1 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * lra-int.h (lra_reg): Change offset from int to poly_int64.
6 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
8 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
9 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
10 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
11 offset as a poly_int64 rather than an int.
12 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
13 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14 * lra-constraints.c (equiv_address_substitution): Track offsets
16 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
17 (curr_insn_transform): Handle the new form of sp_offset.
18 * lra-eliminations.c (lra_elim_table): Change previous_offset
19 and offset from HOST_WIDE_INT to poly_int64.
20 (print_elim_table, update_reg_eliminate): Update accordingly.
21 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
22 (get_elimination): Update accordingly.
23 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
24 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
25 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
26 poly_int64 offsets generally.
27 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
28 (mark_not_eliminable, init_elimination): Update accordingly.
29 (remove_reg_equal_offset_note): Return a bool and pass the new
30 offset back by pointer as a poly_int64.
31 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
32 rather than a HOST_WIDE_INT.
33 (do_remat): Track offsets poly_int64s.
34 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
36 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
37 Alan Hayward <alan.hayward@arm.com>
38 David Sherwood <david.sherwood@arm.com>
40 * rtl.h (mem_attrs): Add a default constructor. Change size and
41 offset from HOST_WIDE_INT to poly_int64.
42 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
43 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
44 (widen_memory_access): Take the sizes and offsets as poly_int64s
45 rather than HOST_WIDE_INTs.
46 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
47 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
49 (adjust_offset_for_component_ref): Change the offset from a
50 HOST_WIDE_INT to a poly_int64.
51 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
52 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
53 * dce.c (find_call_stack_args): Likewise.
54 * dse.c (record_store): Likewise.
55 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
56 * print-rtl.c (rtx_writer::print_rtx): Likewise.
57 * read-rtl-function.c (test_loading_mem): Likewise.
58 * rtlanal.c (may_trap_p_1): Likewise.
59 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
60 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
61 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
62 (mem_attrs::mem_attrs): New function.
63 (set_mem_attributes_minus_bitpos): Change bitpos from a
64 HOST_WIDE_INT to poly_int64.
65 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
66 (clear_mem_offset, clear_mem_size, change_address)
67 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
69 (set_mem_offset, set_mem_size, adjust_address_1)
70 (adjust_automodify_address_1, offset_address, widen_memory_access):
71 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
73 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
74 Alan Hayward <alan.hayward@arm.com>
75 David Sherwood <david.sherwood@arm.com>
77 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
78 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
79 rather than 0 to represent an unknown size. Assert that the size
80 is known when the mode isn't BLKmode.
81 (may_trap_p_1): Use -1 for unknown sizes.
82 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
84 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
85 Alan Hayward <alan.hayward@arm.com>
86 David Sherwood <david.sherwood@arm.com>
88 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
89 to poly_int64. Update commentary for positions_needed.large.
90 (read_info_type): Change offset and width from HOST_WIDE_INT
92 (set_usage_bits): Likewise.
93 (canon_address): Return the offset as a poly_int64 rather than
94 a HOST_WIDE_INT. Use strip_offset_and_add.
95 (set_all_positions_unneeded, any_positions_needed_p): Use
96 positions_needed.large to track stores with non-constant widths.
97 (all_positions_needed_p): Likewise. Take the offset and width
98 as poly_int64s rather than ints. Assert that rhs is nonnull.
99 (record_store): Cope with non-constant offsets and widths.
100 Nullify the rhs of an earlier store if we can't tell which bytes
102 (find_shift_sequence): Take the access_size and shift as poly_int64s
104 (get_stored_val): Take the read_offset and read_width as poly_int64s
105 rather than HOST_WIDE_INTs.
106 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
107 non-constant offsets and widths.
109 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
110 Alan Hayward <alan.hayward@arm.com>
111 David Sherwood <david.sherwood@arm.com>
113 * inchash.h (inchash::hash::add_poly_int): New function.
114 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
115 Use poly_int64 rather than HOST_WIDE_INT.
116 (ao_ref::max_size_known_p): New function.
117 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
118 rather than HOST_WIDE_INT.
119 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
120 to temporaries until its interface is adjusted to match.
121 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
122 (aliasing_component_refs_p, decl_refs_may_alias_p)
123 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
124 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
125 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
127 * alias.c (ao_ref_from_mem): Likewise.
128 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
129 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
130 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
131 (maybe_trim_complex_store, maybe_trim_constructor_store)
132 (live_bytes_read, dse_classify_store): Likewise.
133 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
134 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
135 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
136 (vn_reference_lookup_3): Likewise.
137 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
139 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
140 Alan Hayward <alan.hayward@arm.com>
141 David Sherwood <david.sherwood@arm.com>
143 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
144 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
145 instead of ranges_overlap_p.
147 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
152 sizes as poly_int64s rather than HOST_WIDE_INTs.
154 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
155 Alan Hayward <alan.hayward@arm.com>
156 David Sherwood <david.sherwood@arm.com>
158 * gimple-fold.h (fold_ctor_reference): Take the offset and size
159 as poly_uint64 rather than unsigned HOST_WIDE_INT.
160 * gimple-fold.c (fold_ctor_reference): Likewise.
162 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * target.def (dwarf_poly_indeterminate_value): New hook.
167 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
168 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
169 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
170 * doc/tm.texi: Regenerate.
171 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
172 offset as a poly_int64.
173 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
174 offset as a poly_int64.
175 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
176 Take the offset as a poly_int64.
177 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
178 (frame_pointer_fb_offset): Change to a poly_int64.
179 (int_loc_descriptor): Take the offset as a poly_int64. Use
180 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
181 (based_loc_descr): Take the offset as a poly_int64.
182 Use strip_offset_and_add to handle (plus X (const)).
183 Use new_reg_loc_descr instead of an open-coded version of the
184 previous implementation.
185 (mem_loc_descriptor): Handle CONST_POLY_INT.
186 (compute_frame_pointer_to_fb_displacement): Take the offset as a
187 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
189 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
190 Alan Hayward <alan.hayward@arm.com>
191 David Sherwood <david.sherwood@arm.com>
193 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
194 (gen_rtx_REG_offset): Take the offset as a poly_int64.
195 * inchash.h (inchash::hash::add_poly_hwi): New function.
196 * gengtype.c (main): Register poly_int64.
197 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
198 offset as a poly_int.
199 (reg_attr_hasher::equal): Use must_eq to compare offsets.
200 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
201 offset as a poly_int64.
202 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
203 * print-rtl.c (print_poly_int): New function.
204 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
206 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
208 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
209 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
210 (vt_get_decl_and_offset): Return the offset as a poly_int64.
211 Enforce track_offset_p for parts of a PARALLEL.
212 (vt_add_function_parameter): Use const_offset for the final
213 offset to track. Use get_tracked_reg_offset for the parts
216 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
217 Alan Hayward <alan.hayward@arm.com>
218 David Sherwood <david.sherwood@arm.com>
220 * target.def (truly_noop_truncation): Take poly_uint64s instead of
221 unsigned ints. Change default to hook_bool_puint64_puint64_true.
222 * doc/tm.texi: Regenerate.
223 * hooks.h (hook_bool_uint_uint_true): Delete.
224 (hook_bool_puint64_puint64_true): Declare.
225 * hooks.c (hook_bool_uint_uint_true): Delete.
226 (hook_bool_puint64_puint64_true): New function.
227 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
228 instead of unsigned ints.
229 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
230 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
232 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
233 Alan Hayward <alan.hayward@arm.com>
234 David Sherwood <david.sherwood@arm.com>
236 * optabs.h (expand_operand): Add an int_value field.
237 (create_expand_operand): Add an int_value parameter and use it
238 to initialize the new expand_operand field.
239 (create_integer_operand): Replace with a declaration of a function
240 that accepts poly_int64s. Move the implementation to...
241 * optabs.c (create_integer_operand): ...here.
242 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
243 the mode preserves the value of int_value, instead of calling
244 const_int_operand on the rtx. Use gen_int_mode to generate
247 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
248 Alan Hayward <alan.hayward@arm.com>
249 David Sherwood <david.sherwood@arm.com>
251 * dumpfile.h (dump_dec): Declare.
252 * dumpfile.c (dump_dec): New function.
253 * pretty-print.h (pp_wide_integer): Turn into a function and
254 declare a poly_int version.
255 * pretty-print.c (pp_wide_integer): New function for poly_ints.
257 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
258 Alan Hayward <alan.hayward@arm.com>
259 David Sherwood <david.sherwood@arm.com>
261 * doc/generic.texi (POLY_INT_CST): Document.
262 * tree.def (POLY_INT_CST): New tree code.
263 * treestruct.def (TS_POLY_INT_CST): New tree layout.
264 * tree-core.h (tree_poly_int_cst): New struct.
265 (tree_node): Add a poly_int_cst field.
266 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
267 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
268 instead of a wide_int_ref.
269 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
271 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
272 instead of an unsigned HOST_WIDE_INT.
273 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
274 (ptrdiff_tree_p): Declare.
275 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
276 extern inline implementations if the target doesn't use POLY_INT_CST.
277 (poly_int_tree_p): New function.
278 (wi::unextended_tree): New class.
279 (wi::int_traits <unextended_tree>): New override.
280 (wi::extended_tree): Add a default constructor.
281 (wi::extended_tree::get_tree): New function.
282 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
283 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
284 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
285 (wi::tree_to_poly_wide_ref): New typedefs.
286 (wi::ints_for): Provide overloads for extended_tree and
288 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
289 (wi::to_wide): New functions.
290 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
291 * tree.c (poly_int_cst_hasher): New struct.
292 (poly_int_cst_hash_table): New variable.
293 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
294 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
296 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
297 (init_ttree): Initialize poly_int_cst_hash_table.
298 (build_int_cst, build_int_cst_type, build_invariant_address): Take
299 a poly_int64 instead of a HOST_WIDE_INT.
300 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
301 instead of an unsigned HOST_WIDE_INT.
302 (wide_int_to_tree): Rename to...
303 (wide_int_to_tree_1): ...this.
304 (build_new_poly_int_cst, build_poly_int_cst): New functions.
305 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
306 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
307 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
308 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
309 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
311 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
312 (streamer_read_tree_body): Likewise.
313 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
314 (streamer_write_tree_body): Likewise.
315 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
316 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
317 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
318 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
319 * gimple-expr.h (is_gimple_constant): Likewise.
320 * gimplify.c (maybe_with_size_expr): Likewise.
321 * print-tree.c (print_node): Likewise.
322 * tree-data-ref.c (data_ref_compare_tree): Likewise.
323 * tree-pretty-print.c (dump_generic_node): Likewise.
324 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
325 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
326 * tree-vrp.c (compare_values_warnv): Likewise.
327 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
328 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
329 (force_expr_to_var_cost): Likewise.
330 * tree-ssa-loop.c (for_each_index): Likewise.
331 * fold-const.h (build_invariant_address, size_int_kind): Take a
332 poly_int64 instead of a HOST_WIDE_INT.
333 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
334 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
336 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
337 (int_const_binop_2): New function, split out from...
338 (int_const_binop_1): ...here. Handle POLY_INT_CST.
339 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
340 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
341 * gimple-ssa-strength-reduction.c (slsr_process_add)
342 (slsr_process_mul): Check for INTEGER_CSTs before using them
344 * stor-layout.c (bits_from_bytes): New function.
345 (bit_from_pos): Use it.
346 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
347 by BITS_PER_UNIT to get the TYPE_SIZE.
348 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
349 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
351 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
352 Alan Hayward <alan.hayward@arm.com>
353 David Sherwood <david.sherwood@arm.com>
355 * doc/rtl.texi (const_poly_int): Document. Also document the
356 rtl sharing behavior.
357 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
358 * rtl.h (const_poly_int_def): New struct.
359 (rtx_def::u): Add a cpi field.
360 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
361 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
362 (wi::rtx_to_poly_wide_ref): New typedef
363 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
364 (poly_int_rtx_p): New functions.
365 (trunc_int_for_mode): Declare a poly_int64 version.
366 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
367 (immed_wide_int_const): Take a poly_wide_int_ref rather than
369 (strip_offset): Declare.
370 (strip_offset_and_add): New function.
371 * rtl.def (CONST_POLY_INT): New rtx code.
372 * rtl.c (rtx_size): Handle CONST_POLY_INT.
373 (shared_const_p): Use poly_int_rtx_p.
374 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
376 (gen_int_shift_amount): Likewise.
377 * emit-rtl.c (const_poly_int_hasher): New class.
378 (const_poly_int_htab): New variable.
379 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
380 (const_poly_int_hasher::hash): New function.
381 (const_poly_int_hasher::equal): Likewise.
382 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
383 (immed_wide_int_const): Rename to...
384 (immed_wide_int_const_1): ...this and make static.
385 (immed_wide_int_const): New function, taking a poly_wide_int_ref
386 instead of a wide_int_ref.
387 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
388 (gen_lowpart_common): Handle CONST_POLY_INT.
389 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
390 * cselib.c (cselib_hash_rtx): Likewise.
391 * dwarf2out.c (const_ok_for_output_1): Likewise.
392 * expr.c (convert_modes): Likewise.
393 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
394 * rtlhash.c (add_rtx): Likewise.
395 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
396 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
397 Handle existing CONST_POLY_INT rtxes.
398 * expmed.h (expand_shift): Take a poly_int64 instead of a
400 * expmed.c (expand_shift): Likewise.
401 * rtlanal.c (strip_offset): New function.
402 (commutative_operand_precedence): Give CONST_POLY_INT the same
403 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
405 * rtl-tests.c (const_poly_int_tests): New struct.
406 (rtl_tests_c_tests): Use it.
407 * simplify-rtx.c (simplify_const_unary_operation): Handle
409 (simplify_const_binary_operation): Likewise.
410 (simplify_binary_operation_1): Fold additions of symbolic constants
412 (simplify_subreg): Handle extensions and truncations of
414 (simplify_const_poly_int_tests): New struct.
415 (simplify_rtx_c_tests): Use it.
416 * wide-int.h (storage_ref): Add default constructor.
417 (wide_int_ref_storage): Likewise.
418 (trailing_wide_ints): Use GTY((user)).
419 (trailing_wide_ints::operator[]): Add a const version.
420 (trailing_wide_ints::get_precision): New function.
421 (trailing_wide_ints::extra_size): Likewise.
423 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
424 Alan Hayward <alan.hayward@arm.com>
425 David Sherwood <david.sherwood@arm.com>
427 * emit-rtl.h (gen_int_shift_amount): Declare.
428 * emit-rtl.c (gen_int_shift_amount): New function.
429 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
431 * calls.c (shift_return_value): Likewise.
432 * cse.c (fold_rtx): Likewise.
433 * dse.c (find_shift_sequence): Likewise.
434 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
435 (expand_shift, expand_smod_pow2): Likewise.
436 * lower-subreg.c (shift_cost): Likewise.
437 * optabs.c (expand_superword_shift, expand_doubleword_mult)
438 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
439 (expand_vec_perm_var): Likewise.
440 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
441 (simplify_binary_operation_1): Likewise.
442 * combine.c (try_combine, find_split_point, force_int_to_mode)
443 (simplify_shift_const_1, simplify_shift_const): Likewise.
444 (change_zero_ext): Likewise. Use simplify_gen_binary.
446 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
448 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
450 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
452 * doc/invoke.texi (ARM Options): Document accepted extension options
453 for -march=armv8.3-a.
455 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
458 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
459 or --with-float=softfp, set the default CPU to arm10e.
461 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
463 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
464 * config/visium/predicates.md (const_shift_operand): Likewise.
465 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
466 (visium_legitimize_reload_address): Likewise.
468 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
471 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
473 2017-12-20 Jakub Jelinek <jakub@redhat.com>
476 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
478 2017-12-20 Julia Koval <julia.koval@intel.com>
480 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
481 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
482 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
483 * config.gcc: Include vpclmulqdqintrin.h.
484 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
485 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
487 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
488 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
489 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
490 * config/i386/i386.c (isa2_opts): Add -mcx16.
491 (isa_opts): Add -mpclmulqdq, remove -mcx16.
492 (ix86_option_override_internal): Move mcx16 to flags2.
493 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
494 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
495 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
496 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
497 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
498 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
499 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
500 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
501 * doc/invoke.texi: Add -mvpclmulqdq.
503 2017-12-20 Tom de Vries <tom@codesourcery.com>
506 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
507 * calls.c (rtx_for_static_chain): ... here. New function.
508 * calls.h (rtx_for_static_chain): Declare.
509 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
510 instead of targetm.calls.static_chain.
511 * df-scan.c (df_get_entry_block_def_set): Same.
513 2017-12-19 Tom de Vries <tom@codesourcery.com>
515 PR tree-optimization/83493
516 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
519 2017-12-19 Jakub Jelinek <jakub@redhat.com>
521 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
522 inform with hardcoded english plural handling.
524 2017-12-18 Jeff Law <law@redhat.com>
526 PR tree-optimization/83477
527 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
528 a non-virtual PHI, always push a new range.
530 2017-12-19 Martin Sebor <msebor@redhat.com>
533 * builtins.c (compute_objsize): Handle non-constant offsets.
535 2017-12-19 Jakub Jelinek <jakub@redhat.com>
537 PR tree-optimization/83444
538 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
539 character load case, if get_stridx on MEM_REF's operand doesn't
540 look usable, retry with get_addr_stridx.
542 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
545 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
546 (variable_tracking_main_1): Keep markers even when VTA fails.
549 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
550 even if there are markers before it.
551 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
553 2017-12-19 Jakub Jelinek <jakub@redhat.com>
555 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
556 typical order conditions.
557 * sel-sched.c (extract_new_fences_from): Likewise.
558 * config/visium/constraints.md (J, K, L): Likewise.
559 * config/visium/predicates.md (const_shift_operand): Likewise.
560 * config/visium/visium.c (visium_legitimize_address,
561 visium_legitimize_reload_address): Likewise.
562 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
563 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
564 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
565 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
566 SET_DUMP_DETAIL): Likewise.
567 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
568 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
569 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
570 avr_set_core_architecture, avr_set_current_function,
571 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
572 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
573 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
574 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
575 avr_map_decompose, avr_fold_builtin): Likewise.
576 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
577 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
578 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
579 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
580 * config/m32c/m32c.c (m32c_conditional_register_usage,
581 m32c_address_cost): Likewise.
582 * config/m32c/predicates.md (shiftcount_operand,
583 longshiftcount_operand): Likewise.
584 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
585 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
586 can_use_cdx_ldstw): Likewise.
587 * config/nios2/nios2.h (CDX_REG_P): Likewise.
588 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
590 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
591 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
592 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
593 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
594 * config/vax/vax.c (adjacent_operands_p): Likewise.
595 * config/ft32/constraints.md (L, b, KA): Likewise.
596 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
598 * cfgexpand.c (expand_stack_alignment): Likewise.
599 * gcse.c (insert_expr_in_table): Likewise.
600 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
601 * cgraphunit.c (cgraph_node::expand): Likewise.
602 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
603 * emit-rtl.c (add_insn): Likewise.
604 * input.c (dump_location_info): Likewise.
605 * passes.c (NEXT_PASS): Likewise.
606 * read-rtl-function.c (parse_note_insn_name,
607 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
609 * sched-rgn.c (sched_rgn_init): Likewise.
610 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
611 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
612 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
613 * explow.c (eliminate_constant_term): Likewise.
614 * final.c (leaf_renumber_regs_insn): Likewise.
615 * cfgrtl.c (print_rtl_with_bb): Likewise.
616 * genhooks.c (emit_init_macros): Likewise.
617 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
618 * tree-data-ref.c (conflict_fn): Likewise.
619 * selftest.c (assert_streq): Likewise.
620 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
621 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
622 fold_binary_loc, multiple_of_p): Likewise.
623 * reload.c (push_reload, find_equiv_reg): Likewise.
624 * et-forest.c (et_nca, et_below): Likewise.
625 * dbxout.c (dbxout_symbol_location): Likewise.
626 * reorg.c (relax_delay_slots): Likewise.
627 * dojump.c (do_compare_rtx_and_jump): Likewise.
628 * gengtype-parse.c (type): Likewise.
629 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
630 simplify_const_relational_operation): Likewise.
631 * reload1.c (do_output_reload): Likewise.
632 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
633 * gengtype.c (type_for_name): Likewise.
634 * gimple-ssa-sprintf.c (format_directive): Likewise.
636 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
639 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
640 accessing it. Adjust comment.
642 2017-12-19 Jakub Jelinek <jakub@redhat.com>
645 * predict.c (zero_one_minusone): New function.
646 (apply_return_prediction): Avoid return prediction for functions
647 returning only -1, 0 and 1 values, unless they only return -1 and 0
650 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
652 * config/arc/arc.c (legitimate_scaled_address_p): Clean
653 fall-through warning.
654 (arc_compute_frame_size): Remove unused variables.
655 (arc_print_operand): Fix fprintif format.
656 (arc_can_follow_jump): Clean fall-through warning.
658 2017-12-19 Marek Polacek <polacek@redhat.com>
661 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
664 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
666 * config/arc/arc.c (overriderregs): New variable.
667 (arc_override_options): Track fixed/call saved/call options.
668 (arc_conditional_register_usage): Check against overrideregs
669 variable whenever we change register properties.
671 2017-12-19 Nathan Sidwell <nathan@acm.org>
673 * opts.c (finish_options): Don't prefix dump_base_name if it
674 already contains directories.
676 2017-12-19 Martin Liska <mliska@suse.cz>
678 PR rtl-optimization/82675
679 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
680 more element in sbitmap.
682 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
684 * gimplify.c (gimplify_expr): Use error_operand_p.
686 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
689 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
692 2017-12-19 Daniel Cederman <cederman@gaisler.com>
694 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
695 the jump is to a label.
697 2017-12-19 Jakub Jelinek <jakub@redhat.com>
699 PR tree-optimization/83444
700 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
705 * ipa-inline.c (flatten_remove_node_hook): New function.
706 (ipa_inline): Keep only nodes with flatten attribute at the end of
707 the array in the order from ipa_reverse_postorder, only walk that
708 portion of array for flattening, if there is more than one such
709 node, temporarily register a removal hook and ignore removed nodes.
711 PR tree-optimization/80631
712 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
713 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
714 IFN_REDUC_MAX or IFN_REDUC_MIN.
716 2017-12-18 Jeff Law <law@redhat.com>
718 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
719 record symbolic equivalences from backedges in the CFG.
722 2017-11-19 Jeff Law <law@redhat.com>
724 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
725 of degenerates resulting from ignoring an edge.
727 2017-12-18 Martin Sebor <msebor@redhat.com>
730 PR tree-optimization/78450
731 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
732 (handle_builtin_strlen): Call it.
734 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
736 PR rtl-optimization/83424
737 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
739 2017-12-18 Marek Polacek <polacek@redhat.com>
742 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
743 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
746 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
748 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
749 (umaddsidi4, umaddsidi_split): Likewise.
751 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
753 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
756 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
759 * config/s390/s390.c (s390_option_override): Avoid strncpy.
761 2017-12-18 Richard Biener <rguenther@suse.de>
763 PR tree-optimization/81877
764 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
765 (outermost_indep_loop): Adjust.
766 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
767 (can_sm_ref_p): Adjust.
769 2017-12-18 Richard Biener <rguenther@suse.de>
772 * tree.c (array_at_struct_end_p): Return true if the underlying
773 object has space for at least one element in excess of what
774 the array domain specifies.
776 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
778 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
779 example to avoid overfull hbox.
780 * doc/invoke.texi (Option Summary): Add missing @gol.
781 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
784 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
785 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
787 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
788 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
789 (AArch64 Options): Document -mverbose-cost-dump.
790 (ARM Options): Likewise, plus -mflip-thumb.
792 2017-12-17 Martin Sebor <msebor@redhat.com>
795 * gimple-ssa-warn-restrict.c
796 (builtin_memref::offset_out_of_bounds): Correct the handling of
799 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
801 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
803 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
805 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
807 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
809 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
812 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
814 * doc/invoke.texi: Fix some typos.
816 2017-12-16 Martin Sebor <msebor@redhat.com>
818 PR tree-optimization/78918
819 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
820 * builtins.c (check_sizes): Rename...
821 (check_access): ...to this. Rename function arguments for clarity.
822 (check_memop_sizes): Adjust names.
823 (expand_builtin_memchr, expand_builtin_memcpy): Same.
824 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
825 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
826 (check_strncat_sizes, expand_builtin_strncat): Same.
827 (expand_builtin_strncpy, expand_builtin_memset): Same.
828 (expand_builtin_bzero, expand_builtin_memcmp): Same.
829 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
830 (maybe_emit_sprintf_chk_warning): Same.
831 (expand_builtin_strcpy): Adjust.
832 (expand_builtin_stpcpy): Same.
833 (expand_builtin_with_bounds): Detect out-of-bounds accesses
834 in pointer-checking forms of memcpy, memmove, and mempcpy.
835 (gcall_to_tree_minimal, max_object_size): Define new functions.
836 * builtins.h (max_object_size): Declare.
837 * calls.c (alloc_max_size): Call max_object_size instead of
838 hardcoding ssizetype limit.
839 (get_size_range): Handle new argument.
840 * calls.h (get_size_range): Add a new argument.
841 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
842 * doc/invoke.texi (-Wrestrict): Adjust, add example.
843 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
845 (gimple_fold_builtin_memory_chk): Same.
846 (gimple_fold_builtin_stxcpy_chk): New function.
847 * gimple-ssa-warn-restrict.c: New source.
848 * gimple-ssa-warn-restrict.h: New header.
849 * gimple.c (gimple_build_call_from_tree): Propagate location.
850 * passes.def (pass_warn_restrict): Add new pass.
851 * tree-pass.h (make_pass_warn_restrict): Declare.
852 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
854 (handle_builtin_strcat): Same.
855 (strlen_optimize_stmt): Rename...
856 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
857 stpncpy, strncpy, and their checking forms.
859 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
861 PR rtl-optimization/82849
862 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
863 and get_max_loop_iterations_int.
865 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
866 Alan Hayward <alan.hayward@arm.com>
867 David Sherwood <david.sherwood@arm.com>
869 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
870 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
871 size as a poly_uint64.
872 (mode_for_vector, mode_for_int_vector): Take the number of vector
873 elements as a poly_uint64.
874 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
875 size as a poly_uint64.
876 (mode_for_vector, mode_for_int_vector): Take the number of vector
877 elements as a poly_uint64.
879 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
880 Alan Hayward <alan.hayward@arm.com>
881 David Sherwood <david.sherwood@arm.com>
883 * machmode.h (MACRO_MODE): New macro.
884 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
885 * targhooks.c (default_libcall_value, default_secondary_reload)
886 (default_memory_move_cost, default_register_move_cost)
887 (default_class_max_nregs): Likewise.
889 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
890 Alan Hayward <alan.hayward@arm.com>
891 David Sherwood <david.sherwood@arm.com>
893 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
894 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
896 * genautomata.c (main): Likewise.
897 * genconditions.c (write_header): Likewise.
898 * genemit.c (main): Likewise.
899 * genextract.c (print_header): Likewise.
900 * genopinit.c (main): Likewise.
901 * genoutput.c (output_prologue): Likewise.
902 * genpeep.c (main): Likewise.
903 * genpreds.c (write_insn_preds_c): Likewise.
904 * genrecog.c (writer_header): Likewise.
905 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
906 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
907 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
908 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
909 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
910 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
911 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
912 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
913 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
914 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
915 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
916 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
917 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
918 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
919 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
920 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
921 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
922 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
923 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
924 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
925 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
926 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
927 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
928 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
929 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
930 * config/darwin.c (IN_TARGET_CODE): Likewise.
931 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
932 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
933 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
934 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
935 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
936 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
937 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
938 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
939 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
940 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
941 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
942 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
943 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
944 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
945 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
946 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
947 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
948 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
949 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
950 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
951 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
952 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
953 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
954 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
955 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
956 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
957 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
958 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
959 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
960 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
961 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
962 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
963 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
964 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
965 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
966 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
967 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
968 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
969 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
970 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
971 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
972 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
973 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
974 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
975 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
976 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
977 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
978 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
979 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
980 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
981 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
982 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
983 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
984 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
985 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
986 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
987 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
988 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
989 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
990 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
991 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
992 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
993 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
994 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
995 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
996 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
997 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
998 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
999 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1000 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1001 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1002 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1003 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1004 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1005 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1006 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1007 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1008 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1009 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1010 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1011 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1012 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1013 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1014 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1015 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1016 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1017 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1018 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1019 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1020 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1021 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1022 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1023 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1024 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1025 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1026 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1027 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1028 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1029 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1030 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1031 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1032 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1033 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1034 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1035 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1036 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1038 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1039 Alan Hayward <alan.hayward@arm.com>
1040 David Sherwood <david.sherwood@arm.com>
1042 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1045 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1046 Alan Hayward <alan.hayward@arm.com>
1047 David Sherwood <david.sherwood@arm.com>
1049 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1050 * doc/md.texi (vec_series@var{m}): Document.
1051 * tree.def (VEC_SERIES_EXPR): New tree code.
1052 * tree.h (build_vec_series): Declare.
1053 * tree.c (build_vec_series): New function.
1054 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1055 * tree-pretty-print.c (dump_generic_node): Likewise.
1056 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1057 * tree-inline.c (estimate_operator_cost): Likewise.
1058 * expr.c (expand_expr_real_2): Likewise.
1059 * optabs-tree.c (optab_for_tree_code): Likewise.
1060 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1061 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1062 * expmed.c (make_tree): Handle VEC_SERIES.
1063 * optabs.def (vec_series_optab): New optab.
1064 * optabs.h (expand_vec_series_expr): Declare.
1065 * optabs.c (expand_vec_series_expr): New function.
1066 * tree-vect-generic.c (expand_vector_operations_1): Check that
1067 the operands also have vector type.
1069 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1070 Alan Hayward <alan.hawyard@arm.com>
1071 David Sherwood <david.sherwood@arm.com>
1073 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1074 (VEC_COND_EXPR): Add missing @tindex.
1075 * doc/md.texi (vec_duplicate@var{m}): Document.
1076 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1077 * tree.c (build_vector_from_val): Add stubbed-out handling of
1078 variable-length vectors, using VEC_DUPLICATE_EXPR.
1079 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1080 * cfgexpand.c (expand_debug_expr): Likewise.
1081 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1082 * tree-inline.c (estimate_operator_cost): Likewise.
1083 * tree-pretty-print.c (dump_generic_node): Likewise.
1084 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1085 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1086 (test_vec_duplicate_folding): New function.
1087 (fold_const_c_tests): Call it.
1088 * optabs.def (vec_duplicate_optab): New optab.
1089 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1090 * optabs.h (expand_vector_broadcast): Declare.
1091 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1092 vec_duplicate_optab.
1093 * expr.c (store_constructor): Try using vec_duplicate_optab for
1095 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1097 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1100 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1101 div/mod latencies a bit.
1103 2017-12-15 Jeff Law <law@redhat.com>
1105 PR tree-optimization/36550
1106 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1107 (mark_threaded_blocks): Rewrite code to avoid block copying when
1108 optimizing for size. Don't pessimize blocks which will be
1109 copied, but all the statements will be dead.
1111 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1113 PR tree-optimization/81165
1114 * tree-ssa-threadupdate.c (uses_in_bb): New.
1115 (estimate_threading_killed_stmts): New.
1116 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1117 * tree-ssa-threadedge.c
1118 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1121 2017-12-15 Jeff Law <law@redhat.com>
1123 PR tree-optimization/83410
1124 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1125 threads when parallelizing loops.
1127 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1129 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1131 * config/alpha/alpha.c (vms_attribute_table): Swap
1132 affects_type_identity and handler fields, adjust comments.
1133 * config/mips/mips.c (mips_attribute_table): Likewise.
1134 * config/visium/visium.c (visium_attribute_table): Likewise.
1135 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1136 * config/microblaze/microblaze.c (microblaze_attribute_table):
1138 * config/spu/spu.c (spu_attribute_table): Likewise.
1139 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1140 * config/arc/arc.c (arc_attribute_table): Likewise.
1141 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1142 * config/v850/v850.c (v850_handle_interrupt_attribute,
1143 v850_handle_data_area_attribute): Formatting fixes.
1144 (v850_attribute_table): Swap affects_type_identity and handler
1145 fields, adjust comments.
1146 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1147 * config/arm/arm.c (arm_attribute_table): Likewise.
1148 * config/avr/avr.c (avr_attribute_table): Likewise.
1149 * config/s390/s390.c (s390_attribute_table): Likewise.
1150 * config/sh/sh.c (sh_attribute_table): Likewise.
1151 * config/i386/i386.c (ix86_handle_cconv_attribute,
1152 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1153 (ix86_attribute_table): Swap affects_type_identity and handler
1154 fields, adjust comments.
1155 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1156 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1157 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1158 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1159 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1160 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1161 * config/rx/rx.c (rx_attribute_table): Likewise.
1162 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1163 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1164 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1165 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1166 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1167 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1168 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1169 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1170 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1171 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1172 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1174 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1176 (register_scoped_attributes, decl_attributes): Formatting fixes.
1178 PR tree-optimization/83269
1179 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1180 subtraction in arg0's type if type is signed and arg0 is unsigned.
1184 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1185 convert? on @0 instead of convert. Check type of @1, not @0.
1186 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1187 convert? on @0 instead of convert. Check type of @1, not @0.
1188 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1189 only optimize if either both @1 and @2 types are narrower
1190 precision, or both are wider or equal precision, and in the former
1191 case only if both have undefined overflow.
1193 2017-12-15 Richard Biener <rguenther@suse.de>
1196 * internal-fn.def (IFN_NOP): Add.
1197 * internal-fn.c (expand_NOP): Do nothing.
1198 * lto-streamer-in.c (input_function): Instead of removing
1199 sanitizer calls replace them with IFN_NOP calls.
1201 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1202 Alan Hayward <alan.hayward@arm.com>
1203 David Sherwood <david.sherwood@arm.com>
1205 * dse.c (store_info, read_info_type): Replace begin and end with
1207 (print_range): New function.
1208 (set_all_positions_unneeded, any_positions_needed_p)
1209 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1211 (record_store): Likewise. Optimize the case in which all positions
1213 (get_stored_val): Replace read_begin and read_end with read_offset
1215 (replace_read): Update call accordingly.
1217 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1219 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1220 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1221 (loop_cand::loop_cand): Initialize above members.
1222 (loop_cand::supported_operations): Delete.
1223 (loop_cand::can_interchange_p): Inline above function.
1224 (loop_cand::classify_simple_reduction): Record number of constant
1225 initialized simple reductions.
1226 (should_interchange_loops): New parameters. Check stmt cost of loops
1228 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1229 Update call to should_interchange_loops.
1230 (should_interchange_loop_nest): Update call to
1231 should_interchange_loops.
1233 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1236 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1237 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1238 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1240 2017-12-15 Julia Koval <julia.koval@intel.com>
1242 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1243 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1244 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1245 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1246 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1248 2017-12-15 Julia Koval <julia.koval@intel.com>
1250 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1251 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1252 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1253 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1254 _mm_aesenc_epi128): New intrinsics.
1256 2017-12-15 Julia Koval <julia.koval@intel.com>
1258 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1259 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1260 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1261 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1262 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1264 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1266 * gimple-ssa-strength-reduction.c (analyze_increments):
1267 Distinguish replacement costs for constant strides from those for
1270 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1272 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1274 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1276 * doc/invoke.texi: Document -Wcast-function-type.
1277 * recog.h (stored_funcptr): Change signature.
1278 * tree-dump.c (dump_node): Avoid warning.
1279 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1281 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1284 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1287 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1289 PR tree-optimization/83312
1290 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1291 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1292 "val" param, and to cope with arbitrary basic blocks.
1293 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1294 handle NULL_TREE for "val", dropping "bb" param.
1295 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1296 drop "bb" param. Handle NULL_TREE for "val".
1297 (find_case_label_for_value): Make "switch_stmt" param const.
1298 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1300 (vrp_prop::check_all_array_refs): Reimplement as...
1301 (check_array_bounds_dom_walker::before_dom_children): ...this new
1302 vfunc. Replace linear search through BB block list, excluding
1303 those with non-executable in-edges via dominator walk.
1305 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1307 * config/arm/arm.opt (mverbose-cost-dump): New option.
1308 * config/arm/arm.c (arm_rtx_costs): Use it.
1310 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1313 * reload1.c (emit_input_reload_insns): Skip debug markers.
1315 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1317 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1321 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1322 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1323 * config/sh/sh.c: Skip debug insns besides notes.
1324 * config/sh/sh.md: Likewise.
1325 * config/sh/sh_treg_combine.cc: Likewise.
1326 * config/sh/sync.md: Likewise.
1328 2017-12-14 Tom de Vries <tom@codesourcery.com>
1330 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1331 Add item for weak_undefined.
1333 2017-12-14 Richard Biener <rguenther@suse.de>
1335 PR tree-optimization/67842
1336 * sese.h (bb_in_region): Remove #if 0'ed code.
1338 2017-12-14 Richard Biener <rguenther@suse.de>
1340 PR tree-optimization/83326
1341 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1342 may_be_zero parameter and handle it by not marking the first
1343 peeled copy as not exiting the loop.
1344 (try_peel_loop): Likewise.
1345 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1346 to handle the case of constant or zero iterations and perform
1347 loop header copying on-the-fly.
1349 2017-12-14 Richard Biener <rguenther@suse.de>
1351 PR tree-optimization/83418
1352 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1353 Instead of asserting we don't get unfolded comparisons deal with
1356 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1359 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1360 basic blocks. Assert debug bind insns don't appear outside of bbs,
1361 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1364 PR tree-optimization/83198
1365 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1366 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1367 value if arg is a REAL_CST with incompatible type.
1369 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1370 Bin Cheng <bin.cheng@arm.com>
1373 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1375 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1378 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1380 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1381 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1383 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1384 Alan Hayward <alan.hayward@arm.com>
1385 David Sherwood <david.sherwood@arm.com>
1387 * poly-int.h: New file.
