/i386.c (ix86_target_string): Add missing options
21707 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
21708 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
21709 flags_other and ix86_target_other to flags2_other. Display unknown
21711 (ix86_valid_target_attribute_inner_p): Add missing options and
21712 reorder options by implied ISAs, as in ix86_target_string.
21714 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
21716 * hash-table.h (hash_table::too_empty_p): New function.
21717 (hash_table::expand): Use it.
21718 (hash_table::traverse): Likewise.
21719 (hash_table::empty_slot): Use sizeof (value_type) instead of
21720 sizeof (PTR) to convert bytes to elements. Shrink the table
21721 if the current size is excessive for the current number of
21724 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
21726 * ira-costs.c (record_reg_classes): Break from the inner loop
21727 early once alt_fail is known to be true. Update outer loop
21728 handling accordingly.
21730 2017-01-13 Jeff Law <law@redhat.com>
21732 * tree-ssa-dse.c (decrement_count): New function.
21733 (increment_start_addr, maybe_trim_memstar_call): Likewise.
21734 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
21735 when we know the partially dead statement is a mem* function.
21737 PR tree-optimization/61912
21738 PR tree-optimization/77485
21739 * tree-ssa-dse.c: Include expr.h.
21740 (maybe_trim_constructor_store): New function.
21741 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
21743 PR tree-optimization/33562
21744 PR tree-optimization/61912
21745 PR tree-optimization/77485
21746 * doc/invoke.texi: Document new dse-max-object-size param.
21747 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
21748 * tree-ssa-dse.c: Include params.h.
21749 (dse_store_status): New enum.
21750 (initialize_ao_ref_for_dse): New, partially extracted from
21752 (valid_ao_ref_for_dse, normalize_ref): New.
21753 (setup_live_bytes_from_ref, compute_trims): Likewise.
21754 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
21755 (maybe_trim_partially_dead_store): Likewise.
21756 (maybe_trim_complex_store): Likewise.
21757 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
21758 Track what bytes live from the original store. Return tri-state
21759 for dead, partially dead or live.
21760 (dse_dom_walker): Add constructor, destructor and new private members.
21761 (delete_dead_call, delete_dead_assignment): New extracted from
21763 (dse_optimize_stmt): Make a member of dse_dom_walker.
21764 Use initialize_ao_ref_for_dse.
21766 PR tree-optimization/33562
21767 PR tree-optimization/61912
21768 PR tree-optimization/77485
21769 * sbitmap.h (bitmap_count_bits): Prototype.
21770 (bitmap_clear_range, bitmap_set_range): Likewise.
21771 * sbitmap.c (bitmap_clear_range): New function.
21772 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
21774 2017-01-13 Martin Liska <mliska@suse.cz>
21777 * function.c (set_cfun): Add new argument force.
21778 * function.h (set_cfun): Likewise.
21779 * ipa-inline-transform.c (inline_call): Use the function when
21780 strict alising from is dropped for function we inline to.
21782 2017-01-13 Richard Biener <rguenther@suse.de>
21784 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
21785 for dumping GIMPLE INTEGER_CSTs.
21787 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21789 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
21790 to 201112L since C++17.
21792 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21795 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
21796 if -fsanitize=kernel-address is present.
21798 2017-01-13 Richard Biener <rguenther@suse.de>
21800 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
21801 as _Literal ( type ) number in case usual suffixes do not
21802 preserve all information.
21804 2017-01-13 Richard Biener <rguenther@suse.de>
21806 PR tree-optimization/77283
21807 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
21808 and ssa-iterators.h.
21809 (is_feasible_trace): Implement a cost model based on joiner
21812 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21815 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
21816 char or short to __float128/_Float128 directly.
21818 2017-01-12 Martin Sebor <msebor@redhat.com>
21820 to -Wformat-overflow.
21821 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
21822 (min_bytes_remaining): Same.
21823 (get_string_length): Same.
21824 (format_string): Same.
21825 (format_directive): Same.
21827 (pass_sprintf_length::handle_gimple_call): Same.
21829 2017-01-12 Jakub Jelinek <jakub@redhat.com>
21831 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
21832 info.nowrite calls with no lhs that can't throw. Return bool
21833 whether gsi_remove has been called or not.
21834 (pass_sprintf_length::handle_gimple_call): Return bool whether
21835 try_substitute_return_value called gsi_remove. Formatting fix.
21836 (pass_sprintf_length::execute): Don't use gsi_remove if
21837 handle_gimple_call returned true.
21840 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
21841 be removed due to side-effects, don't remove following barrier nor
21842 turn the successor edge into fallthru edge.
21844 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21847 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
21848 element-reversing loads and stores as not swappable.
21850 2017-01-12 Nathan Sidwell <nathan@acm.org>
21851 Nicolai Stange <nicstange@gmail.com>
21853 * combine.c (try_combine): Don't ignore result of overlap checking
21854 loop. Combine overlap & asm check into single loop.
21856 2017-01-12 Richard Biener <rguenther@suse.de>
21858 * tree-pretty-print.c (dump_generic_node): Provide -gimple
21859 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
21861 2017-01-12 Richard Biener <rguenther@suse.de>
21863 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
21864 and TS_TARGET_OPTION directly derive from TS_BASE.