1388 * poly-int-types.h: Likewise.
1389 * coretypes.h: Include them.
1390 (POLY_INT_CONVERSION): Define.
1391 * target.def (estimated_poly_value): New hook.
1392 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1393 * doc/tm.texi: Regenerate.
1394 * doc/poly-int.texi: New file.
1395 * doc/gccint.texi: Include it.
1396 * doc/rtl.texi: Describe restrictions on subreg modes.
1397 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1398 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1399 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1400 * targhooks.h (default_estimated_poly_value): Declare.
1401 * targhooks.c (default_estimated_poly_value): New function.
1402 * target.h (estimated_poly_value): Likewise.
1403 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1404 (wi::unary_traits): Delete.
1405 (wi::binary_traits::signed_shift_result_type): Define for
1406 offset_int << HOST_WIDE_INT, etc.
1407 (generic_wide_int::operator <<=): Define for all types and use
1408 wi::lshift instead of <<.
1409 (wi::hwi_with_prec): Add a default constructor.
1410 (wi::ints_for): New class.
1411 (operator <<): Define for all wide-int types.
1412 (operator /): New function.
1413 (operator %): Likewise.
1414 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1415 (ASSERT_MAYBE_NE_AT): New macros.
1417 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1418 Dominik Vogt <vogt@linux.vnet.ibm.com>
1421 * emit-rtl.c (init_emit): Remove ??? comment.
1422 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1423 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1424 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1425 alignment of 3 virtual registers to BITS_PER_WORD.
1427 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1429 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1431 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1432 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1434 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1437 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1439 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1441 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1442 with the exception of debug begin stmt markers appear before
1446 * final.c (rest_of_handle_final): Call variable_tracking_main only
1447 if !flag_var_tracking.
1449 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1450 Jakub Jelinek <jakub@redhat.com>
1454 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1455 labels debug stmts that can only appear after labels.
1457 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1459 PR rtl-optimization/82398
1460 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1461 EXPR_USEFULNESS in priority comparison.
1463 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1465 PR rtl-optimization/83393
1466 * combine.c (move_deaths): If reg_stat points to a too new insn in
1467 last_death, do not use it: find the proper insn instead.
1469 2017-12-12 Jeff Law <law@redhat.com>
1471 PR tree-optimization/83298
1472 PR tree-optimization/83362
1473 PR tree-optimization/83383
1474 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1475 push_value_range a public interface. Add new argument to
1476 record_ranges_from_stmt.
1477 * gimple-ssa-evrp-analyze.c
1478 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1479 Update comments. Handle recording temporary equivalences.
1480 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1481 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1482 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1483 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1484 gimple-ssa-evrp-analyze.h.
1485 (record_temporary_equivalences_from_phis): Add new argument. When
1486 the PHI arg is an SSA_NAME, set the result's range to the range
1488 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1489 from statements too.
1490 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1491 Pass it down to children as needed.
1492 (thread_outgoing_edges): Likewise.
1493 (thread_across_edge): Likewise. Push/pop range state as needed.
1494 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1496 2017-12-12 Julia Koval <julia.koval@intel.com>
1498 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1499 (PTA_CANNONLAKE): Remove PTA_CLWB.
1501 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1504 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1505 (vcondv2div2df): Likewise.
1506 (vconduv2dfv2di): Likewise.
1508 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1510 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1511 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1512 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1513 ATTR_MATHFN_FPROUNDING_ERRNO.
1515 2017-12-12 Richard Biener <rguenther@suse.de>
1517 PR tree-optimization/83385
1518 * graphite-scop-detection.c (get_order, order): Remove.
1519 (bb_to_rpo): New global.
1521 (build_scops): Sort pbbs in RPO order.
1523 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1525 * combine.c (simplify_set): Do not transform subregs to zero_extends
1526 if the destination mode is a vector mode.
1528 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1530 PR tree-optimization/83359
1531 * tree-cfg.h (fold_loop_internal_call): Declare.
1532 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1533 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1534 (find_loop_dist_alias): New function.
1535 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1536 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1539 PR tree-optimization/80631
1540 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1541 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1542 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1543 hardcoding zero as the value if COND_EXPR is never true. For
1544 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1545 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1546 hardcoding MAX_EXPR as the reduction operation.
1547 (is_nonwrapping_integer_induction): Allow negative step.
1548 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1549 vect_create_epilog_for_reduction, if no value is suitable, don't
1550 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1552 2017-12-12 Richard Biener <rguenther@suse.de>
1554 PR tree-optimization/81889
1555 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1556 range info from the non-wrapping IV instead of just the range
1559 2017-12-12 Julia Koval <julia.koval@intel.com>
1561 * config.gcc: Add vaesintrin.h.
1562 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1564 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1565 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1567 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1568 * config/i386/immintrin.h: Include vaesintrin.h.
1569 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1570 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1571 _mm_aesdec_epi128): New intrinsics.
1573 2017-12-12 Julia Koval <julia.koval@intel.com>
1575 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1576 OPTION_MASK_ISA_VAES_UNSET): New.
1577 (ix86_handle_option): Handle -mvaes.
1578 * config/i386/cpuid.h: Define bit_VAES.
1579 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1580 * config/i386/i386-c.c (__VAES__): New.
1581 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1582 (ix86_valid_target_attribute_inner_p): Ditto.
1583 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1584 * config/i386/i386.opt: Add -mvaes.
1585 * doc/invoke.texi: Ditto.
1587 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1589 * debug.h (gcc_debug_hooks): Add inline_entry.
1590 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1591 * debug.c (do_nothing_debug_hooks): Likewise.
1592 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1593 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1594 (dwarf2_lineno_debug_hooks): Likewise.
1596 * common.opt (gstatement-frontiers): New, setting
1597 debug_nonbind_markers_p.
1598 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1599 * toplev.c (process_options): Autodetect value for debug statement
1601 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1602 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1604 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1605 markers. Integrate source bind into debug stmt expand loop.
1606 (pass_expand::execute): Check debug marker limit. Avoid deep
1607 TER and expand debug locations for debug bind insns only.
1608 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1610 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1611 * final.c (reemit_insn_block_notes): Take current block from
1612 nonbind markers. Declare note where it's first set.
1613 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1614 begin stmt markers if enabled.
1615 (notice_source_line): Handle nonbind markers. Fail if their
1616 location is unknown or that of builtins.
1617 (rest_of_handle_final): Convert begin stmt markers to notes if
1618 var-tracking didn't run.
1619 (rest_of_clean_state): Skip begin stmt markers.
1620 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1622 * function.c (allocate_struct_function): Set begin_stmt_markers.
1623 * function.h (struct function): Add debug_marker_count counter
1624 and debug_nonbind_markers flag.
1625 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1626 * gimple-low.c (lower_function_body): Adjust
1627 debug_nonbind_markers.
1628 (lower_stmt): Drop or skip gimple debug stmts.
1629 (lower_try_catch): Skip debug stmts.
1630 * gimple.c (gimple_build_debug_begin_stmt): New.
1631 (gimple_copy): Increment debug_marker_count if copying one.
1632 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1633 * gimplify.c (rexpr_location): New.
1634 (rexpr_has_location): New.
1635 (warn_switch_unreachable_r): Handle gimple debug stmts.
1636 (shortcut_cond_r): Call expr_location.
1638 (find_goto_label): New.
1639 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1641 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1643 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1644 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1645 (LANG_HOOKS_INITIALIZER): ... this.
1646 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1647 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1649 (update_ebb_live_info): Skip debug insn markers.
1650 * lra.c (debug_insn_static_data): Rename to...
1651 (debug_bind_static_data): ... this.
1652 (debug_marker_static_data): New.
1653 (lra_set_insn_recog_data): Select one of the above depending
1655 (lra_update_isn_regno_info): Don't assume debug insns have
1657 (push_insns): Skip debug insns.
1658 * lto-streamer-in.c (input_function): Drop debug stmts
1659 depending on active options. Adjust debug_nonbind_markers.
1660 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1661 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1662 begin stmt marker notes.
1663 (print_insn): Likewise.
1664 * recog.c (extract_insn): Recognize rtl for debug markers.
1665 * rtl.def (DEBUG_MARKER): New.
1666 * tree-inline.c: Include params.h.
1667 (remap_gimple_stmt): Handle nonbind markers.
1668 (maybe_move_debug_stmts_to_successors): Likewise.
1669 (copy_debug_stmt): Likewise.
1670 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1671 markers regardless of no side effects.
1672 (tsi_link_before): Don't update container's side effects when adding
1673 a begin stmt marker.
1674 (tsi_link_after): Likewise.
1675 (expr_first): Skip begin stmt markers.
1676 (expr_last): Likewise.
1677 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1678 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1679 Disregard nonbind markers.
1680 * tree.c (make_node_stat): Don't set side effects for begin stmt
1682 (build1_stat): Likewise.
1683 * tree.def (DEBUG_BEGIN_STMT): New.
1684 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1685 * var-tracking.c (delete_debug_insns): Renamed to...
1686 (delete_vta_debug_insns): ... this.
1687 (reemit_marker_as_note): New.
1688 (vt_initialize): Reemit markers.
1689 (delete_vta_debug_insns): Likewise.
1690 (vt_debug_insns_local): Reemit or delete markers.
1691 (variable_tracking_main_1): Likewise.
1692 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1693 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1694 (gimple_debug_nonbind_marker_p): New.
1695 (gimple_build_debug_bind): Adjust.
1696 (gimple_build_debug_begin_stmt): New.
1697 * doc/invoke.texi (max-debug-marker-count): New param.
1698 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
1699 (debug_parameter_ref, debug_marker): New.
1700 (NOTE_INSN_BEGIN_STMT): New.
1701 (DEBUG_INSN): Describe begin stmt markers.
1703 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
1704 without debug insns, we wouldn't, but clean up debug insns
1705 after a control flow insn nevertheless.
1707 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
1708 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
1709 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
1710 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
1711 (rtl_tidy_fallthru_edge): Likewise.
1712 (rtl_verify_fallthru): Likewise.
1713 (rtl_verify_bb_layout): Likewise.
1714 (skip_insns_after_block): Likewise.
1715 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
1716 * dwarf2out.c: Include print-rtl.h.
1717 (dwarf2out_next_real_insn): New.
1718 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1719 Dump debug binds in asm comments.
1720 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1721 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1722 callers to use gsi_start_nondebug_bb instead.
1723 (gsi_after_labels): Skip gimple debug stmts.
1724 (gsi_start_nondebug): New.
1725 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1726 (proper_loop_form_for_interchange): Adjust.
1727 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1728 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1729 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1730 (collect_fallthrough_labels): Likewise.
1731 (should_warn_for_implicit_fallthrough): Likewise.
1732 (warn_implicit_fallthrough_r): Likewise.
1733 (expand_FALLTHROUGH_r): Likewise.
1734 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1735 (graphite_copy_stmts_from_block): Skip nonbind markers.
1736 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1737 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1738 * jump.c (clean_barriers): Skip debug insns.
1739 * omp-expand.c (expand_parallel_call): Skip debug insns.
1740 (expand_task_call): Likewise.
1741 (remove_exit_barrier): Likewise.
1742 (expand_omp_taskreg): Likewise.
1743 (expand_omp_for_init_counts): Likewise.
1744 (expand_omp_for_generic): Likewise.
1745 (expand_omp_for_static_nochunk): Likewise.
1746 (expand_omp_for_static_chunk): Likewise.
1747 (expand_omp_simd): Likewise.
1748 (expand_omp_taskloop_for_outer): Likewise.
1749 (expand_omp_taskloop_for_inner): Likewise.
1750 (expand_oacc_for): Likewise.
1751 (expand_omp_sections): Likewise.
1752 (expand_omp_single): Likewise.
1753 (expand_omp_synch): Likewise.
1754 (expand_omp_atomic_load): Likewise.
1755 (expand_omp_atomic_store): Likewise.
1756 (expand_omp_atomic_fetch_op): Likewise.
1757 (expand_omp_atomic_pipeline): Likewise.
1758 (expand_omp_atomic_mutex): Likewise.
1759 (expand_omp_target): Likewise.
1760 (grid_expand_omp_for_loop): Likewise.
1761 (grid_expand_target_grid_body): Likewise.
1762 (build_omp_regions_1): Likewise.
1763 * omp-low.c (check_combined_parallel): Skip debug stmts.
1764 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1765 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1766 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1768 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1769 (make_edges): Likewise.
1770 (cleanup_dead_labels): Likewise.
1771 (gimple_can_merge_blocks_p): Likewise.
1772 (stmt_starts_bb_p): Likewise.
1773 (gimple_block_label): Likewise.
1774 (gimple_redirect_edge_and_branch): Likewise.
1775 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1777 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1779 * tree-pretty-print (print_declaration): Omit initializer in slim
1781 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1783 (eliminate_unnecessary_stmts): Stabilize block removal order.
1784 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1785 * var-tracking.c (get_first_insn): New.
1786 (vt_emit_notes): Call it.
1787 (vt_initialize): Walk any insns before the first BB.
1788 (delete_debug_insns): Likewise.
1790 * gimple.h (enum gimple_debug_subcode): Add
1791 GIMPLE_DEBUG_BEGIN_STMT.
1792 (gimple_debug_begin_stmt_p): New.
1793 (gimple_debug_nonbind_marker_p): New.
1794 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1795 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1796 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1797 * insn-notes.def (BEGIN_STMT): New.
1798 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1799 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1800 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1801 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1802 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1803 (INSN_DEBUG_MARKER_KIND): New.
1804 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1805 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1806 (INSN_VAR_LOCATION_PTR): New.
1807 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1809 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1810 deps only if debug bind insns are enabled.
1811 (pass_expand::execute): Avoid deep TER and expand
1812 debug locations for debug bind insns only.
1813 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1814 debug stmts special handling down to debug bind stmts.
1815 * combine.c (try_combine): Narrow debug insns special handling
1816 down to debug bind insns.
1817 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1818 Narrow debug insns preexisting special handling down to debug
1820 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1821 handling down to debug bind insns.
1822 * function.c (instantiate_virtual_regs): Skip debug markers,
1823 adjust handling of debug binds.
1824 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1825 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1826 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1827 insns to debug bind insns.
1828 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1829 special handling of debug stmts to debug bind stmts.
1830 * ipa-split.c (split_function): Likewise.
1831 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1832 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1834 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1835 * regrename.c (build_def_use): Likewise.
1836 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1837 (pass_cprop_hardreg): Narrow special casing of debug insns to
1839 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1840 * reload1.c (reload): Likewise.
1841 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1842 casing of debug stmts to debug bind stmts.
1843 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1844 * ssa-iterators.h (num_imm_uses): Likewise.
1845 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1846 debug stmts to debug bind stmts.
1847 * tree-inline.c (tree_function_versioning): Narrow special casing
1848 of debug stmts to debug bind stmts.
1849 * tree-loop-distribution.c (generate_loops_for_partition):
1850 Narrow special casing of debug stmts to debug bind stmts.
1851 * tree-sra.c (analyze_access_subtree): Narrow special casing
1852 of debug stmts to debug bind stmts.
1853 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1854 stmts to debug bind stmts.
1855 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1856 casing of debug stmts to debug bind stmts.
1857 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1858 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1859 casing of debug stmts to debug bind stmts.
1860 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1862 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1863 debug stmts to debug bind stmts.
1864 (gimple_replace_ssa_lhs): Likewise.
1865 (insert_debug_temp_for_var_def): Likewise.
1866 (insert_debug_temps_for_defs): Likewise.
1867 (reset_debug_uses): Likewise.
1868 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1869 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1870 (adjust_debug_stmts): Likewise.
1871 (adjust_phi_and_debug_stmts): Likewise.
1872 (vect_do_peeling): Likewise.
1873 * tree-vect-loop.c (vect_transform_loop): Likewise.
1874 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1875 * var-tracking.c (adjust_mems): Narrow special casing of debug
1876 insns to debug bind insns.
1877 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1878 (compute_bb_dataflow, vt_find_locations): Likewise.
1879 (vt_expand_loc, emit_notes_for_changes): Likewise.
1880 (vt_init_cfa_base): Likewise.
1881 (vt_emit_notes): Likewise.
1882 (vt_initialize): Likewise.
1883 (vt_finalize): Likewise.
1885 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1886 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1887 (next_nonnote_nondebug_insn_bb): New.
1888 (prev_nonnote_nondebug_insn_bb): New.
1889 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1890 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1891 (prev_nonnote_nondebug_insn_bb): Declare.
1892 (next_nonnote_nondebug_insn_bb): Declare.
1893 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1894 * cfgrtl.c (get_last_bb_insn): Likewise.
1895 * lra.c (push_insns): Likewise.
1897 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1900 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1902 * selftest.c (selftest::test_runner::test_runner): New ctor.
1903 (selftest::test_runner::~test_runner): New dtor.
1904 * selftest.h (class selftest::test_runner): New class.
1906 2017-12-11 Carl Love <cel@us.ibm.com>
1908 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1909 vec_extract_fp32_from_shortl]): Add #defines.
1910 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1911 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1912 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1913 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1914 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1915 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1916 * doc/extend.texi: Add documentation for the added builtins.
1918 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1920 PR rtl-optimization/80693
1921 PR rtl-optimization/81019
1922 PR rtl-optimization/81020
1923 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1924 are not mentioned in i3. Place the REG_UNUSED note on i2,
1925 possibly modified to REG_DEAD, if it did not originate in i3.
1927 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1929 * recog.c (store_data_bypass_p_1): New function.
1930 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1931 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1933 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1935 PR rtl-optimization/83361
1936 * ifcvt.c (if_convert): Call fixup_partitions.
1938 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1940 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1941 early folding of splat_u{8,16,32}.
1943 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1945 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1946 output_operand_lossage first argument with capital letter.
1947 (aarch64_override_options): Don't start error and sorry first argument
1948 with capital letter.
1950 2017-12-11 Andi Kleen <ak@linux.intel.com>
1952 PR gcov-profile/83355
1953 * auto-profile.c (string_table::get_index_by_decl): Don't
1954 recurse when abstract origin points to itself.
1956 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1958 PR tree-optimization/83320
1959 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1960 (prune_datarefs_not_in_loop): Ditto.
1962 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1964 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1966 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1968 PR tree-optimization/83337
1969 * gimple-loop-interchange.cc (compute_access_stride): Handle
1970 bitfield DRs properly.
1972 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1974 PR tree-optimization/83338
1975 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1976 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1977 vector of unsigned integers to vector of signed integers.
1979 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1981 PR rtl-optimization/83317
1982 * lra-constraints.c (process_address_1): Add insn code check.
1984 2017-12-08 Michael Matz <matz@suse.de>
1986 Fix PR tree-optimization/83323
1987 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1988 head-controlled loops and loop BBs.
1989 * common.opt (funroll-and-jam): Remove, instead ...
1990 (floop-unroll-and-jam): ... reuse this option.
1991 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1992 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1993 (-floop-unroll-and-jam): ... this option.
1995 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1997 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1998 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1999 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2001 2017-12-08 Richard Biener <rguenther@suse.de>
2004 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2005 handle accesses outside of zero-sized vars.
2007 2017-12-08 Martin Jambor <mjambor@suse.cz>
2009 PR tree-optimization/83141
2010 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2011 test for MEM_REFs implicitely changing types with padding. Remove
2013 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2015 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2017 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2019 * config/avr/avr.c (avr_attribute_table): Likewise.
2020 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2021 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2022 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2023 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2024 of struct attribute_spec.
2025 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2027 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2030 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2032 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2035 * profile-count.c (profile_count::from_gcov_type): Move from
2036 profile-count.h; handle overflow.
2037 * profile-count.h (profile_count::from_gcov_type): Move offline.
2039 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2041 PR rtl-optimization/83304
2042 * combine.c (move_deaths): If we do not know where a register died,
2045 2017-12-08 Richard Biener <rguenther@suse.de>
2047 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2048 Provide -fopt-info-loop feedback when we interchange in a nest.
2050 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2052 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2053 for armv6 ARM CPU IDs.
2055 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2057 * common/config/arm/arm-common.c: Include <algorithm>.
2058 (INCLUDE_VECTOR): Define.
2059 (compare_opt_names): New function.
2060 (arm_rewrite_selected_arch): Only strip out extensions that can be
2061 expressed through -mfpu. Sort the remaining extensions
2064 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2066 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2067 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2068 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2069 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2071 2017-06-08 Tristan Gingold <gindold@adacore.com>
2074 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2075 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2077 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2079 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2080 of struct attribute_spec.
2082 2017-12-08 Julia Koval <julia.koval@intel.com>
2084 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2085 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2086 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2087 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2088 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2089 _mm_maskz_dpwssds_epi32): Ditto.
2091 2017-12-08 Richard Biener <rguenther@suse.de>
2093 PR tree-optimization/81303
2094 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2095 conditions try to create a comparison vector type matching
2096 the data vector type.
2097 (vectorizable_condition): Adjust.
2098 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2099 Leave invariant conditions alone in case we can vectorize those.
2101 2017-12-08 Julia Koval <julia.koval@intel.com>
2103 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2104 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2105 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2106 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2107 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2109 2017-12-08 Julia Koval <julia.koval@intel.com>
2111 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2112 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2113 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2114 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2115 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2116 _mm_maskz_dpbusds_epi32): New intrinsics.
2118 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2120 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2122 (ld<bh>io_signed): Likewise.
2123 (st<bhw>io): Likewise.
2124 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2125 SMALL_INT12 constant integer operand.
2127 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2129 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2131 (nios2_large_constant_allowed): ...this. Adjust uses.
2132 (nios2_plus_symbolic_constant_p): Rename to...
2133 (nios2_plus_large_constant_p): ...this. Adjust uses.
2134 (nios2_legitimate_address_p): Correct CONST_INT handling.
2135 (nios2_symbolic_memory_operand_p): Rename to...
2136 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2137 (nios2_large_constant_p): Check for large constant integers too.
2138 (nios2_split_large_constant): Handle constant integers.
2139 (nios2_split_symbolic_memory_operand): Rename to...
2140 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2141 (nios2_legitimize_constant_address): Handle constant integers.
2142 (r0rel_constant_p): Handle small constant integers.
2143 (nios2_print_operand_address): Handle r0-relative integer addresses.
2144 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2145 * config/nios2/nios2.md: Adjust for renamed functions.
2147 2017-12-07 Andrew Waterman <andrew@sifive.com>
2149 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2150 (TARGET_HAVE_SRODATA_SECTION): New define.
2151 (riscv_select_section): New function.
2153 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2154 Alexander Monakov <amonakov@ispras.ru>
2155 Jakub Jelinek <jakub@redhat.com>
2158 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2160 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2162 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2163 a bool success value. Don't call output_operand_lossage here.
2164 (aarch64_print_ldpstp_address): Return a bool success value.
2165 (aarch64_print_operand_address): Call output_addr_const if
2166 aarch64_print_address_internal fails.
2167 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2168 'y'; call output_operand_lossage instead. Call output_operand_lossage
2169 if aarch64_print_ldpstp_address fails.
2171 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2173 * tree-vector-builder.h
2174 (tree_vector_builder::binary_encoded_nelts): Declare.
2175 * tree-vector-builder.c
2176 (tree_vector_builder::binary_encoded_nelts): New function.
2177 * fold-const.c (negate_expr_p): Likewise.
2178 (operand_equal_p, fold_checksum_tree): Likewise.
2179 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2180 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2181 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2182 (uniform_vector_p): Likewise.
2183 * varasm.c (const_hash_1, compare_constant): Likewise.
2184 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2185 (valid_lattice_transition): Operate directly on the VECTOR_CST
2187 * ipa-icf.c: Include tree-vector-builder.h.
2188 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2189 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2191 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2193 * tree.c (build_vector): Delete.
2194 * tree.h (build_vector): Make static and move into the self-testing
2197 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2199 * vector-builder.h (vector_builder::derived): New const overload.
2200 (vector_builder::elt): New function.
2201 * tree-vector-builder.h (tree_vector_builder::type): New function.
2202 (tree_vector_builder::apply_step): Declare.
2203 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2205 * gimple-fold.h (tree_vector_builder): Declare.
2206 (gimple_build_vector): Take a tree_vector_builder instead of a
2207 type and vector of elements.
2208 * gimple-fold.c (gimple_build_vector): Likewise.
2209 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2211 (get_initial_defs_for_reduction): Likewise.
2212 (vectorizable_induction): Likewise.
2214 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2216 * tree-vector-builder.h
2217 (tree_vector_builder::new_binary_operation): Declare.
2218 * tree-vector-builder.c
2219 (tree_vector_builder::new_binary_operation): New function.
2220 * fold-const.c (fold_relational_const): Use it.
2221 (const_binop): Likewise. Check that both input vectors have
2222 the same number of elements, thus excluding things like WIDEN_SUM.
2223 Check whether it is possible to operate directly on the encodings
2226 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2228 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2229 new_unary_operation, operating only on the encoded elements.
2230 (const_unop): Likewise.
2231 (exact_inverse): Likewise.
2232 (distributes_over_addition_p): New function.
2233 (const_binop): Use tree_vector_builder and new_unary_operation
2234 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2235 on the encoded elements unless the encoding is strided and the
2236 operation does not distribute over addition.
2237 (fold_convert_const): Use tree_vector_builder and
2238 new_unary_operation. Operate only on the encoded elements
2239 for truncating integer conversions, or for non-stepped encodings.
2241 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2243 * config/sparc/sparc.c: Include tree-vector-builder.h.
2244 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2245 * expmed.c: Include tree-vector-builder.h.
2246 (make_tree): Use tree_vector_builder instead of build_vector.
2247 * fold-const.c: Include tree-vector-builder.h.
2248 (const_binop): Use tree_vector_builder instead of build_vector.
2249 (const_unop): Likewise.
2250 (native_interpret_vector): Likewise.
2251 (fold_vec_perm): Likewise.
2252 (fold_ternary_loc): Likewise.
2253 * gimple-fold.c: Include tree-vector-builder.h.
2254 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2256 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2257 (simplify_vector_constructor): Use tree_vector_builder instead
2259 * tree-vect-generic.c: Include tree-vector-builder.h.
2260 (add_rshift): Use tree_vector_builder instead of build_vector.
2261 (expand_vector_divmod): Likewise.
2262 (optimize_vector_constructor): Likewise.
2263 * tree-vect-loop.c: Include tree-vector-builder.h.
2264 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2265 of build_vector. Explicitly use a stepped encoding for
2267 * tree-vect-slp.c: Include tree-vector-builder.h.
2268 (vect_get_constant_vectors): Use tree_vector_builder instead
2270 (vect_transform_slp_perm_load): Likewise.
2271 (vect_schedule_slp_instance): Likewise.
2272 * tree-vect-stmts.c: Include tree-vector-builder.h.
2273 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2274 (vect_gen_perm_mask_any): Likewise.
2275 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2276 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2278 (build_vector_from_val): Likewise. Explicitly use a duplicate
2281 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2283 * doc/generic.texi (VECTOR_CST): Describe new representation of
2285 * vector-builder.h: New file.
2286 * tree-vector-builder.h: Likewise.
2287 * tree-vector-builder.c: Likewise.
2288 * Makefile.in (OBJS): Add tree-vector-builder.o.
2289 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2290 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2291 (tree_vector): Change the number of elements to
2292 vector_cst_encoded_nelts.
2293 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2294 (VECTOR_CST_ELTS): Delete.
2295 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2296 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2297 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2298 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2299 (VECTOR_CST_ENCODED_ELT): Likewise.
2300 (vector_cst_encoded_nelts): New function.
2301 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2302 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2303 (vector_cst_int_elt, vector_cst_elt): Declare.
2304 * tree.c: Include tree-vector-builder.h.
2305 (tree_code_size): Abort if passed VECTOR_CST.
2306 (tree_size): Update for new VECTOR_CST layout.
2307 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2308 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2309 (build_vector): Use tree_vector_builder.
2310 (vector_cst_int_elt, vector_cst_elt): New functions.
2311 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2312 encoded elements and then create the vector in the canonical form.
2313 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2314 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2315 (tree_c_tests): Call test_vector_cst_patterns.
2316 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2318 (hash_tree): Likewise.
2319 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2320 (streamer_write_tree_header): Likewise.
2321 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2322 (streamer_alloc_tree): Likewise. Update call to make_vector.
2323 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2325 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2327 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2328 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2329 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2331 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2332 Richard Biener <rguenther@suse.de>
2334 PR tree-optimization/81303
2335 * Makefile.in (gimple-loop-interchange.o): New object file.
2336 * common.opt (floop-interchange): Reuse the option from graphite.
2337 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2338 -floop-interchange and mention it for -O3.
2339 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2340 * gimple-loop-interchange.cc: New file.
2341 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2342 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2343 * passes.def (pass_linterchange): New pass.
2344 * timevar.def (TV_LINTERCHANGE): New time var.
2345 * tree-pass.h (make_pass_linterchange): New declaration.
2346 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2347 interchange. Record IV before/after increment in new parameters.
2348 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2349 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2351 (check_reduction_path): ...New function here.
2352 * tree-vectorizer.h (check_reduction_path): New declaration.
2354 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2357 PR rtl-optimization/80818
2358 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2359 always early clobbered.
2360 * lra-lives.c (process_bb_lives): Check input hard regs for early
2361 clobbered non-operand hard reg.
2363 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2366 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2367 types_compatible_p, just that TYPE_MODE is the same.
2369 2017-12-07 Martin Sebor <msebor@redhat.com>
2372 * attribs.c (empty_attribute_table): Initialize new member of
2373 struct attribute_spec.
2374 (decl_attributes): Add argument. Handle mutually exclusive
2375 combinations of attributes.
2376 (selftests::test_attribute_exclusions): New function.
2377 (selftests::attribute_c_tests): Ditto.
2378 * attribs.h (decl_attributes): Add default argument.
2379 * selftest.h (attribute_c_tests): Declare.
2380 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2381 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2383 * doc/extend.texi (Common Function Attributes): Update const and pure.
2384 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2385 of struct attribute_spec.
2386 * config/arc/arc.c (arc_attribute_table): Same.
2387 * config/arm/arm.c (arm_attribute_table): Same.
2388 * config/avr/avr.c ( avr_attribute_table): Same.
2389 * config/bfin/bfin.c (bfin_attribute_table): Same.
2390 * config/cr16/cr16.c (cr16_attribute_table): Same.
2391 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2392 * config/h8300/h8300.c (h8300_attribute_table): Same.
2393 * config/i386/i386.c (ix86_attribute_table): Same.
2394 * config/ia64/ia64.c (ia64_attribute_table): Same.
2395 * config/m32c/m32c.c (m32c_attribute_table): Same.
2396 * config/m32r/m32r.c (m32r_attribute_table): Same.
2397 * config/m68k/m68k.c (m68k_attribute_table): Same.
2398 * config/mcore/mcore.c (mcore_attribute_table): Same.
2399 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2400 * config/mips/mips.c (mips_attribute_table): Same.
2401 * config/msp430/msp430.c (msp430_attribute_table): Same.
2402 * config/nds32/nds32.c (nds32_attribute_table): Same.
2403 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2404 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2405 * config/rl78/rl78.c (rl78__attribute_table): Same.
2406 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2407 * onfig/rx/rx.c (rx_attribute_table): Same.
2408 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2409 * config/sh/sh.c (sh_attribute_table): Same.
2410 * config/sparc/sparc.c (sparc_attribute_table): Same.
2411 * config/spu/spu.c (spu_attribute_table): Same.
2412 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2413 * config/v850/v850.c (v850_attribute_table): Same.
2414 * config/visium/visium.c (visium_attribute_table): Same.
2416 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2419 * config/arm/arm.c (INCLUDE_STRING): Define.
2420 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2421 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2424 2017-12-07 Michael Matz <matz@suse.de>
2426 Add unroll and jam pass
2428 * gimple-loop-jam.c: New file.
2429 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2430 * common.opt (funroll-and-jam): New option.
2431 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2432 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2433 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2434 * passes.def: Add pass_loop_jam.
2435 * timevar.def (TV_LOOP_JAM): Add.
2436 * tree-pass.h (make_pass_loop_jam): Declare.
2437 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2438 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2439 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2440 to flow_loop_tree_node_add.
2441 (duplicate_subloops, copy_loops_to): Append to sibling list.
2442 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2443 * doc/invoke.texi (-funroll-and-jam): Document new option.
2444 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2446 2017-12-07 Richard Biener <rguenther@suse.de>
2448 PR tree-optimization/83296
2449 PR tree-optimization/67769
2450 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2451 flow sensitive info in an unrelated BB.
2452 (value_replacement): Use reset_flow_sensitive_info.
2453 (minmax_replacement): Reset flow sensitive info on the def
2454 we move. Do not reset flow sensitive info in the whole BB
2455 we move the stmt to.
2456 (abs_replacement): Likewise.
2458 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2462 rs6000_cpu to the given -mcpu=, or to the default processor.
2464 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2466 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2467 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2468 instead of rs6000_cpu_attr.
2469 (rs6000_adjust_cost): Ditto.
2470 (is_microcoded_insn): Ditto.
2471 (rs6000_adjust_priority): Ditto.
2472 (rs6000_issue_rate): Ditto.
2473 (rs6000_use_sched_lookahead): Ditto.
2474 (rs6000_use_sched_lookahead_guard): Ditto.
2475 (rs6000_sched_reorder): Ditto.
2476 (force_new_group): Ditto.
2477 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2478 (group_ending_nop): Ditto.
2480 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2482 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2483 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2484 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2485 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2487 (rs6000_reassociation_width): Ditto.
2488 (rs6000_emit_epilogue): Ditto.
2489 (rs6000_adjust_cost): Ditto.
2490 (is_microcoded_insn): Ditto.
2491 (is_cracked_insn): Ditto.
2492 (rs6000_adjust_priority): Ditto.
2493 (rs6000_sched_reorder): Ditto.
2494 (rs6000_sched_reorder2): Ditto.
2495 (insn_must_be_first_in_group): Ditto.
2496 (insn_must_be_last_in_group): Ditto.
2497 (rs6000_register_move_cost): Ditto.
2498 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2501 2017-12-07 Julia Koval <julia.koval@intel.com>
2503 * config.gcc: Add vaesintrin.h.
2504 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2506 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2507 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2509 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2510 * config/i386/immintrin.h: Include vaesintrin.h.
2511 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2512 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2513 _mm_aesdec_epi128): New intrinsics.
2515 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2517 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2518 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2519 c-family/c-spellcheck.cc.
2520 (best_macro_match::best_macro_match): Likewise.
2522 (struct edit_distance_traits<cpp_hashnode *>): Move to
2523 c-family/c-spellcheck.h.
2524 (class best_macro_match): Likewise.
2526 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2528 PR tree-optimization/83293
2529 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2530 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2531 might insert into empty bb.
2534 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2535 simplify for plus with :c added, and pointer_plus without that.
2536 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2537 with undefined overflow and the conversion is not widening,
2538 perform negation in utype and only convert to type afterwards.
2539 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2540 simplify for plus with :c added, and pointer_plus without that.
2541 If type is integral with undefined overflow and the conversion is
2542 not widening, perform minus in utype and only convert to type
2543 afterwards. Move the last pointer_diff_expr simplify into the
2546 2017-12-06 Martin Sebor <msebor@redhat.com>
2548 PR tree-optimization/82646
2549 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2550 strncpy, not maxlen.
2552 2017-12-06 Martin Sebor <msebor@redhat.com>
2554 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2557 PR tree-optimization/83075
2558 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2559 strncat/strncpy don't change length of source string.
2561 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2564 2017-11-29 Martin Aberg <maberg@gaisler.com>
2566 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2567 to prevent b2bst errata sequence.
2568 (sqrtdf2_fix): Likewise.
2570 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2572 PR tree-optimization/81945
2573 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2574 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2575 to dest_cfun has orig_loop_num set, either remap it to the new
2576 loop number if the loop got moved too, or clear it.
2578 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2580 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2582 (thunderx2t99_nothing): New insn reservation.
2583 (thunderx2t99_mrs): New insn reservation.
2584 (thunderx2t99_multiple): New insn reservation.
2585 (thunderx2t99_alu_basi): Add bfx to reservation.
2586 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2588 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2591 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2593 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2595 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2596 tree-ssa-pre.c::remove_dead_inserted_code.
2597 * tree-ssa-dce.h: New file.
2598 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2599 (remove_dead_inserted_code): Move and rename to function
2600 tree-ssa-dce.c::simple_dce_from_worklist.
2601 (pass_pre::execute): Update use.
2603 2017-12-05 Richard Biener <rguenther@suse.de>
2605 PR tree-optimization/83277
2606 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2607 to code-gen liveout vars.
2609 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2611 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2612 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2613 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2616 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2618 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2620 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2621 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2624 2017-12-05 Richard Biener <rguenther@suse.de>
2626 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2627 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2628 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2629 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2631 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2633 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2635 2017-12-05 Martin Liska <mliska@suse.cz>
2636 Jakub Jelinek <jakub@redhat.com>
2638 * doc/invoke.texi: Document the options.
2639 * flag-types.h (enum sanitize_code): Add
2640 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2641 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2642 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2643 * opts.c: Define new sanitizer options.
2644 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2645 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2647 2017-12-05 Julia Koval <julia.koval@intel.com>
2649 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2650 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2651 (ix86_handle_option): Handle -mavx512vnni.
2652 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2653 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2654 * config/i386/i386-c (__AVX512VNNI__): New.
2655 * config/i386/i386.c (ix86_target_string): Handle new option.
2656 (ix86_valid_target_attribute_inner_p): Handle new option.
2657 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2658 * config/i386/i386.opt (mavx512vnni): New option.
2660 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2663 * athlon.md: Disable for generic.
2664 * haswell.md: Enable for generic.
2665 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2666 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2668 (ix86_adjust_cost): Move generic to haswell path.
2670 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2672 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2673 instead of MEM_P in a couple more places. Fix formatting issues.