21865 * tree-core.h (tree_optimization_option): Derive from tree_base.
21866 (tree_target_option): Likewise.
21868 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
21870 * config/i386/i386.c (memory_address_length): Increase len
21871 only when rip_relative_addr_p returns false.
21873 2017-01-11 Julia Koval <julia.koval@intel.com>
21875 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
21876 (OPTION_MASK_ISA_SGX_SET): New.
21877 (ix86_handle_option): Handle OPT_msgx.
21878 * config.gcc: Added sgxintrin.h.
21879 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
21880 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
21881 * config/i386/i386.c (ix86_target_string): Add -msgx.
21883 (ix86_option_override_internal): Handle new options.
21884 (ix86_valid_target_attribute_inner_p): Add sgx.
21885 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
21886 * config/i386/i386.opt: Add msgx.
21887 * config/i386/sgxintrin.h: New file.
21888 * config/i386/x86intrin.h: Add sgxintrin.h.
21890 2017-01-11 Jakub Jelinek <jakub@redhat.com>
21893 * fold-const.c (maybe_nonzero_address): Return 1 for function
21895 (tree_single_nonzero_warnv_p): Don't handle function local objects
21899 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
21902 2017-01-11 David Malcolm <dmalcolm@redhat.com>
21905 * opts.c: Include "spellcheck.h"
21906 (struct string_fragment): New struct.
21907 (struct edit_distance_traits<const string_fragment &>): New
21909 (get_closest_sanitizer_option): New function.
21910 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
21912 2017-01-11 Jakub Jelinek <jakub@redhat.com>
21914 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
21916 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
21917 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
21918 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
21919 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
21920 for initial die_offset if dwarf_split_debug_info.
21921 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
21922 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
21924 (output_skeleton_debug_sections): Formatting fix. Use
21925 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
21926 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
21928 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
21930 * config/arm/cortex-a53.md: Add bypasses for
21931 cortex_a53_r2f_cvt.
21932 (cortex_a53_r2f): Only use for transfers.
21933 (cortex_a53_f2r): Likewise.
21934 (cortex_a53_r2f_cvt): Add reservation for conversions.
21935 (cortex_a53_f2r_cvt): Likewise.
21937 2017-01-11 Tamar Christina <tamar.christina@arm.com>
21939 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
21940 to all inlined functions, change static to extern.
21942 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
21945 * config/arm/arm.c (legitimize_pic_address): Handle reference to
21947 (arm_assemble_integer): Likewise.
21949 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
21951 * config.gcc: Use new awk script to check CPU, FPU and architecture
21952 parameters for --with-... options.
21953 * config/arm/parsecpu.awk: New file
21954 * config/arm/arm-cpus.in: New file.
21955 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
21957 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
21959 * config/arm/t-arm: Update dependency rules.
21960 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
21961 of processing .def files.
21962 * config/arm/genopt.sh: Deleted.
21963 * config/arm/gentune.sh: Deleted.
21964 * config/arm/arm-cores.def: Deleted.
21965 * config/arm/arm-arches.def: Deleted.
21966 * config/arm/arm-fpus.def: Deleted.
21967 * config/arm/arm-tune.md: Regenerated.
21968 * config/arm/arm-tables.opt: Regenerated.
21969 * config/arm/arm-cpu.h: New generated file.
21970 * config/arm/arm-cpu-data.h: New generated file.
21971 * config/arm/arm-cpu-cdata.h: New generated file.
21973 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
21976 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
21978 (input_varpool_node): Unpack dynamically_initialized bit.
21980 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
21982 PR rtl-optimization/79032
21983 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
21984 the alignment of the adjusted memory reference against that of MODE,
21985 instead of the alignment of the original memory reference.
21987 2017-01-11 Martin Jambor <mjambor@suse.cz>
21989 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
21991 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
21992 decorated functions.
21994 2017-01-11 Richard Biener <rguenther@suse.de>
21996 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
21997 set range/nonnull info for PHI results. Do not set it on
21998 stmts marked for removal.
22000 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22002 * expr.c (store_field): In the bitfield case, fetch the return value
22003 from the registers before applying a single big-endian adjustment.
22004 Always do a final load for a BLKmode value not larger than a word.
22006 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22009 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22010 that we correctly handle column numbers greater than
22011 LINE_MAP_MAX_COLUMN_NUMBER.
22013 2017-01-10 Martin Sebor <msebor@redhat.com>
22015 PR middle-end/78245
22016 * gimple-ssa-sprintf.c (get_destination_size): Call
22017 {init,fini}object_sizes.
22018 * tree-object-size.c (addr_object_size): Adjust.
22019 (pass_through_call): Adjust.
22020 (pass_object_sizes::execute): Adjust.
22021 * tree-object-size.h (fini_object_sizes): Declare.
22023 2017-01-10 Martin Sebor <msebor@redhat.com>
22025 PR tree-optimization/78775
22026 * builtins.c (get_size_range): Move...
22027 * calls.c: ...to here.
22028 (alloc_max_size): Accept zero argument.
22029 (operand_signed_p): Remove.
22030 (maybe_warn_alloc_args_overflow): Call get_size_range.
22031 * calls.h (get_size_range): Declare.
22033 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22035 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22036 from TI's devices.csv file as of September 2016.
22037 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22039 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22041 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22042 * doc/invoke.texi: Likewise.
22043 * doc/md.texi: Likewise.
22044 * doc/objc.texi: Likewise.
22046 2017-01-10 Joshua Conner <joshconner@google.com>
22048 * config/arm/fuchsia-elf.h: New file.
22049 * config/fuchsia.h: New file.
22050 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22051 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22053 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22055 2016-01-10 Richard Biener <rguenther@suse.de>
22057 PR tree-optimization/79034
22058 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22059 Propagate out degenerate PHIs in the joiner.
22061 2017-01-10 Martin Liska <mliska@suse.cz>
22063 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22064 (sort_congruence_classes_by_decl_uid): Likewise.
22065 (sort_congruence_class_groups_by_decl_uid): Likewise.
22066 (sem_item_optimizer::merge_classes): Sort class, groups in these
22067 classes and members in the groups by DECL_UID of declarations.
22068 This would make merge operations stable.
22070 2017-01-10 Martin Liska <mliska@suse.cz>
22072 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22073 usage of m_classes_vec.
22074 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22075 (sem_item_optimizer::get_group_by_hash): Likewise.
22076 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22077 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22078 (sem_item_optimizer::verify_classes): Likewise.
22079 (sem_item_optimizer::process_cong_reduction): Likewise.
22080 (sem_item_optimizer::dump_cong_classes): Likewise.
22081 (sem_item_optimizer::merge_classes): Likewise.
22082 * ipa-icf.h (congruence_class_hash): Rename from
22083 congruence_class_group_hash. Remove declaration of m_classes_vec.
22085 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22087 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22088 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22089 * config.gcc: Add avx512vpopcntdqintrin.h.
22090 * config/i386/avx512vpopcntdqintrin.h: New.
22091 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22092 * config/i386/i386-builtin-types.def: Add new types.
22093 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22094 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22095 __builtin_ia32_vpopcountq_v8di_mask): New.
22096 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22097 __AVX512VPOPCNTDQ__.
22098 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22099 (PTA_AVX512VPOPCNTDQ): Define.
22100 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22101 TARGET_AVX512VPOPCNTDQ_P): Define.
22102 * config/i386/i386.opt: Add mavx512vpopcntdq.
22103 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22104 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22106 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22108 PR middle-end/77484
22109 * predict.def (PRED_CALL): Set to 67.
22111 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22113 * expr.c (store_field): In the bitfield case, if the value comes from
22114 a function call and is of an aggregate type returned in registers, do
22115 not modify the field mode; extract the value in all cases if the mode
22116 is BLKmode and the size is not larger than a word.
22118 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22121 * config/i386/cpuid.h: Fix undefined behavior.
22123 2017-01-04 Jeff Law <law@redhat.com>
22125 PR tree-optimization/79007
22126 PR tree-optimization/67955
22127 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22128 conservative for pt.null when flag_non_call_exceptions is on.
22130 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22132 PR translation/79019
22133 PR translation/79020
22134 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22135 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22136 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22138 * config/avr/avr.opt (maccumulate-args): Likewise.
22139 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22140 * common.opt (freport-bug): Likewise.
22141 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22142 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22143 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22144 translatable string.
22145 * config/i386/i386.c (function_value_32): Likewise.
22146 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22147 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22149 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22150 * common/config/msp430/msp430-common.c (msp430_handle_option):
22152 * symtab.c (symtab_node::verify_base): Likewise.
22153 * opts.c (set_debug_level): Likewise.
22154 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22155 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22156 missing whitespace to translatable strings.
22157 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22158 * config/sh/superh.h: Likewise.
22159 * config/i386/xopintrin.h: Likewise.
22160 * config/i386/znver1.md: Likewise.
22161 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22162 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22163 * double-int.h (struct double_int): Likewise.
22164 * double-int.c (div_and_round_double): Likewise.
22165 * wide-int.cc: Likewise.
22166 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22167 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22168 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22169 (crossjumps_occurred): ... this.
22170 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22173 PR tree-optimization/78899
22174 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22175 returning bool return struct loop *, NULL for failure and the new
22177 (versionable_outer_loop_p): Don't version outer loop if it has
22178 dont_vectorized bit set.
22179 (tree_if_conversion): When versioning outer loop, ensure
22180 tree_if_conversion is performed also on the inner loop of the
22181 non-vectorizable outer loop copy.
22182 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22183 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22184 prevent vectorization of it.
22185 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22186 the outer loop vectorization of the non-scalar version is attempted
22187 before vectorization of the inner loop in scalar version. If
22188 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22189 vectorization of its inner loop.
22190 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22191 has 2 inner loops, rename also on edges from bb whose single pred
22192 is outer_loop->header. Fix typo in function comment.
22194 2017-01-09 Martin Sebor <msebor@redhat.com>
22197 * asan.c (asan_emit_stack_protection): Increase local buffer size
22198 to avoid snprintf truncation warning.
22200 2017-01-09 Andrew Pinski <apinski@cavium.com>
22202 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22203 to reference thunderx2t99 for the tuning structure
22204 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22206 (thunderx2t99_extra_costs): This.
22207 * config/aarch64/aarch64-tune.md: Regenerate.