2675 2017-12-04 Jim Wilson <jimw@sifive.com>
2677 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2678 instead of GP_REG_LAST-1.
2679 (riscv_adjust_libcall_cfi_prologue): Likewise.
2680 (riscv_adjust_libcall_cri_epilogue): Likewise.
2681 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2684 2017-12-04 Luis Machado <luis.machado@linaro.org>
2686 * ipa-pure-const.c (check_decl): Add missing newline.
2687 (state_from_flags): Likewise.
2689 2017-12-04 Jeff Law <law@redhat.com>
2691 PR tree-optimizatin/78496
2692 * gimple-ssa-evrp-analyze.h
2693 (evrp_range_analyzer::get_vr_values): Simplify.
2694 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2695 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
2696 and gimple-ssa-evrp-analyze.h.
2697 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
2698 (simplify_stmt_for_jump_threading): Copy a blob of code from
2699 tree-vrp.c to use ranges to simplify statements.
2700 (dom_opt_dom_walker::before_dom_children): Call
2701 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
2702 (dom_opt_dom_walker::after_dom_children): Similarly for
2703 evrp_range_analyzer::leave.
2704 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
2707 * gimple-ssa-evrp-analyze.c
2708 (evrp_range_analyzer::extract_range_from_stmt): Always use
2709 vr_values::update_value_range so preexisting range info is
2710 medged with new range info, even if the new range is VR_VARYING.
2712 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2714 * combine.c: Adjust comment.
2715 (use_crosses_set_p): Delete.
2716 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
2717 (try_combine): Ditto.
2719 2017-12-04 Richard Biener <rguenther@suse.de>
2721 PR tree-optimization/83255
2722 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2723 Re-add zero-iteration check.
2725 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2727 PR rtl-optimization/83245
2728 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2729 hard registers as earlyclobber, also if not in an asm.
2731 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2735 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2739 rs6000_cpu based on cpu_index, not tune_index.
2741 2017-12-04 Richard Biener <rguenther@suse.de>
2743 PR tree-optimization/83238
2744 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2745 code match comment, rejecting invalid SESE regions.
2747 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2749 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2750 require base operand is a REG_POINTER prior to reload on targets
2751 with non-equivalent space registers.
2753 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2755 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2756 (update_specialized_profile): Fix updating of counts.
2757 (perhaps_add_new_callers): Likewise.
2759 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2762 * x86-tune.def: Remove obsolette FIXMEs.
2763 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2764 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2765 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2767 (X86_TUNE_PAD_RETURNS): Disable for generic.
2769 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2771 PR tree-optimization/83170
2772 PR tree-optimization/83241
2773 * gimple-ssa-store-merging.c
2774 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2775 gimple_vuse (ins_stmt) in case it has changed.
2776 (imm_store_chain_info::output_merged_store): Likewise.
2778 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2782 * tree-cfg.c (pass_warn_function_return::execute): Handle
2783 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2784 with BUILTINS_LOCATION.
2788 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2789 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2790 instead of DECL_MODE.
2792 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2793 last argument to rtx pointer.
2794 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2795 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2796 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2797 depending on the chosen ISAs.
2798 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2799 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2800 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2802 * config/i386/sse.md (mov<mode>_internal): Likewise.
2803 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2805 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2807 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2809 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2811 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2812 parameters from prototype.
2813 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2814 parameters. Don't print a comment.
2815 (emit_fusion_gpr_load): Adjust.
2816 (emit_fusion_load_store): Adjust.
2817 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2818 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2819 comment on the second line.
2821 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2824 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2825 rs6000_cpu based on cpu_index, not tune_index.
2827 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2829 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2830 which_alternative instead of which_alternative + 1.
2831 (output_asm_insn): Print an extra tab if the template is short.
2833 2017-12-01 Jim Wilson <jimw@sifive.com>
2835 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2837 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2838 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2840 * doc/tm.texi: Regenerate.
2842 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2845 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2846 whether we can allocate pseudos before trying to fix an address.
2847 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2848 memory address is indexed or indirect.
2849 (floatuns_<mode>si2_hw2): Likewise.
2851 2017-12-01 Jason Merrill <jason@redhat.com>
2853 * Makefile.in (TAGS): Add c-family/*.cc.
2855 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2857 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2858 (call_value_insn): Likewise.
2859 (sibcall_insn): Likewise.
2860 (sibcall_value_insn): Likewise.
2861 (movsi_aarch64): Likewise.
2862 (movdi_aarch64): Likewise.
2863 (add_losym_): Likewise.
2864 (ldr_got_small_): Likewise.
2865 (ldr_got_small_sidi): Likewise.
2866 (ldr_got_small_28k_): Likewise.
2867 (ldr_got_small_28k_sidi): Likewise.
2868 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2869 Move output_addr_const to symbolic case. Add error check.
2871 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2873 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2874 * config/s390/s390.md: Disable MVC merging peephole if it would
2875 disable operand forwarding.
2876 (new peephole2): Split MVCs if it would turn them into up to 2
2879 2017-12-01 Richard Biener <rguenther@suse.de>
2881 PR tree-optimization/83232
2882 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2883 detection of same access. Instead of breaking the group here
2884 do not consider the duplicate. Add comment explaining real fix.
2886 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2888 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2890 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2892 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2894 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2896 * function.h (struct function): Remove cilk_frame_decl,
2897 is_cilk_function and calls_cilk_spawn fields.
2898 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2900 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2901 cilk_elemental field.
2902 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2903 * target.def: Adjust comment.
2904 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2905 Don't test cilk_elemental.
2907 PR tree-optimization/83233
2908 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2909 bswap_stat name for the struct.
2912 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2913 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2914 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2915 SWITCH_BREAK_LABEL_P set on the label.
2916 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2917 added for default case if it was missing and not all cases covered.
2918 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2919 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2921 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2922 to avoid -Wimplicit-fallthrough warning.
2923 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2924 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2926 PR tree-optimization/83221
2927 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2929 (init_reassoc): Formatting fix.
2932 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2933 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2937 * tree-cfg.c: Include asan.h.
2938 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2940 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2942 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2944 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2946 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2947 vpcmpeqd instruction.
2949 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2951 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2952 registers type for 128bit mode.
2954 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2956 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2958 * gengtype-state.c (read_state_pair): Likewise.
2959 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2960 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2961 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2962 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2963 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2964 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2965 * builtins.c (fold_builtin_3): Likewise.
2966 * graphite-scop-detection.c
2967 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2968 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2971 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2974 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2975 parameter. Return true if ignore_decl_rtl_set_p is true and other
2976 conditions are satisfied.
2977 * asan.h (asan_protect_global): Add new parameter.
2978 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2979 to asan_protect_global calls.
2981 2017-11-30 Jim Wilson <jimw@sifive.com>
2983 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2984 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2985 -mstrict-align, add info on default value. Delete redundant lines for
2986 -mabi. Add missing -mexplicit-relocs docs.
2988 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2990 * config/arc/arc.md (trap): New pattern.
2992 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2994 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2995 instruction to end into a delay slot.
2996 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2997 can be placed into a delay slot against reg_note.
2999 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3001 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3002 labels number of usages.
3004 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3006 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3008 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3010 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3013 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3014 multiplication by power of 2 constant into two shifts + comparison.
3016 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3019 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3021 2017-11-30 Richard Biener <rguenther@suse.de>
3023 PR tree-optimization/83202
3024 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3025 allow_peel argument and guard peeling.
3026 (canonicalize_loop_induction_variables): Likewise.
3027 (canonicalize_induction_variables): Pass false.
3028 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3029 peeling from cunrolli.
3031 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3033 * combine.c (try_combine): Print a message to dump file whenever
3034 I0, I1, or I2 cannot be combined into I3.
3036 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3038 PR rtl-optimization/83156
3039 PR rtl-optimization/82621
3040 * combine.c (try_combine): Don't split an I2 if one of the dests is
3041 set again before I3. Allow unused dests.
3043 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3045 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3047 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3049 PR rtl-optimization/80818
3050 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3051 recursively. Use insn code for clobber.
3052 (lra_set_insn_recog_data): Pass the new arg to
3053 collect_non_operand_hard_regs.
3054 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3056 (lra_update_insn_regno_info): Pass insn to
3057 add_regs_to_insn_regno_info.
3059 2017-11-29 Jim Wilson <jimw@sifive.com>
3060 Andrew Waterman <andrew@sifive.com>
3062 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3063 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3064 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3065 use SINGLE_SHIFT_COST.
3066 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3067 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3069 2017-11-29 Julia Koval <julia.koval@intel.com>
3071 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3072 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3073 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3074 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3075 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3076 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3077 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3078 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3079 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3080 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3081 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3082 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3083 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3084 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3085 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3086 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3087 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3088 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3089 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3090 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3091 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3092 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3093 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3094 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3095 __builtin_ia32_vpshldv_v2di_mask,
3096 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3097 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3098 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3100 2017-11-29 Julia Koval <julia.koval@intel.com>
3102 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3103 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3104 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3105 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3106 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3107 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3108 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3109 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3110 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3111 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3112 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3113 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3114 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3115 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3116 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3117 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3118 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3119 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3120 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3121 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3122 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3124 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3126 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3127 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3130 2017-11-29 Martin Aberg <maberg@gaisler.com>
3132 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3133 to prevent b2bst errata sequence.
3134 (sqrtdf2_fix): Likewise.
3136 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3138 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3139 (div_sqrt_insn_p): New function.
3140 (sparc_do_work_around_errata): Insert NOP instructions to
3141 prevent sequences that could trigger the TN-0013 errata for
3142 certain LEON3 processors.
3143 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3144 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3145 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3146 (in_branch_delay): Prevent div and sqrt in delay slot if
3148 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3150 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3152 * config/sparc/sparc.c (atomic_insn_p): New function.
3153 (sparc_do_work_around_errata): Insert NOP instructions to
3154 prevent sequences that could trigger the TN-0010 errata for
3156 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3157 instruction referable in atomic_insns_p.
3159 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3161 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3162 (atomic_compare_and_swap_leon3_1): Likewise.
3165 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3167 * config/sparc/sparc.c (fpop_insn_p): New function.
3168 (sparc_do_work_around_errata): Insert NOP instructions to
3169 prevent sequences that could trigger the TN-0012 errata for
3171 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3172 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3173 (in_branch_annul_delay): Prevent floating-point instructions
3174 in delay slot of annulled integer branch.
3176 2017-11-29 Richard Biener <rguenther@suse.de>
3178 PR tree-optimization/83202
3179 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3181 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3182 nodes vectorized to the same stmts multiple times.
3183 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3184 (vect_analyze_slp_instance): Adjust.
3185 (scalar_stmts_to_slp_tree_map_t): New typedef.
3186 (vect_schedule_slp_instance): Add a map recording the SLP node
3187 representing the vectorized stmts for a set of scalar stmts.
3188 Avoid code-generating redundancies.
3189 (vect_schedule_slp): Allocate map and pass it down.
3191 2017-11-29 Nathan Sidwell <nathan@acm.org>
3194 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3197 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3199 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3200 cases for printing LDP/STP memory addresses.
3201 (aarch64_print_address_internal): Renamed from
3202 aarch64_print_operand_address, added parameter, add Pmode check.
3203 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3204 (aarch64_print_operand_address): Indirect to
3205 aarch64_print_address_internal.
3206 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3207 'y' operand output specifier.
3209 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3212 * tree.c (build_simple_mem_ref_loc): Handle
3213 get_addr_base_and_unit_offset returning a MEM_REF.
3216 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3219 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3222 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3224 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3226 (layout::move_to_column): Likewise.
3228 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3231 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3232 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3234 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3235 Martin Jambor <mjambor@suse.cz>
3238 * tree.h (expr_type_first_operand_type_p): Declare
3239 * tree.c (expr_type_first_operand_type_p): New function.
3240 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3241 (ipa_value_from_jfunc): Adjust declaration.
3242 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3243 Use it as result type for arithmetics, unless it is NULL in which case
3244 be more conservative.
3245 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3246 ipa_get_jf_pass_through_result.
3247 (propagate_vals_across_pass_through): Likewise.
3248 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3249 is to propagate_vals_across_pass_through.
3250 (propagate_constants_across_call): Pass PARM_TYPE to
3251 propagate_scalar_across_jump_function.
3252 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3253 ipa_value_from_jfunc.
3254 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3255 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3256 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3257 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3258 target_type, pass it to ipa_value_from_jfunc.
3259 (update_indirect_edges_after_inlining): Pass parameter type to
3260 try_make_edge_direct_simple_call.
3262 2017-11-28 Jeff Law <law@redhat.com>
3264 * gimple-ssa-evrp-analyze.c
3265 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3266 refine ranges if scev_initialized_p returns true.
3267 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3269 2017-11-28 Julia Koval <julia.koval@intel.com>
3271 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3272 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3273 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3274 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3275 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3276 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3277 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3278 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3279 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3280 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3281 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3282 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3283 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3284 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3285 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3286 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3287 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3288 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3289 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3290 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3291 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3292 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3294 2017-11-28 Julia Koval <julia.koval@intel.com>
3296 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3297 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3298 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3299 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3300 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3301 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3302 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3303 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3304 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3305 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3306 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3307 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3308 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3309 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3310 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3311 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3312 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3313 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3314 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3315 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3316 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3317 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3318 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3319 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3320 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3321 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3322 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3323 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3324 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3325 __builtin_ia32_vpshld_v2di_mask): New builtins.
3326 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3327 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3329 2017-11-28 Richard Biener <rguenther@suse.de>
3331 PR tree-optimization/80776
3332 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3334 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3336 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3337 If the incoming edge is an effective fallthru because the other
3338 edge only reaches a __builtin_unreachable () then record ranges
3339 derived from the controlling condition in SSA info.
3340 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3341 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3343 2017-11-28 Olivier Hainque <hainque@adacore.com>
3345 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3346 on mingw build hosts.
3348 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3351 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3353 2017-11-28 Richard Biener <rguenther@suse.de>
3356 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3357 copies generated from memcpy use a character array as reference
3360 2017-11-28 Julia Koval <julia.koval@intel.com>
3361 Sebastian Peryt <sebastian.peryt@intel.com>
3363 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3364 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3365 cilk-common.o, cilk.h, cilk-common.c): Remove.
3367 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3368 * builtins.c (is_builtin_name): Remove cilkplus condition.
3369 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3370 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3371 cilk-builtins.def, cilkplus.def): Remove.
3372 * cif-code.def (CILK_SPAWN): Remove.
3373 * cilk-builtins.def: Delete.
3374 * cilk-common.c: Ditto.
3376 * cilkplus.def: Ditto.
3377 * config/darwin.h (fcilkplus): Delete.
3378 * cppbuiltin.c: Ditto.
3379 * doc/extend.texi: Remove cilkplus doc.
3380 * doc/generic.texi: Ditto.
3381 * doc/invoke.texi: Ditto.
3382 * doc/passes.texi: Ditto.
3383 * gcc.c (fcilkplus): Remove.
3384 * gengtype.c (cilk.h): Remove.
3385 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3387 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3389 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3391 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3392 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3393 cilkplus conditions.
3394 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3395 inline_read_section): Ditto.
3396 * ipa-inline-analysis.c (cilk.h): Remove.
3397 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3398 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3399 append_linker_options): Remove condition for fcilkplus.
3400 * lto/lto-lang.c (cilk.h): Remove.
3401 (lto_init): Remove condition for fcilkplus.
3402 * omp-expand.c (expand_cilk_for_call): Delete.
3403 (expand_omp_taskreg, expand_omp_for_static_chunk,
3404 expand_omp_for): Remove cilkplus
3406 (expand_cilk_for): Delete.
3407 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3408 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3409 execute_lower_omp, diagnose_sb_0): Ditto.
3410 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3411 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3412 * tree-nested.c: Ditto.
3413 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3414 (dump_generic_node): Ditto.
3415 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3416 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3418 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3420 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3422 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3423 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3424 for vector mode and !TARGET_SIMD.
3426 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3428 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3430 * tree.h (SWITCH_LABELS): Remove.
3431 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3432 assert SWITCH_BODY is non-NULL.
3433 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3435 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3437 PR tree-optimization/80788
3438 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3439 has TREE_OVERFLOW set, call drop_tree_overflow.
3441 2017-11-28 Richard Biener <rguenther@suse.de>
3443 PR tree-optimization/83158
3444 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3446 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3449 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3450 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3452 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3454 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3455 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3457 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3460 * varasm.c (compare_constant): Take the mode of the constants into
3461 account when comparing floating point constants.
3463 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3465 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3466 from explicit instantiation of debug_helper.
3467 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3469 2017-11-27 Richard Biener <rguenther@suse.de>
3471 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3474 2017-11-27 Richard Biener <rguenther@suse.de>
3476 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3477 (type_hash_canon): Also clear int_cst_hash_table entry for
3479 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3481 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3483 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3484 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3485 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3487 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3489 * hash-map.h (gt_cleare_cache): Avoid UB.
3491 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3493 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3494 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3495 (decide_unroll_runtime_iterations): Likewise.
3496 (decide_unroll_stupid): Likewise.
3498 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3501 * config/i386/i386.md: Add a loop with incssp.
3503 2017-11-27 Martin Jambor <mjambor@suse.cz>
3505 PR tree-optimization/81248
3506 * tree-sra.c (splice_param_accesses): Remove size check.
3507 (decide_one_param_reduction): Fix size check.
3508 * gimple-pretty-print.c (dump_profile): Silence warning.
3509 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3511 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3514 * dbxout.c (lastlineno): New variable.
3515 (dbx_debug_hooks): Use dbxout_switch_text_section as
3516 switch_text_section debug hook.
3517 (dbxout_function_end): Switch to current_function_section
3518 rather than function_section. If crtl->has_bb_partition,
3519 output just one N_FUN, depending on in_cold_section_p.
3520 (dbxout_source_line): Remember last lineno in lastlineno.
3521 (dbxout_switch_text_section): New function.
3522 (dbxout_function_decl): Adjust dbxout_block caller.
3523 (dbx_block_with_cold_children): New function.
3524 (dbxout_block): Return true if any LBRAC/RBRAC have been
3525 emitted. Use dbx_block_with_cold_children at depth == 0
3526 in second partition. Add PARENT_BLOCKNUM argument, pass
3527 it optionally adjusted to children. Output LBRAC/RBRAC
3528 around recursive call only if the block is in the current
3529 partition, if not and anything was output, emit empty
3531 * final.c (final_scan_insn): Compute cold_function_name
3532 before calling switch_text_section debug hook. Call
3533 that hook even if dwarf2out_do_frame if not emitting
3537 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3538 TREE_READONLY decls.
3540 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3542 PR rtl-optimization/82488
3543 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3545 2017-11-26 Julia Koval <julia.koval@intel.com>
3547 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3549 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3552 2017-11-26 Julia Koval <julia.koval@intel.com>
3554 * config/i386/driver-i386.c (host_detect_local_cpu):
3555 Detect skylake-avx512.
3557 2017-11-26 Julia Koval <julia.koval@intel.com>
3559 * config.gcc: Add -march=cannonlake.
3560 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3561 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3562 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3563 (PTA_CANNONLAKE): New.
3564 (processor_target_table): Add cannonlake.
3565 (ix86_option_override_internal): Ditto.
3566 (fold_builtin_cpu): Ditto.
3567 (get_builtin_code_for_version): Handle cannonlake.
3568 (M_INTEL_COREI7_CANNONLAKE): New.
3569 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3570 * doc/invoke.texi: Add -march=cannonlake.
3572 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3574 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3575 (try_init_one_plugin): Alternative implementation for MinGW.
3576 * Makefile.in (plugin_implib): New.
3577 (gengtype-lex.c): Fix broken AIX workaround.
3578 * configure: Regenerate.
3579 * doc/plugins.texi: Document support for MinGW.
3581 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3583 PR rtl-optimization/81553
3584 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3585 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3586 is a shift where C1 has different mode than the whole shift, use C1's
3587 mode for MULT rather than the shift's mode.
3590 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3591 builtins not enabled in the currently selected ISA.
3593 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3595 PR tree-optimization/71026
3596 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3597 (insert_reciprocals): Change to insert reciprocals before a division
3598 by a square and to insert the square of a reciprocal.
3599 (execute_cse_reciprocals_1): Change to consider division by a square.
3600 (register_division_in): Add importance parameter.
3602 2017-11-24 Richard Biener <rguenther@suse.de>
3604 PR tree-optimization/82402
3605 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3606 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3608 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3610 * match.pd (0-ptr): New transformation.
3612 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3615 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3616 when skipping checking.
3618 2017-11-24 Richard Biener <rguenther@suse.de>
3620 PR tree-optimization/83128
3621 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3622 (vn_reference_lookup_3): Likewise.
3624 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3627 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3628 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3631 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3632 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3633 check for the latter with gimple_call_builtin_p. Do not handle
3634 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3636 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3638 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3641 2017-11-23 Julia Koval <julia.koval@intel.com>
3643 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3644 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3645 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3646 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3647 _mm512_maskz_expandloadu_epi16): New intrinsics.
3648 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3649 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3650 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3651 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3652 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3653 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3654 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3655 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3656 _mm256_maskz_expandloadu_epi8): New intrinsics.
3657 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3658 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3659 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3660 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3661 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3662 * config/i386/sse.md (VI248_VLBW): New iterator.
3663 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3665 2017-11-23 Julia Koval <julia.koval@intel.com>
3667 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3668 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3669 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3670 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3671 _mm512_mask_compressstoreu_epi16): New.
3672 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3673 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3674 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3675 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3676 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3677 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3678 _mm256_mask_compressstoreu_epi8): New.
3679 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3680 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3681 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3682 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3683 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3684 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3685 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3686 __builtin_ia32_compresshi128_mask,
3687 __builtin_ia32_compressstoreuqi512_mask,
3688 __builtin_ia32_compressstoreuhi512_mask,
3689 __builtin_ia32_compressstoreuqi256_mask,
3690 __builtin_ia32_compressstoreuqi128_mask,
3691 __builtin_ia32_compressstoreuhi256_mask,
3692 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3693 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3695 (ix86_expand_special_args_builtin): Handle new types.
3696 (s4fma_expand): Handle new builtin array.
3697 * config/i386/immintrin.h: Include new headers.
3698 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
3699 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
3701 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3703 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
3704 fixes. Declare temp and g variables at the top in order to avoid
3705 {} in most of the cases.
3707 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
3709 * match.pd (ptr-0): New transformation.
3711 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3713 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
3714 (struct addr_mode_cost_table): New.
3715 (struct tune_params): Add field addr_mode_costs.
3716 * config/arm/arm.c (generic_addr_mode_costs): New.
3717 (arm_slowmul_tune): Initialise addr_mode_costs field.
3718 (arm_fastmul_tune): Likewise.
3719 (arm_strongarm_tune): Likewise.
3720 (arm_xscale_tune): Likewise.
3721 (arm_9e_tune): Likewise.
3722 (arm_marvell_pj4_tune): Likewise.
3723 (arm_v6t2_tune): Likewise.
3724 (arm_cortex_tune): Likewise.
3725 (arm_cortex_a8_tune): Likewise.
3726 (arm_cortex_a7_tune): Likewise.
3727 (arm_cortex_a15_tune): Likewise.
3728 (arm_cortex_a35_tune): Likewise.
3729 (arm_cortex_a53_tune): Likewise.
3730 (arm_cortex_a57_tune): Likewise.
3731 (arm_exynosm1_tune): Likewise.
3732 (arm_xgene1_tune): Likewise.
3733 (arm_cortex_a5_tune): Likewise.
3734 (arm_cortex_a9_tune): Likewise.
3735 (arm_cortex_a12_tune): Likewise.
3736 (arm_cortex_a73_tune): Likewise.
3737 (arm_v7m_tune): Likewise.
3738 (arm_cortex_m7_tune): Likewise.
3739 (arm_v6m_tune): Likewise.
3740 (arm_fa726te_tune): Likewise.
3741 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3743 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3745 * config/arm/arm.c (arm_mem_costs): New function.
3746 (arm_rtx_costs_internal): Use arm_mem_costs.
3748 2017-11-23 Mark Wielaard <mark@klomp.org>
3750 * dwarf2out.c (init_sections_and_labels): Use generation to create
3751 unique ranges_section_label and ranges_base_label. Return generation.
3752 (output_rnglists): Add generation argument. Use generation to create
3753 unique ranges labels.
3754 (dwarf2out_finish): Get generation from init_sections_and_labels
3755 and pass generation to output_rnglists.
3757 2017-11-23 Mike Stump <mikestump@comcast.net>
3758 Eric Botcazou <ebotcazou@adacore.com>
3760 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3761 * cfgloop.h (struct loop): Add unroll field.
3762 * function.h (struct function): Add has_unroll bitfield.
3763 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3764 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3765 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3766 (pass_rtl_unroll_loops::gate): Likewise.
3767 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3768 for which loop->unroll==1.
3769 (decide_unroll_constant_iterations): Use note for consistency and
3770 take loop->unroll into account. Return early if loop->unroll is set.
3771 Fix thinko in existing test.
3772 (decide_unroll_runtime_iterations): Use note for consistency and
3773 take loop->unroll into account.
3774 (decide_unroll_stupid): Likewise.
3775 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3776 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3777 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3779 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3780 (print_loop): Print loop->unroll if set.
3781 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3782 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3783 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3785 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3786 loop->unroll is set and smaller than the trip count. Otherwise bypass
3787 entirely the heuristics if loop->unroll is set. Remove dead note.
3788 Fix off-by-one bug in other note.
3789 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3790 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3792 (tree_unroll_loops_completely): Make static.
3793 (pass_complete_unroll::execute): Use correct type for variable.
3794 (pass_complete_unrolli::execute): Fix formatting.
3795 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3797 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3799 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3800 enable when TARGET_PREFER_AVX128 is set.
3802 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3804 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3806 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3807 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3809 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3811 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3812 frequencies. Fix estimation of aggregate parameters.
3814 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3816 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3819 2017-11-23 Tom de Vries <tom@codesourcery.com>
3821 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3823 2017-11-23 Tom de Vries <tom@codesourcery.com>
3825 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3828 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3831 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3832 sibcall_value_pcrel_fdpic): Use local variable instead of
3834 (calli_tbr_rel): Add missing operand 2.
3835 (call_valuei_tbr_rel): Add missing operand 3.
3837 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3840 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3841 bitpos/bitsize covering the whole destination, use store_expr only if
3842 the complex mode is the same. Otherwise, use expand_normal and if
3843 it returns CONCAT, subreg each part separately instead of trying to
3844 subreg the whole result.
3846 2017-11-23 Richard Biener <rguenther@suse.de>
3848 PR tree-optimization/23094
3849 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3850 come here from walking over backedges in the first iteration.
3851 (vn_reference_lookup_3): Skip clobbers that store the same value.
3853 2017-11-23 Richard Biener <rguenther@suse.de>
3855 PR tree-optimization/81403
3856 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3857 a block we need a leader relative to.
3858 (phi_translate_1): For nary processing require a leader from
3859 get_representative_for given we run expression simplification
3860 using match-and-simplify. Remove previous fix.
3862 2017-11-22 Jeff Law <law@redhat.com>
3864 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3865 Use new method allocate_value_range rather than accessing the
3866 vrp_value_range_pool data member directly.
3867 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3868 to use extract_range_from_stmt method to avoid need for
3869 extract_range_from_assignment method.
3870 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3871 method rather than setting values_propgated data member directly.
3872 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3873 and values propagated data members and extract_range_from_assignment
3874 method. Reorder private data members to conform to standards.
3875 Add new methods set_lattice_propagation_complete and
3876 allocate_value_range.
3878 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3880 PR rtl-optimization/83030
3881 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3882 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3883 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3884 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3885 CROSSING_JUMP_P flag is already set before setting it.
3886 * cfgrtl.c (fixup_partition_crossing): Likewise.
3887 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3890 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3892 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3893 Handle the case where both arguments are using gen_const_vec_series.
3895 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3898 * pretty-print.c (pp_format): Move quoting implementation to
3899 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3900 to pass address of "quote" local.
3901 (pp_begin_quote): New function.
3902 (pp_end_quote): New function.
3903 * pretty-print.h (printer_fn): Convert penultimate param from bool
3905 (pp_begin_quote): New decl.
3906 (pp_end_quote): New decl.
3907 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3908 param from bool to bool *.
3909 * tree-diagnostic.h (default_tree_printer): Likewise.
3911 2017-11-22 Jeff Law <law@redhat.com>
3913 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3914 Initialize vr_values.
3915 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3916 vr_values via vr_values class instance rather than delegators.
3917 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3918 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3919 (evrp_range_analyzer::push_value_range): Likewise.
3920 (evrp_range_analyzer::pop_value_range): Likewise.
3921 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3922 most delegators. Those remaining are exposed as public interfaces.
3923 Make vr_values a pointer and private.
3924 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3926 (evrp_range_analyzer::get_vr_value): New method.
3927 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3928 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3929 (class evrp_dom_walker): Attach evrp_folder class, initialize
3930 it in the ctor. Remove temporary delegators.
3931 (evrp_dom_walker::before_dom_children): Call methods in attached
3932 evrp_range_analyzer class via class instance pointer. Use
3933 free value_range_constant_singleton to remove need for
3934 op_with_constant_singleton_value delegator method. Do not
3935 create a vrp_prop class instance for every call! Narrow
3936 scope of a couple variables.
3937 (evrp_dom_walker::cleanup): Call methods in attached
3938 evrp_range_analyzer class via class instance pointer.
3939 * vr-values.h (class vr_values): Privatize many methods and
3942 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3944 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3945 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3946 * expr.c (expand_expr_real_2): Likewise.
3947 * fold-const.c (const_unop): Likewise.
3948 * optabs-tree.c (optab_for_tree_code): Likewise.
3949 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3950 * tree-inline.c (estimate_operator_cost): Likewise.
3951 * tree-pretty-print.c (dump_generic_node): Likewise.
3952 (op_code_prio): Likewise.
3953 (op_symbol_code): Likewise.
3954 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3955 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3956 * internal-fn.c (direct_internal_fn_optab): New function.
3957 (direct_internal_fn_array, direct_internal_fn_supported_p
3958 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3959 * fold-const-call.c (fold_const_reduction): New function.
3960 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3962 * tree-vect-loop.c: Include internal-fn.h.
3963 (reduction_code_for_scalar_code): Rename to...
3964 (reduction_fn_for_scalar_code): ...this and return an internal
3966 (vect_model_reduction_cost): Take an internal_fn rather than
3968 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3970 (vectorizable_reduction): Use internal functions rather than tree
3971 codes for the reduction operation. Update calls to the functions
3973 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3974 Use calls to internal functions rather than REDUC tree codes.
3975 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3977 2017-11-22 Olivier Hainque <hainque@adacore.com>
3979 * config/vxworks.c (vxworks_override_options): Pick default
3980 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3981 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3982 DWARF_GNAT_ENCODINGS_DEFAULT.
3983 * config/vxworksae.h: Likewise.
3985 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3987 PR tree-optimization/83104
3988 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3991 2017-11-22 Marek Polacek <polacek@redhat.com>
3992 H.J. Lu <hongjiu.lu@intel.com>
3993 Jason Merrill <jason@redhat.com>
3998 * calls.c (initialize_argument_information): Call
3999 warn_parameter_passing_abi target hook.
4000 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4001 argument onto stack.
4002 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4003 * common.opt: Update -fabi-version description.
4004 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4005 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4007 (ix86_is_empty_record): New function.
4008 (ix86_warn_parameter_passing_abi): New function.
4009 (TARGET_EMPTY_RECORD_P): Redefine.
4010 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4011 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4012 * doc/tm.texi: Regenerated.
4013 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4014 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4015 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4016 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4017 instead of int_size_in_bytes.
4018 * expr.c (copy_blkmode_to_reg): Likewise.
4019 * function.c (aggregate_value_p): Return 0 for empty types.
4020 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4022 (locate_and_pad_parm): Call arg size_in_bytes instead
4024 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4025 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4026 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4028 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4029 (std_gimplify_va_arg_expr): Skip empty records. Call
4030 arg_size_in_bytes instead size_in_bytes.
4031 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4032 * tree-core.h (tree_type_common): Add empty_flag.
4033 (tree_decl_common): Update comments.
4034 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4036 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4037 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4039 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4040 * tree.c (default_is_empty_type): New function.
4041 (default_is_empty_record): New function.
4042 (arg_int_size_in_bytes): New function.
4043 (arg_size_in_bytes): New function.
4044 (get_ultimate_context): New function.
4045 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4046 TRANSLATION_UNIT_WARN_EMPTY_P.
4047 (default_is_empty_record, arg_int_size_in_bytes,
4048 arg_size_in_bytes, get_ultimate_context): Declare.
4050 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4052 * config/arm/arm.c (cmse_clear_registers): New function.
4053 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4054 code by call to cmse_clear_registers.
4055 (cmse_nonsecure_entry_clear_before_return): Likewise.
4057 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4059 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4060 (vdot_s32, vdotq_s32): New.
4061 (vdot_lane_u32, vdotq_lane_u32): New.
4062 (vdot_lane_s32, vdotq_lane_s32): New.
4065 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4068 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4069 for unsigned values with fewer HWIs than the precision.
4070 (test_overflow): New function.
4071 (wide_int_cc_tests): Call it.
4073 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4075 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4076 has the same mode class as Pmode.
4078 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4080 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4081 padding_bits_to_clear_ptr.
4082 (cmse_nonsecure_entry_clear_before_return): Likewise.
4084 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4086 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4087 auto_sbitap instead of integer bitfield to control register needing
4090 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4092 PR tree-optimization/83044
4093 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4094 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4095 that need the upper bound. Subtract offset from
4096 get_addr_base_and_unit_offset only if positive and subtract it
4097 before division by eltsize rather than after it.
4100 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4101 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4102 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4106 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4107 expand_binop with *mul_widen_optab, make sure at least one of the
4108 operands doesn't have VOIDmode.
4111 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4113 PR rtl-optimization/82044
4114 PR tree-optimization/82042
4115 * dse.c (record_store): Check for overflow.
4116 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4117 add_wild_read instead of clear_rhs_from_active_local_stores on
4118 overflow and log it into dump_file.
4120 2017-11-22 Richard Biener <rguenther@suse.de>
4122 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4123 fake edges to exit when looking for a place to insert.
4124 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4126 (insert_into_preds_of_block): Commit edge insertion immediately,
4127 assert that doesn't require new BBs.
4128 (fini_pre): Release expressions.
4129 (pass_pre::execute): Shuffle things around a bit, if the fn
4130 is too large do not compute AVAIL either as this is really the
4133 2017-11-22 Richard Biener <rguenther@suse.de>
4135 PR tree-optimization/83089
4136 * tree-if-conv.c (pass_if_conversion::execute): If anything
4137 changed reset SCEV and free the number of iteration estimates.
4139 2017-11-21 Martin Sebor <msebor@redhat.com>
4141 PR tree-optimization/82945
4142 * calls.h (warn_nonstring_bound): Remove unused function.
4144 2017-11-21 Martin Sebor <msebor@redhat.com>
4146 PR tree-optimization/82945
4147 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4148 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4149 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4151 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4152 * calls.h (get_attr_nonstring_decl): Declare new function.
4153 * doc/extend.texi (attribute nonstring): Update.
4154 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4155 get_attr_nonstring_decl and handle it.
4156 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4157 detection of nul-termination.
4158 (strlen_to_stridx): Change to a pointer.
4159 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4160 (pass_strlen::execute): Same.
4162 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4164 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4165 for the new option -mprefer-vector-width=[none|128|256|512].
4166 * config/i386/i386.c (ix86_target_string): remove old style options
4167 -mprefer-avx256 and make -mprefer-avx128 as alias.
4168 (ix86_option_override_internal): Apply defaults for the
4169 -mprefer-vector-width=[128|256] option.
4170 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4171 Implement macros to work with -mprefer-vector-width=.
4172 * config/i386/i386.opt: Implemented option
4173 -mprefer-vector-width=[none|128|256|512].
4174 * doc/invoke.texi: Documentation for
4175 -mprefer-vector-width=[none|128|256|512].
4177 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4179 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4181 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4183 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4185 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4187 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4189 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4190 * expr.c (expand_expr_real_2): Likewise.
4191 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4192 fold_binary_loc): Likewise.
4193 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4194 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4195 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4196 MINUS_EXPR transformations.
4197 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4198 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4199 * tree-inline.c (estimate_operator_cost): Likewise.
4200 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4201 op_symbol_code): Likewise.
4202 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4203 * vr-values.c (extract_range_from_binary_expr): Likewise.
4204 * varasm.c (initializer_constant_valid_p_1): Likewise.
4205 * tree.def: New tree code POINTER_DIFF_EXPR.
4207 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4209 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4210 integer suffix to movbe mnemonic.
4211 (*bswaphi2_movbe): Ditto.
4212 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4214 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4217 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4218 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4219 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4221 2017-11-21 Martin Liska <mliska@suse.cz>
4223 * tree-inline.c (expand_call_inline): Remove not needed
4226 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4227 Jakub Jelinek <jakub@redhat.com>
4230 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4231 Remove static keyword from f variable.
4233 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4235 PR tree-optimization/83086
4236 * gimple-ssa-store-merging.c
4237 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4238 rather than n.base_addr.
4240 2017-11-21 Martin Liska <mliska@suse.cz>
4242 PR rtl-optimization/82044
4243 PR tree-optimization/82042
4244 * dse.c (check_mem_read_rtx): Check for overflow.
4246 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4248 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4251 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4253 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4254 and its corresponding call arg location note.
4256 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4257 Andrew Burgess <andrew.burgess@embecosm.com>
4259 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4261 (arc_return_slot_offset): Likewise.
4262 (arc_eh_return_address_location): New declaration.
4263 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4264 (MUST_SAVE_REGISTER): Add exception handler case.
4265 (MUST_SAVE_RETURN_ADDR): Likewise.
4266 (arc_frame_pointer_required): Likewise.
4267 (arc_frame_pointer_needed): New function.
4268 (arc_compute_frame_size): Changed.
4269 (arc_expand_prologue): Likewise.
4270 (arc_expand_epilogue): Likewise.
4271 (arc_initial_elimination_offset): Likewise.
4272 (arc_return_slot_offset): Delete.
4273 (arc_eh_return_address_location): New function.
4274 (arc_builtin_setjmp_frame_value): Likewise.
4275 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4276 (EH_RETURN_STACKADJ_RTX): Define.
4277 (EH_RETURN_HANDLER_RTX): Likewise.
4278 * config/arc/arc.md (eh_return): Delete.
4280 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4282 * print-tree.h (debug_vec_tree): Remove prototype.
4283 * gdbinit.in (pvt): Remove macro.
4285 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4287 PR tree-optimization/83047
4288 * gimple-ssa-store-merging.c
4289 (imm_store_chain_info::output_merged_store): If the loads with the
4290 same vuse are in different basic blocks, for load_gsi pick a load
4291 location that is dominated by the other loads.
4294 * config/i386/i386.c (ix86_memmodel_check): Start
4295 -Winvalid-memory-model diagnostics with lowercase letter.
4298 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4299 set in_cold_section_p to the partition containing loc_list->first.
4300 When seeing loc_list->last_before_switch node, update secname and
4301 perform range_across_switch second partition handling only after that.
4304 * run-rtl-passes.c: Include debug.h.
4305 (run_rtl_passes): Call debug_hooks->assembly_start.
4306 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4310 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4311 OPTAB_DIRECT in calls to expand_simple_binop.
4313 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4316 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4318 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4321 * spellcheck.h (best_match::blithely_get_best_candidate): New
4324 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4326 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4327 with lower case letter.
4329 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4331 * config/i386/i386.md (bswaphi2): New expander.
4332 (*bswaphi2_movbe): New insn pattern.
4333 (bswaphi -> rorhi pepehole2): New peephole pattern.
4335 2017-11-20 Jeff Law <law@redhat.com>
4337 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4338 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4339 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4340 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4341 gimple-ssa-evrp-analyze.h.
4343 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4344 set BB_VISITED here.
4345 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4347 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4349 * tree-predcom.c: Add general comment on Store-Store chains.
4350 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4352 (get_chain_last_ref_at): Rename into...
4353 (get_chain_last_write_at): ...this.
4354 (get_chain_last_write_before_load): New function.
4355 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4356 CT_STORE_STORE when write reference is added.
4357 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4358 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4359 (initialize_root_vars_store_elim_1): Ditto.
4360 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4361 definition is created.
4362 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4363 chain by replacing it with dominant stored value.
4365 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4367 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4369 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4371 * vr-values.c (extract_range_from_binary_expr): Use a full range
4374 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4376 * config/arm/arm.md (R4_REGNUM): Define constant.
4377 (nonsecure_call_internal): Remove r4 clobber.
4378 (nonsecure_call_value_internal): Likewise.
4379 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4380 clobber and resequence match_operands.
4381 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4382 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4383 (nonsecure_call_value_reg_thumb2): Likewise.
4385 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4387 PR tree-optimization/78821
4388 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4389 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4390 to the address of the base rather than the base itself.
4391 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4392 (find_bswap_or_nop_finalize): New function.
4393 (find_bswap_or_nop): Use it.
4394 (bswap_replace): Return a tree rather than bool, change first
4395 argument from gimple * to gimple_stmt_iterator, allow inserting
4396 into an empty sequence, allow ins_stmt to be NULL - then emit
4397 all stmts into gsi. Fix up MEM_REF address gimplification.
4398 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4399 (struct store_immediate_info): Add N and INS_STMT non-static
4401 (store_immediate_info::store_immediate_info): Initialize them
4402 from newly added ctor args.
4403 (merged_store_group::apply_stores): Formatting fixes. Sort by
4405 (stmts_may_clobber_ref_p): For stores call also
4406 refs_anti_dependent_p.
4407 (gather_bswap_load_refs): New function.
4408 (imm_store_chain_info::try_coalesce_bswap): New method.
4409 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4410 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4411 (imm_store_chain_info::output_merged_store): Fail if number of
4412 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4413 and NOP_EXPR rhs_code.
4414 (pass_store_merging::process_store): Compute n and ins_stmt, if
4415 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4416 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4418 (pass_store_merging::execute): Calculate dominators.
4420 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4421 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4422 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4423 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4424 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4425 class pass_optimize_bswap, bswap_replace,
4426 pass_optimize_bswap::execute): Moved to ...
4427 * gimple-ssa-store-merging.c: ... this file.
4428 Include optabs-tree.h.
4429 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4430 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4431 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4432 anonymous namespace, remove static keywords.
4433 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4434 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4436 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4439 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4441 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4443 * vec.h (debug_helper): New function.
4444 (DEFINE_DEBUG_VEC): New macro.
4445 * hash-set.h (debug_helper): New function.
4446 (DEFINE_DEBUG_HASH_SET): New macro.
4447 * cfg.c (debug_slim (edge)): New function.
4448 Call DEFINE_DEBUG_VEC for edges.
4449 Call DEFINE_DEBUG_HASH_SET for edges.
4450 * cfghooks.c (debug_slim (basic_block)): New function.
4451 Call DEFINE_DEBUG_VEC for basic blocks.
4452 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4453 * print-tree.c (debug_slim): New function to handle trees.
4454 Call DEFINE_DEBUG_VEC for trees.
4455 Call DEFINE_DEBUG_HASH_SET for trees.
4456 (debug (vec<tree, va_gc>) &): Remove.
4457 (debug (<vec<tree, va_gc>) *): Remove.
4458 * print-rtl.c (debug_slim): New function to handle const_rtx.
4459 Call DEFINE_DEBUG_VEC for rtx_def.
4460 Call DEFINE_DEBUG_VEC for rtx_insn.
4461 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4462 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4463 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4464 (debug (vec<rtx_insn *> *ptr): Remove.
4465 (debug_insn_vector): Remove.
4466 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4468 2017-11-20 Tom de Vries <tom@codesourcery.com>
4470 PR rtl-optimization/82020
4471 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4472 IF_THEN_ELSE condition.
4474 2017-11-19 Jeff Law <law@redhat.com>
4476 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4477 of degenerates resulting from ignoring an edge.
4479 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4482 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4484 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4487 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4488 for uninitialied counts.
4490 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4493 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4495 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4498 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4500 2017-11-19 Tom de Vries <tom@codesourcery.com>
4502 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4505 2017-11-19 Tom de Vries <tom@codesourcery.com>
4507 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4510 2017-11-19 Tom de Vries <tom@codesourcery.com>
4512 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4515 2017-11-19 Tom de Vries <tom@codesourcery.com>
4517 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4519 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4521 2017-11-19 Tom de Vries <tom@codesourcery.com>
4523 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4525 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4526 semicolon after MCORE_EXPORT_NAME call.
4528 2017-11-19 Tom de Vries <tom@codesourcery.com>
4531 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4532 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4534 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4536 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4537 (cgraph_edge::clone): Cleanup updating of profile.
4538 * ipa-cp.c (update_profiling_info): Likewise.
4539 * ipa-inline-transform.c (inline_transform): Likewise.
4540 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4541 * ipa-split.c (execute_split_functions): Do not split when function
4543 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4544 * profile-count.c (profile_count::dump): Add global0.
4545 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4547 (profile_count::to_sreal_scale): Likewise.
4548 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4549 (profile_count::combine_with_ipa_count): New function.
4550 * profile-count.h (profile_guessed_global0adjusted): New.
4551 (profile_count::adjusted_zero): New.
4552 (profile_count::global0adjusted): New.
4553 (profile_count::combine_with_ipa_count): New.
4554 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4555 correct profile of return block of split functions.
4556 (copy_cfg_body): Remove unused profile_count.
4557 (copy_body): Likewise.
4558 (expand_call_inline): Update.
4559 (tree_function_versioning): Update.
4561 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4563 * hash-set.h (hash_set::empty): New.
4564 * tree-ssa-threadbackward.h: Delete.
4565 * tree-ssa-threadbackward.c (class thread_jumps): New.
4566 Move max_threaded_paths into class.
4567 (fsm_find_thread_path): Remove arguments that are now in class.
4568 (profitable_jump_thread_path): Rename to...
4569 (thread_jumps::profitable_jump_thread_path): ...this.
4570 (convert_and_register_jump_thread_path): Rename to...
4571 (thread_jumps::convert_and_register_current_path): ...this.
4572 (check_subpath_and_update_thread_path): Rename to...
4573 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4574 (register_jump_thread_path_if_profitable): Rename to...
4575 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4576 (handle_phi): Rename to...
4577 (thread_jumps::handle_phi): ...this.
4578 (handle_assignment): Rename to...
4579 (thread_jumps::handle_assignment): ...this.
4580 (fsm_find_control_statement_thread_paths): Rename to...
4581 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4582 (find_jump_threads_backwards): Rename to...
4583 (thread_jumps::find_jump_threads_backwards): ...this.
4584 Initialize path local data.
4585 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4586 from within thread_jumps class.
4587 (pass_early_thread_jumps::execute): Same.
4589 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4591 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4593 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4595 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4596 to canonical location.
4598 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4601 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4603 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4605 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4607 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4608 rcrt1.o%s/grcrt1.o%s for -static-pie.
4610 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4612 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4613 ix86_shift_rotate_cost): Break out from ...
4614 (ix86_rtx_costs): ... here.
4615 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4618 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4620 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4621 when entry block was promoted unlikely.
4622 (estimate_bb_frequencies): Increase frequency scale.
4623 * profile-count.h (profile_count): Export precision info.
4625 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4627 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4628 disturbin profile of entry block.
4630 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4632 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4635 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4637 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4638 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4640 2017-11-17 Jeff Law <law@redhat.com>
4642 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4643 from evrp_dom_walker class. Various methods moved into new class.
4644 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4645 (evrp_range_analyzer::enter): New method.
4646 (evrp_range_analyzer::leave): New method.
4647 (evrp_dom_walker): Remove delegators no longer needed by this class.
4648 Replace vr_values data member with evrp_range_analyzer
4650 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4651 method extracted from evrp_dom_walker::before_dom_children.
4652 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4653 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4655 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4656 Add private copy constructor and move assignment operators.
4657 Privatize methods and class data where trivially possible.
4658 (evrp_dom_walker::cleanup): New function, extracted from
4659 execute_early_vrp. Simplify access to class data.
4661 * vr-values.h (get_output_for_vrp): Prototype.
4662 * vr-values.c (get_output_for_vrp): New function extracted from
4663 vrp_visit_assignment_or_call and extract_range_from_stmt.
4664 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4666 2017-11-17 Luis Machado <luis.machado@linaro.org>
4668 * config/aarch64/aarch64.c
4669 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4670 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4671 tune_params::AUTOPREFETCHER_WEAK.
4673 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4676 * config/arm/arm.c (arm_valid_target_attribute_rec):
4677 Parse "arch=" and "+<ext>".
4678 (arm_valid_target_attribute_tree): Re-init global options.
4679 (arm_option_override): Make non-static.
4680 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4681 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4682 (__ARM_FEATURE_CRC32): Support undef.
4683 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4684 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4686 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4688 * gdbinit.in (break-on-diagnostic): New command.
4690 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4692 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4693 return to indirect jump for EH return if control-flow
4694 protection is enabled. Change explicit 'false' argument in
4695 pro_epilogue_adjust_stack with a value of
4697 * config/i386/i386.md (simple_return_indirect_internal):
4698 Remove SImode restriction to support 64-bit.
4700 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4702 * combine.c (added_notes_insn): New.
4703 (try_combine): Handle added_notes_insn like added_links_insn.
4704 Rewrite return value code.
4705 (distribute_notes): Set added_notes_insn to the earliest insn we added
4708 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4710 PR rtl-optimization/82621
4711 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
4712 dest of one of those SETs is unused.
4714 2017-11-17 Richard Biener <rguenther@suse.de>
4717 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4718 * tree-pretty-print.c (dump_generic_node): Handle
4719 annot_expr_parallel_kind.
4720 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4721 * gimplify.c (gimple_boolify): Likewise.
4723 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4725 * config.gcc (extra_headers): Add cet.h for x86 targets.
4726 * config/i386/cet.h: New file.
4727 * doc/install.texi: Add --enable-cet/--disable-cet.
4729 2017-11-17 Richard Biener <rguenther@suse.de>
4731 PR tree-optimization/83017
4732 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4733 (gen_parallel_loop): Properly count iterations.
4734 (parallelize_loops): Handle loop->can_be_parallel independent
4735 of flag_loop_parallelize_all. Make static profitability test match
4737 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4738 * invoke.texi (parloops-min-per-thread): Document.
4740 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4742 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4743 upstreaming review comments.
4745 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4747 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4748 with fast unaligned access.
4749 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4751 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4753 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4754 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4756 2017-11-17 Richard Biener <rguenther@suse.de>
4758 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4759 folding of references.
4761 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4764 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4765 of replacing call to strncmp with corresponding call to strcmp when
4768 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4770 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4771 option prefer-avx256 for skylake-avx512 configuration.
4772 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4773 (get_builtin_code_for_version): Ditto.
4775 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4776 Monk Chiang <sh.chiang04@gmail.com>
4778 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4779 (FIXED_REGISTERS): Reserve more register numbers.
4780 (CALL_USED_REGISTERS): Likewise.
4781 (REG_ALLOC_ORDER): Likewise.
4782 (REG_CLASS_CONTENTS): Likewise.
4783 (REGISTER_NAMES): Likewise.
4785 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4786 Kito Cheng <kito.cheng@gmail.com>
4788 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4790 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4792 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4794 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4795 parameter list for vec_splats.
4797 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4799 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4801 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4803 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4806 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4807 with the required version of automake 1.11.6.
4809 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4811 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4813 (DU_C2_3_power9): Correct reservation combinations.
4814 (FP_DIV_power9, VEC_DIV_power9): New.
4815 (power9-alu): Split out rotate/shift...
4816 (power9-rot): ...to here, correct dispatch resource.
4817 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4819 (power9-fp): Correct latency.
4820 (power9-sdiv): Add div/sqrt resource.
4821 (power9-ddiv): Correct latency, add div/sqrt resource.
4822 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4823 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4825 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4827 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4829 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4830 switch statement mapping KF built-ins to TF built-ins if we don't
4831 have the proper ISA 3.0 assembler support.
4833 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4835 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4836 (lower_emutls_function_body): Do not compute it.
4838 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4840 * ipa-split.c (split_bb_info): Turn time to sreal.
4841 (split_point): Likewise.
4842 (dump_split_point): Likewise.
4843 (fine_split_points): Likewise.
4844 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4846 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4848 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4851 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4853 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4855 (expensive_function_p): Remove useless assert.
4856 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4858 2017-11-16 Martin Sebor <msebor@redhat.com>
4860 PR tree-optimization/82588
4861 PR tree-optimization/82583
4862 * tree-vrp.c (check_array_ref): Handle flexible array members,
4863 string literals, and inner indices.
4864 (search_for_addr_array): Add detail to diagnostics.
4866 2017-11-16 Nathan Sidwell <nathan@acm.org>
4870 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4871 (SET_DECL_ASSEMBLER_NAME): Forward to
4872 overwrite_decl_assembler_name.
4873 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4874 (overwrite_decl_assembler_name): Declare.
4875 * tree.c (overwrite_decl_assembler_name): New.
4876 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4877 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4878 (LANG_HOOKS_INITIALIZER): Add it.
4879 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4880 * langhooks.c (lhd_set_decl_assembler_name): Use
4881 SET_DECL_ASSEMBLER_NAME.
4882 (lhd_overwrite_decl_assembler_name): Default implementation.
4884 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4885 Jackson Woodruff <jackson.woodruff@arm.com>
4887 PR tree-optimization/71026
4888 * match.pd: Canonicalize constant multiplies in division.
4890 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4892 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4893 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4894 Remove OPT_fomit_frame_pointer entry.
4895 * common/config/arc/arc-common.c: Likewise.
4896 * common/config/arm/arm-common.c: Likewise.
4897 * common/config/avr/avr-common.c: Likewise.
4898 * common/config/c6x/c6x-common.c: Likewise.
4899 * common/config/cr16/cr16-common.c: Likewise.
4900 * common/config/cris/cris-common.c: Likewise.
4901 * common/config/epiphany/epiphany-common.c: Likewise.
4902 * common/config/fr30/fr30-common.c: Likewise.
4903 * common/config/frv/frv-common.c: Likewise.
4904 * common/config/ia64/ia64-common.c: Likewise.
4905 * common/config/iq2000/iq2000-common.c: Likewise.
4906 * common/config/lm32/lm32-common.c: Likewise.
4907 * common/config/m32r/m32r-common.c: Likewise.
4908 * common/config/mcore/mcore-common.c: Likewise.
4909 * common/config/microblaze/microblaze-common.c: Likewise.
4910 * common/config/mips/mips-common.c: Likewise.
4911 * common/config/mmix/mmix-common.c: Likewise.
4912 * common/config/mn10300/mn10300-common.c: Likewise.
4913 * common/config/nios2/nios2-common.c: Likewise.
4914 * common/config/pa/pa-common.c: Likewise.
4915 * common/config/pdp11/pdp11-common.c: Likewise.
4916 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4917 * common/config/riscv/riscv-common.c: Likewise.
4918 * common/config/rs6000/rs6000-common.c: Likewise.
4919 * common/config/rx/rx-common.c: Likewise.
4920 * common/config/s390/s390-common.c: Likewise.
4921 * common/config/sh/sh-common.c: Likewise.
4922 * common/config/sparc/sparc-common.c: Likewise.
4923 * common/config/tilegx/tilegx-common.c: Likewise.
4924 * common/config/tilepro/tilepro-common.c: Likewise.
4925 * common/config/v850/v850-common.c: Likewise.
4926 * common/config/visium/visium-common.c: Likewise.
4927 * common/config/xstormy16/xstormy16-common.c: Likewise.
4928 * common/config/xtensa/xtensa-common.c: Likewise.
4929 * invoke.texi (-fomit-frame-pointer): Update documentation.
4931 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4933 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4935 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4937 * cfg.c (scale_bbs_frequencies_int,
4938 cale_bbs_frequencies_gcov_type): Remove.
4939 * cfg.h (scale_bbs_frequencies_int,
4940 cale_bbs_frequencies_gcov_type): Remove.
4942 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4944 * tree-ssa-loop-manip.c
4945 (scale_dominated_blocks_in_loop): Update to profile counts.
4946 (tree_transform_and_unroll_loop): Likewise.
4948 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4950 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4951 scale_bbs_frequencies_int.
4953 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4955 * final.c (compute_alignments): Use counts rather than frequencies.
4957 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4959 * cfgloopanal.c: Include sreal.h
4960 (average_num_loop_insns): Use counts and sreal for accounting.
4962 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4964 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4967 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4968 Kito Cheng <kito.cheng@gmail.com>
4970 * config/nds32/constraints.md: Provide more constraints.
4971 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4972 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4973 support constraints usage.
4975 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4977 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4979 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4980 Kito Cheng <kito.cheng@gmail.com>
4982 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4983 * config/nds32/nds32.opt: Refine the layout.
4984 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4985 TARGET_EXT_STRING): Support new options.
4986 * config/nds32/nds32.h: Likewise.
4987 * config/nds32/nds32.md: Likewise.
4988 * config/nds32/nds32-predicates.c: Likewise.
4989 * config/nds32/constraints.md: Likewise.
4990 * common/config/nds32/nds32-common.c: Likewise.
4992 2017-11-16 Julia Koval <julia.koval@intel.com>
4995 * config/i386/gfniintrin.h: Add sse check.
4996 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4998 2017-11-16 Julia Koval <julia.koval@intel.com>
5000 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5001 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5002 (ix86_handle_option): Handle -mavx512vbmi2.
5003 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5004 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5005 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5006 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5007 (ix86_valid_target_attribute_inner_p): Ditto.
5008 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5009 * config/i386/i386.opt (mavx512vbmi2): New option.
5010 * doc/invoke.texi: Add new option.
5012 2017-11-16 Julia Koval <julia.koval@intel.com>
5014 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5015 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5016 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5017 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5018 _mm512_gf2p8mul_epi8): New intrinsics.
5019 * config/i386/i386-builtin-types.def
5020 (V64QI_FTYPE_V64QI_V64QI): New type.
5021 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5022 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5023 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5024 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5025 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5026 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5028 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5030 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5031 explicitly as a stream of bytes.
5033 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5035 * config/rs6000/altivec.h (vec_xst_be): New #define.
5036 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5037 and externalize from *altivec_vperm_<mode>_internal.
5038 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5040 (XL_BE_V8HI): Likewise.
5041 (XL_BE_V4SI): Likewise.
5042 (XL_BE_V4SI): Likewise.
5043 (XL_BE_V2DI): Likewise.
5044 (XL_BE_V4SF): Likewise.
5045 (XL_BE_V2DF): Likewise.
5046 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5047 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5048 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5049 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5050 VSX_BUILTIN_VEC_XST_BE.
5051 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5052 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5054 (altivec_init_builtins): Replace conditional calls to def_builtin
5055 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5056 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5057 with unconditional calls. Remove calls to def_builtin for
5058 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5059 __builtin_vec_xst_be.
5060 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5061 to define_expand, and add alternate RTL generation for P8.
5062 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5063 vsx_ld_elemrev_v8hi.
5064 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5065 add alternate RTL generation for P8.
5066 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5067 vsx_ld_elemrev_v16qi.
5068 (vsx_st_elemrev_v8hi): Convert define_insn
5069 to define_expand, and add alternate RTL generation for P8.
5070 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5071 vsx_st_elemrev_v8hi.
5072 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5073 add alternate RTL generation for P8.
5074 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5075 vsx_st_elemrev_v16qi.
5077 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5080 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5081 TARGET_AVX512ER check.
5082 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5083 neither -mzeroupper nor -mno-zeroupper is used and
5084 TARGET_EMIT_VZEROUPPER is set.
5085 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5086 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5088 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5090 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5091 folding of vector compares.
5092 (fold_build_vec_cmp): New helper function.
5093 (fold_compare_helper): New helper function.
5094 (builtin_function_type): Add compare builtins to the list of functions
5095 having unsigned arguments. Cosmetic updates to comment indentation.
5096 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5097 the not+eq combination.
5099 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5101 PR tree-optimization/82726
5102 PR tree-optimization/70754
5103 * tree-predcom.c (order_drefs_by_pos): New function.
5104 (combine_chains): Move code setting has_max_use_after to...
5105 (try_combine_chains): ...here. New parameter. Sort combined chains
5106 according to position information.
5107 (tree_predictive_commoning_loop): Update call to above function.
5108 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5110 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5112 PR tree-optimization/82726
5114 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5116 PR tree-optimization/70754
5117 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5118 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5119 combined stmt before it if not NULL.
5120 (combine_chains): Process refs reversely and compute dominance point
5124 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5126 PR tree-optimization/79663
5127 * tree-predcom.c (combine_chains): Process refs in reverse order
5128 only for ZERO length chains, and add explaining comment.
5130 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5132 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5133 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5134 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5136 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5138 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5140 2017-11-15 Martin Liska <mliska@suse.cz>
5142 * tree-cfg.c (pass_warn_function_return::execute):
5143 Compare warn_return_type for greater than zero.
5145 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5149 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5150 to return true on Xeon and not on Xeon Phi.
5151 (ix86_check_avx256_register): Changed to ...
5152 (ix86_check_avx_upper_register): ... this. Add extra check for
5153 VALID_AVX512F_REG_OR_XI_MODE.
5154 (ix86_avx_u128_mode_needed): Changed
5155 ix86_check_avx256_register to ix86_check_avx_upper_register.
5156 (ix86_check_avx256_stores): Changed to ...
5157 (ix86_check_avx_upper_stores): ... this. Changed
5158 ix86_check_avx256_register to ix86_check_avx_upper_register.
5159 (ix86_avx_u128_mode_after): Changed
5160 avx_reg256_found to avx_upper_reg_found. Changed
5161 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5162 (ix86_avx_u128_mode_entry): Changed
5163 ix86_check_avx256_register to ix86_check_avx_upper_register.
5164 (ix86_avx_u128_mode_exit): Ditto.
5165 * config/i386/i386.h: (host_detect_local_cpu): New define.
5167 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5169 * config/arm/xgene1.md (xgene1): Split into automatons
5170 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5171 (xgene1_f_load): Adjust reservations and/or types.
5172 (xgene1_f_store): Likewise.
5173 (xgene1_load_pair): Likewise.
5174 (xgene1_store_pair): Likewise.
5175 (xgene1_fp_load1): Likewise.
5176 (xgene1_load1): Likewise.
5177 (xgene1_store1): Likewise.
5178 (xgene1_move): Likewise.
5179 (xgene1_alu): Likewise.
5180 (xgene1_simd): Likewise.
5181 (xgene1_bfm): Likewise.
5182 (xgene1_neon_load1): Likewise.
5183 (xgene1_neon_store1): Likewise.
5184 (xgene1_neon_logic): Likewise.
5185 (xgene1_neon_st1): Likewise.
5186 (xgene1_neon_ld1r): Likewise.
5187 (xgene1_alu_cond): Added.
5188 (xgene1_shift_reg): Likwise.
5189 (xgene1_bfx): Likewise.
5190 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5192 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5195 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5197 (can_widen_mult_without_libcall): New function.
5198 (expand_mul_overflow): If only checking unsigned mul overflow,
5199 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5200 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5201 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5202 (expand_DIVMOD): Formatting fix.
5203 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5204 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5205 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5207 PR tree-optimization/82977
5208 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5209 constructed temporary to strlen_to_stridx.put.
5211 2017-11-15 Martin Liska <mliska@suse.cz>
5213 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5214 * configure: Regenerate.
5216 2017-11-15 Martin Liska <mliska@suse.cz>
5219 * config/sh/sh-mem.cc: Use proper probability for
5222 2017-11-14 Jeff Law <law@redhat.com>
5224 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5225 the red zone for stack_clash_protection_final_dynamic_probe targets
5226 when the total dynamic stack size is zero bytes.
5228 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5229 blocks is post order.
5231 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5233 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5234 * final.c (rest_of_clean_state): Set it for the
5235 -fcompare-debug dump.
5236 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5237 class when TDF_COMPARE_DEBUG is set.
5239 * dwarf2out.c (gen_producer_string): Discard
5242 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5245 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5246 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5247 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5248 (__TGMATH_REAL_2_3): Remove macros.
5249 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5250 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5252 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5254 (remquo): Define using __TGMATH_REAL_3.
5256 2017-11-14 Jeff Law <law@redhat.com>
5258 * vr-values.c: New file with contents extracted from tree-vrp.c.
5259 * Makefile.in (OBJS): Add vr-values.o
5260 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5261 (set_value_range, set_and_canonicalize_value_range): Likewise.
5262 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5263 (value_range_constant_singleton, symbolic_range_p): Likewise.
5264 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5265 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5266 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5267 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5268 (find_case_label_range, find_case_label_index): Likewise.
5269 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5270 (range_int_cst_singleton_p, value_inside_range): Likewise.
5271 (get_single_symbol): Likewise.
5272 (switch_update): Move structure definition here.
5273 (to_remove_edges, to_update_switch_stmts): Provide externs.
5274 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5275 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5276 (vrp_val_is_min, set_value_range): Likewise.
5277 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5278 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5279 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5280 (range_is_nonnull, range_int_cst_p): Likewwise.
5281 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5282 (get_single_symbol, operand_less_p): Likewise
5283 (compare_values_warnv, compare_values): Likewise.
5284 (value_inside_range, value_range_constant_singleton): Likewise.
5285 (zero_nonzero_bitgs_from_vr): Likewise.
5286 (extract_range_from_binary_expr_1): Likewise.
5287 (overflow_comparison_p): Likewise.
5288 (to_remove_edges, to_update_switch_stmts): Likewise.
5289 (find_case_label-index, find_case_label_range): Likewise.
5290 (switch_update, set_value_range_to_nonnegative): Remove.
5291 (set_value_range_to_truthvalue): Likewise.
5292 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5293 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5294 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5295 (find_case_label_ranges, test_for_singularity): Likewise.
5296 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5297 (x_vr_values): Move to its remaining use site.
5299 2017-11-10 Jeff Law <law@redhat.com>
5301 * vr-values.h (VR_INITIALIZER): Move #define here.
5302 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5303 * Makefile.in (OBJS): Add tree-evrp.o
5304 * tree-vrp.h (assert_info): Move structure definition here.
5305 (set_value_range_to_varying): Prototype.
5306 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5307 (infer_value_range, register_edge_assert_for): Likewise.
5308 (stmt_interesting_for_vrp): Likewise.
5309 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5310 (set_value_range_to_varying): No longer static.
5311 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5312 (infer_value_range, register_edge_assert_for): Likewise.
5314 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5316 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5317 generate the XXBRD instruction.
5319 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5320 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5321 as being compatible if -mabi=ieeelongdouble.
5322 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5323 to setup float128 built-ins with hardware support.
5324 (BU_FLOAT128_HW_2): Likewise.
5325 (BU_FLOAT128_HW_3): Likewise.
5326 (BU_FLOAT128_HW_VSX_1): Likewise.
5327 (BU_FLOAT128_HW_VSX_2): Likewise.
5328 (scalar_extract_expq): Change float128 built-in functions to
5329 accommodate having both KFmode and TFmode functions. Use the
5330 KFmode variant as the default.
5331 (scalar_extract_sigq): Likewise.
5332 (scalar_test_neg_qp): Likewise.
5333 (scalar_insert_exp_q): Likewise.
5334 (scalar_insert_exp_qp): Likewise.
5335 (scalar_test_data_class_qp): Likewise.
5336 (sqrtf128_round_to_odd): Delete processing the round to odd
5337 built-in functions as special built-in functions, and define them
5338 as float128 built-ins. Use the KFmode variant as the default.
5339 (truncf128_round_to_odd): Likewise.
5340 (addf128_round_to_odd): Likewise.
5341 (subf128_round_to_odd): Likewise.
5342 (mulf128_round_to_odd): Likewise.
5343 (divf128_round_to_odd): Likewise.
5344 (fmaf128_round_to_odd): Likewise.
5345 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5346 support for KFmode and TFmode xststdcqp calls.
5347 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5348 point, switch the built-in handlers for the get/set float128
5349 exponent, get float128 mantissa, float128 test built-ins, and the
5350 float128 round to odd built-in functions. Eliminate creating the
5351 float128 round to odd built-in functions as special built-ins.
5352 (rs6000_init_builtins): Eliminate special creation of the float128
5353 round to odd built-in functions.
5354 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5355 function insns to support both TFmode and KFmode variants.
5356 (xsxsigqp_<mode>): Likewise.
5357 (xsiexpqpf_<mode>): Likewise.
5358 (xsiexpqp_<mode>): Likewise.
5359 (xststdcqp_<mode>): Likewise.
5360 (xststdcnegqp_<mode>): Likewise.
5361 (xststdcqp_<mode>): Likewise.
5363 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5365 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5366 unused path_in_freq_ptr parameter.
5367 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5369 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5371 * ipa-inline.c (edge_badness): Dump sreal frequency.
5372 (compute_inlined_call_time): Match natural implementaiton ...
5373 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5374 forgotten division by CGRAPH_FREQ_BASE.
5376 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5378 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5379 Solaris 11. Update comment.
5380 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5382 * config/sol2.h (STARTFILE_SPEC): Likewise.
5383 * configure: Regenerate.
5385 2017-11-14 Carl Love <cel@us.ibm.com>
5387 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5388 le_ and be_ prefixes to swap* variables. Remove
5389 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5392 2017-11-14 Jason Merrill <jason@redhat.com>
5394 Support GTY((cache)) on hash_map.
5395 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5396 (ggc_cache_remove): Override it instead of ggc_mx.
5397 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5398 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5399 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5400 (simple_cache_map_traits): Override maybe_mx.
5401 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5402 (hash_map): Friend gt_cleare_cache.
5403 (gt_cleare_cache): New.
5404 * tree.h (tree_cache_traits): New hash_map traits class.
5405 (tree_cache_map): New typedef.
5407 2017-11-14 Richard Biener <rguenther@suse.de>
5409 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5410 paramter and handling.
5411 (cleanup_control_flow_bb): Likewise.
5412 (cleanup_control_flow_pre): New helper performing a DFS walk
5413 to call cleanup_control_flow_bb in PRE order.
5414 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5415 via cleanup_control_flow_pre.
5417 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5419 * config/aarch64/aarch64-simd.md
5420 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5421 (*aarch64_simd_bsl<mode>_alt): Likewise.
5422 (aarch64_simd_bsldi_internal): New.
5423 (aarch64_simd_bsldi_alt): Likewise.
5425 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5427 * tracer.c (better_p): Do not compare frequencies.
5428 * reg-stack.c (better_edge): Likewise.
5429 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5432 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5434 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5435 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5436 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5437 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5438 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5439 * predict.c (maybe_hot_frequency_p): Inline to ...
5440 (maybe_hot_count_p): ... here; rewrite to counts.
5441 (counts_to_freqs): Rename to ...
5442 (update_max_bb_count): ... this one.
5443 (expensive_function_p): Use counts.
5444 (estimate_bb_frequencies): Update.
5445 (rebuild_frequencies): Update.
5446 * predict.h (counts_to_freqs): Rename to ...
5447 (update_max_bb_count): ... this one.
5448 * profile.c (compute_branch_probabilities): Add debug info
5449 * tree-inline.c (expand_call_inline): Update debug info.
5450 (optimize_inline_calls): Use update_max_bb_count..
5451 (tree_function_versioning): Use update_max_bb_count..
5452 * value-prof.c (gimple_value_profile_transformations):
5453 Do not use update_max_bb_count.
5455 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5457 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5458 always use frequencies.
5460 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5462 * bb-reorder.c: Remove frequencies from comments.
5463 (better_edge_p): Use profile counts.
5464 (find_traces): Dump profile counts.
5465 (rotate_loop): Use profile counts.
5466 (find_traces_1_round): Likewise.
5467 (connect_better_edge_p): Use counts instead of probabilities for
5469 (copy_bb_p): Drop early check for non-0 frequency.
5470 (sanitize_hot_paths): Update comments.
5472 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5474 * ipa-split.c (struct split_point): Add count.
5475 (consider_split): Do not compute incoming frequency; compute incoming
5476 count and store it to split_point.
5477 (split_function): Set count of the call to split part correctly.
5479 2017-11-13 Carl Love <cel@us.ibm.com>
5481 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5483 2017-11-13 Tom Tromey <tom@tromey.com>
5485 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5487 2017-11-13 Carl Love <cel@us.ibm.com>
5489 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5490 Add support for builtins:
5491 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5492 vector {un,}signed {char,int,short},
5493 vector {un,}signed {char,int,short}) arguments.
5494 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5495 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5496 Add BU_P9V_AV_2 expansions for the builtins.
5497 * config/rs6000/altivec.h (vec_first_match_index,
5498 vec_first_mismatch_index, vec_first_match_or_eos_index,
5499 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5500 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5501 new extern declaration.
5502 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5503 * config/rs6000/vsx.md (first_match_index_<mode>,
5504 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5505 first_mismatch_or_eos_index_<mode>): Add define expand.
5506 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5507 * doc/extend.texi: Update the built-in documenation file for the new
5510 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5512 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5513 into the min/max operations for _Float<N> and _Float<N>X types.
5515 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5518 * dwarf2out.c (do_eh_frame): New static variable.
5519 (dwarf2out_begin_prologue): Set it.
5520 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5522 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5524 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5526 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5528 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5531 * debug.h (dwarf2out_do_eh_frame): Declare.
5532 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5533 (dwarf2out_do_frame): Use it.
5534 (dwarf2out_do_cfi_asm): Likewise.
5535 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5536 (dwarf2out_assembly_start): Likewise.
5537 (dwarf2out_begin_prologue): Fix comment.
5538 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5539 if the target needs either debug or unwind DWARF2 info.
5540 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5541 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5542 -ftrapping-math, -ftrapv and -fwrapv.
5544 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5546 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5547 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5548 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5549 (estimate_edge_size_and_time): Likewise.
5550 (ipa_merge_fn_summary_after_inlining): Likewise.
5551 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5552 (compute_uninlined_call_time): Use sreal_frequency.
5553 (compute_inlined_call_time): Likewise.
5554 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5555 * profile-count.c: Include sreal.h.
5556 (profile_count::to_sreal_scale): New.
5557 * profile-count.h: Forward declare sreal.
5558 (profile_count::to_sreal_scale): Declare.
5560 2017-11-13 Nathan Sidwell <nathan@acm.org>
5562 * diagnostic.c (maybe_line_and_column): New.
5563 (diagnostic_get_location_text): Use it.
5564 (diagnostic_report_current_module): Likewise.
5565 (test_diagnostic_get_location_text): Add tests.
5567 2017-11-13 Luis Machado <luis.machado@linaro.org>
5569 * doc/md.texi (Specifying processor pipeline description): Fix
5570 incorrect latency for the div instruction example.
5572 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5574 PR tree-optimization/78821
5575 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5576 that bit_not_p is the same.
5577 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5578 (split_group): Count precisely bit_not_p bits in each statement.
5579 (invert_op): New function.
5580 (imm_store_chain_info::output_merged_store): Use invert_op to
5581 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5582 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5584 2017-11-13 Martin Liska <mliska@suse.cz>
5586 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5587 (struct source_info): Likewise.
5588 (add_branch_counts): Likewise.
5589 (add_line_counts): Likewise.
5590 (function_summary): Likewise.
5591 (output_intermediate_line): Likewise.
5592 (generate_results): Likewise.
5594 2017-11-13 Martin Liska <mliska@suse.cz>
5596 * gcov.c (struct block_info): Remove typedef for block_t.
5597 (struct line_info): Likewise.
5598 (line_info::has_block): Likewise.