22208 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22209 (vulcan_addrcost_table): This.
22210 (vulcan_regmove_cost): Rename to ...
22211 (thunderx2t99_regmove_cost): This.
22212 (vulcan_vector_cost): Rename to ...
22213 (thunderx2t99_vector_cost): this.
22214 (vulcan_branch_cost): Rename to ...
22215 (thunderx2t99_branch_cost): This.
22216 (vulcan_tunings): Rename to ...
22217 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22218 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22220 2017-01-09 Martin Jambor <mjambor@suse.cz>
22224 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22225 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22226 (propagate_vr_accross_jump_function): Use the above function for all
22227 value range computations for pass-through jump functions and type
22228 converasion from explicit value range values.
22229 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22230 parameters from TYPE_ARG_TYPES.
22231 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22232 (ipa_write_node_info): Stream type of the actual argument.
22233 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22235 2017-01-09 Martin Liska <mliska@suse.cz>
22238 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22239 (lookup_compiler): Do not show error message with have_E.
22241 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22243 PR tree-optimization/78938
22244 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22245 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22246 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22247 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22250 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22252 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22255 2017-01-09 Richard Biener <rguenther@suse.de>
22257 PR tree-optimization/78997
22258 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22259 name condition properly.
22261 2017-01-09 Richard Biener <rguenther@suse.de>
22264 * dwarf2out.c (is_cxx): New overload with context.
22265 (is_naming_typedef_decl): Use it.
22267 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22269 * invoke.texi (Option Summary): Correct spacing in option lists
22270 and add line breaks to fix over-long lines.
22272 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22274 PR middle-end/17660
22276 * extend.texi (Common Variable Attributes): Add xref to GCC
22277 Internals manual to explain mode attribute keywords.
22279 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22282 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22283 and Preprocessor Options.
22284 (Options for Linking): Document -pthread here....
22285 (RS/6000 and PowerPC Options): ...not here.
22286 (Solaris 2 Options): ...or here.
22287 * doc/cppopts.texi: Document -pthread.
22289 2017-01-08 Martin Sebor <msebor@redhat.com>
22291 PR middle-end/77708
22292 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22293 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22294 New member functions.
22295 (format_directive): Used them.
22297 (pass_sprintf_length::handle_gimple_call): Same.
22298 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22299 to avoid truncation for any argument.
22300 (extract_affine_mul): Same.
22301 * tree.c (get_file_function_name): Same.
22303 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22305 PR middle-end/77484
22306 * predict.def (PRED_INDIR_CALL): Set to 86.
22308 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22310 PR preprocessor/54124
22311 * doc/cppopts.texi: Reformat -d subtable to list the full name
22312 of the options. Add cross-reference to the docs for the general
22313 compiler -d options.
22314 * doc/invoke.texi (Developer Options): Add cross-reference to the
22315 preprocessor-specific -d option documentation.
22317 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22319 PR preprocessor/13498
22320 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22321 redudant material, and reflect new command-line options.
22322 (System Headers): Likewise.
22324 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22326 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22327 -isystem, and -idirafter. Copy-edit.
22328 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22329 default for -ftrack-macro-expansion. Delete obsolete and
22330 badly-formatted implementation details about -fdebug-cpp output.
22331 * doc/cppwarnopts.texi: Copy-edit.
22333 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22336 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22337 that the transition from a max line width >= 1<<10 to narrower
22338 lines works correctly.
22340 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22342 * doc/options.texi (PerFunction): New.
22343 * opt-functions.awk (switch_flags): Map both Optimization and
22344 PerFunction to CL_OPTIMIZATION.
22345 * opth-gen.awk: Test for PerFunction flag along with
22347 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22348 it only when the latter is present. Skip those that don't in
22349 the hash function generator.
22350 * common.opt (fvar-tracking): Mark as PerFunction instead of
22352 (fvar-tracking-assignments): Likewise.
22353 (fvar-tracking-assignments-toggle): Likewise.
22354 (fvar-tracking-uninit): Likewise.
22356 2017-01-07 Jakub Jelinek <jakub@redhat.com>
22358 PR translation/79018
22359 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
22362 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
22365 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
22366 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
22367 TARGET_LONG_JUMP_TABLE_OFFSETS.
22368 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
22369 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
22370 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
22371 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
22372 * config/m68k/m68k.md (tablejump expander): Likewise.
22373 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
22374 TARGET_LONG_JUMP_TABLE_OFFSETS.
22375 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
22376 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
22378 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
22379 David Holsgrove <david.holsgrove@xilinx.com>
22381 * common/config/microblaze/microblaze-common.c
22382 (TARGET_EXCEPT_UNWIND_INFO): Remove.
22383 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
22385 * config/microblaze/microblaze.c (microblaze_must_save_register)
22386 (microblaze_expand_epilogue, microblaze_return_addr): Handle
22388 (microblaze_eh_return): New function.
22389 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
22390 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
22391 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
22392 * config/microblaze/microblaze.md (eh_return): New pattern.
22394 2017-01-06 Jakub Jelinek <jakub@redhat.com>
22396 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
22397 GCC_DIAGNOSTIC_STRINGIFY): Define.
22399 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
22401 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22403 * config/arm/arm.md (<mcrr>): New.
22405 * config/arm/arm.c (arm_arch5te): New.