5599 (EXIT_BLOCK): Likewise.
5600 (unblock): Likewise.
5601 (circuit): Likewise.
5602 (get_cycles_count): Likewise.
5603 (process_file): Likewise.
5604 (read_graph_file): Likewise.
5605 (solve_flow_graph): Likewise.
5606 (find_exception_blocks): Likewise.
5607 (add_line_counts): Likewise.
5608 (accumulate_line_info): Likewise.
5609 (output_line_details): Likewise.
5611 2017-11-13 Martin Liska <mliska@suse.cz>
5613 * gcov.c (struct arc_info): Remove typedef for arc_t.
5614 (struct line_info): Likewise.
5615 (add_branch_counts): Likewise.
5616 (output_branch_count): Likewise.
5617 (function_info::~function_info): Likewise.
5618 (circuit): Likewise.
5619 (output_intermediate_line): Likewise.
5620 (read_graph_file): Likewise.
5621 (solve_flow_graph): Likewise.
5622 (find_exception_blocks): Likewise.
5623 (add_line_counts): Likewise.
5624 (accumulate_line_info): Likewise.
5625 (output_line_details): Likewise.
5626 (output_function_details): Likewise.
5628 2017-11-13 Martin Liska <mliska@suse.cz>
5630 * gcov.c (struct function_info): Remove typedef for function_t.
5631 (struct source_info): Likewise.
5632 (source_info::get_functions_at_location): Likewise.
5633 (solve_flow_graph): Likewise.
5634 (find_exception_blocks): Likewise.
5635 (add_line_counts): Likewise.
5636 (output_intermediate_file): Likewise.
5637 (process_file): Likewise.
5638 (generate_results): Likewise.
5639 (release_structures): Likewise.
5640 (read_graph_file): Likewise.
5641 (read_count_file): Likewise.
5642 (accumulate_line_counts): Likewise.
5643 (output_lines): Likewise.
5645 2017-11-13 Martin Liska <mliska@suse.cz>
5647 * gcov.c (function_info::function_info): Remove num_counts
5648 and add vector<gcov_type>.
5649 (function_info::~function_info): Use the vector.
5650 (process_file): Likewise.
5651 (read_graph_file): Likewise.
5652 (read_count_file): Likewise.
5653 (solve_flow_graph): Likewise.
5655 2017-11-13 Martin Liska <mliska@suse.cz>
5657 * gcov.c (function_info::is_artificial): New function.
5658 (process_file): Erase all artificial early.
5659 (generate_results): Skip as all artificial are already
5662 2017-11-13 Martin Liska <mliska@suse.cz>
5664 * gcov.c (read_graph_file): Store to global vector of functions.
5665 (read_count_file): Iterate the vector.
5666 (process_file): Likewise.
5667 (generate_results): Likewise.
5668 (release_structures): Likewise.
5670 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5672 PR tree-optimization/82954
5673 * gimple-ssa-store-merging.c
5674 (imm_store_chain_info::coalesce_immediate_stores): If
5675 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5677 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5679 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5680 Upddate call to ENDIAN_LANE_N.
5681 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5682 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5683 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5684 and use aarch64_endian_lane_rtx.
5685 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5687 2017-11-12 Tom de Vries <tom@codesourcery.com>
5689 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5691 2017-11-12 Tom de Vries <tom@codesourcery.com>
5693 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5695 2017-11-12 Tom de Vries <tom@codesourcery.com>
5697 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
5699 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5700 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5701 * defaults.h (ASM_OUTPUT_LABELREF): Same.
5703 2017-11-11 Martin Sebor <msebor@redhat.com>
5706 * doc/extend.texi (attribute nonstring): Remove spurious argument.
5709 * prefic.c (translate_name): Replace strncpy with memcpy to
5710 avoid -Wstringop-truncation.
5712 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5714 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
5716 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5718 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5720 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5722 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5725 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5727 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5728 enable generating XXBRH if the value is in a vector register.
5729 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5730 value is in a vector register.
5731 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5732 register to register bswap64's instead of doing the GPR sequence
5733 used on previous machines.
5734 (bswapdi2_xxbrd): New insn.
5735 (bswapdi2_reg): Disallow on ISA 3.0.
5736 (register to register bswap64 splitter): Do not split the insn on
5737 ISA 3.0 systems that use XXBRD.
5739 2017-11-10 Martin Sebor <msebor@redhat.com>
5742 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5744 (find_subframework_file): Same.
5746 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5748 * auto-profile.c (afdo_indirect_call): Drop frequency.
5749 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5750 (cgraph_node::create_edge): Drop frequency argument.
5751 (cgraph_node::create_indirect_edge): Drop frequency argument.
5752 (cgraph_edge::make_speculative): Drop frequency arguments.
5753 (cgraph_edge::resolve_speculation): Do not update frequencies
5754 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5755 (cgraph_node::dump): Check consistency in IPA mode.
5756 (cgraph_edge::maybe_hot_p): Use IPA counter.
5757 (cgraph_edge::verify_count_and_frequency): Rename to ...
5758 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5759 (cgraph_node::verify_node): Update.
5760 * cgraph.h (struct cgraph_edge): Drop frequency.
5761 (cgraph_edge::frequency): New function.
5762 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5764 (cgraph_edge::rebuild_edges): Likewise.
5765 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5766 (duplicate_thunk_for_node): Do not pass frequency.
5767 (cgraph_node::create_clone): Scale only counts.
5768 (cgraph_node::create_virtual_clone): Do not pass frequency.
5769 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5770 (cgraph_node::create_version_clone): Do not pass frequency.
5771 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5772 (cgraph_node::expand_thunk): Do not pass frequency.
5773 (cgraph_node::create_wrapper): Do not pass frequency.
5774 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5776 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5777 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5778 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5779 (gather_caller_stats): Use frequency function.
5780 (ipcp_cloning_candidate_p): Use frequency function.
5781 (ipcp_propagate_stage): Use frequency function.
5782 (get_info_about_necessary_edges): Use frequency function.
5783 (update_profiling_info): Update only IPA profile.
5784 (update_specialized_profile): Use frequency functoin.
5785 (perhaps_add_new_callers): Update only IPA profile.
5786 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5787 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5788 (dump_ipa_call_summary): Use frequency function.
5789 (estimate_edge_size_and_time): Use frequency function.
5790 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5791 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5792 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5793 (update_noncloned_counts): ... ths one; scale counts only.
5794 (clone_inlined_nodes): Do not scale frequency.
5795 (inline_call): Do not pass frequency.
5796 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5797 (compute_inlined_call_time): Use IPA profile.
5798 (want_inline_small_function_p): Use IPA profile.
5799 (want_inline_self_recursive_call_p): Use IPA profile.
5800 (edge_badness): Use IPA profile.
5801 (lookup_recursive_calls): Use IPA profile.
5802 (recursive_inlining): Do not pass frequency.
5803 (resolve_noninline_speculation): Do not update frequency.
5804 (inline_small_functions): Collect max of IPA profile.
5805 (dump_overall_stats): Dump IPA porfile.
5806 (dump_inline_stats): Dump IPA porfile.
5807 (ipa_inline): Collect IPA stats.
5808 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5809 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5810 (ipa_propagate_frequency): Use frequency function.
5811 (ipa_profile): Cleanup.
5812 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5813 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5814 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5815 (input_node): Do not stream frequency.
5816 (input_edge): Do not stream frequency.
5817 (merge_profile_summaries): Scale only IPA profiles.
5818 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5819 * predict.c (drop_profile): Do not recompute frequency.
5820 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5821 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5822 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5823 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5824 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5825 * tree-inline.c (copy_bb): Do not scale frequency.
5826 (expand_call_inline): Do not scale frequency.
5827 (tree_function_versioning): Do not scale frequency.
5828 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5830 2017-11-10 Julia Koval <julia.koval@intel.com>
5832 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5833 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5834 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5835 (_mm256_maskz_gf2p8affine_epi64_epi8)
5836 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5837 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5838 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5839 (__builtin_ia32_vgf2p8affineqb_v32qi)
5840 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5841 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5843 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5847 (arm_option_override): Refactor.
5848 (arm_option_reconfigure_globals): New.
5849 (arm_options_perform_arch_sanity_checks): New.
5850 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5852 (arm_options_perform_arch_sanity_checks): Likewise
5854 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5856 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5857 (power9-qpmul): New.
5858 * rs6000/rs6000.md ("type" attr): Add qmul.
5859 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5860 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5861 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5863 2017-11-10 Martin Sebor <msebor@redhat.com>
5866 * builtins.c (compute_objsize): Handle arrays that
5867 compute_builtin_object_size likes to fail for. Make extern.
5868 * builtins.h (compute_objsize): Declare.
5869 (check_strncpy_sizes): New function.
5870 (expand_builtin_strncpy): Call check_strncpy_sizes.
5871 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5872 -Wstringop-truncation.
5873 (gimple_fold_builtin_strncat): Same.
5874 * gimple.c (gimple_build_call_from_tree): Set call location.
5875 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5876 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5877 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5878 (handle_builtin_strlen): Use strlen_to_stridx.
5879 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5881 Use strlen_to_stridx.
5882 (pass_strlen::execute): Release strlen_to_stridx.
5883 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5884 (-Wstringop-truncation): Document new option.
5886 2017-11-10 Martin Liska <mliska@suse.cz>
5888 PR gcov-profile/82702
5889 * gcov.c (main): Handle intermediate files in a different
5891 (get_gcov_intermediate_filename): New function.
5892 (output_gcov_file): Remove support of intermediate files.
5893 (generate_results): Allocate intermediate file.
5894 (release_structures): Clean-up properly fn_end.
5895 (output_intermediate_file): Start iterating with line 1.
5897 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5899 PR tree-optimization/82929
5900 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5901 ops_swapped_p non-static data member.
5902 (store_immediate_info::store_immediate_info): Clear it.
5903 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5904 ops set ops_swapped_p.
5905 (count_multiple_uses): Handle ops_swapped_p.
5907 2017-11-10 Martin Liska <mliska@suse.cz>
5909 * coverage.c (coverage_init): Stream information about
5910 support of has_unexecuted_blocks.
5911 * doc/gcov.texi: Document that.
5912 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5913 * gcov.c (read_graph_file): Likewise.
5914 (output_line_beginning): Fix a small issue with
5917 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5919 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5920 reference of trivial component.
5922 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5925 * gimple-ssa-store-merging.c
5926 (pass_store_merging::terminate_all_aliasing_chains): For
5927 gimple_store_p stmts also call refs_output_dependent_p.
5929 PR rtl-optimization/82913
5930 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5933 2017-11-09 Jeff Law <law@redhat.com>
5935 * vr-values.h: New file with vr_values class.
5936 * tree-vrp.c: Include vr-values.h
5937 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5938 data objects into the vr_values class.
5939 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5940 (get_value_range): Make it a member function within vr_values class.
5941 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5942 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5943 (op_with_constant_singleton_value_range): Likewise.
5944 (extract_range_for_var_from_comparison_expr): Likewise.
5945 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5946 (extract_range_from_binary_expr): Likewise.
5947 (extract_range_from_unary_expr): Likewise.
5948 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5949 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5950 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5951 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5952 (compare_name_with_value, compare_names): Likewise.
5953 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5954 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5955 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5956 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5957 (extract_range_from_phi_node): Likewise.
5958 (simplify_truth_ops_using_ranges): Likewise.
5959 (simplify_div_or_mod_using_ranges): Likewise.
5960 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5961 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5962 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5963 (simplify_float_conversion_using_ranges): Likewise.
5964 (simplify_internal_call_using_ranges): Likewise.
5965 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5966 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5967 poitner via x_vr_values for calls into gimple folder.
5968 (vrp_initialize_lattice): Make this the vr_values ctor.
5969 (vrp_free_lattice): Make this the vr_values dtor.
5970 (set_vr_value): New function.
5971 (class vrp_prop): Add vr_values data member. Add various member
5972 functions as well as member functions that delegate to vr_values.
5973 (check_array_ref): Make a member function within vrp_prop class.
5974 (search_for_addr_array, vrp_initialize): Likewise.
5975 (vrp_finalize): Likewise. Revamp to avoid direct access to
5976 vr_value, values_propagated, etc.
5977 (check_array_bounds): Extract vrp_prop class instance pointer from
5978 walk info structure. Use it to call member functions.
5979 (check_all_array_refs): Make a member function within vrp_prop class.
5980 Smuggle class instance pointer via walk info structure.
5981 (x_vr_values): New local static.
5982 (vrp_valueize): Use x_vr_values to get class instance.
5983 (vr_valueize_1): Likewise.
5984 (class vrp_folder): Add vr_values data member. Add various member
5985 functions as well as member functions that delegate to vr_values.
5986 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5987 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5988 class instance from vr_values. Use it to call member functions.
5989 (vrp_dom_walker): Add vr_values data member.
5990 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5991 instance via x_vr_values.
5992 (identify_jump_threads): Accept vr_values as argument. Store
5993 it into the walker structure.
5994 (evrp_dom_walker): Add vr_values class data member. Add various
5996 (evrp_dom_walker::try_find_new_range): Use vr_values data
5997 member to access the memory allocator.
5998 (evrp_dom_walker::before_dom_children): Store vr_values class
5999 instance into the vrp_folder class.
6000 (evrp_dom_walker::push_value_range): Rework to avoid direct
6001 access to num_vr_values and vr_value.
6002 (evrp_dom_walker::pop_value_range): Likewise.
6003 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6004 Use vr_values to get to dump_all_value_ranges member function.
6005 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6006 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6007 Pass vr_values class instance down to identify_jump_threads.
6008 Remove call to vrp_free_lattice.
6009 (debug_all_value_ranges): Remove.
6011 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6012 (vrp_folder): Likewise.
6014 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6015 Get it from the existing bitmap instead.
6016 (vrp_intersect_ranges_1): Likewise.
6018 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6020 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6022 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6023 set bit_not_p to it.
6024 (imm_store_chain_info::coalesce_immediate_stores): Break group
6025 if bit_not_p is different.
6026 (count_multiple_uses, split_group,
6027 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6028 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6029 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6030 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6031 to store_immediate_info ctor.
6033 2017-11-09 Jim Wilson <jimw@sifive.com>
6035 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6036 (scan_prog_file): Likewise.
6038 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6040 * bb-reorder.c (max_entry_frequency): Remove.
6041 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6042 connect_traces, push_to_next_round_p): Remove prototypes.
6043 (find_traces_1_round): Use counts only.
6044 (push_to_next_round_p): Likewise.
6045 (find_traces): Likewise.
6046 (rotate_loop): Likewise.
6047 (find_traces_1_round): Likewise.
6048 (connect_traces): Likewise.
6049 (edge_order): Likewise.
6051 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6053 * config/arm/arm.c (output_return_instruction): Add comments to
6054 indicate requirement for cmse_nonsecure_entry return to account
6055 for the size of clearing instruction output here.
6056 (thumb_exit): Likewise.
6057 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6058 return in hardfloat mode.
6060 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6062 * config/rs6000/rs6000.c (machine_function): Add a bool,
6063 "toc_is_wrapped_separately".
6064 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6065 if it wasn't explicitly set or unset, we are optimizing for speed, and
6066 doing separate shrink-wrapping.
6067 (rs6000_get_separate_components): Enable the TOC component if
6068 saving the TOC register in the prologue.
6069 (rs6000_components_for_bb): Handle the TOC component.
6070 (rs6000_emit_prologue_components): Store the TOC register where needed.
6071 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6072 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6074 2017-11-09 Martin Jambor <mjambor@suse.cz>
6076 * ipa-param-manipulation.c: New file.
6077 * ipa-param-manipulation.h: Likewise.
6078 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6079 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6080 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6081 (ipa_parm_adjustment): Likewise.
6082 (ipa_parm_adjustment_vec): Likewise.
6083 (ipa_get_vector_of_formal_parms): Moved declaration to
6084 ipa-param-manipulation.h.
6085 (ipa_get_vector_of_formal_parm_types): Likewise.
6086 (ipa_modify_formal_parameters): Likewise.
6087 (ipa_modify_call_arguments): Likewise.
6088 (ipa_combine_adjustments): Likewise.
6089 (ipa_dump_param_adjustments): Likewise.
6090 (ipa_modify_expr): Likewise.
6091 (ipa_get_adjustment_candidate): Likewise.
6092 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6093 ipa-param-manipulation.c.
6094 (ipa_get_vector_of_formal_parm_types): Likewise.
6095 (ipa_modify_formal_parameters): Likewise.
6096 (ipa_modify_call_arguments): Likewise.
6097 (ipa_modify_expr): Likewise.
6098 (get_ssa_base_param): Likewise.
6099 (ipa_get_adjustment_candidate): Likewise.
6100 (index_in_adjustments_multiple_times_p): Likewise.
6101 (ipa_combine_adjustments): Likewise.
6102 (ipa_dump_param_adjustments): Likewise.
6103 * tree-sra.c: Also include ipa-param-manipulation.h
6104 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6107 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6108 Alan Hayward <alan.hayward@arm.com>
6109 David Sherwood <david.sherwood@arm.com>
6111 * doc/sourcebuild.texi (vect_masked_store): Document.
6113 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6114 Alan Hayward <alan.hayward@arm.com>
6115 David Sherwood <david.sherwood@arm.com>
6117 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6119 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6120 Alan Hayward <alan.hayward@arm.com>
6121 David Sherwood <david.sherwood@arm.com>
6123 * doc/sourcebuild.texi (vect_variable_length): Document.
6125 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6126 Alan Hayward <alan.hayward@arm.com>
6127 David Sherwood <david.sherwood@arm.com>
6129 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6131 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6132 Alan Hayward <alan.hayward@arm.com>
6133 David Sherwood <david.sherwood@arm.com>
6135 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6137 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6138 Alan Hayward <alan.hayward@arm.com>
6139 David Sherwood <david.sherwood@arm.com>
6141 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6142 previously undocumented selectors.
6143 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6145 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6147 * doc/rtl.texi (const_vector): Say that elements can be
6148 const_wide_ints too.
6149 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6150 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6151 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6152 * optabs.c (expand_vector_broadcast): Likewise.
6154 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6155 Alan Hayward <alan.hayward@arm.com>
6156 David Sherwood <david.sherwood@arm.com>
6158 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6159 scaled index even if the unscaled address was invalid.
6160 Don't increase the complexity of using a scale in that case.
6162 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6163 Alan Hayward <alan.hayward@arm.com>
6164 David Sherwood <david.sherwood@arm.com>
6166 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6167 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6168 UNITS_PER_WORD bytes.
6169 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6170 based on REGMODE_NATURAL_SIZE of the inner mode.
6171 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6172 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6173 * expmed.c (lowpart_bit_field_p): Divide the value up into
6174 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6175 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6176 whether something is likely to occupy more than one register.
6178 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6181 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6182 profile_count::adjust_for_ipa_scaling.
6183 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6184 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6186 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6188 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6190 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6191 (split_group): Add total_orig and total_new arguments, estimate the
6192 number of statements related to the store group without store merging
6193 and with store merging.
6194 (imm_store_chain_info::output_merged_store): Adjust split_group
6195 callers, punt if estimated number of statements with store merging
6196 is not smaller than estimated number of statements without it.
6198 (handled_load): Remove has_single_use checks.
6199 (pass_store_merging::process_store): Likewise.
6201 2017-11-09 Richard Biener <rguenther@suse.de>
6203 PR tree-optimization/82902
6204 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6206 2017-11-09 Martin Liska <mliska@suse.cz>
6209 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6212 2017-11-09 Martin Liska <mliska@suse.cz>
6214 PR tree-optimization/82669
6215 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6217 2017-11-09 Martin Liska <mliska@suse.cz>
6219 PR gcov-profile/48463
6220 * coverage.c (coverage_begin_function): Output also end locus
6221 of a function and information whether the function is
6223 * gcov-dump.c (tag_function): Parse and print the information.
6224 * gcov.c (INCLUDE_MAP): Add include.
6225 (INCLUDE_SET): Likewise.
6226 (struct line_info): Move earlier in the source file because
6227 of vector<line_info> in function_info structure.
6228 (line_info::line_info): Likewise.
6229 (line_info::has_block): Likewise.
6230 (struct source_info): Add new member index.
6231 (source_info::get_functions_at_location): New function.
6232 (function_info::group_line_p): New function.
6233 (output_intermediate_line): New function.
6234 (output_intermediate_file): Use the mentioned function.
6235 (struct function_start): New.
6236 (struct function_start_pair_hash): Likewise.
6237 (process_file): Add code that identifies group functions.
6238 Assign lines either to global or function scope.
6239 (generate_results): Skip artificial functions.
6240 (find_source): Assign index for each source file.
6241 (read_graph_file): Read new flag artificial and end_line.
6242 (add_line_counts): Assign it either to global of function scope.
6243 (accumulate_line_counts): Isolate core of the function to
6244 accumulate_line_info and call it for both function and global
6246 (accumulate_line_info): New function.
6247 (output_line_beginning): Fix GNU coding style.
6248 (print_source_line): New function.
6249 (output_line_details): Likewise.
6250 (output_function_details): Likewise.
6251 (output_lines): Iterate both source (global) scope and function
6253 (struct function_line_start_cmp): New class.
6254 * doc/gcov.texi: Reflect changes in documentation.
6256 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6259 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6260 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6261 and similarly for not instead of neg.
6263 2017-11-08 Andi Kleen <ak@linux.intel.com>
6265 * config/i386/i386.opt: Add -mforce-indirect-call.
6266 * config/i386/predicates.md: Check for flag_force_indirect_call.
6267 * doc/invoke.texi: Document -mforce-indirect-call
6269 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6271 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6273 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6274 config/riscv/riscv.c (predict.h): New include.
6275 (riscv_slow_unaligned_access_p): No longer static.
6276 (riscv_block_move_straight): Add require.
6277 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6279 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6282 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6283 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6284 nonimmediate_operand predicate for operand 1 instead of
6287 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6289 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6291 * config/aarch64/constraints.md (Uml): New constraint.
6292 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6295 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6297 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6298 of two vec_duplicates into a vec_concat.
6300 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6302 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6303 Simplify vec_merge of vec_duplicate and vec_concat.
6304 * config/aarch64/constraints.md (Utq): New constraint.
6305 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6308 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6310 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6311 Simplify vec_merge of vec_duplicate and const_vector.
6312 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6314 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6315 mode iterator. Update predicate on operand 1 to
6316 handle non-const_vec constants. Delete constraints.
6317 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6319 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6321 PR tree-optimization/78821
6322 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6324 (store_operand_info::store_operand_info): Initialize it to false.
6325 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6326 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6327 store in the group, and if chain_info is non-NULL, to ignore altogether
6329 (compatible_load_p): Fail if bit_not_p does not match.
6330 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6331 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6332 (pass_store_merging::process_store): Adjust
6333 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6334 call terminate_all_aliasing_chains newly when adding a store into
6335 a chain with non-NULL chain_info.
6337 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6339 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6341 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6343 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6345 (aarch64_layout_frame): Initialise emit_frame_chain.
6346 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6347 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6349 2017-11-08 Martin Liska <mliska@suse.cz>
6351 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6352 of gimple_call_internal_p.
6354 2017-11-07 Tom de Vries <tom@codesourcery.com>
6356 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6358 2017-11-07 Tom de Vries <tom@codesourcery.com>
6360 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6363 2017-11-08 Martin Liska <mliska@suse.cz>
6366 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6368 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6370 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6372 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6375 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6376 (*cmp<mode>_ccz_1): New insn with $k alternative.
6379 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6380 RTX_COMM_COMPARE as commutative as well.
6381 (ix86_binary_operator_ok): Formatting fix.
6382 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6383 *<code><mode>3<mask_name><round_saeonly_name>,
6384 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6385 *<s>mul<mode>3_highpart<mask_name>,
6386 *vec_widen_umult_even_v16si<mask_name>,
6387 *vec_widen_umult_even_v8si<mask_name>,
6388 *vec_widen_umult_even_v4si<mask_name>,
6389 *vec_widen_smult_even_v16si<mask_name>,
6390 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6391 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6392 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6393 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6394 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6395 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6396 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6397 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6398 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6399 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6400 ix86_binary_operator_ok. Formatting fixes.
6401 (*<plusminus_insn><mode>3<mask_name><round_name>,
6402 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6405 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6407 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6408 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6409 eq and ne if TARGET_ISEL.
6410 (cmp): New code_iterator.
6411 (UNS, UNSU_, UNSIK): New code_attrs.
6412 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6413 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6415 ("eq<mode>3"): ... this.
6416 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6418 ("ne<mode>3"): ... this.
6420 2017-11-07 Julia Koval <julia.koval@intel.com>
6423 * common/config/i386/i386-common.c
6424 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6425 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6426 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6427 * config/i386/i386.opt: Ditto.
6428 * config/i386/i386.c (ix86_target_string): Ditto.
6429 (ix86_option_override_internal): Ditto.
6430 (ix86_init_mpx_builtins): Move MPX to args2.
6431 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6432 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6433 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6434 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6435 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6436 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6437 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6439 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6442 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6443 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6444 (zero-extendsidi peephole2): Remove peephole.
6446 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6449 * stor-layout.c: Include attribs.h.
6450 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6451 explicit lookup of "aligned" attribute.
6453 2017-11-07 Andrew Waterman <andrew@sifive.com>
6455 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6456 (riscv_expand_block_move): Likewise.
6457 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6459 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6460 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6461 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6462 (riscv_adjust_block_mem): Likewise.
6463 (riscv_block_move_loop): Likewise.
6464 (riscv_expand_block_move): Likewise.
6465 * config/riscv/riscv.md (movmemsi): New pattern.
6467 2017-11-07 Michael Clark <michaeljclark@mac.com>
6469 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6470 (MUSL_DYNAMIC_LINKER): Likewise.
6472 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6474 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6475 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6478 2017-11-07 Richard Biener <rguenther@suse.de>
6480 * match.pd: Fix build.
6482 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6483 Jackson Woodruff <jackson.woodruff@arm.com>
6485 PR tree-optimization/71026
6486 * match.pd: Canonicalize negate in division.
6488 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6491 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6493 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6495 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6496 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6498 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6500 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6501 non-scalar integral types.
6502 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6503 (-(A-B), -(~A)): New transformations.
6505 2017-11-07 Tom de Vries <tom@codesourcery.com>
6507 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6508 semicolon after "do {} while (0)".
6509 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6510 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6511 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6512 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6513 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6515 2017-11-07 Tom de Vries <tom@codesourcery.com>
6517 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6518 after "do {} while (0)".
6519 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6520 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6521 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6522 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6523 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6525 2017-11-07 Tom de Vries <tom@codesourcery.com>
6528 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6530 (arm_rtx_costs_internal): Add missing semicolon after
6531 HANDLE_NARROW_SHIFT_ARITH call.
6533 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6536 disable isel if it was not set explicitly.
6538 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6540 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6541 (add_type_attribute) likewise.
6543 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6545 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6546 of struct ix86_frame.
6547 (ix86_initial_elimination_offset): Likewise.
6548 (ix86_expand_split_stack_prologue): Likewise.
6550 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6552 * tree-vrp.h (enum value_range_type): Update stale comment.
6554 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6555 Alan Hayward <alan.hayward@arm.com>
6556 David Sherwood <david.sherwood@arm.com>
6558 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6559 (aarch64_expand_vec_perm_const): Take the number of units too.
6560 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6561 (aarch64_expand_vec_perm_const): Likewise.
6562 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6563 (vec_perm<mode>): Update accordingly.
6565 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6566 Alan Hayward <alan.hayward@arm.com>
6567 David Sherwood <david.sherwood@arm.com>
6569 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6570 Take the number of units too.
6571 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6572 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6573 but check for a vector mode before rather than after the call.
6574 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6575 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6576 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6577 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6578 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6579 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6580 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6581 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6582 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6583 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6584 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6585 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6586 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6587 (aarch64_sqdmull2_n<mode>): Update accordingly.
6589 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6590 Alan Hayward <alan.hayward@arm.com>
6591 David Sherwood <david.sherwood@arm.com>
6593 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6594 the number of units too.
6595 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6596 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6597 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6598 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6599 (vec_store_lanesxi<mode>): Update accordingly.
6601 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6602 Alan Hayward <alan.hayward@arm.com>
6603 David Sherwood <david.sherwood@arm.com>
6605 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6606 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6607 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6608 of units rather than the mode.
6609 * config/aarch64/iterators.md (nunits): New mode attribute.
6610 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6611 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6612 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6613 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6614 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6615 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6616 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6617 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6618 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6619 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6620 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6621 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6622 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6623 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6624 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6625 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6626 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6627 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6628 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6629 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6630 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6631 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6632 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6633 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6634 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6635 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6636 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6637 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6638 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6639 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6640 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6641 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6642 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6643 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6644 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6645 (aarch64_simd_vec_setv2di): Likewise.
6647 2017-11-06 Carl Love <cel@us.ibm.com>
6649 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6651 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6652 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6654 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6656 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6657 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6658 Add power 8 macro expansions.
6659 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6660 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6661 power 8 instructions. (VSX_XXBR): Add iterator.
6663 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6665 * config/arm/arm.md (predicable_short_it): Change default to "no",
6666 improve documentation, remove uses that are identical to the default.
6667 (enabled_for_depr_it): Rename to enabled_for_short_it.
6668 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6669 Remove default uses.
6670 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6671 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6672 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6673 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6675 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6678 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6679 float128 helper macros, which are no longer used after deleting
6680 the old 'q' built-in functions, and moving the round to odd
6681 built-in functions to being special built-in functions.
6682 (BU_FLOAT128_2): Likewise.
6683 (BU_FLOAT128_1_HW): Likewise.
6684 (BU_FLOAT128_2_HW): Likewise.
6685 (BU_FLOAT128_3_HW): Likewise.
6686 (FABSQ): Delete old 'q' built-in functions.
6687 (COPYSIGNQ): Likewise.
6688 (SQRTF128_ODD): Move round to odd built-in functions to be
6689 special built-in functions, so that we can handle
6690 -mabi=ieeelongdouble.
6691 (TRUNCF128_ODD): Likewise.
6692 (ADDF128_ODD): Likewise.
6693 (SUBF128_ODD): Likewise.
6694 (MULF128_ODD): Likewise.
6695 (DIVF128_ODD): Likewise.
6696 (FMAF128_ODD): Likewise.
6697 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
6698 built-in names to 'f128'.
6699 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
6700 old 'q' built-in functions, as the machine independent code for
6701 'f128' built-in functions handles this.
6702 (rs6000_expand_builtin): Add expansion for float128 round to odd
6703 functions, keying off on -mabi=ieeelongdouble of whether to use
6704 the KFmode or TFmode variant.
6705 (rs6000_init_builtins): Initialize the _Float128 round to odd
6707 * doc/extend.texi (PowerPC Built-in Functions): Document the old
6708 _Float128 'q' built-in functions are now mapped into the new
6709 'f128' built-in functions.
6711 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
6713 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
6714 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
6716 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6719 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6720 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6721 * toplev.c: Include "ipa-fnsummary.h".
6722 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6724 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6726 PR tree-optimization/82838
6727 * gimple-ssa-store-merging.c
6728 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6729 on a separate gimple_seq which is then appended to seq.
6731 2017-11-06 Jeff Law <law@redhat.com>
6734 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6735 (get_probe_interval): New functions.
6736 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6737 (ix86_adjust_stack_and_probe): Likewise.
6738 (output_adjust_stack_and_probe): Likewise.
6739 (ix86_emit_probe_stack_range): Likewise.
6740 (ix86_expand_prologue): Likewise.
6742 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6744 PR tree-optimization/82816
6745 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6746 if the modes of the two types are the same.
6747 (convert_plusminus_to_widen): Likewise.
6749 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6751 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6753 (usadv16qi): New define_expand.
6754 (usadv8hi): New define_expand.
6756 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6759 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6761 (inline_transform): Likewise.
6762 * predict.c (counts_to_freqs): Remove useless conditional.
6763 * profile-count.h (profile_count::apply_scale): Move sanity check.
6764 * tree-inline.c (copy_bb): Always scale.
6765 (copy_cfg_body): Likewise.
6767 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6770 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6772 (*cmp_ite0): Add enabled_for_depr_it attribute.
6773 (*cmp_ite1): Likewise.
6775 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6777 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6780 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6782 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6783 return the value by pointer.
6784 (extract_range_from_multiplicative_op_1): Update accordingly.
6785 Return as soon as an operation fails.
6787 2017-11-05 Tom de Vries <tom@codesourcery.com>
6790 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6791 (DEF_SANITIZER_BUILTIN): ... here.
6792 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6793 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6795 2017-11-05 Tom de Vries <tom@codesourcery.com>
6798 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6800 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6801 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6802 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6804 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6805 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6806 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6808 2017-11-05 Tom de Vries <tom@codesourcery.com>
6811 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6814 2017-11-04 Michael Clark <michaeljclark@mac.com>
6816 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6817 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6819 (*addsi3_extended): Likewise.
6820 (*addsi3_extended2): Likewise.
6821 (<optab>si3): Likewise.
6822 (<optab>di3): Likewise.
6823 (<optab><mode>3): Likewise.
6824 (<*optabe>si3_internal): Likewise.
6825 (zero_extendqi<SUPERQI:mode>2): Likewise.
6826 (*add<mode>hi3): Likewise.
6827 (*xor<mode>hi3): Likewise.
6828 (<optab>di3): Likewise.
6829 (*<optab>si3_extend): Likewise.
6830 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6831 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6832 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6834 2017-11-04 Andrew Waterman <andrew@sifive.com>
6836 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6837 TARGET_STRICT_ALIGN based upon -mtune argument.
6839 2017-11-04 Andrew Waterman <andrew@sifive.com>
6841 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6843 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6845 * config/i386/i386.c (choose_basereg): Use optional scratch
6846 register and add assertion.
6847 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6848 needed, and don't allocate stack.
6849 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6850 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6851 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6853 2017-11-03 Jeff Law <law@redhat.com>
6855 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6856 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6857 to probe at the start of a noreturn function.
6859 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6861 PR tree-optimization/78821
6862 * gimple-ssa-store-merging.c: Update the file comment.
6863 (MAX_STORE_ALIAS_CHECKS): Define.
6864 (struct store_operand_info): New type.
6865 (store_operand_info::store_operand_info): New constructor.
6866 (struct store_immediate_info): Add rhs_code and ops data members.
6867 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6868 and op1r arguments to the ctor, initialize corresponding data members.
6869 (struct merged_store_group): Add load_align_base and load_align
6871 (merged_store_group::merged_store_group): Initialize them.
6872 (merged_store_group::do_merge): Update them.
6873 (merged_store_group::apply_stores): Pick the constant for
6874 encode_tree_to_bitpos from one of the two operands, or skip
6875 encode_tree_to_bitpos if neither operand is a constant.
6876 (class pass_store_merging): Add process_store method decl. Remove
6877 bool argument from terminate_all_aliasing_chains method decl.
6878 (pass_store_merging::terminate_all_aliasing_chains): Remove
6879 var_offset_p argument and corresponding handling.
6880 (stmts_may_clobber_ref_p): New function.
6881 (compatible_load_p): New function.
6882 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6883 if there is overlap and rhs_code is not INTEGER_CST. For
6884 non-overlapping stores terminate group if rhs is not mergeable.
6885 (get_alias_type_for_stmts): Change first argument from
6886 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6887 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6888 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6890 (get_location_for_stmts): Change first argument from
6891 auto_vec<gimple *> & to vec<gimple *> &.
6892 (struct split_store): Remove orig_stmts data member, add orig_stores.
6893 (split_store::split_store): Create orig_stores rather than orig_stmts.
6894 (find_constituent_stmts): Renamed to ...
6895 (find_constituent_stores): ... this. Change second argument from
6896 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6897 to info structures rather than the statements.
6898 (split_group): Rename ALLOW_UNALIGNED argument to
6899 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6900 it. Adjust find_constituent_stores caller.
6901 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6902 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6903 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6904 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6905 (mem_valid_for_store_merging): New function.
6906 (handled_load): New function.
6907 (pass_store_merging::process_store): New method.
6908 (pass_store_merging::execute): Use process_store method. Adjust
6909 terminate_all_aliasing_chains caller.
6911 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6913 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6914 Return true for more constants, symbols and label references.
6915 (aarch64_valid_floating_const): Remove unused function.
6917 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6920 * config/aarch64/aarch64.c (aarch64_layout_frame):
6921 Undo forcing of LR at bottom of frame.
6923 2017-11-03 Jeff Law <law@redhat.com>
6926 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6927 for int_registers_saved.
6929 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6930 extracted from tree-ssa-dom.c.
6931 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6932 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6933 (record_equivalences_from_incoming_edge): Add additional argument
6934 to single_pred_edge_ignoring_loop_edges call.
6935 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6936 (uncprop_dom_walker::before_dom_children): Add additional argument
6937 to single_pred_edge_ignoring_loop_edges call.
6938 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6939 single_pred_edge_ignoring_loop_edges rather than open coding.
6940 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6942 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6944 * match.pd (-(-A)): Rewrite.
6946 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6948 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6949 (rs6000_emit_int_cmove): New declaration.
6950 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6951 (rs6000_emit_sISEL): Delete.
6952 (rs6000_emit_int_cmove): Make non-static.
6953 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6954 instead of rs6000_emit_sISEL.
6956 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6958 * asan.c (create_cond_insert_point): Maintain profile.
6959 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6961 * basic-block.h (struct basic_block_def): Remove frequency.
6962 (EDGE_FREQUENCY): Use to_frequency
6963 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6965 (find_traces): Update to use to_frequency.
6966 (find_traces_1_round): Likewise; use only IPA counts.
6967 (bb_to_key): Likewise.
6968 (connect_traces): Use IPA counts only.
6969 (copy_bb_p): Update to use to_frequency.
6970 (fix_up_crossing_landing_pad): Likewise.
6971 (sanitize_hot_paths): Likewise.