22406 (arm_option_override): Set arm_arch5te.
22407 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
22409 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
22410 (arm_mcrr_qualifiers): ... this. New.
22411 (MRRC_QUALIFIERS): Define to...
22412 (arm_mrrc_qualifiers): ... this. New.
22413 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
22415 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
22416 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
22417 (MRRCI, mrrc, MRRC): New.
22418 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
22419 VUNSPEC_MRRC2): New.
22421 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22423 * config/arm/arm.md (<mcr>): New.
22425 * config/arm/arm.c (arm_coproc_builtin_available): Add
22426 support for mcr, mrc, mcr2 and mrc2.
22427 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
22428 (arm_mcr_qualifiers): ... this. New.
22429 (MRC_QUALIFIERS): Define to ...
22430 (arm_mrc_qualifiers): ... this. New.
22431 (MCR_QUALIFIERS): Define to ...
22432 (arm_mcr_qualifiers): ... this. New.
22433 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
22435 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
22436 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
22437 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
22438 VUNSPEC_MRC2): New.
22440 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22442 * config/arm/arm.md (*ldc): New.
22446 * config/arm/arm.c (arm_coproc_builtin_available): Add
22447 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
22448 (arm_coproc_ldc_stc_legitimate_address): New.
22449 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
22450 'qualifier_const_pointer'.
22451 (LDC_QUALIFIERS): Define to...
22452 (arm_ldc_qualifiers): ... this. New.
22453 (STC_QUALIFIERS): Define to...
22454 (arm_stc_qualifiers): ... this. New.
22455 * config/arm/arm-protos.h
22456 (arm_coproc_ldc_stc_legitimate_address): New.
22457 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
22458 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
22459 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
22460 stc2, stcl, stc2l): New.
22461 * config/arm/constraints.md (Uz): New.
22462 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
22463 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
22464 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
22465 VUNSPEC_STC2L): New.
22467 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22469 * config/arm/arm.md (<cdp>): New.
22470 * config/arm/arm.c (neon_const_bounds): Rename this ...
22471 (arm_const_bounds): ... this.
22472 (arm_coproc_builtin_available): New.
22473 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
22474 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
22475 (CDP_QUALIFIERS): Define to...
22476 (arm_cdp_qualifiers): ... this. New.
22478 (arm_expand_builtin_args): Add case for 6 arguments.
22479 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
22480 (arm_const_bounds): ... this.
22481 (arm_coproc_builtin_available): New.
22482 * config/arm/arm_acle.h (__arm_cdp): New.
22484 * config/arm/arm_acle_builtins.def (cdp): New.
22486 * config/arm/iterators.md (CDPI,CDP,cdp): New.
22487 * config/arm/neon.md: Rename all 'neon_const_bounds' to
22488 'arm_const_bounds'.
22489 * config/arm/types.md (coproc): New.
22490 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
22491 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
22492 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
22493 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
22495 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22497 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
22498 (UBINOP_QUALIFIERS): New.
22500 (acle_builtin_data): New. Change comment.
22501 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
22502 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
22503 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
22504 arm_acle_builtins.def.
22505 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
22506 (arm_init_acle_builtins): New.
22507 (CRC32_BUILTIN): Remove.
22508 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
22509 crc32cb, crc32ch and crc32cw.
22510 (arm_init_crc32_builtins): Remove.
22511 (arm_init_builtins): Use arm_init_acle_builtins rather
22512 than arm_init_crc32_builtins.
22513 (arm_expand_acle_builtin): New.
22514 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
22515 * config/arm/arm_acle_builtins.def: New.
22517 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
22519 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
22520 (arm_builtin_datum): ... this.
22521 (arm_init_neon_builtin): Rename to ...
22522 (arm_init_builtin): ... this. Add a new parameters PREFIX
22523 and USE_SIG_IN_NAME.
22524 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
22525 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
22526 'arm_builtin_datum'.
22527 (arm_init_vfp_builtins): Likewise.
22528 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
22529 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
22530 (arm_expand_neon_args): Rename to ...
22531 (arm_expand_builtin_args): ... this. Rename builtin_arg
22532 enum values and differentiate between ARG_BUILTIN_MEMORY
22533 and ARG_BUILTIN_NEON_MEMORY.
22534 (arm_expand_neon_builtin_1): Rename to ...
22535 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
22536 values, arm_expand_builtin_args and add bool parameter NEON.
22537 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
22538 (arm_expand_vfp_builtin): Likewise.
22539 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
22541 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22543 PR middle-end/77484
22544 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
22545 * predict.c (tree_estimate_probability_bb): Reverse direction of
22546 polymorphic call predictor.
22548 2017-01-06 David Malcolm <dmalcolm@redhat.com>
22550 * passes.c (execute_one_pass): Split out pass-skipping logic into...
22551 (determine_pass_name_match): ...this new function and...
22552 (should_skip_pass_p): ...this new function.
22554 2017-01-06 Nathan Sidwell <nathan@acm.org>
22556 * ipa-visibility.c (function_and_variable_visibility): Reformat
22557 comments and long lines. Remove extrneous if.
22558 * symtab.c (symtab_node::make_decl_local): Fix code format.
22559 (symtab_node::set_section_for_node): Fix comment typo.