6972 * bt-load.c (basic_block_freq): Likewise.
6973 * cfg.c (init_flow): Set count_max to uninitialized.
6974 (check_bb_profile): Remove frequencies; check counts.
6975 (dump_bb_info): Do not dump frequencies.
6976 (update_bb_profile_for_threading): Update counts only.
6977 (scale_bbs_frequencies_int): Likewise.
6978 (MAX_SAFE_MULTIPLIER): Remove.
6979 (scale_bbs_frequencies_gcov_type): Update counts only.
6980 (scale_bbs_frequencies_profile_count): Update counts only.
6981 (scale_bbs_frequencies): Update counts only.
6982 * cfg.h (struct control_flow_graph): Add count-max.
6983 (update_bb_profile_for_threading): Update prototype.
6984 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6985 (find_many_sub_basic_blocks): Likewise.
6986 * cfgcleanup.c (try_forward_edges): Likewise.
6987 (try_crossjump_to_edge): Likewise.
6988 * cfgexpand.c (expand_gimple_cond): Likewise.
6989 (expand_gimple_tailcall): Likewise.
6990 (construct_init_block): Likewise.
6991 (construct_exit_block): Likewise.
6992 * cfghooks.c (verify_flow_info): Check consistency of counts.
6993 (dump_bb_for_graph): Do not dump frequencies.
6994 (split_block_1): Do not update frequencies.
6995 (split_edge): Do not update frequencies.
6996 (make_forwarder_block): Do not update frequencies.
6997 (duplicate_block): Do not update frequencies.
6998 (account_profile_record): Do not update frequencies.
6999 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7000 for global heuristics.
7001 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7002 (expected_loop_iterations_unbounded): Use counts only.
7003 * cfgloopmanip.c (scale_loop_profile): Simplify.
7004 (create_empty_loop_on_edge): Simplify
7006 (duplicate_loop_to_header_edge): Simplify
7007 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7008 (update_br_prob_note): Take care of removing note when profile
7010 (relink_block_chain): Do not dump frequency.
7011 (rtl_account_profile_record): Use to_frequency.
7012 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7013 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7014 (cgraph_update_edges_for_call_stmt_node): Likewise.
7015 (cgraph_edge::verify_count_and_frequency): Update.
7016 (cgraph_node::verify_node): Temporarily disable frequency verification.
7017 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7018 to_cgraph_frequency.
7019 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7020 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7022 (cgraph_node::expand_thunk): Update profile.
7023 * except.c (dw2_build_landing_pads): Do not update frequency.
7024 * final.c (compute_alignments): Use to_frequency.
7025 (dump_basic_block_info): Do not dump frequency.
7026 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7027 (dump_gimple_bb_header): Do not dump frequency.
7028 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7030 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7031 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7032 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7033 (init_before_recovery): Do not update frequency.
7034 (sched_create_recovery_edges): Do not update frequency.
7035 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7036 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7037 (ipa_cp_c_finalize): Set max_count to uninitialized.
7038 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7039 (param_change_prob): Use counts.
7040 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7042 * ipa-split.c (consider_split): Use to_frequency.
7043 (split_function): Use to_frequency.
7044 * ira-build.c (loop_compare_func): Likewise.
7045 (mark_loops_for_removal): Likewise.
7046 (mark_all_loops_for_removal): Likewise.
7047 * loop-doloop.c (doloop_modify): Do not update frequency.
7048 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7050 * lto-streamer-in.c (input_function): Update count_max.
7051 * omp-expand.c (expand_omp_taskreg): Update count_max.
7052 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7053 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7054 (maybe_hot_count_p): Use ipa counts only.
7055 (maybe_hot_bb_p): Simplify.
7056 (maybe_hot_edge_p): Simplify.
7057 (probably_never_executed): Do not take frequency argument.
7058 (probably_never_executed_bb_p): Do not pass frequency.
7059 (probably_never_executed_edge_p): Likewise.
7060 (combine_predictions_for_bb): Check that profile is nonzero.
7061 (propagate_freq): Do not set frequency.
7062 (drop_profile): Simplify.
7063 (counts_to_freqs): Simplify.
7064 (expensive_function_p): Use to_frequency.
7065 (propagate_unlikely_bbs_forward): Simplify.
7066 (determine_unlikely_bbs): Simplify.
7067 (estimate_bb_frequencies): Add hack to silence graphite issues.
7068 (compute_function_frequency): Use ipa counts.
7069 (pass_profile::execute): Update.
7070 (rebuild_frequencies): Use counts only.
7071 (force_edge_cold): Use counts only.
7072 * profile-count.c (profile_count::dump): Dump new count types.
7073 (profile_count::differs_from_p): Check compatiblity.
7074 (profile_count::to_frequency): New function.
7075 (profile_count::to_cgraph_frequency): New function.
7076 * profile-count.h (struct function): Declare.
7077 (enum profile_quality): Add profile_guessed_local and
7078 profile_guessed_global0.
7079 (class profile_proability): Decrease number of bits to 29;
7080 update from_reg_br_prob_note and to_reg_br_prob_note.
7081 (class profile_count: Update comment; decrease number of bits
7082 to 61. Check compatibility.
7083 (profile_count::compatible_p): New private member function.
7084 (profile_count::ipa_p): New member function.
7085 (profile_count::operator<): Handle global zero correctly.
7086 (profile_count::operator>): Handle global zero correctly.
7087 (profile_count::operator<=): Handle global zero correctly.
7088 (profile_count::operator>=): Handle global zero correctly.
7089 (profile_count::nonzero_p): New member function.
7090 (profile_count::force_nonzero): New member function.
7091 (profile_count::max): New member function.
7092 (profile_count::apply_scale): Handle IPA scalling.
7093 (profile_count::guessed_local): New member function.
7094 (profile_count::global0): New member function.
7095 (profile_count::ipa): New member function.
7096 (profile_count::to_frequency): Declare.
7097 (profile_count::to_cgraph_frequency): Declare.
7098 * profile.c (OVERLAP_BASE): Delete.
7099 (compute_frequency_overlap): Delete.
7100 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7101 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7102 * sched-ebb.c (rank): Use counts only.
7103 * shrink-wrap.c (handle_simple_exit): Use counts only.
7104 (try_shrink_wrapping): Use counts only.
7105 (place_prologue_for_one_component): Use counts only.
7106 * tracer.c (find_best_predecessor): Use to_frequency.
7107 (find_trace): Use to_frequency.
7108 (tail_duplicate): Use to_frequency.
7109 * trans-mem.c (expand_transaction): Do not update frequency.
7110 * tree-call-cdce.c: Do not update frequency.
7111 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7112 (gimple_merge_blocks): Likewise.
7113 (gimple_split_edge): Likewise.
7114 (gimple_duplicate_sese_region): Likewise.
7115 (gimple_duplicate_sese_tail): Likewise.
7116 (move_sese_region_to_fn): Likewise.
7117 (gimple_account_profile_record): Likewise.
7118 (insert_cond_bb): Likewise.
7119 * tree-complex.c (expand_complex_div_wide): Likewise.
7120 * tree-eh.c (lower_resx): Update profile.
7121 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7123 (initialize_cfun): Do not initialize frequencies
7124 (freqs_to_counts): Delete.
7125 (copy_cfg_body): Ignore count parameter.
7126 (copy_body): Update.
7127 (expand_call_inline): Update count_max.
7128 (optimize_inline_calls): Update count_max.
7129 (tree_function_versioning): Update count_max.
7130 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7131 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7133 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7134 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7135 (try_peel_loop): Likewise.
7136 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7138 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7139 (tree_transform_and_unroll_loop): Do not use frequencies
7140 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7141 Use reliable prediction only.
7142 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7143 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7144 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7145 probability scaling.
7146 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7147 not update frequency
7148 (any_remaining_duplicated_blocks): Likewise.
7149 (update_profile): Likewise.
7150 (estimated_freqs_path): Delete.
7151 (freqs_to_counts_path): Delete.
7152 (clear_counts_path): Delete.
7153 (ssa_fix_duplicate_block_edges): Likewise.
7154 (duplicate_thread_path): Likewise.
7155 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7156 * tree-tailcall.c (decrease_profile): Do not update frequency.
7157 (eliminate_tail_call): Likewise.
7158 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7159 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7160 (optimize_mask_stores): Likewise.
7161 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7162 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7163 (ubsan_expand_ptr_ifn): Update profile.
7164 * value-prof.c (gimple_ic): Simplify.
7165 * value-prof.h (gimple_ic): Update prototype.
7166 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7167 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7169 (want_inline_self_recursive_call_p): Likewise.
7170 (resolve_noninline_speculation): Only cummulate defined counts.
7171 (inline_small_functions): Use nonzero_p.
7172 (ipa_inline): Do not access freed node.
7174 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7176 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7177 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7179 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7181 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7182 non-legitimate address.
7184 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7186 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7187 (*lt0_<mode>di, *lt0_<mode>si): New.
7189 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7191 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7192 TARGET_PAIRED_FLOAT.
7194 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7195 Jim Wilson <jim.wilson@linaro.org>
7197 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7198 * config/aarch64/aarch64-tune.md: Regenerated.
7199 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7200 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7202 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7204 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7205 (arc_expand_prologue): Restore blink for millicode.
7206 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7208 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7211 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7212 gen_vec_duplicate after forcing the scalar into a register.
7214 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7216 * combine (try_combine): Print the insns input to try_combine to the
7219 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7222 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7223 Remove second argument from aarch64_process_target_attr call.
7224 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7226 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7228 (aarch64_handle_attr_arch): Remove second argument.
7229 (aarch64_handle_attr_cpu): Ditto.
7230 (aarch64_handle_attr_tune): Ditto.
7231 (aarch64_handle_attr_isa_flags): Ditto.
7232 (aarch64_process_one_target_attr): Ditto.
7233 (aarch64_process_target_attr): Ditto.
7234 (aarch64_option_valid_attribute_p): Remove second argument.
7235 on aarch64_process_target_attr call.
7237 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7239 * diagnostic.c: Include "selftest-diagnostic.h".
7240 (selftest::assert_location_text): New function.
7241 (selftest::test_diagnostic_get_location_text): New function.
7242 (selftest::diagnostic_c_tests): Call it.
7244 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7246 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7247 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7248 (class selftest::test_diagnostic_context): Move to...
7249 * selftest-diagnostic.c: New file.
7250 * selftest-diagnostic.h: New file.
7252 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7254 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7255 offset range for FT32B.
7256 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7257 * config/ft32/ft32.md: Add TARGET_NOPM.
7258 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7259 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7261 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7263 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7265 2017-11-02 Jeff Law <law@redhat.com>
7267 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7268 virtual keyword on FINAL OVERRIDE members.
7270 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7271 virtuals together. Add virtual destructor.
7272 (substitute_and_fold_engine): Similarly.
7274 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7276 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7278 2017-11-02 Richard Biener <rguenther@suse.de>
7280 PR tree-optimization/82795
7281 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7283 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7285 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7287 (gcc_GAS_CHECK_FEATURE): Remove.
7288 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7289 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7290 * configure: Regenerate.
7292 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7294 * config/arc/arc.c (hwloop_optimize): Account for empty
7297 2017-11-02 Richard Biener <rguenther@suse.de>
7300 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7301 Truncate ARRAY_REF index and element size.
7303 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7305 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7307 2017-11-01 Jeff Law <law@redhat.com>
7309 * tree-ssa-ccp.c (ccp_folder): New class derived from
7310 substitute_and_fold_engine.
7311 (ccp_folder::get_value): New member function.
7312 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7313 (ccp_fold_stmt): Remove prototype.
7314 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7315 * tree-ssa-copy.c (copy_folder): New class derived from
7316 substitute_and_fold_engine.
7317 (copy_folder::get_value): Renamed from get_value.
7318 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7319 * tree-vrp.c (vrp_folder): New class derived from
7320 substitute_and_fold_engine.
7321 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7322 (vrp_folder::get_value): New member function.
7323 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7324 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7325 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7326 provide a class interface to folder/substitute routines.
7327 (ssa_prop_fold_stmt_fn): Remove typedef.
7328 (ssa_prop_get_value_fn): Likewise.
7329 (subsitute_and_fold): Remove prototype.
7330 (replace_uses_in): Likewise.
7331 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7332 Renamed from replace_uses_in. Call the virtual member function
7333 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7334 (substitute_and_fold_dom_walker): Remove initialization of
7335 data member entries for calbacks. Add substitute_and_fold_engine
7336 member and initialize it.
7337 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7338 member functions for get_value, replace_phi_args_in c
7339 replace_uses_in, and fold_stmt calls.
7340 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7341 substitute_and_fold. Remove assert. Update ctor call.
7343 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7344 (ssa_prop_visit_phi_fn): Likewise.
7345 (class ssa_propagation_engine): New class to provide an interface
7347 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7349 (ssa_prop_visit_phi): Likewise.
7350 (ssa_propagation_engine::simulate_stmt): Moved into class.
7351 Call visit_phi/visit_stmt from the class rather than via
7352 file scoped static variables.
7353 (ssa_propagation_engine::simulate_block): Moved into class.
7354 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7355 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7356 set file scoped statics for the visit_stmt/visit_phi callbacks.
7357 * tree-complex.c (complex_propagate): New class derived from
7358 ssa_propagation_engine.
7359 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7360 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7361 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7363 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7364 ssa_propagation_engine.
7365 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7366 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7367 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7368 * tree-ssa-copy.c (copy_prop): New class derived from
7369 ssa_propagation_engine.
7370 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7371 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7372 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7373 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7374 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7375 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7376 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7378 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7380 PR rtl-optimization/82778
7381 PR rtl-optimization/82597
7382 * compare-elim.c (struct comparison): Add in_a_setter field.
7383 (find_comparison_dom_walker::before_dom_children): Remove killed
7384 bitmap and df_simulate_find_defs call, instead walk the defs.
7385 Compute last_setter and initialize in_a_setter. Merge definitions
7386 with first initialization for a few variables.
7387 (try_validate_parallel): Use insn_invalid_p instead of
7388 recog_memoized. Return insn rather than just the pattern.
7389 (try_merge_compare): Fix up comment. Don't uselessly test if
7390 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7392 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7395 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7396 Alan Hayward <alan.hayward@arm.com>
7397 David Sherwood <david.sherwood@arm.com>
7399 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7400 aarch64_hard_regno_nregs to get the number of registers
7403 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7404 Alan Hayward <alan.hayward@arm.com>
7405 David Sherwood <david.sherwood@arm.com>
7407 * config/aarch64/constraints.md (Upl): Rename to...
7409 * config/aarch64/aarch64.md
7410 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7413 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7414 Alan Hayward <alan.hayward@arm.com>
7415 David Sherwood <david.sherwood@arm.com>
7417 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7418 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7421 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7422 Alan Hayward <alan.hayward@arm.com>
7423 David Sherwood <david.sherwood@arm.com>
7425 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7426 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7427 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7429 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7430 of a single element.
7431 * config/aarch64/iterators.md: Add a comment above the permute
7432 unspecs to say that they are generated directly by
7433 aarch64_expand_vec_perm_const.
7434 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7436 2017-11-01 Nathan Sidwell <nathan@acm.org>
7438 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7440 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7442 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7443 and medany code models, and describe what they do.
7445 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7447 Revert accidental duplicate:
7449 * combine.c (can_change_dest_mode): Reject changes in
7450 REGMODE_NATURAL_SIZE.
7452 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7454 PR rtl-optimization/64682
7455 PR rtl-optimization/69567
7456 PR rtl-optimization/69737
7457 PR rtl-optimization/82683
7458 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7459 register mentioned in the note, drop the note, unless it came from I3,
7460 in which case it should go to I3 again.
7462 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7464 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7465 and return false if not.
7466 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7468 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7470 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7471 range is known to be empty.
7473 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7474 Alan Hayward <alan.hayward@arm.com>
7475 David Sherwood <david.sherwood@arm.com>
7477 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7478 and CONST_VECTOR_NUNITS instead of computing the number of units from
7479 the byte sizes of the vector and element.
7480 (simplify_binary_operation_1): Likewise.
7481 (simplify_const_binary_operation): Likewise.
7482 (simplify_ternary_operation): Likewise.
7484 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7485 Alan Hayward <alan.hayward@arm.com>
7486 David Sherwood <david.sherwood@arm.com>
7488 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7489 (int_mem_offset): ...this new function.
7490 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7491 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7492 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7495 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7496 Alan Hayward <alan.hayward@arm.com>
7497 David Sherwood <david.sherwood@arm.com>
7499 * lower-subreg.c (interesting_mode_p): New function.
7500 (compute_costs, find_decomposable_subregs, decompose_register)
7501 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7502 (resolve_clobber, dump_choices): Use it.
7504 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7505 Alan Hayward <alan.hayward@arm.com>
7506 David Sherwood <david.sherwood@arm.com>
7508 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7510 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7511 Alan Hayward <alan.hayward@arm.com>
7512 David Sherwood <david.sherwood@arm.com>
7514 * alias.c (find_base_value, find_base_term): Only process integer
7515 truncations. Check the precision rather than the size.
7517 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7518 Alan Hayward <alan.hayward@arm.com>
7519 David Sherwood <david.sherwood@arm.com>
7521 * machmode.h (is_narrower_int_mode): New function
7522 * optabs.c (expand_float, expand_fix): Use it.
7523 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7525 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7526 Alan Hayward <alan.hayward@arm.com>
7527 David Sherwood <david.sherwood@arm.com>
7529 * rtl.h (narrower_subreg_mode): New function.
7530 * ira-color.c (update_costs_from_allocno): Use it.
7532 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7533 Alan Hayward <alan.hayward@arm.com>
7534 David Sherwood <david.sherwood@arm.com>
7536 * optabs-query.h (convert_optab_p): New function, split out from...
7537 (convert_optab_handler): ...here.
7538 (widening_optab_handler): Delete.
7539 (find_widening_optab_handler): Remove permit_non_widening parameter.
7540 (find_widening_optab_handler_and_mode): Likewise. Provide an
7541 override that operates on mode class wrappers.
7542 * optabs-query.c (widening_optab_handler): Delete.
7543 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7544 parameter. Assert that the two modes are the same class and that
7545 the "from" mode is narrower than the "to" mode. Use
7546 convert_optab_handler instead of widening_optab_handler.
7547 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7548 instead of widening_optab_handler.
7549 * expr.c (expand_expr_real_2): Update calls to
7550 find_widening_optab_handler.
7551 * optabs.c (expand_widen_pattern_expr): Likewise.
7552 (expand_binop_directly): Take the insn_code as a parameter.
7553 (expand_binop): Only call find_widening_optab_handler for
7554 conversion optabs; use optab_handler otherwise. Update calls
7555 to find_widening_optab_handler and expand_binop_directly.
7556 Use convert_optab_handler instead of widening_optab_handler.
7557 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7558 find_widening_optab_handler and use scalar_mode rather than
7560 (convert_plusminus_to_widen): Likewise.
7562 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7563 Alan Hayward <alan.hayward@arm.com>
7564 David Sherwood <david.sherwood@arm.com>
7566 * machmode.h (fixed_size_mode): New class.
7567 * rtl.h (get_pool_mode): Return fixed_size_mode.
7568 * gengtype.c (main): Add fixed_size_mode.
7569 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7570 (get_raw_arg_mode): Likewise.
7571 * doc/tm.texi: Regenerate.
7572 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7573 * targhooks.c (default_get_reg_raw_mode): Likewise.
7574 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7575 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7576 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7577 (msp430_get_raw_result_mode): Likewise.
7578 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7579 * dbxout.c (dbxout_parms): Require fixed-size modes.
7580 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7581 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7582 * omp-low.c (lower_oacc_reductions): Likewise.
7583 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7584 (simplify_subreg): Update accordingly.
7585 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7586 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7587 that aren't fixed-size.
7588 (get_pool_mode): Return a fixed_size_mode.
7589 (output_constant_pool_2): Take a fixed_size_mode.
7591 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7592 Alan Hayward <alan.hayward@arm.com>
7593 David Sherwood <david.sherwood@arm.com>
7595 * doc/rtl.texi (vec_series): Document.
7596 (const): Say that the operand can be a vec_series.
7597 * rtl.def (VEC_SERIES): New rtx code.
7598 * rtl.h (const_vec_series_p_1): Declare.
7599 (const_vec_series_p): New function.
7600 * emit-rtl.h (gen_const_vec_series): Declare.
7601 (gen_vec_series): Likewise.
7602 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7603 (gen_vec_series): Likewise.
7604 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7605 * simplify-rtx.c (simplify_unary_operation): Handle negations
7607 (simplify_binary_operation_series): New function.
7608 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7609 (test_vector_ops_series): New function.
7610 (test_vector_ops): Call it.
7611 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7612 gen_const_vec_series.
7613 (altivec_lvsr): Likewise.
7614 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7616 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7617 Alan Hayward <alan.hayward@arm.com>
7618 David Sherwood <david.sherwood@arm.com>
7620 * doc/rtl.texi (const): Update description of address constants.
7621 Say that vector constants are allowed too.
7622 * common.md (E, F): Use CONSTANT_P instead of checking for
7624 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7625 checking for CONST_VECTOR.
7626 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7628 * expr.c (expand_expr_real_2): Check for vector modes instead
7629 of checking for CONST_VECTOR.
7630 * rtl.h (const_vec_p): New function.
7631 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7632 (unwrap_const_vec_duplicate): Handle them here too.
7634 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7635 David Malcolm <dmalcolm@redhat.com>
7636 Alan Hayward <alan.hayward@arm.com>
7637 David Sherwood <david.sherwood@arm.com>
7639 * rtl.h (vec_duplicate_p): New function.
7640 * selftest-rtl.c (assert_rtx_eq_at): New function.
7641 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7642 (assert_rtx_eq_at): Declare.
7643 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7644 * selftest-run-tests.c (selftest::run_tests): Call it.
7645 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7646 (simplify_unary_operation_1): Recursively handle vector duplicates.
7647 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7649 (simplify_subreg): Handle subregs of vector duplicates.
7650 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7651 (selftest::simplify_rtx_c_tests): New functions.
7653 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7654 Alan Hayward <alan.hayward@arm.com>
7655 David Sherwood <david.sherwood@arm.com>
7657 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7658 (gen_vec_duplicate): Likewise.
7659 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7661 (gen_const_vector): ...here.
7662 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7663 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7664 whose elements are all equal.
7665 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7666 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7667 (simplify_relational_operation): Likewise.
7668 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7670 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7671 (aarch64_expand_vector_init): Likewise.
7672 * config/arm/arm.c (neon_vdup_constant): Likewise.
7673 (neon_expand_vector_init): Likewise.
7674 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7675 (arm_block_set_unaligned_vect): Likewise.
7676 (arm_block_set_aligned_vect): Likewise.
7677 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7678 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7679 (expand_vec_perm_even_odd_pack): Likewise.
7680 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7681 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7682 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7683 gen_const_vec_duplicate.
7684 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7685 * config/mips/mips.c (mips_gen_const_int_vector): Use
7686 gen_const_vec_duplicate.
7687 (mips_expand_vector_init): Use CONST0_RTX.
7688 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7689 (define_split): Use gen_const_vec_duplicate.
7690 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7691 (define_split): Use gen_const_vec_duplicate.
7692 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7693 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7694 * config/spu/spu.c (spu_const): Likewise.
7696 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7697 Alan Hayward <alan.hayward@arm.com>
7698 David Sherwood <david.sherwood@arm.com>
7700 * combine.c (can_change_dest_mode): Reject changes in
7701 REGMODE_NATURAL_SIZE.
7703 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
7705 * configure.ac (--enable-libssp): New.
7706 (gcc_cv_libc_provides_ssp): Check for explicit setting before
7707 trying to determine target-specific default. Adjust indentation.
7708 * configure: Regenerated.
7709 * doc/install.texi (Configuration): Expand --disable-libssp
7712 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
7714 config/i386/i386.c (ix86_expand_epilogue): Correct stack
7717 2017-10-31 Martin Jambor <mjambor@suse.cz>
7720 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7722 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7724 * auto-profile.c (autofdo_source_profile::read): Use
7725 UNKNOWN_LOCATION rather than 0.
7726 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7727 (warning_at): ...this overload.
7728 (warning_at_rich_loc_n): Rename to...
7729 (warning_n): ...this overload.
7730 (error_at_rich_loc): Rename to...
7731 (error_at): ...this overload.
7732 (pedwarn_at_rich_loc): Rename to...
7733 (pedwarn): ...this overload.
7734 (permerror_at_rich_loc): Rename to...
7735 (permerror): ...this overload.
7736 (inform_at_rich_loc): Rename to...
7737 (inform): ...this overload.
7738 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7739 (diagnostic_n_impl_richloc): Rename to...
7740 (diagnostic_n_impl): ...this rich_location *-based decl.
7741 (inform_at_rich_loc): Rename to...
7742 (inform): ...this, and add an assertion.
7743 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7744 (warning_at_rich_loc): Rename to...
7745 (warning_at): ...this, and add an assertion.
7746 (warning_at_rich_loc_n): Rename to...
7747 (warning_n): ...this, and add an assertion.
7748 (warning_n): Update location_t-based implementation for removal of
7749 location_t-based diagnostic_n_impl.
7750 (pedwarn_at_rich_loc): Rename to...
7751 (pedwarn): ...this, and add an assertion.
7752 (permerror_at_rich_loc): Rename to...
7753 (permerror): ...this, and add an assertion.
7754 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7755 (error_at_rich_loc): Rename to...
7756 (error_at): ...this, and add an assertion.
7757 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7758 (driver::do_spec_on_infiles): Likewise.
7759 * substring-locations.c (format_warning_va): Update for renaming
7760 of inform_at_rich_loc.
7762 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7764 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7765 _Float<N>X built-in functions so that the variant without the
7766 "__builtin_" prefix is only enabled for the GNU C and Objective C
7767 languages when they are in non-strict ANSI/ISO mode.
7768 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7769 * target.def (floatn_builtin_p): Add a target hook to control
7770 whether _Float<N> and _Float<N>X built-in functions without the
7771 "__builtin_" prefix are enabled, and return true for C and
7772 Objective C in the default hook. Include langhooks.h in
7774 * targhooks.h (default_floatn_builtin_p): Likewise.
7775 * targhooks.c (default_floatn_builtin_p): Likewise.
7776 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7777 floatn_builtin_p target hook.
7778 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7780 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7781 Eric Botcazou <ebotcazou@adacore.com>
7783 PR rtl-optimization/81803
7784 * lra-constraints.c (curr_insn_transform): Also reload the whole
7785 register for a strict subreg no wider than a word if this is for
7786 a WORD_REGISTER_OPERATIONS target.
7788 2017-10-31 Jason Merrill <jason@redhat.com>
7790 * gdbinit.in: Skip over inlines from timevar.h.
7792 2017-10-31 Martin Liska <mliska@suse.cz>
7794 * doc/gcov.texi: Document new option.
7795 * gcov.c (print_usage): Likewise print it.
7796 (process_args): Support the argument.
7797 (format_count): New function.
7798 (format_gcov): Use the function.
7800 2017-10-31 Martin Liska <mliska@suse.cz>
7802 * gcov.c (struct name_map): do not use typedef.
7803 Define operator== and operator<.
7804 (name_search): Remove.
7805 (name_sort): Remove.
7806 (main): Do not allocate names.
7807 (process_file): Add vertical space.
7808 (generate_results): Use std::find.
7809 (release_structures): Do not release memory.
7810 (find_source): Use std::find.
7812 2017-10-31 Martin Liska <mliska@suse.cz>
7814 * gcov.c (struct line_info): Remove it's typedef.
7815 (line_info::line_info): Add proper ctor.
7816 (line_info::has_block): Do not use a typedef.
7817 (struct source_info): Do not use typedef.
7818 (circuit): Likewise.
7819 (get_cycles_count): Likewise.
7820 (output_intermediate_file): Iterate via vector iterator.
7821 (add_line_counts): Use std::vector methods.
7822 (accumulate_line_counts): Likewise.
7823 (output_lines): Likewise.
7825 2017-10-31 Martin Liska <mliska@suse.cz>
7827 * gcov.c (struct source_info): Remove typedef.
7828 (source_info::source_info): Add proper ctor.
7829 (accumulate_line_counts): Use struct, not it's typedef.
7830 (output_gcov_file): Likewise.
7831 (output_lines): Likewise.
7832 (main): Do not allocate an array.
7833 (output_intermediate_file): Use size of vector container.
7834 (process_file): Resize the vector.
7835 (generate_results): Do not preallocate, use newly added vector
7837 (release_structures): Do not release sources.
7838 (find_source): Use vector methods.
7839 (add_line_counts): Do not use typedef.
7841 2017-10-31 Martin Liska <mliska@suse.cz>
7843 * doc/gcov.texi: Document that.
7844 * gcov.c (add_line_counts): Mark lines with a non-executed
7846 (output_line_beginning): Handle such lines.
7847 (output_lines): Pass new argument.
7848 (output_intermediate_file): Print it in intermediate format.
7850 2017-10-31 Martin Liska <mliska@suse.cz>
7852 * color-macros.h: New file.
7853 * diagnostic-color.c: Factor out color related to macros to
7855 * doc/gcov.texi: Document -k option.
7856 * gcov.c (INCLUDE_STRING): Include string.h.
7857 (print_usage): Add -k option.
7858 (process_args): Parse it.
7859 (pad_count_string): New function.
7860 (output_line_beginning): Likewise.
7861 (DEFAULT_LINE_START): New macro.
7862 (output_lines): Support color output.
7864 2017-10-31 Martin Liska <mliska@suse.cz>
7866 PR gcov-profile/82633
7867 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7868 their interaction with GCOV infrastructure.
7869 * configure.ac: Add -fkeep-{inline,static}-functions to
7871 * configure: Regenerate.
7873 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7876 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7878 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7881 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7882 into a register if it does not fit into an immediate offset field.
7884 2017-10-31 Olivier Hainque <hainque@adacore.com>
7886 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7888 2017-10-31 Julia Koval <julia.koval@intel.com>
7890 * config.gcc: Add gfniintrin.h.
7891 * config/i386/gfniintrin.h: New.
7892 * config/i386/i386-builtin-types.def
7893 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7894 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7895 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7896 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7897 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7898 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7899 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7900 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7901 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7902 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7903 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7904 * config/i386/immintrin.h: Include gfniintrin.h.
7905 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7907 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7909 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7911 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7913 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7914 (arm_ashldi3_1bit): Remove pattern.
7915 (ashrdi3): Remove shift by 1 expansion.
7916 (arm_ashrdi3_1bit): Remove pattern.
7917 (lshrdi3): Remove shift by 1 expansion.
7918 (arm_lshrdi3_1bit): Remove pattern.
7919 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7920 cost of ashldi3 by 1.
7921 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7922 (<shift>di3_neon): Likewise.
7924 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7926 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7927 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7928 and (*aarch64_simd_mov<VQ:mode>).
7929 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7930 pattern alternative.
7931 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7932 attributes to match pattern alternative.
7934 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7936 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7937 * config/rs6000/emmintrin.h: New file.
7938 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7940 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7942 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7943 * (movdi_vfp_cortexa8): Remove pattern.
7945 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7947 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7949 (Specific, alpha*-dec-osf5.1): Remove.
7950 (Specific, mips-sgi-irix5): Remove.
7951 (Specific, mips-sgi-irix6): Remove.
7953 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7956 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7957 arguments to clear_bit_region_be.
7959 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7961 * gimplify.c: Include memmodel.h.
7963 2017-10-30 Martin Jambor <mjambor@suse.cz>
7965 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7966 condition whether loop should be executed at all.
7968 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7970 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7971 gimple folding of vec_madd() intrinsics.
7972 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7973 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7974 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7976 2017-10-30 Richard Biener <rguenther@suse.de>
7978 PR tree-optimization/82762
7980 2017-10-23 Richard Biener <rguenther@suse.de>
7982 PR tree-optimization/82129
7984 2017-08-01 Richard Biener <rguenther@suse.de>
7986 PR tree-optimization/81181
7987 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7988 (compute_antic): ... end of iteration here.
7990 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7992 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7993 -std=iso9899:2017 and -std=gnu17.
7994 * doc/standards.texi (C Language): Document C17 support.
7995 * doc/cpp.texi (Overview): Mention -std=c17.
7996 (Standard Predefined Macros): Document C11 and C17 values of
7997 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7998 * doc/extend.texi (Inline): Do not list individual options for
7999 standards newer than C99.
8000 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8002 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8005 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8007 * asan.c (asan_finish_file): Align asan globals array by shadow
8010 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8013 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8014 (struct store_immediate_info): Add bitregion_start and bitregion_end
8016 (store_immediate_info::store_immediate_info): Add brs and bre
8017 arguments and initialize bitregion_{start,end} from those.
8018 (struct merged_store_group): Add bitregion_start, bitregion_end,
8019 align_base and mask fields. Drop unnecessary struct keyword from
8020 struct store_immediate_info. Add do_merge method.
8021 (clear_bit_region_be): Use memset instead of loop storing zeros.
8022 (merged_store_group::do_merge): New method.
8023 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8024 stores as long as the surrounding bitregions have no gaps.
8025 (merged_store_group::merge_overlapping): Use do_merge.
8026 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8027 is byte aligned, rather than requiring that start and width are
8028 byte aligned. Drop unnecessary struct keyword from
8029 struct store_immediate_info. Allocate and populate also mask array.
8030 Make start of the arrays relative to bitregion_start rather than
8031 start and size them according to bitregion_{end,start} difference.
8032 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8033 struct store_immediate_info.
8034 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8035 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8036 struct keyword from struct store_immediate_info.
8037 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8038 between stores as long as the surrounding bitregions have no gaps.
8040 (struct split_store): Add orig non-static data member.
8041 (split_store::split_store): Initialize orig to false.
8042 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8043 if there is exactly a single original stmt. Change stmts argument
8044 to pointer from reference, if NULL, don't push anything to it. Add
8045 first argument, use it to optimize skipping over orig stmts that
8046 are known to be before bitpos already. Simplify.
8047 (split_group): Return unsigned int count how many stores are or
8048 would be needed rather than a bool. Add allow_unaligned argument.
8049 Change split_stores argument from reference to pointer, if NULL,
8050 only do a dry run computing how many stores would be produced.
8051 Rewritten algorithm to use both alignment and misalign if
8052 !allow_unaligned and handle bitfield stores with gaps.
8053 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8054 from bitregion_start instead of start. Compute allow_unaligned
8055 here, if true, do 2 split_group dry runs to compute which one
8056 produces fewer stores and prefer aligned if equal. Punt if
8057 new count is bigger or equal than original before emitting any
8058 statements, rather than during that. Remove no longer needed
8059 new_ssa_names tracking. Replace num_stmts with
8060 split_stores.length (). Use 32-bit stack allocated entries
8061 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8062 if possible. Handle bitfields with gaps.
8063 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8064 Compute bitregion_{start,end} for the stores and construct
8065 store_immediate_info with that. Formatting fixes.
8067 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8070 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8071 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8073 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8075 * gimplify.c: Include tm_p.h.
8077 * common.opt (gcoff): Re-add as ignored option.
8078 (gcoff1, gcoff2, gcoff3): Likewise.
8080 * Makefile.in (OBJS): Delete sdbout.o.
8081 (GTFILES): Delete $(srcdir)/sdbout.c.
8082 * debug.h: Delete sdb_debug_hooks.
8083 * final.c: Delete sdbout.h include.
8084 (final_scan_insn): Delete SDB_DEBUG check.
8085 (rest_of_clean_state): Likewise.
8086 * output.h: Delete sdb_begin_function_line.
8089 * toplev.c: Delete sdbout.h include.
8090 (process_options): Delete SDB_DEBUG check.
8091 * tree-core.h (tree_type_common): Delete pointer field of
8093 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8094 TYPE_SYMTAB_POINTER.
8095 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8096 (TYPE_SYMTAB_IS_POINTER): Delete.
8097 (TYPE_SYMTAB_IS_DIE): Renumber.
8098 * xcoffout.c: Refer to former sdbout.c file.
8099 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8101 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8102 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8103 (Debugging Options): Delete -gcoff.
8104 (-gstabs): Delete SDB reference.
8106 (-gcoff@var{level}): Delete.
8107 * doc/passes.texi (Debugging information output): Delete SDB and
8108 sdbout.c references.
8109 * doc/tm.texi: Regenerate.
8110 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8111 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8113 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8114 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8115 (SDB_DEBUGGING_INFO): Delete.
8116 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8117 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8118 * target.def (output_source_filename): Delete COFF reference.
8120 * common.opt (gcoff): Delete.
8121 (gxcoff+): Update Negative chain.
8122 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8124 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8125 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8126 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8128 (expand_function_start): Change sdb reference to past tense.
8129 (expand_function_end): Change sdb reference to past tense.
8130 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8131 * opts.c (debug_type_names): Delete coff entry.
8132 (common_handle_option): Delete OPT_gcoff case.
8133 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8135 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8136 * config/cris/cris.h: Delete SDB reference in comment.
8137 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8138 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8139 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8140 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8142 (ix86_expand_prologue): Likewise.
8143 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8144 * config/ia64/ia64.h: Likewise.
8145 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8146 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8148 * config/mmix/mmix.h: Likewise.
8149 * config/nds32/nds32.c: Likewise.
8150 * config/stormy/storym16.h: Likewise.
8151 * config/visium/visium.h: Likewise.
8152 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8154 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8156 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8157 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8158 FRAME_POINTER_REGNUM point at high end of local var area.
8160 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8162 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8163 Move comment around. Do not reset best_edge for a copiable
8164 destination if the copy would cause a partition change.
8165 (better_edge_p): Remove redundant check.
8167 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8169 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8171 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8173 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8174 for math functions that have _Float<N> and _Float<N>X variants.
8175 (mathfn_built_in_2): Add support for math functions that have
8176 _Float<N> and _Float<N>X variants.
8177 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8178 (expand_builtin_mathfn_ternary): Add support for fma with
8179 _Float<N> and _Float<N>X variants.
8180 (expand_builtin): Likewise.
8181 (fold_builtin_3): Likewise.
8182 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8183 create math function _Float<N> and _Float<N>X variants as external
8185 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8186 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8187 the __builtin_ prefix and if not strict ansi, without the prefix.
8188 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8189 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8190 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8191 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8192 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8193 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8194 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8195 function signatures for fma _Float<N> and _Float<N>X variants.
8196 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8197 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8198 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8199 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8200 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8201 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8202 * gencfn-macros.c (print_case_cfn): Add support for math functions
8203 that have _Float<N> and _Float<N>X variants.
8204 (print_define_operator_list): Likewise.
8205 (fltfn_suffixes): Likewise.
8207 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8208 for math functions that have _Float<N> and _Float<N>X variants.
8209 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8210 and _Float<N>X variants.
8211 (COPYSIGN): Likewise.
8214 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8215 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8217 (integer_valued_read_call_p): Likewise.
8218 * fold-const-call.c (fold_const_call_ss): Likewise.
8219 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8220 _Float<N> and _Float<N>X variants.
8221 (fold_const_call_ssss): Add support for fma _Float<N> and
8222 _Float<N>X variants.
8223 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8224 support for copysign and fma _Float<N> and _Float<N>X variants.
8225 (backprop::process_builtin_call_use): Likewise.
8226 * tree-call-cdce.c (can_test_argument_range); Add support for
8227 sqrt _Float<N> and _Float<N>X variants.
8228 (edom_only_function): Likewise.
8229 (get_no_error_domain): Likewise.
8230 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8231 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8232 copysign _Float<N> and _Float<N>X variants.
8233 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8234 handled by machine independent code.
8235 (FMAF128): Likewise.
8236 * doc/cpp.texi (Common Predefined Macros): Document defining
8237 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8238 fma _Float<N> and _Float<N>X variants.
8240 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8243 * config/i386/i386-modes.def (CCFPU): Remove definition.
8244 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8245 (ix86_cc_modes_compatible): Ditto.
8246 (ix86_expand_carry_flag_compare): Ditto.
8247 (ix86_expand_int_movcc): Ditto.
8248 (ix86_expand_int_addcc): Ditto.
8249 (ix86_reverse_condition): Ditto.
8250 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8251 Return true/false for unordered/ordered fp comparisons.
8252 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8253 (ix86_prepare_fp_compare_args): Update for rename.
8254 (ix86_expand_fp_compare): Update for rename. Generate unordered
8255 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8256 (ix86_expand_sse_compare_and_jump): Ditto.
8257 * config/i386/predicates.md (fcmov_comparison_operator):
8258 Remove CCFPU mode handling.
8259 (ix86_comparison_operator): Ditto.
8260 (ix86_carry_flag_operator): Ditto.
8261 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8262 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8263 (*cmpu<mode>_cc_i387): Ditto.
8264 (FPCMP): Remove mode iterator.
8265 (unord): Remove mode attribute.
8266 (unord_subst): New define_subst transformation
8267 (unord): New define_subst attribute.
8269 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8270 (*cmpi<unord>xf_i387): Ditto.
8271 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8272 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8273 using unord_subst transformation.
8274 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8275 (round_saeonly): Also handle CCFP mode.
8276 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8277 Remove UNSPEC_SAHF unspec handling.
8279 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8281 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8283 2017-10-27 Jeff Law <law@redhat.com>
8285 * gimple-ssa-sprintf.c: Include domwalk.h.
8286 (class sprintf_dom_walker): New class, derived from dom_walker.
8287 (sprintf_dom_walker::before_dom_children): New function.
8288 (struct call_info): Moved into sprintf_dom_walker class
8289 (compute_formath_length, handle_gimple_call): Likewise.
8290 (sprintf_length::execute): Call the dominator walker rather
8291 than walking the statements.
8293 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8294 gimple statement locations.
8295 (check_array_bounds): Corresponding changes. Get the statement's
8296 location directly from wi->stmt.
8298 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8301 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8303 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8305 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8306 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8308 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8311 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8312 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8313 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8314 maybe_get_pool_constant.
8315 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8318 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8320 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8321 2.26 caveat. Update gas and gld versions.
8322 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8325 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8327 * cgraph.h (set_malloc_flag): Declare.
8328 * cgraph.c (set_malloc_flag_1): New function.
8329 (set_malloc_flag): Likewise.
8330 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8331 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8333 (read_ipa_call_summary): Add support for reading is_return_callee.
8334 (write_ipa_call_summary): Stream is_return_callee.
8335 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8336 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8337 ipa-prop.h, ipa-fnsummary.h.
8338 (pure_const_names): Change to static.
8339 (malloc_state_e): Define.
8340 (malloc_state_names): Define.
8341 (funct_state_d): Add field malloc_state.
8342 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8343 (check_retval_uses): New function.
8344 (malloc_candidate_p): Likewise.
8345 (analyze_function): Add support for malloc attribute.
8346 (pure_const_write_summary): Stream malloc_state.
8347 (pure_const_read_summary): Add support for reading malloc_state.
8348 (dump_malloc_lattice): New function.
8349 (propagate_malloc): New function.
8350 (warn_function_malloc): New function.
8351 (ipa_pure_const::execute): Call propagate_malloc and
8352 ipa_free_fn_summary.
8353 (pass_local_pure_const::execute): Add support for malloc attribute.
8354 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8355 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8357 2017-10-27 Martin Liska <mliska@suse.cz>
8359 PR gcov-profile/82457
8360 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8361 for fork-like functions to be properly instrumented.
8363 2017-10-27 Richard Biener <rguenther@suse.de>
8366 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8367 info when we redirected EH.
8369 2017-10-26 Michael Collison <michael.collison@arm.com>
8371 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8373 (<optab>_trunchf<GPI:mode>2: New pattern.
8374 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8375 * config/aarch64/iterators.md (wv): New mode attribute.
8376 (vf, VF): New mode attributes.
8377 (vgp, VGP): New mode attributes.
8378 (s): Update attribute with SImode and DImode prefixes.
8380 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8382 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8383 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8384 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8385 (nios2_option_overide): Initialize it. Don't allow R0-relative
8386 addressing with PIC.
8387 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8388 (nios2_symbolic_constant_p): Likewise.
8389 (nios2_legitimate_address_p): Likewise.
8390 (nios2_r0rel_section_name_p): New.
8391 (nios2_symbol_ref_in_r0rel_data_p): New.
8392 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8393 (r0rel_constant_p): New.
8394 (nios2_print_operand_address): Handle r0rel_constant_p.
8395 (nios2_cdx_narrow_form_p): Likewise.
8396 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8397 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8398 (Nios II Options): Document -mr0rel-sec.
8400 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8402 * config/nios2/nios2.c: Include xregex.h.
8403 (nios2_gprel_sec_regex): New.
8404 (nios2_option_overide): Initialize it. Don't allow GP-relative
8405 addressing with PIC.
8406 (nios2_small_section_name_p): Check for regex match.
8407 * config/nios2/nios2.opt (mgprel-sec=): New option.
8408 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8409 (Nios II Options): Document -mgprel-sec.
8411 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8413 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8415 2017-10-26 Tom de Vries <tom@codesourcery.com>
8417 PR tree-optimization/82707
8418 * gimple.c (gimple_copy): Fix unsharing of
8419 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8421 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8423 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8424 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8425 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8426 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8427 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8428 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8429 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8430 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8431 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8432 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8433 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8434 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8435 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8436 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8437 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8438 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8439 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8441 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8443 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8445 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8446 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8447 warning to rs6000.c. Remove the Undocumented flag, since it has
8449 (-mabi=ibmlongdouble): Likewise.
8450 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8451 already set, set the default format for long double.
8452 (rs6000_debug_reg_global): Print whether long double is IBM or
8454 (rs6000_option_override_internal): Rework setting long double
8455 format. Only warn if the user is changing the long double default
8456 and they did not use -Wno-psabi.
8457 * doc/invoke.texi (PowerPC options): Update the documentation for
8458 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8460 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8461 Alan Hayward <alan.hayward@arm.com>
8462 David Sherwood <david.sherwood@arm.com>
8464 * rtl.h (wider_subreg_mode): New function.
8465 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8466 rather than an unsigned int *.
8467 * ira-color.c (regno_max_ref_width): Replace with...
8468 (regno_max_ref_mode): ...this new variable.
8469 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8470 Use wider_subreg_mode.
8471 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8472 rather than an unsigned int *.
8473 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8474 (process_alt_operands): Likewise.
8475 (invariant_p): Likewise.
8476 * lra-spills.c (assign_mem_slot): Likewise.
8477 (add_pseudo_to_slot): Likewise.
8478 * lra.c (collect_non_operand_hard_regs): Likewise.
8479 (add_regs_to_insn_regno_info): Likewise.
8480 * reload1.c (regno_max_ref_width): Replace with...
8481 (regno_max_ref_mode): ...this new variable.
8482 (reload): Update accordingly. Update call to
8483 ira_sort_regnos_for_alter_reg.
8484 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8485 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8486 (scan_paradoxical_subregs): Likewise.
8488 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8490 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8491 (aarch64_frame): Add emit_frame_chain boolean.
8492 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8493 Move eh_return case to aarch64_layout_frame.
8494 (aarch64_layout_frame): Initialize emit_frame_chain.
8495 (aarch64_expand_prologue): Use emit_frame_chain.
8497 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8499 * config/aarch64/aarch64.c (aarch64_layout_frame):
8500 Ensure LR is always stored at the bottom of the callee-saves.
8501 Remove rarely used frame layout which saves callee-saves at top of
8502 frame, so the store of LR can be used as a valid probe in all cases.
8504 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8506 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8507 Improve unaligned TImode/TFmode base/offset split.
8509 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8510 Alan Hayward <alan.hayward@arm.com>
8511 David Sherwood <david.sherwood@arm.com>
8513 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8514 * combine.c (find_single_use_1): Likewise.
8515 (expand_field_assignment): Likewise.
8516 (move_deaths): Likewise.
8517 * lra-constraints.c (simplify_operand_subreg): Likewise.
8518 (curr_insn_transform): Likewise.
8519 * lra.c (collect_non_operand_hard_regs): Likewise.
8520 (add_regs_to_insn_regno_info): Likewise.
8521 * rtlanal.c (reg_referenced_p): Likewise.
8522 (covers_regno_no_parallel_p): Likewise.
8524 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8526 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8527 Don't print any bits outside the precision of the value.
8528 * wide-int.cc (test_printing): Add some new tests.
8530 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8532 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8533 supports -xbrace_comment option.
8534 * configure: Regenerate.
8535 * config.in: Regenerate.
8536 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8537 (ASM_CPU_SPEC): Use it.
8539 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8541 * target.def (static_rtx_alignment): New hook.
8542 * targhooks.h (default_static_rtx_alignment): Declare.
8543 * targhooks.c (default_static_rtx_alignment): New function.
8544 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8545 * doc/tm.texi: Regenerate.
8546 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8547 instead of targetm.constant_alignment. Remove call to
8549 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8550 (cris_preferred_mininum_alignment): New function, split out from...
8551 (cris_constant_alignment): ...here.
8552 (cris_static_rtx_alignment): New function.
8553 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8555 (ix86_constant_alignment): ...here.
8556 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8557 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8558 (mmix_static_rtx_alignment): New function.
8559 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8560 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8562 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8565 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8566 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8568 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8570 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8571 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8572 mark as a sign-extending load.
8573 (local_pic_load_u): Define.
8575 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8578 * fold-const.c (operand_equal_for_comparison_p): Also return true
8579 if ARG0 is a simple variant of ARG1 with narrower precision.
8580 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8582 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8584 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8586 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8587 scatter_static, scatter_per_elt.
8588 * x86-tune-costs.h: Add new cost entries.
8590 2017-10-25 Richard Biener <rguenther@suse.de>
8592 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8593 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8594 class pass_fre): Move to ...
8595 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8596 class pass_fre): ... here and adjust for statistics.
8598 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8601 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8602 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8603 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8604 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8607 2017-10-25 Richard Biener <rguenther@suse.de>
8609 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8610 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8611 eliminate_push_avail, eliminate_insert): Move inside...
8612 (class eliminate_dom_walker): ... this class in preparation
8614 (fini_eliminate): Remove by merging with ...
8615 (eliminate): ... this function. Adjust for class changes.
8616 (pass_pre::execute): Remove fini_eliminate call.
8617 (pass_fre::execute): Likewise.
8619 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8622 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8623 (VPERMI2, VPERMI2I): New mode iterators.
8624 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8625 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8627 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8628 mode iterator. Remove 3 old define_insn patterns.
8629 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8630 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8631 VPERMI2 mode iterator, remove the other two expanders.
8632 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8633 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8634 instructions, remove the other two patterns.
8635 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8636 mode iterator, remove the other two patterns.
8637 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8638 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8639 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8640 and adjust argument order accordingly.
8641 (ix86_expand_vec_perm): Adjust caller.
8642 (expand_vec_perm_1): Likewise.
8643 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8644 (expand_vec_perm_vpermt2_vpshub2): ... this.
8645 (ix86_expand_vec_perm_const_1): Adjust caller.
8646 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8649 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8650 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8651 (vec_shl_<mode>): Remove unused expander.
8652 (avx512bw_<shift_insn><mode>3): New define_insn.
8653 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8654 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8656 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8659 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8662 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8664 PR rtl-optimization/82396
8665 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8666 (autopref_multipass_init): Simplify initialization.
8667 (autopref_rank_data): Simplify sort order.
8668 * gcc/sched-int.h (autopref_multipass_data_): Remove
8669 multi_mem_insn_p, min_offset and max_offset.
8671 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8674 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8675 Check special value of flag_omit_frame_pointer.
8676 (aarch64_can_eliminate): Likewise.
8677 (aarch64_override_options_after_change_1): Simplify handling of
8678 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8680 2017-10-24 Richard Biener <rguenther@suse.de>
8682 PR tree-optimization/82697
8683 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8684 zero for conditional load and unconditional store.
8686 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8688 * doc/install.texi: Document bootstrap-cet.
8690 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8693 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8694 ENDBR instruction at function entrance if function is only
8697 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8700 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
8701 patterns to better describe from which operation the CF is computed.
8702 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
8703 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
8704 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
8705 is 0, use _0 suffixed expanders instead of emitting a comparison
8708 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
8710 * config/i386/i386.md(*movsf_internal, *movdf_internal):
8711 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
8713 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
8716 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
8717 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8718 * loop-iv.c (iv_get_reaching_def): Likewise.
8719 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8720 variable is promoted and the partition contains undefined values.
8722 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8724 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8726 (nios2_address_cost): Define.
8727 (nios2_legitimize_address): Recognize (exp + constant) directly.
8728 (TARGET_ADDRESS_COST): Define.
8730 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8732 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8733 (nios2_symbolic_memory_operand_p): Declare.
8734 (nios2_split_large_constant): Declare.
8735 (nios2_split_symbolic_memory_operand): Declare.
8736 * config/nios2/nios2.c: Adjust includes.
8737 (nios2_symbolic_constant_allowed): New.
8738 (nios2_symbolic_constant_p): New.
8739 (nios2_plus_symbolic_constant_p): New.
8740 (nios2_valid_addr_expr_p): Recognize addresses involving
8742 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8743 (nios2_symbolic_memory_operand_p): New.
8744 (nios2_large_constant_p): New.
8745 (nios2_split_large_constant): New.
8746 (nios2_split_plus_large_constant): New.
8747 (nios2_split_symbolic_memory_operand): New.
8748 (nios2_legitimize_address): Code refactoring. Handle addresses
8749 involving symbolic constants.
8750 (nios2_emit_move_sequence): Likewise.
8751 (nios2_print_operand): Improve error output.
8752 (nios2_print_operand_address): Handle LO_SUM.
8753 (nios2_cdx_narrow_form_p): Likewise.
8754 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8755 operands involving symbolic constants.
8756 (movhi_internal, movsi_internal): Likewise.
8757 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8758 (extendhisi2, extendqi<mode>2): Likewise.
8760 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8762 * tree-pass.h (PROP_rtl_split_insns): Define.
8763 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8765 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8767 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8769 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8772 * target.def (const_not_ok_for_debug_p): Default to
8773 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8774 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8775 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8776 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8777 which targetm.const_not_ok_for_debug_p returned true.
8778 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8780 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8782 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8783 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8784 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8786 (ix86_const_not_ok_for_debug_p): New function.
8787 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8788 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8790 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8793 * system.h: Conditionally include "unique-ptr.h" if
8794 INCLUDE_UNIQUE_PTR is defined.
8795 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8796 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8798 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8800 * config/rl78/rl78.md: New define_expand "subdi3".
8802 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8805 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8806 DF_REF_INSN if DF_REF_INSN_INFO is false.
8808 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8810 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8811 xmm_move instead of sse_move.
8812 (sse_store_index): New function.
8813 (ix86_register_move_cost): Be more sensible about mismatch stall;
8814 model AVX moves correctly; make difference between sse->integer and
8816 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8817 moves; make difference between SSE and AVX.
8818 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8819 and zmm_move. Increase size of sse load and store tables;
8820 add unaligned load and store tables; add ssemmx_to_integer.
8821 * x86-tune-costs.h: Update all entries according to real
8822 move latencies from Agner Fog's manual and chip documentation.
8824 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8827 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8828 * config/i386/constraints.md (Wf): New constraint.
8829 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8830 (cmp<dwi>_doubleword): Removed.
8831 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8832 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8833 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8834 expand with cmp<dwi>_doubleword. For LTU and GEU use
8835 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8837 * common.opt (gcolumn-info): Enable by default.
8838 * doc/invoke.texi (gcolumn-info): Document new default.
8840 2017-10-23 Richard Biener <rguenther@suse.de>
8842 PR tree-optimization/82672
8843 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8844 Fold the stmt if we propagated into it.
8846 2017-10-23 Richard Biener <rguenther@suse.de>
8848 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8849 (bitmap_remove_expr_from_set): ... this. All callers call this
8850 for non-constant values.
8851 (bitmap_set_subtract): Rename to...
8852 (bitmap_set_subtract_expressions): ... this. Adjust and
8854 (bitmap_set_contains_value): Remove superfluous check.
8855 (bitmap_set_replace_value): Inline into single caller ...
8856 (bitmap_value_replace_in_set): ... here and simplify.
8857 (dependent_clean): Merge into ...
8858 (clean): ... this using an overload. Adjust.
8859 (prune_clobbered_mems): Adjust.
8860 (compute_antic_aux): Likewise.
8861 (compute_partial_antic_aux): Likewise.
8863 2017-10-23 Richard Biener <rguenther@suse.de>
8865 PR tree-optimization/82129
8867 2017-08-01 Richard Biener <rguenther@suse.de>
8869 PR tree-optimization/81181
8870 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8871 (compute_antic): ... end of iteration here.
8873 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8875 * target.def (starting_frame_offset): New hook.
8876 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8877 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8878 * doc/tm.texi.in: Regenerate.
8879 * hooks.h (hook_hwi_void_0): Declare.
8880 * hooks.c (hook_hwi_void_0): New function.
8881 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8882 STARTING_FRAME_OFFSET.
8883 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8884 * reload1.c (reload): Likewise.
8885 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8886 instead of STARTING_FRAME_OFFSET.
8887 * function.c (try_fit_stack_local): Likewise.
8888 (assign_stack_local_1): Likewise
8889 (instantiate_virtual_regs): Likewise.
8890 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8891 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8892 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8893 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8894 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8895 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8896 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8897 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8898 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8899 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8900 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8901 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8902 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8903 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8904 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8905 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8906 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8907 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8908 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8909 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8910 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8911 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8912 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8913 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8914 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8915 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8916 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8917 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8918 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8919 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8920 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8921 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8922 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8923 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8924 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8925 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8926 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8927 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8928 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8929 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8930 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8931 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8932 return a HOST_WIDE_INT.
8933 (avr_builtin_setjmp_frame_value): Use it instead of
8934 STARTING_FRAME_OFFSET.
8935 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8936 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8937 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8939 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8940 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8941 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8942 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8943 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8944 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8945 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8946 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8947 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8948 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8949 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8950 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8952 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8953 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8954 * config/mips/mips.c (mips_compute_frame_info): Refer to
8955 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8956 (mips_starting_frame_offset): New function.
8957 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8958 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8959 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8960 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8961 and return a HOST_WIDE_INT.
8962 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8963 (mmix_initial_elimination_offset): Refer to
8964 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8965 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8966 * config/pa/pa.c (pa_starting_frame_offset): New function.
8967 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8968 (pa_expand_prologue): Likewise.
8969 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8970 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8971 !FRAME_GROWS_DOWNWARD handling to...
8972 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8973 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8974 !FRAME_GROWS_DOWNWARD handling to...
8975 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8976 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8977 !FRAME_GROWS_DOWNWARD handling to...
8978 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8979 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8981 (rs6000_starting_frame_offset): New function.
8982 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8983 !FRAME_GROWS_DOWNWARD handling to...
8984 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8985 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8986 !FRAME_GROWS_DOWNWARD handling to...
8987 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8988 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8989 !FRAME_GROWS_DOWNWARD handling to...
8990 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8991 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8992 (rs6000_starting_frame_offset): New function.
8993 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8994 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8995 * config/vax/vax.c (vax_starting_frame_offset): New function.
8996 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8997 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8998 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8999 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9000 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9001 * system.h (STARTING_FRAME_OFFSET): Poison.
9003 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9005 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9006 SCALAR_TYPE_MODE instead of TYPE_MODE.
9008 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9009 Alan Hayward <alan.hayward@arm.com>
9010 David Sherwood <david.sherwood@arm.com>
9012 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9014 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9015 Alan Hayward <alan.hayward@arm.com>
9016 David Sherwood <david.sherwood@arm.com>
9018 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9020 2017-10-23 Richard Biener <rguenther@suse.de>
9022 PR tree-optimization/82129
9023 * tree-ssa-pre.c (bitmap_set_and): Remove.
9024 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9025 canonicalizing expressions in the set to those with lowest
9026 ID rather than taking that from the first edge.
9028 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9030 * combine.c (rtx_equal_for_field_assignment_p): Use
9031 byte_lowpart_offset.
9033 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9034 Alan Hayward <alan.hayward@arm.com>
9035 David Sherwood <david.sherwood@arm.com>
9037 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9038 to a SUBREG_PROMOTED_VAR.
9040 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9041 Alan Hayward <alan.hayward@arm.com>
9042 David Sherwood <david.sherwood@arm.com>
9044 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9045 (expand_debug_source_expr): Likewise.
9046 * combine.c (combine_simplify_rtx): Likewise.
9047 * cse.c (fold_rtx): Likewise.
9048 * optabs.c (expand_float): Likewise.
9049 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9050 (simplify_binary_operation_1): Likewise.
9052 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9053 Alan Hayward <alan.hayward@arm.com>
9054 David Sherwood <david.sherwood@arm.com>
9056 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9057 (record_promoted_value): Likewise.
9058 * expr.c (expand_expr_real_2): Likewise.
9059 * ree.c (update_reg_equal_equiv_notes): Likewise.
9060 (combine_set_extension): Likewise.
9061 * rtlanal.c (low_bitmask_len): Likewise.
9062 * simplify-rtx.c (neg_const_int): Likewise.
9063 (simplify_binary_operation_1): Likewise.
9065 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9066 Alan Hayward <alan.hayward@arm.com>
9067 David Sherwood <david.sherwood@arm.com>
9069 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9070 * regcprop.c (maybe_mode_change): Likewise.
9071 * reload1.c (alter_reg): Likewise.
9073 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9075 * inchash.h (inchash::hash::add_wide_int): New function.
9076 * lto-streamer-out.c (hash_tree): Use it.
9078 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9080 * inchash.h (inchash::hash::add_wide_int): Rename to...
9081 (inchash::hash::add_hwi): ...this.
9082 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9083 (polymorphic_call_target_hasher::hash): Likewise.
9084 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9085 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9086 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9087 * lto-streamer-out.c (hash_tree): Likewise.
9088 * optc-save-gen.awk: Likewise.
9089 * tree.c (add_expr): Likewise.
9091 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9094 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9095 for ordered inequality comparisons even with TARGET_IEEE_FP.
9097 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9100 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9101 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9102 Expand with cmp<dwi>_doubleword.
9104 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9106 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9107 List CET intrinsics.
9108 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9109 specific to -fcf-protection option.
9111 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9113 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9114 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9115 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9116 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9117 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9118 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9119 (extra_objs): Add cet.o for Linux/x86 targets.
9120 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9121 * config/i386/cet.c: New file.
9122 * config/i386/cetintrin.h: Likewise.
9123 * config/i386/t-cet: Likewise.
9124 * config/i386/cpuid.h (bit_SHSTK): New.
9125 (bit_IBT): Likewise.
9126 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9127 pass IBT and SHSTK bits.
9128 * config/i386/i386-builtin-types.def
9129 (VOID_FTYPE_UNSIGNED_PVOID): New.
9130 (VOID_FTYPE_UINT64_PVOID): Likewise.
9131 * config/i386/i386-builtin.def: Add CET intrinsics.
9132 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9133 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9134 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9135 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9137 * config/i386/i386.c (rest_of_insert_endbranch): New.
9138 (pass_data_insert_endbranch): Likewise.
9139 (pass_insert_endbranch): Likewise.
9140 (make_pass_insert_endbranch): Likewise.
9141 (ix86_notrack_prefixed_insn_p): Likewise.
9142 (ix86_target_string): Add -mibt, -mshstk flags.
9143 (ix86_option_override_internal): Add flag_cf_protection
9145 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9146 (ix86_print_operand): Add 'notrack' prefix output.
9147 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9148 (ix86_expand_builtin): Expand CET intrinsics.
9149 (x86_output_mi_thunk): Add 'endbranch' instruction.
9150 * config/i386/i386.h (TARGET_IBT): New.
9151 (TARGET_IBT_P): Likewise.
9152 (TARGET_SHSTK): Likewise.
9153 (TARGET_SHSTK_P): Likewise.
9154 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9155 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9156 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9157 (builtin_setjmp_setup): New pattern.
9158 (builtin_longjmp): Likewise.
9159 (rdssp<mode>): Likewise.
9160 (incssp<mode>): Likewise.
9161 (saveprevssp): Likewise.
9162 (rstorssp): Likewise.
9163 (wrss<mode>): Likewise.
9164 (wruss<mode>): Likewise.
9165 (setssbsy): Likewise.
9166 (clrssbsy): Likewise.
9167 (nop_endbr): Likewise.
9168 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9170 * config/i386/immintrin.h: Include <cetintrin.h>.
9171 * config/i386/linux-common.h
9172 (file_end_indicate_exec_stack_and_cet): New prototype.
9173 (TARGET_ASM_FILE_END): New.
9175 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9177 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9178 latencies instead of having separate table; make difference between
9179 integer and float costs.
9180 * i386.h (processor_costs): Remove scalar_stmt_cost,
9181 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9182 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9184 * x86-tune-costs.h: Remove entries which has been removed in
9185 procesor_costs from all tables; make cond_taken_branch_cost
9186 and cond_not_taken_branch_cost COST_N_INSNS based.
9188 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9190 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9192 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9194 * config/i386/i386.md (isa): Remove fma_avx512f.
9195 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9196 <avx512>_fmadd_<mode>_mask3<round_name>,
9197 <avx512>_fmsub_<mode>_mask<round_name>,
9198 <avx512>_fmsub_<mode>_mask3<round_name>,
9199 <avx512>_fnmadd_<mode>_mask<round_name>,
9200 <avx512>_fnmadd_<mode>_mask3<round_name>,
9201 <avx512>_fnmsub_<mode>_mask<round_name>,
9202 <avx512>_fnmsub_<mode>_mask3<round_name>,
9203 <avx512>_fmaddsub_<mode>_mask<round_name>,
9204 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9205 <avx512>_fmsubadd_<mode>_mask<round_name>,
9206 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9207 (*vec_widen_umult_even_v16si<mask_name>,
9208 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9209 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9211 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9213 * extend.texi: Add 'nocf_check' documentation.
9214 * gimple.texi: Add second parameter to
9215 gimple_build_call_from_tree.
9216 * invoke.texi: Add -fcf-protection documentation.
9217 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9219 2017-10-20 Richard Biener <rguenther@suse.de>
9221 PR tree-optimization/82473
9222 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9223 the largest input type.
9225 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9227 * c-attribs.c (handle_nocf_check_attribute): New function.
9228 (c_common_attribute_table): Add 'nocf_check' handling.
9229 * gimple-parser.c: Add second argument NULL to
9230 gimple_build_call_from_tree.
9231 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9232 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9234 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9235 * common.opt: Add fcf-protection flag.
9236 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9237 * flag-types.h: Add enum cf_protection_level.
9238 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9239 Add 'nocf_check' attribute propagation to gimple call.
9240 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9241 (gimple_build_call_from_tree): Update prototype.
9242 (gimple_call_nocf_check_p): New function.
9243 (gimple_call_set_nocf_check): Likewise.
9244 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9245 * ipa-icf.c: Add nocf_check attribute in statement hash.
9246 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9247 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9248 * toplev.c (process_options): Add flag_cf_protection handling.
9250 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9252 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9254 2017-10-20 Richard Biener <rguenther@suse.de>
9256 PR tree-optimization/82603
9257 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9258 remove false predicated stores.
9260 2017-10-20 Richard Biener <rguenther@suse.de>
9262 * graphite-isl-ast-to-gimple.c
9263 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9264 Remove return value and simplify, dump copied stmt after lhs
9266 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9267 Reduce dump verbosity.
9268 (gsi_insert_earliest): Likewise.
9269 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9270 * graphite.c (print_global_statistics): Adjust dumping.
9271 (print_graphite_scop_statistics): Likewise.
9272 (print_graphite_statistics): Do not dump loops here.
9273 (graphite_transform_loops): But here.
9275 2017-10-20 Nicolas Roche <roche@adacore.com>
9277 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9278 * configure: Regenerate.
9280 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9283 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9284 functions when optimizing replace GIMPLE_RETURN stmts with
9285 calls to __builtin_unreachable ().
9288 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9289 for -fsanitize=thread link of executables.
9290 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9291 link of executables.
9294 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9296 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9297 define_insns for logical vector shifts to use VI248_AVX512BW
9298 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9299 condition, useless isa and prefix attributes. Change the first
9300 2 of these define_insns to ...
9301 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9302 define_insn for avx512vl.
9303 (<shift_insn><mode>3): ... and this, new define_insn without
9304 masking for non-avx512vl.
9307 * config/i386/sse.md (*andnot<mode>3,
9308 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9309 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9310 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9311 not applied use empty suffix even for TARGET_AVX512VL.
9312 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9313 is applied, supply evex,evex or evex,evex,evex instead of just
9316 2017-10-20 Julia Koval <julia.koval@intel.com>
9318 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9319 (OPTION_MASK_ISA_GFNI_UNSET): New.
9320 (ix86_handle_option): Handle OPT_mgfni.
9321 * config/i386/cpuid.h (bit_GFNI): New.
9322 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9323 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9324 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9325 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9326 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9327 * config/i386/i386.opt: Add mgfni.
9329 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9331 * config/msp430/msp430.c (msp430_option_override): Disable
9332 -fdelete-null-pointer-checks.
9333 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9335 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9337 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9338 of x87 and SSE instructions.
9340 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9342 * asan.c (create_cond_insert_point): Do not update edge count.
9343 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9344 (afdo_propagate_circuit): Likewise.
9345 (afdo_calculate_branch_prob): Likewise.
9346 (afdo_annotate_cfg): Likewise.
9347 * basic-block.h (struct edge_def): Remove count.
9348 (edge_def::count): New accessor.
9349 * bb-reorder.c (rotate_loop): Update.
9350 (find_traces_1_round): Update.
9351 (connect_traces): Update.
9352 (sanitize_hot_paths): Update.
9353 * cfg.c (unchecked_make_edge): Update.
9354 (make_single_succ_edge): Update.
9355 (check_bb_profile): Update.
9356 (dump_edge_info): Update.
9357 (update_bb_profile_for_threading): Update.
9358 (scale_bbs_frequencies_int): Update.
9359 (scale_bbs_frequencies_gcov_type): Update.
9360 (scale_bbs_frequencies_profile_count): Update.
9361 (scale_bbs_frequencies): Update.
9362 * cfganal.c (connect_infinite_loops_to_exit): Update.
9363 * cfgbuild.c (compute_outgoing_frequencies): Update.
9364 (find_many_sub_basic_blocks): Update.
9365 * cfgcleanup.c (try_forward_edges): Update.
9366 (try_crossjump_to_edge): Update
9367 * cfgexpand.c (expand_gimple_cond): Update
9368 (expand_gimple_tailcall): Update
9369 (construct_exit_block): Update
9370 * cfghooks.c (verify_flow_info): Update
9371 (redirect_edge_succ_nodup): Update
9372 (split_edge): Update
9373 (make_forwarder_block): Update
9374 (duplicate_block): Update
9375 (account_profile_record): Update
9376 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9377 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9378 * cfgloopmanip.c (scale_loop_profile): Update.
9380 (lv_adjust_loop_entry_edge): Update.
9381 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9382 (force_nonfallthru_and_redirect): Update.
9383 (purge_dead_edges): Update.
9384 (rtl_flow_call_edges_add): Update.
9385 * cgraphunit.c (init_lowered_empty_function): Update.
9386 (cgraph_node::expand_thunk): Update.
9387 * gimple-pretty-print.c (dump_probability): Update.
9388 (dump_edge_probability): Update.
9389 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9390 * haifa-sched.c (sched_create_recovery_edges): Update.
9391 * hsa-gen.c (convert_switch_statements): Update.
9392 * ifcvt.c (dead_or_predicable): Update.
9393 * ipa-inline-transform.c (inline_transform): Update.
9394 * ipa-split.c (split_function): Update.
9395 * ipa-utils.c (ipa_merge_profiles): Update.
9396 * loop-doloop.c (add_test): Update.
9397 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9398 * lto-streamer-in.c (input_cfg): Update.
9399 (input_function): Update.
9400 * lto-streamer-out.c (output_cfg): Update.
9401 * modulo-sched.c (sms_schedule): Update.
9402 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9403 * predict.c (maybe_hot_edge_p): Update.
9404 (unlikely_executed_edge_p): Update.
9405 (probably_never_executed_edge_p): Update.
9406 (dump_prediction): Update.
9407 (drop_profile): Update.
9408 (propagate_unlikely_bbs_forward): Update.
9409 (determine_unlikely_bbs): Update.
9410 (force_edge_cold): Update.
9411 * profile.c (compute_branch_probabilities): Update.
9412 * reg-stack.c (better_edge): Update.
9413 * shrink-wrap.c (handle_simple_exit): Update.
9414 * tracer.c (better_p): Update.
9415 * trans-mem.c (expand_transaction): Update.
9416 (split_bb_make_tm_edge): Update.
9417 * tree-call-cdce.c: Update.
9418 * tree-cfg.c (gimple_find_sub_bbs): Update.
9419 (gimple_split_edge): Update.
9420 (gimple_duplicate_sese_region): Update.
9421 (gimple_duplicate_sese_tail): Update.
9422 (gimple_flow_call_edges_add): Update.
9423 (insert_cond_bb): Update.
9424 (execute_fixup_cfg): Update.
9425 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9426 * tree-complex.c (expand_complex_div_wide): Update.
9427 * tree-eh.c (lower_resx): Update.
9428 (unsplit_eh): Update.
9429 (cleanup_empty_eh_move_lp): Update.
9430 * tree-inline.c (copy_edges_for_bb): Update.
9431 (freqs_to_counts): Update.
9432 (copy_cfg_body): Update.
9433 * tree-ssa-dce.c (remove_dead_stmt): Update.
9434 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9435 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9436 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9437 (unloop_loops): Update.
9438 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9439 * tree-ssa-loop-split.c (connect_loops): Update.
9440 (split_loop): Update.
9441 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9442 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9443 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9444 * tree-ssa-reassoc.c (branch_fixup): Update.
9445 * tree-ssa-tail-merge.c (replace_block_by): Update.
9446 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9447 (compute_path_counts): Update.
9448 (update_profile): Update.
9449 (recompute_probabilities): Update.
9450 (update_joiner_offpath_counts): Update.
9451 (estimated_freqs_path): Update.
9452 (freqs_to_counts_path): Update.
9453 (clear_counts_path): Update.
9454 (ssa_fix_duplicate_block_edges): Update.
9455 (duplicate_thread_path): Update.
9456 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9457 (case_bit_test_cmp): Update.
9458 (collect_switch_conv_info): Update.
9459 (gen_inbound_check): Update.
9460 (do_jump_if_equal): Update.
9461 (emit_cmp_and_jump_insns): Update.
9462 * tree-tailcall.c (decrease_profile): Update.
9463 (eliminate_tail_call): Update.
9464 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9465 (vect_do_peeling): Update.
9466 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9467 * ubsan.c (ubsan_expand_null_ifn): Update.
9468 (ubsan_expand_ptr_ifn): Update.
9469 * value-prof.c (gimple_divmod_fixed_value): Update.
9470 (gimple_mod_pow2): Update.
9471 (gimple_mod_subtract): Update.
9472 (gimple_ic): Update.
9473 (gimple_stringop_fixed_value): Update.
9475 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9478 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9480 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9482 PR rtl-optimization/82395
9483 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9484 based on non_spilled_static_chain_regno_p.
9486 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9488 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9489 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9490 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9492 2017-10-19 Martin Sebor <msebor@redhat.com>
9494 PR tree-optimization/82596
9495 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9497 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9499 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9500 (get_mem_refs_of_builtin_call): Likewise.
9501 * builtins.c (expand_builtin_apply): Adjust call to
9502 allocate_dynamic_stack_space.
9503 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9504 the third argument to allocate_dynamic_stack_space, otherwise -1.
9505 (expand_builtin): Deal with all alloca variants.