22561 2017-01-06 Martin Liska <mliska@suse.cz>
22564 * lra-constraints.c: Rename invariant to lra_invariant.
22565 * predict.c (set_even_probabilities): Initialize e to NULL.
22567 2017-01-05 Martin Sebor <msebor@redhat.com>
22569 PR tree-optimization/78910
22570 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
22571 (format_integer): Correct off-by-one error in the handling
22572 of precision with negative numbers in signed conversions..
22574 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
22576 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
22578 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22580 PR tree-optimization/71016
22581 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
22582 factor_out_conditional_conversion. Formatting fix.
22583 (factor_out_conditional_conversion): Add cond_stmt argument.
22584 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
22585 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
22588 2017-01-05 David Malcolm <dmalcolm@redhat.com>
22590 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
22591 read-rtl-function.o, and selftest-rtl.o.
22592 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
22593 (selftest::aarch64_test_loading_full_dump): New function.
22594 (selftest::aarch64_run_selftests): New function.
22595 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
22596 selftest::aarch64_run_selftests.
22597 * config/i386/i386.c
22598 (selftest::ix86_test_loading_dump_fragment_1): New function.
22599 (selftest::ix86_test_loading_call_insn): New function.
22600 (selftest::ix86_test_loading_full_dump): New function.
22601 (selftest::ix86_test_loading_unspec): New function.
22602 (selftest::ix86_run_selftests): Call the new functions.
22603 * emit-rtl.c (maybe_set_max_label_num): New function.
22604 * emit-rtl.h (maybe_set_max_label_num): New decl.
22605 * function.c (instantiate_decls): Guard call to
22606 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
22607 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
22609 * gensupport.c (gen_reader::gen_reader): Pass "false"
22610 for new "compact" param of rtx_reader.
22611 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
22612 rather than an empty string for NULL strings.
22613 * read-md.c: Potentially include config.h rather than bconfig.h.
22614 Wrap include of errors.h with #ifdef GENERATOR_FILE.
22615 (have_error): New global, copied from errors.c.
22616 (md_reader::read_name): Rename to...
22617 (md_reader::read_name_1): ...this, adding "out_loc" param,
22618 and converting "missing name or number" to returning false, rather
22620 (md_reader::read_name): Reimplement in terms of read_name_1.
22621 (md_reader::read_name_or_nil): New function.
22622 (md_reader::read_string): Handle "(nil)" by returning NULL.
22623 (md_reader::md_reader): Add new param "compact".
22624 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
22625 (md_reader::read_file): New method.
22626 * read-md.h (md_reader::md_reader): Add new param "compact".
22627 (md_reader::read_file): New method.
22628 (md_reader::is_compact): New accessor.
22629 (md_reader::read_name): Convert return type from void to file_location.
22630 (md_reader::read_name_or_nil): New decl.
22631 (md_reader::read_name_1): New decl.
22632 (md_reader::m_compact): New field.
22633 (noop_reader::noop_reader): Pass "false" for new "compact" param
22635 (rtx_reader::rtx_reader): Add new "compact" param.
22636 (rtx_reader::read_rtx_operand): Make virtual and convert return
22637 type from void to rtx.
22638 (rtx_reader::read_until): New decl.
22639 (rtx_reader::handle_any_trailing_information): New virtual function.
22640 (rtx_reader::postprocess): New virtual function.
22641 (rtx_reader::finalize_string): New virtual function.
22642 (rtx_reader::m_in_call_function_usage): New field.
22643 (rtx_reader::m_reuse_rtx_by_id): New field.
22644 * read-rtl-function.c: New file.
22645 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
22646 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
22647 (selftest::verify_three_block_rtl_cfg): New decl.
22648 * read-rtl-function.h: New file.
22649 * read-rtl.c: Potentially include config.h rather than bconfig.h.
22650 For host, include function.h, memmodel.h, and emit-rtl.h.
22651 (one_time_initialization): New function.
22652 (struct compact_insn_name): New struct.
22653 (compact_insn_names): New array.
22654 (find_code): Handle insn codes in compact dumps.
22655 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
22656 (bind_subst_iter_and_attr): Likewise.
22657 (add_condition_to_string): Likewise.
22658 (add_condition_to_rtx): Likewise.
22659 (apply_attribute_uses): Likewise.
22660 (add_current_iterators): Likewise.
22661 (apply_iterators): Likewise.
22662 (initialize_iterators): Guard usage of apply_subst_iterator with
22663 #ifdef GENERATOR_FILE.
22664 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
22665 (md_reader::read_mapping): Likewise.
22666 (add_define_attr_for_define_subst): Likewise.
22667 (add_define_subst_attr): Likewise.
22668 (read_subst_mapping): Likewise.
22669 (check_code_iterator): Likewise.
22670 (rtx_reader::read_rtx): Likewise. Move one-time initialization
22672 (one_time_initialization): New function.
22673 (rtx_reader::read_until): New method.
22674 (read_flags): New function.
22675 (parse_reg_note_name): New function.
22676 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
22677 Handle reuse_rtx ids.
22678 Wrap iterator lookup within #ifdef GENERATOR_FILE.
22679 Add parsing support for RTL dumps, mirroring the special-cases in
22680 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
22681 values, and calling handle_any_trailing_information.
22682 (rtx_reader::read_rtx_operand): Convert return type from void
22683 to rtx, returning return_rtx. Handle case 'e'. Call
22684 finalize_string on XSTR and XTMPL fields.
22685 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
22686 "(nil)" values were omitted. Call the postprocess vfunc on the
22688 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
22689 class ctor. Initialize m_in_call_function_usage. Call
22690 one_time_initialization.
22691 * rtl-tests.c (selftest::test_uncond_jump): Call
22692 set_new_first_and_last_insn.
22693 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
22694 * selftest-rtl.c: New file.
22695 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
22696 (selftest::get_insn_by_uid): New decl.
22697 * selftest-run-tests.c (selftest::run_tests): Call
22698 read_rtl_function_c_tests.
22699 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
22700 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
22703 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
22705 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
22706 operands in a special way. Assert that pos+len <= mode precision.
22708 2017-01-05 Jakub Jelinek <jakub@redhat.com>
22710 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
22711 3 argument Alias with unlimited for the negative form.
22712 (fno-vect-cost-model): Removed.
22714 2017-01-05 Martin Liska <mliska@suse.cz>
22716 * hsa-gen.c (gen_hsa_divmod): New function.
22717 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
22719 2017-01-05 Martin Liska <mliska@suse.cz>
22722 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
22725 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22727 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
22728 small constant length operands.
22730 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22732 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
22733 between loop iterations.
22735 2017-01-05 Martin Liska <mliska@suse.cz>
22738 * gimplify.c (gimplify_decl_expr): Compare to
22739 asan_poisoned_variables instread of checking flags.
22740 (gimplify_target_expr): Likewise.
22741 (gimplify_expr): Likewise.
22742 (gimplify_function_tree): Conditionally initialize
22743 asan_poisoned_variables.
22745 2017-01-04 Jeff Law <law@redhat.com>
22747 PR tree-optimizatin/78812
22748 * rtl.h (contains_mem_rtx_p): Prototype.
22749 * ifcvt.c (containts_mem_rtx_p): Move from here to...
22750 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
22751 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
22752 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
22753 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
22755 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22757 * input.c (assert_char_at_range): Default-initialize actual_range.
22759 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22761 * df-scan.c (df_ref_create_structure): Make regno unsigned,
22762 to match the caller.
22764 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22766 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
22767 insns after final jump in test to emit dummy move.
22769 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22771 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
22772 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
22774 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
22776 * multiple_target.c (create_dispatcher_calls): Init e_next.
22777 * tree-ssa-loop-split.c (split_loop): Init border.
22778 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
22781 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
22786 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
22787 (altivec_register_operand): Do not return true if the operand
22788 contains a SUBREG mixing SImode and SFmode.
22789 (vsx_register_operand): Likewise.
22790 (vsx_reg_sfsubreg_ok): New predicate.
22791 (vfloat_operand): Do not return true if the operand contains a
22792 SUBREG mixing SImode and SFmode.
22793 (vint_operand): Likewise.
22794 (vlogical_operand): Likewise.
22795 (gpc_reg_operand): Likewise.
22796 (int_reg_operand): Likewise.
22797 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
22798 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
22799 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
22801 (rs6000_emit_move_si_sf_subreg): New helper function.
22802 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
22803 fixup SUBREGs involving SImode and SFmode.
22804 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
22805 numbers for the new peephole2 optimization.
22806 (peephole2 for SFmode unions): New peephole2 to optimize cases in
22807 the GLIBC math library that do AND/IOR/XOR operations on single
22808 precision floating point.
22809 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
22810 target macros to say whether we need to avoid SUBREGs mixing
22812 (TARGET_ALLOW_SF_SUBREG): Likewise.
22813 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
22814 (UNSPEC_SI_FROM_SF): Likewise.
22815 (iorxor): Change spacing.
22816 (and_ior_xor): New iterator for AND, IOR, and XOR.
22817 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
22818 (movdi_from_sf_zero_ext): Likewise.
22819 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
22820 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
22821 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
22822 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
22823 (fms<mode>4): Likewise.
22824 (fnma<mode>4): Likewise.
22825 (fnms<mode>4): Likewise.
22826 (nfma<mode>4): Likewise.
22827 (nfms<mode>4): Likewise.
22829 2017-01-04 Marek Polacek <polacek@redhat.com>
22832 * doc/invoke.texi: Document -Wpointer-compare.
22834 2017-01-04 Jakub Jelinek <jakub@redhat.com>
22836 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
22839 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
22840 descriptions for -gdwarf-5 and emit them as uleb128 instead of
22843 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
22846 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
22847 documentation of the powerpc_popcntb_ok attribute.
22848 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22849 code to issue warning messages if a requested CPU configuration is
22850 not supported by the binary (assembler and loader) toolchain.
22851 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
22852 made to define a built-in function that has been disabled.
22853 (paired_init_builtins): Add assertion to prevent ICE if attempt is
22854 made to define a built-in function that has been disabled.
22855 (altivec_init_builtins): Add comment explaining why definition
22856 of the DST built-in functions is not preceded by an assertion
22857 check. Add assertions to prevent ICE if attempts are made to
22858 define an altivec predicate or an abs* built-in function that has
22860 (htm_init_builtins): Add comment explaining why definition of the
22861 htm built-in functions is not preceded by an assertion check.
22863 2017-01-04 Jeff Law <law@redhat.com>
22865 PR tree-optimizatin/67955
22866 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
22867 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
22868 the points-to solution does not include pt_null. Use DECL_PT_UID
22871 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
22873 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
22874 Use gen_int_mode instead of gen_lopwart for const_int operands.
22876 2017-01-04 Jakub Jelinek <jakub@redhat.com>
22878 PR tree-optimization/71563
22879 * match.pd: Simplify X << Y into X if Y is known to be 0 or
22880 out of range value - has low bits known to be zero.
22882 2017-01-04 Alan Modra <amodra@gmail.com>
22884 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
22885 * configure: Regenerate.
22886 * config.in: Regenerate.
22888 2017-01-04 Jakub Jelinek <jakub@redhat.com>
22891 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
22892 a substring of the message, but strcmp with the whole message. Ifdef
22893 ENABLE_NLS, translate the message first using dgettext.
22895 2017-01-03 Jeff Law <law@redhat.com>
22897 PR tree-optimizatin/78856
22898 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
22899 (mark_threaded_blocks): Remove code to truncate thread paths that
22900 cross multiple loop headers. Instead invalidate the cached loop
22901 iteration information and handle case of a thread path walking
22902 into an irreducible region.
22904 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
22907 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
22908 assertions. Add support for doing the signbit if the IEEE 128-bit
22909 floating point value is in a GPR.
22910 * config/rs6000/rs6000.md (Fsignbit): Delete.
22911 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
22912 Update the length attribute if the value is in a GPR.
22913 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
22914 the sign or zero extension instruction, since the value is always 0/1.
22915 (signbit<mode>2_dm2): Delete using <Fsignbit>.
22918 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
22919 extracting SImode to a GPR register so that we can generate a
22920 store, limit the vector to be in a traditional Altivec register
22921 for the vextuwrx instruction.
22923 2017-01-03 Ian Lance Taylor <iant@google.com>
22925 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
22927 2017-01-03 Martin Sebor <msebor@redhat.com>
22929 PR tree-optimization/78696
22930 * gimple-ssa-sprintf.c (format_floating): Correct handling of
22931 precision. Use MPFR for %f for greater fidelity. Correct handling
22933 (pass_sprintf_length::compute_format_length): Set width and precision
22934 specified by asrerisk to void_node for vararg functions.
22935 (try_substitute_return_value): Adjust dump output.
22937 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
22939 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
22941 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
22943 * doc/invoke.texi (SPARC options): Document -mlra as the default.
22944 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
22945 -mlra/-mno-lra was passed to the compiler.
22947 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
22949 PR rtl-optimization/65618
22950 * emit-rtl.c (try_split): Move initialization of "before" and
22951 "after" to just before the call to emit_insn_after_setloc.
22953 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
22955 * doc/md.texi (Standard Names): Remove reference to Java frontend.
22957 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
22959 * dwarf2out.c (gen_enumeration_type_die): When
22960 -gno-strict-dwarf, add a DW_AT_encoding attribute.
22962 2017-01-03 Jakub Jelinek <jakub@redhat.com>
22964 PR tree-optimization/78965
22965 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
22966 Change first argument from const call_info & to call_info &. For %n
22967 set info.nowrite to false.
22969 PR middle-end/78901
22970 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
22971 possibly throwing calls.
22973 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
22974 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
22975 and fns handling, rather than in a separate case SSA_NAME.
22977 2017-01-02 Jeff Law <law@redhat.com>
22979 * config/darwin-driver.c (darwin_driver_init): Const-correctness
22980 fixes for first_period and second_period variables.
22982 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
22985 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
22986 (*insvqi_1): New insn pattern.
22987 (*insvqi_1_mem_rex64): Ditto.
22988 (*insvqi_2): Ditto.
22989 (*insvqi_3): Rename from *insvqi.
22991 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
22993 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
22995 * doc/cfg.texi (Edges): Remove reference to Java.
22996 (Maintaining the CFG): Ditto.
22998 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23000 PR middle-end/77674
23001 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23002 transparent aliases.
23004 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23006 PR middle-end/77484
23007 * predict.def (PRED_CALL): Update hitrate.
23008 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23009 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23010 into direct/indirect/polymorphic variants.
23012 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23014 Update copyright years.
23016 * gcc.c (process_command): Update copyright notice dates.
23017 * gcov-dump.c (print_version): Ditto.
23018 * gcov.c (print_version): Ditto.
23019 * gcov-tool.c (print_version): Ditto.
23020 * gengtype.c (create_file): Ditto.
23021 * doc/cpp.texi: Bump @copying's copyright year.
23022 * doc/cppinternals.texi: Ditto.
23023 * doc/gcc.texi: Ditto.
23024 * doc/gccint.texi: Ditto.
23025 * doc/gcov.texi: Ditto.
23026 * doc/install.texi: Ditto.
23027 * doc/invoke.texi: Ditto.
23029 Copyright (C) 2017 Free Software Foundation, Inc.
23031 Copying and distribution of this file, with or without modification,
23032 are permitted in any medium without royalty provided the copyright
23033 notice and this notice are preserved.