9506 (is_inexpensive_builtin): Likewise.
9507 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9508 * calls.c (special_function_p): Deal with all alloca variants.
9509 (initialize_argument_information): Adjust call to
9510 allocate_dynamic_stack_space.
9511 (expand_call): Likewise.
9512 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9513 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9514 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9515 use it for the stack usage computation.
9516 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9517 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9518 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9519 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9520 (in_loop_p): Remove first argument and useless check.
9521 (pass_walloca::execute): Remove useless test and adjust call to above.
9522 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9523 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9524 (gimplify_call_expr): Deal with all alloca variants.
9525 * hsa-gen.c (gen_hsa_alloca): Likewise.
9526 (gen_hsa_insns_for_call): Likewise.
9527 * ipa-pure-const.c (special_builtin_state): Likewise.
9528 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9529 * tree-object-size.c (alloc_object_size): Likewise.
9530 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9531 (call_may_clobber_ref_p_1): Likewise.
9532 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9533 (ccp_fold_stmt): Likewise.
9534 (optimize_stack_restore): Likewise.
9535 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9536 (mark_all_reaching_defs_necessary_1): Likewise.
9537 (propagate_necessity): Likewise.
9538 (eliminate_unnecessary_stmts): Likewise.
9539 * tree.c (build_common_builtin_nodes): Build
9540 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9541 (build_alloca_call_expr): New function.
9542 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9543 (CASE_BUILT_IN_ALLOCA): Likewise.
9544 (build_alloca_call_expr): Declare.
9545 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9547 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9550 * dwarf2out.c (new_die_raw): New static inline function.
9551 (new_die): Use it to create the DIE.
9552 (add_AT_external_die_ref): Likewise.
9553 (clone_die): Likewise.
9554 (clone_as_declaration): Likewise.
9555 (dwarf2out_vms_debug_main_pointer): Likewise.
9556 (base_type_die): Likewise. Remove early return for corner cases.
9557 Do not call add_pubtype on the DIE here.
9558 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9559 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9560 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9561 native order exists for base types, attach the DIE manually and call
9562 add_pubtype on it. Do not equate a reverse order DIE to the type.
9564 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9566 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9567 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9569 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9571 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9573 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9574 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9575 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9576 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9577 * opts.c (sanitizer_opts): Add builtin.
9578 * ubsan.c (instrument_builtin): New function.
9579 (pass_ubsan::execute): Call it.
9580 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9581 * doc/invoke.texi: Document -fsanitize=builtin.
9583 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9584 builtins, store max (log2 (align), 0) into uchar field instead of
9585 align into uptr field.
9586 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9587 store uchar 0 field instead of uptr 0 field.
9588 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9589 instead of passing one address of struct with 2 locations pass
9590 two addresses of structs with 1 location each.
9591 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9592 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9593 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9594 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9595 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9596 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9597 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9598 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9600 2017-10-19 Martin Liska <mliska@suse.cz>
9603 * file-find.c (remove_prefix): Remove.
9604 * file-find.h (remove_prefix): Likewise.
9605 * gcc-ar.c: Remove smartness of lookup.
9607 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9609 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9610 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9611 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9613 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9616 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9617 (setcc + and to xor + setcc): New peephole2.
9619 2017-10-19 Tom de Vries <tom@codesourcery.com>
9621 * doc/sourcebuild.texi (Test Directives, Variants of
9622 dg-require-support): Add dg-require-stack-size.
9624 2017-10-19 Martin Liska <mliska@suse.cz>
9627 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9628 that have a large alignment.
9629 (gimplify_target_expr): Likewise.
9631 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9633 PR rtl-optimization/82602
9634 * ira.c (rtx_moveable_p): Return false for volatile asm.
9636 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9639 * config/i386/i386-modes.def (CCGZ): New CC mode.
9640 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9641 * config/i386/predicates.md (ix86_comparison_operator):
9643 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9644 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9645 with double-word subtraction.
9646 (put_condition_code): Handle CCGZmode.
9648 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9650 * wide-int.cc (debug (const wide_int &)): New.
9651 (debug (const wide_int *)): New.
9652 (debug (const widest_int &)): New.
9653 (debug (const widest_int *)): New.
9655 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9658 * lra-constraints.c (curr_insn_transform): Use non-input operand
9659 instead of output one for matched reload.
9661 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9663 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9664 (tree-ssa-loop-ivopts.h): New header file.
9665 (struct builtin_info): New fields.
9666 (classify_builtin_1): Compute and record base and offset parts for
9667 memset builtin partition by calling strip_offset.
9668 (offset_cmp, fuse_memset_builtins): New functions.
9669 (finalize_partitions): Fuse adjacent memset partitions by calling
9671 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9672 Expose the interface.
9673 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9675 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9677 PR tree-optimization/82574
9678 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9679 that data reference must be executed exactly once per iteration
9680 against the outermost loop in nest.
9681 (classify_partition): Update call to above function.
9683 2017-10-18 Richard Biener <rguenther@suse.de>
9685 PR tree-optimization/82591
9686 * graphite.c (graphite_transform_loops): Move code gen message
9688 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9689 Here. Handle scop_to_isl_ast failing.
9690 (scop_to_isl_ast): Limit the number of ISL operations.
9692 2017-10-18 Richard Biener <rguenther@suse.de>
9694 * graphite-isl-ast-to-gimple.c
9695 (translate_isl_ast_to_gimple::set_rename): Simplify.
9696 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
9697 (graphite_copy_stmts_from_block): ... here.
9698 (copy_bb_and_scalar_dependences): Simplify.
9699 (add_parameters_to_ivs_params): Canonicalize.
9700 (generate_entry_out_of_ssa_copies): Simplify.
9701 * graphite-sese-to-poly.c (extract_affine_name): Simplify
9702 by passing in ISL dimension.
9703 (parameter_index_in_region_1): Rename to ...
9704 (parameter_index_in_region): ... this.
9705 (extract_affine): Adjust assert, pass down parameter index.
9706 (add_param_constraints): Use range-info when available.
9707 (build_scop_context): Adjust.
9708 * sese.c (new_sese_info): Adjust.
9709 (free_sese_info): Likewise.
9710 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
9711 Remove unused typedefs.
9712 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
9714 2017-10-18 Martin Liska <mliska@suse.cz>
9716 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9718 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9720 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9721 (s390_sched_init): Do not reset s390_sched_state if we entered the
9722 current basic block via a fallthru edge and all others are unlikely.
9724 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9726 * config/s390/s390.c (NUM_SIDES): New variable.
9727 (LONGRUNNING_THRESHOLD): New variable.
9728 (LATENCY_FACTOR): New variable.
9729 (s390_sched_score): Decrease score for long-running instructions on
9731 (s390_sched_variable_issue): Perform bookkeeping for long-running
9734 2017-10-18 Richard Biener <rguenther@suse.de>
9736 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9737 Simplify with removal of the parameter rename map.
9738 (set_rename): Likewise.
9739 (should_copy_to_new_region): Likewise.
9740 (graphite_copy_stmts_from_block): Likewise.
9741 (copy_bb_and_scalar_dependences): Remove initialization of
9742 unused copied_bb_map.
9744 (copy_internal_parameters): Likewise.
9745 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9746 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9747 Use INTEGRAL_TYPE_P.
9748 (parameter_index_in_region_1): Rename to ...
9749 (assign_parameter_index_in_region): ... this. Assert we have
9750 a parameter we handle.
9751 (scan_tree_for_params): Adjust.
9752 * sese.h (parameter_rename_map_t): Remove.
9753 (struct sese_info_t): Remove unused parameter_rename_map and
9754 copied_bb_map members.
9755 * sese.c (new_sese_info): Adjust.
9756 (free_sese_info): Likewise.
9758 2017-10-18 Martin Liska <mliska@suse.cz>
9761 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9762 on an abnormal edge.
9764 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9766 * doc/invoke.texi (ffunction-sections and fdata-sections):
9769 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9771 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9772 the use statement can throw internally.
9774 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9776 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9777 any RTX present on the RHS of a SET.
9778 * compare-elim.c (try_eliminate_compare): Restore comment.
9780 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9782 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9783 may be also called on tcc_type nodes.
9784 * langhooks.c (lhd_tree_size): Likewise.
9786 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9788 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9789 format_warning_at_substring.
9790 (maybe_warn): Convert source_range * param to a location_t. Pass
9791 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9792 (format_directive): Remove code to extract source_ranges and
9793 source_range * in favor of just a location_t.
9794 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9796 * substring-locations.c (format_warning_va): Convert
9797 source_range * param to a location_t.
9798 (format_warning_at_substring): Likewise.
9799 * substring-locations.h (format_warning_va): Likewise.
9800 (format_warning_at_substring): Likewise.
9802 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9804 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9806 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9807 and scatter/gather ops.
9809 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9810 vec_gather_load and vec_scatter_store.
9811 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9812 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9813 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9814 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9815 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9816 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9818 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9820 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9821 Detect FTST instruction and handle its register pops. Only pop
9822 second operand if can_pop_second_op is true.
9823 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9824 set can_pop_second_op to false in the compare_for_stack_reg call.
9826 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9827 output_fp_compare for stack register operands.
9828 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9829 instructions here. Do not emit stack register pops here. Assert
9830 that FCOMPP pops next to top stack register. Rewrite function.
9832 2017-10-17 Nathan Sidwell <nathan@acm.org>
9835 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9836 use DECL_ASSEMBLER_NAME_RAW.
9839 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9842 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9843 Wilco Dijkstra <wilco.dijkstra@arm.com>
9845 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9846 converstion to Pmode from the buf_addr.
9848 2017-10-17 Richard Biener <rguenther@suse.de>
9850 * graphite-dependences.c (scop_get_reads_and_writes): Change
9851 output parameters to references.
9853 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9855 PR 71026/tree-optimization
9856 * fold-const.c (distribute_real_division): Removed.
9857 (fold_binary_loc): Remove calls to distribute_real_divison.
9859 2017-10-17 Richard Biener <rguenther@suse.de>
9861 * graphite-scop-detection.c
9862 (scop_detection::stmt_has_simple_data_refs_p): Always use
9863 the full nest as region.
9864 (try_generate_gimple_bb): Likewise.
9865 * sese.c (scalar_evolution_in_region): Simplify now that
9866 SCEV can handle instantiation in regions.
9867 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9868 in the non-loop part of a function if requested.
9870 2017-10-17 Richard Biener <rguenther@suse.de>
9872 PR tree-optimization/82563
9873 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9875 (graphite_regenerate_ast_isl): Call it.
9876 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9878 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9880 PR tree-optimization/82549
9881 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9882 Formatting fixes. Instead of calling make_bit_field_ref with negative
9885 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9887 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9888 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9889 _mm_maskz_reduce_ss): New.
9890 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9891 __builtin_ia32_reducess_mask): Ditto..
9892 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9893 * config/i386/sse.md (reduces<mode>): Renamed to ...
9894 (reduces<mode><mask_scalar_name>): ... this.
9895 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9897 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9898 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9900 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9902 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9903 * selftest-run-tests.c (selftest::run_tests): Call
9904 selftest::unique_ptr_tests_cc_tests.
9905 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9906 * unique-ptr-tests.cc: New file.
9908 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9911 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9913 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9914 (make_hard_regno_born, make_hard_regno_dead): Update
9915 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9917 2017-10-16 Jeff Law <law@redhat.com>
9919 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9921 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9923 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9925 2017-10-16 Olivier Hainque <hainque@adacore.com>
9927 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9928 with_cpu if we were configured for an e500v2 target cpu name.
9930 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9932 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9933 * doc/invoke.texi: Document +nodsp as a valid extension for
9936 2017-10-16 Martin Liska <mliska@suse.cz>
9938 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9939 (test_set_range): Likewise.
9940 (test_range_functions): Rename to ...
9941 (test_bit_in_range): ... this.
9942 (sbitmap_c_tests): Add new test.
9944 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9946 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9948 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9949 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9951 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9953 * config/aarch64/aarch64-builtins.c
9954 (aarch64_types_quadopu_lane_qualifiers): New.
9955 (TYPES_QUADOPU_LANE): New.
9956 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9957 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9958 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9959 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9960 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9961 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9962 (Vdottype, DOTPROD): New.
9963 (sur): Add SDOT and UDOT.
9965 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9967 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9968 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9969 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9971 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9972 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9973 Enable TARGET_DOTPROD.
9974 (cortex-a75.cortex-a55): Likewise.
9975 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9977 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9979 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9980 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9982 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9984 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9985 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9986 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9987 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9988 * config/arm/types.md (neon_dot, neon_dot_q): New.
9989 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9991 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9993 * config/arm/arm.h (TARGET_DOTPROD): New.
9994 * config/arm/arm.c (arm_arch_dotprod): New.
9995 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9996 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9997 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9998 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9999 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10000 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10001 * doc/invoke.texi (armv8.2-a): Document dotprod
10003 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10005 * i386.c (ix86_vec_cost): New function.
10006 (ix86_rtx_costs): Handle vector operations better.
10007 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10008 * x86-tune-costs.h: Add new costs to all tables.
10010 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10012 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10014 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10015 divsd, sqrtss and sqrtsd
10016 * x86-tune-costs.h: Add new entries to all costs.
10017 (znver1_cost): Fix to match real instruction latencies.
10019 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10020 Michael Collison <michael.collison@arm.com>
10022 * compare-elim.c: Include emit-rtl.h.
10023 (can_merge_compare_into_arith): New function.
10024 (try_validate_parallel): Likewise.
10025 (try_merge_compare): Likewise.
10026 (try_eliminate_compare): Call the above when no previous clobber
10028 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10031 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10033 PR middle-end/62263
10034 PR middle-end/82498
10035 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10036 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10038 PR middle-end/62263
10039 PR middle-end/82498
10040 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10041 to be any operand_equal_p operands. For & (B - 1) require
10042 B to be power of 2. Recognize
10043 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10045 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10048 * optabs.c (expand_memory_blockage): Fix call of
10049 targetm.have_memory_blockage.
10051 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10054 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10055 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10056 objects to extra_objs instead of overwriting it.
10058 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10060 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10061 Use any_fp_register_operand as operand[3] predicate. Simplify
10062 equality test for operands[2] and operands[4] memory location.
10063 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10064 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10065 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10066 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10067 any_fp_register_operand as operand[1] predicate. Simplify
10068 equality test for operands[0] and operands[3] memory location.
10069 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10070 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10071 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10073 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10075 * target-insns.def: Add memory_blockage.
10076 * optabs.c (expand_memory_blockage): New function.
10077 (expand_asm_memory_barrier): Rename ...
10078 (expand_asm_memory_blockage): ... to this.
10079 (expand_mem_thread_fence): Call expand_memory_blockage
10080 instead of expand_asm_memory_barrier.
10081 (expand_mem_singnal_fence): Ditto.
10082 (expand_atomic_load): Ditto.
10083 (expand_atomic_store): Ditto.
10084 * doc/md.texi (Standard Pattern Names For Generation):
10085 Document memory_blockage instruction pattern.
10087 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10089 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10090 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10091 * config/rl78/rl78.md: New define_expand "adddi3".
10093 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10095 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10098 2017-10-13 Jeff Law <law@redhat.com>
10100 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10102 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10105 * internal-fn.c (expand_mul_overflow): If both operands have
10106 the same highpart of -1 or 0 and the topmost bit of lowpart
10107 is different, overflow is if res <= 0 rather than res < 0.
10109 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10111 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10112 TARGET_P9_VECTOR code for unaligned_load case.
10114 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10116 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10118 2017-10-13 Nathan Sidwell <nathan@acm.org>
10120 * tree-core.h (tree_contains_struct): Make bool.
10121 * tree.c (tree_contains_struct): Likewise.
10122 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10123 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10124 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10125 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10126 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10128 2017-10-13 Richard Biener <rguenther@suse.de>
10130 * graphite-isl-ast-to-gimple.c
10131 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10132 parameters and dominance check.
10133 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10134 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10135 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10136 Do not update SSA form here or do intermediate IL verification.
10137 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10138 (graphite_initialize): Remove check on the number of loops in
10139 the function and inline into graphite_transform_loops.
10140 (graphite_finalize): Inline into graphite_transform_loops.
10141 (graphite_transform_loops): Perform SSA update and IL verification
10143 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10145 2017-10-13 Richard Biener <rguenther@suse.de>
10147 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10148 graphite_expression_type_precision): Avoid global constructor
10150 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10151 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10152 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10153 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10154 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10155 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10157 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10160 * config/i386/i386.h (ix86_red_zone_size): New.
10161 * config/i386/i386.md (push peephole2s): Replace
10162 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10164 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10165 Alan Hayward <alan.hayward@arm.com>
10166 David Sherwood <david.sherwood@arm.com>
10168 * combine.c (can_change_dest_mode): Reject changes in
10169 REGMODE_NATURAL_SIZE.
10171 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10172 Alan Hayward <alan.hayward@arm.com>
10173 David Sherwood <david.sherwood@arm.com>
10175 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10176 (expand_debug_source_expr): Likewise.
10177 * combine.c (combine_simplify_rtx): Likewise.
10178 * cse.c (fold_rtx): Likewise.
10179 * fwprop.c (canonicalize_address): Likewise.
10180 * targhooks.c (default_shift_truncation_mask): Likewise.
10182 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10183 Alan Hayward <alan.hayward@arm.com>
10184 David Sherwood <david.sherwood@arm.com>
10186 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10187 (widened_mode): Likewise.
10188 (expand_unop): Likewise.
10189 * ree.c (transform_ifelse): Likewise.
10190 (merge_def_and_ext): Likewise.
10191 (combine_reaching_defs): Likewise.
10192 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10194 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10195 Alan Hayward <alan.hayward@arm.com>
10196 David Sherwood <david.sherwood@arm.com>
10198 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10199 * combine.c (gen_lowpart_for_combine): Likewise.
10200 * dwarf2out.c (rtl_for_decl_location): Likewise.
10201 * final.c (alter_subreg): Likewise.
10202 * rtlhooks.c (gen_lowpart_general): Likewise.
10203 (gen_lowpart_if_possible): Likewise.
10205 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10206 Alan Hayward <alan.hayward@arm.com>
10207 David Sherwood <david.sherwood@arm.com>
10209 * calls.c (expand_call): Use subreg_lowpart_offset.
10210 * cse.c (cse_insn): Likewise.
10211 * regcprop.c (copy_value): Likewise.
10212 (copyprop_hardreg_forward_1): Likewise.
10214 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10217 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10218 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10219 =Q constraints to +Q and into insn condition add check
10220 that operands[0] and operands[1] are equal.
10221 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10222 =Q constraints to +Q and into insn condition add check
10223 that operands[0] is equal to either operands[1] or operands[2].
10226 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10227 instead of handling MINUS_EXPR twice (once for each argument),
10228 canonicalize operand order and handle just once, use rtype where
10229 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10232 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10233 any values of __C while still being pattern recognizable as a simple
10234 rotate instruction.
10236 2017-10-13 Richard Biener <rguenther@suse.de>
10238 PR tree-optimization/82451
10240 2017-10-02 Richard Biener <rguenther@suse.de>
10242 PR tree-optimization/82355
10243 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10244 a mapping for the enclosing loop but avoid generating one for
10245 the loop tree root.
10246 (copy_bb_and_scalar_dependences): Remove premature codegen
10247 error on PHIs in blocks duplicated into multiple places.
10248 * graphite-scop-detection.c
10249 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10250 in the region use it as loop and nest to analyze the DR in.
10251 (try_generate_gimple_bb): Likewise.
10252 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10253 (add_loop_constraints): For blocks in a loop not in the region
10254 create a dimension with a single iteration.
10255 * sese.h (gbb_loop_at_index): Remove assert.
10257 * cfgloop.c (loop_preheader_edge): For the loop tree root
10258 return the single successor of the entry block.
10259 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10260 Reset the SCEV hashtable and niters.
10261 * graphite-scop-detection.c
10262 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10263 assert that we only have POLYNOMIAL_CHREC that vary in loops
10264 contained in the region.
10265 (scop_detection::graphite_can_represent_expr): Adjust.
10266 (scop_detection::stmt_has_simple_data_refs_p): For loops
10267 not in the region set loop to NULL. The nest is now the
10268 entry edge to the region.
10269 (try_generate_gimple_bb): Likewise.
10270 * sese.c (scalar_evolution_in_region): Adjust for
10271 instantiate_scev change.
10272 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10273 Make nest parameter the edge into the region.
10274 (create_data_ref): Likewise.
10275 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10276 entry edge into a region and adjust instantiate_scev calls.
10277 (create_data_ref): Likewise.
10278 (graphite_find_data_references_in_stmt): Likewise.
10279 (find_data_references_in_stmt): Pass the loop preheader edge
10280 from the nest argument.
10281 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10282 parameter the edge into the region.
10283 (instantiate_parameters): Use the loop preheader edge as entry.
10284 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10286 (get_instantiated_value_entry): Make instantiate_below parameter
10287 the edge into the region.
10288 (instantiate_scev_name): Likewise. Adjust dominance checks,
10289 when we cannot use loop-based instantiation instantiate by
10290 walking use-def chains.
10291 (instantiate_scev_poly): Adjust.
10292 (instantiate_scev_binary): Likewise.
10293 (instantiate_scev_convert): Likewise.
10294 (instantiate_scev_not): Likewise.
10295 (instantiate_array_ref): Remove.
10296 (instantiate_scev_3): Likewise.
10297 (instantiate_scev_2): Likewise.
10298 (instantiate_scev_1): Likewise.
10299 (instantiate_scev_r): Do not blindly handle N-operand trees.
10300 Do not instantiate array-refs. Handle all constants and invariants.
10301 (instantiate_scev): Make instantiate_below parameter
10302 the edge into the region.
10303 (resolve_mixers): Use the loop preheader edge for the region
10304 parameter to instantiate_scev_r.
10305 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10307 2017-10-13 Richard Biener <rguenther@suse.de>
10309 PR tree-optimization/82525
10310 * graphite-isl-ast-to-gimple.c
10311 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10313 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10314 Fail code generation when we cannot represent the isl integer.
10315 (binary_op_to_tree): Elide modulo operations that are no-ops
10316 in the type we code generate. Remove now superfluous code
10319 2017-10-13 Richard Biener <rguenther@suse.de>
10321 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10322 (scop_detection::harmful_loop_in_region): Remove premature
10323 IV type restriction.
10324 (scop_detection::graphite_can_represent_scev): We can handle
10325 pointer IVs just fine.
10327 2017-10-13 Alan Modra <amodra@gmail.com>
10329 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10330 "Clobbers and Scratch Registers". Add paragraph on
10331 alternative to clobbers for scratch registers and OpenBLAS
10334 2017-10-13 Alan Modra <amodra@gmail.com>
10336 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10337 example of a memory input for a string of known length. Move
10338 commentary out of table. Add a number of new examples
10339 covering array memory inputs.
10341 2017-10-12 Martin Liska <mliska@suse.cz>
10343 PR tree-optimization/82493
10344 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10345 (test_range_functions): New function.
10346 (sbitmap_c_tests): Likewise.
10347 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10348 * selftest.h (sbitmap_c_tests): New function.
10350 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10352 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10354 * config/rs6000/amo.h: Fix spacing issue.
10356 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10359 * config/i386/i386.md (*ashl<mode>3_mask_1,
10360 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10361 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10364 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10366 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10367 (profile_probability): Set max_probability
10368 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10370 * profile-count.c (profile_probability::differs_from_p): Do not
10371 rely on max_probaiblity == 10000
10373 2017-10-12 Jeff Law <law@redhat.com>
10375 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10378 2017-10-12 Martin Sebor <msebor@redhat.com>
10382 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10383 (handle_alias_pairs): Call it.
10384 * common.opt (-Wattribute-alias): New option.
10385 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10386 * doc/invoke.texi (-Wattribute-alias): Document.
10388 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10391 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10393 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10395 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10396 (make_hard_regno_born, make_hard_regno_dead): Update
10397 bb_killed_pseudos and bb_gen_pseudos.
10399 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10401 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10403 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10405 * config/alpha/alpha.c (alpha_split_conditional_move):
10406 Use std::swap instead of manually swapping.
10407 (alpha_stdarg_optimize_hook): Ditto.
10408 (alpha_canonicalize_comparison): Ditto.
10410 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10412 * tree-loop-distribution.c (struct builtin_info): New struct.
10413 (struct partition): Refactor fields into struct builtin_info.
10414 (partition_free): Free struct builtin_info.
10415 (build_size_arg_loc, build_addr_arg_loc): Delete.
10416 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10417 information from struct builtin_info.
10418 (find_single_drs): New function refactored from classify_partition.
10419 Also moved builtin validity checks to this function.
10420 (compute_access_range, alloc_builtin): New functions.
10421 (classify_builtin_st, classify_builtin_ldst): New functions.
10422 (classify_partition): Refactor code into functions find_single_drs,
10423 classify_builtin_st and classify_builtin_ldst.
10424 (distribute_loop): Don't do runtime alias check when distributing
10426 (find_seed_stmts_for_distribution): New function.
10427 (pass_loop_distribution::execute): Refactor code finding seed
10428 stmts into above function. Support distribution for the innermost
10429 two-level loop nest. Adjust dump information.
10431 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10433 * tree-loop-distribution.c: Adjust the general comment.
10434 (NUM_PARTITION_THRESHOLD): New macro.
10435 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10436 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10437 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10438 in call to build_partition_graph.
10439 (finalize_partitions): New parameter. Make loop distribution more
10440 conservative by fusing more partitions.
10441 (distribute_loop): Don't do runtime alias check in case of loop nest
10443 (find_seed_stmts_for_distribution): New function.
10444 (prepare_perfect_loop_nest): New function.
10445 (pass_loop_distribution::execute): Refactor code finding seed stmts
10446 and loop nest into above functions. Support loop nest distribution.
10447 Adjust dump information accordingly.
10449 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10451 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10452 and set PTYPE_SEQUENTIAL for merged partition.
10454 2017-10-12 Richard Biener <rguenther@suse.de>
10456 PR tree-optimization/69728
10458 2017-09-19 Richard Biener <rguenther@suse.de>
10460 PR tree-optimization/69728
10461 * graphite-sese-to-poly.c (schedule_error): New global.
10462 (add_loop_schedule): Handle empty domain by failing the
10464 (build_original_schedule): Handle schedule_error.
10466 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10467 domain by returning an unchanged schedule.
10469 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10471 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10472 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10474 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10476 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10479 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10482 * expr.c (store_field): Don't optimize away bitsize == 0 store
10483 from CALL_EXPR with addressable return type.
10485 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10487 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10488 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10489 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10490 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10491 TARGET_ISEL instead of TARGET_ISEL<sel>.
10493 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10495 * config/rs6000/rs6000.c
10496 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10498 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10500 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10502 (reg_or_cint_operand): Fix comment.
10503 (reg_or_zero_operand): New predicate.
10504 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10505 * config/rs6000/rs6000.c (output_isel): Delete.
10506 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10507 instead of reg_or_cint_operand. Output instruction directly (not via
10509 (isel_unsigned_<mode>): Ditto.
10510 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10511 gpc_reg_operand. Add an instruction alternative for this. Output
10512 instruction directly.
10513 (*isel_reversed_unsigned_<mode>): Ditto.
10515 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10517 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10518 (TARGET_CANONICALIZE_COMPARISON): Define.
10520 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10523 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10524 Check whether the dest is REG before adding REG_EQUIV note.
10526 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10529 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10531 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10532 (make_hard_regno_born, make_hard_regno_dead): Update
10533 bb_killed_pseudos and bb_gen_pseudos.
10535 2017-10-11 Nathan Sidwell <nathan@acm.org>
10537 * incpath.h (enum incpath_kind): Name enum, prefix values.
10538 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10539 * incpath.c (heads, tails): Use INC_MAX.
10540 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10541 (merge_include_chains, split_quote_chain,
10542 register_include_chains): Update incpath_kind names.
10543 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10544 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10546 (add_framework_path, darwin_register_objc_includes): Likewise.
10547 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10549 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10551 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10552 Do not use float_operator operator predicate.
10553 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10554 * config/i386/predicates.md (float_operator): Remove predicate.
10556 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10558 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10559 (*jccxf_i387): Ditto.
10560 (*jcc<mode>_i387): Ditto.
10561 (*jccu<mode>_i387): Ditto.
10562 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10563 (*jcc_*_i387 splitters): Remove.
10564 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10565 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10566 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10569 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10571 * profile-count.h (slow_safe_scale_64bit): New function.
10572 (safe_scale_64bit): New inline.
10573 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10574 * profile-count.c: Include wide-int.h
10575 (slow_safe_scale_64bit): New.
10577 2017-10-11 Nathan Sidwell <nathan@acm.org>
10579 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10580 HAS_DECL_ASSEMBLER_NAME_P.
10581 * gimple-expr.c (gimple_decl_printable_name: Check
10582 HAS_DECL_ASSEMBLER_NAME_P too.
10583 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10584 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10585 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10586 * passes.c (rest_of_decl_compilation): Check
10587 HAS_DECL_ASSEMBLER_NAME_P too.
10588 * recog.c (verify_changes): Likewise.
10589 * tree-pretty-print.c (dump_decl_name): Likewise.
10590 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10592 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10593 (SET_DECL_ASSEMBLER_NAME): Use it.
10594 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10595 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10596 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10598 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10600 * config.gcc (i386, x86_64): Add extra objects.
10601 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10602 (ix86_min_insn_size): Declare.
10603 (ix86_issue_rate): Declare.
10604 (ix86_adjust_cost): Declare.
10605 (ia32_multipass_dfa_lookahead): Declare.
10606 (ix86_macro_fusion_p): Declare.
10607 (ix86_macro_fusion_pair_p): Declare.
10608 (ix86_bd_has_dispatch): Declare.
10609 (ix86_bd_do_dispatch): Declare.
10610 (ix86_core2i7_init_hooks): Declare.
10611 (ix86_atom_sched_reorder): Declare.
10612 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10613 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10614 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10615 (rip_relative_addr_p): Rename to ...
10616 (ix86_rip_relative_addr_p): ... this one; export.
10617 (memory_address_length): Update.
10618 (ix86_issue_rate): Move to x86-tune-sched.c.
10619 (ix86_flags_dependent): Move to x86-tune-sched.c.
10620 (ix86_agi_dependent): Move to x86-tune-sched.c.
10621 (exact_dependency_1): Move to x86-tune-sched.c.
10622 (exact_store_load_dependency): Move to x86-tune-sched.c.
10623 (ix86_adjust_cost): Move to x86-tune-sched.c.
10624 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10625 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10626 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10627 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10628 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10629 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10630 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10631 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10632 (min_insn_size): Rename to ...
10633 (ix86_min_insn_size): ... this one; export.
10634 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10635 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10636 (core2i7_first_cycle_multipass_backtrack): Move to
10637 x86-tune-sched-core.c.
10638 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10639 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10640 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10641 (ix86_avoid_jump_mispredicts): Update.
10642 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10643 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10644 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10645 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10646 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10647 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10648 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10649 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10650 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10651 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10652 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10653 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10654 (BIG): Move to ix86-tune-sched-bd.c.
10655 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10656 (enum insn_path): Move to ix86-tune-sched-bd.c.
10657 (get_mem_group): Move to ix86-tune-sched-bd.c.
10658 (is_cmp): Move to ix86-tune-sched-bd.c.
10659 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10660 (is_branch): Move to ix86-tune-sched-bd.c.
10661 (is_prefetch): Move to ix86-tune-sched-bd.c.
10662 (init_window): Move to ix86-tune-sched-bd.c.
10663 (allocate_window): Move to ix86-tune-sched-bd.c.
10664 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10665 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10666 (process_end_window): Move to ix86-tune-sched-bd.c.
10667 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10668 (find_constant): Move to ix86-tune-sched-bd.c.
10669 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10670 (has_immediate): Move to ix86-tune-sched-bd.c.
10671 (get_insn_path): Move to ix86-tune-sched-bd.c.
10672 (get_insn_group): Move to ix86-tune-sched-bd.c.
10673 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10674 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10675 (add_insn_window): Move to ix86-tune-sched-bd.c.
10676 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10677 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10678 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10679 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10680 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10681 (do_dispatch): Move to ix86-tune-sched-bd.c.
10682 (has_dispatch): Move to ix86-tune-sched-bd.c.
10683 * config/i386/t-i386: Add new object files.
10684 * config/i386/x86-tune-costs.h: New file.
10685 * config/i386/x86-tune-sched-atom.c: New file.
10686 * config/i386/x86-tune-sched-bd.c: New file.
10687 * config/i386/x86-tune-sched-core.c: New file.
10688 * config/i386/x86-tune-sched.c: New file.
10690 2017-10-11 Liu Hao <lh_mouse@126.com>
10692 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10693 the generic version below instead.
10694 (should_colorize): Recognize Windows consoles as terminals
10696 * pretty-print.c [__MINGW32__] (write_all): New function.
10697 [__MINGW32__] (find_esc_head): Likewise.
10698 [__MINGW32__] (find_esc_terminator): Likewise.
10699 [__MINGW32__] (eat_esc_sequence): Likewise.
10700 [__MINGW32__] (mingw_ansi_fputs): New function that handles
10702 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
10705 2017-10-11 Richard Biener <rguenther@suse.de>
10707 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
10708 Properly call analyze_scalar_evolution with the loop of the stmt.
10710 2017-10-11 Richard Biener <rguenther@suse.de>
10712 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
10713 * tree-core.h (tree_base): Add chrec_var union member.
10714 * tree.h (CHREC_VAR): Remove.
10715 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
10716 * tree-chrec.h (build_polynomial_chrec): Adjust.
10717 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10718 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10720 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10722 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10723 * match.pd: ... here.
10724 ((T) X == (T) Y): Relax condition.
10726 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10728 PR tree-optimization/82472
10729 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10731 (break_alias_scc_partitions): Update postorder number.
10733 2017-10-11 Martin Liska <mliska@suse.cz>
10736 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10738 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10740 2017-10-11 Martin Liska <mliska@suse.cz>
10742 * print-rtl.c (print_insn): Move declaration of idbuf
10743 to same scope as name.
10745 2017-10-11 Martin Liska <mliska@suse.cz>
10750 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10751 volatile arguments.
10753 2017-10-11 Martin Liska <mliska@suse.cz>
10756 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10757 volatile arguments.
10759 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10761 * config.gcc (default_gnu_indirect_function): Default to yes for
10762 arm*-*-linux* with glibc.
10764 2017-10-11 Richard Biener <rguenther@suse.de>
10766 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10767 default-defs and types we do not want to analyze.
10768 (interpret_loop_phi): Replace unreachable code with an assert.
10769 (compute_scalar_evolution_in_loop): Remove and inline ...
10770 (analyze_scalar_evolution_1): ... here, replacing condition with
10771 what makes the intent clearer. Remove handling of cases
10772 get_scalar_evolution now handles.
10774 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10776 PR rtl-optimization/81434
10777 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10778 comment for main loop. In sched_group_found if, also add checks for
10779 pass and min_cost_group.
10781 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10783 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10784 (rs6000_insn_cost): New function.
10785 * config/rs6000/rs6000.md (cost): New attribute.
10787 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10788 H.J. Lu <hongjiu.lu@intel.com>
10792 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10793 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10794 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10795 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10796 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10797 OPTION_MASK_ISA_MMX and some other bit set require both
10798 mmx and the ISAs without the mmx bit.
10799 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10800 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10801 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10802 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10803 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10804 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10805 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10806 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10807 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10808 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10809 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10810 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10811 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10812 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10813 Add OPTION_MASK_ISA_MMX.
10815 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10817 * config.gcc (armv7*-*-freebsd*): New target.
10818 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10820 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10822 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10823 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10826 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10828 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10829 a template parameter.
10830 (WIDE_INT_REF_FOR): Update accordingly.
10831 * tree.h (wi::int_traits <const_tree>): Delete.
10832 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10833 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10834 (wi::tree_to_wide_ref): New typedef.
10835 (wi::to_wide): New function.
10836 * calls.c (get_size_range): Use wi::to_wide when operating on
10837 trees as wide_ints.
10838 * cgraph.c (cgraph_node::create_thunk): Likewise.
10839 * config/i386/i386.c (ix86_data_alignment): Likewise.
10840 (ix86_local_alignment): Likewise.
10841 * dbxout.c (stabstr_O): Likewise.
10842 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10843 * expr.c (const_vector_from_tree): Likewise.
10844 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10845 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10846 (fold_negate_expr_1, int_const_binop_1, const_binop)
10847 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10848 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10849 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10850 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10851 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10852 (fold_not_const, round_up_loc): Likewise.
10853 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10854 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10855 (alloca_call_type): Likewise.
10856 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10857 * godump.c (go_output_typedef): Likewise.
10858 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10859 * internal-fn.c (get_min_precision): Likewise.
10860 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10861 * ipa-polymorphic-call.c
10862 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10863 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10864 (ipa_modify_call_arguments): Likewise.
10865 * match.pd: Likewise.
10866 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10867 * print-tree.c (print_node_brief, print_node): Likewise.
10868 * stmt.c (expand_case): Likewise.
10869 * stor-layout.c (layout_type): Likewise.
10870 * tree-affine.c (tree_to_aff_combination): Likewise.
10871 * tree-cfg.c (group_case_labels_stmt): Likewise.
10872 * tree-data-ref.c (dr_analyze_indices): Likewise.
10873 (prune_runtime_alias_test_list): Likewise.
10874 * tree-dump.c (dequeue_and_dump): Likewise.
10875 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10876 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10877 * tree-pretty-print.c (dump_generic_node): Likewise.
10878 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10879 (simple_iv_with_niters): Likewise.
10880 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10881 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10882 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10883 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10884 (refine_value_range_using_guard, number_of_iterations_ne_max)
10885 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10886 (get_cst_init_from_scev, record_nonwrapping_iv)
10887 (scev_var_range_cant_overflow): Likewise.
10888 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10889 * tree-ssa-pre.c (compute_avail): Likewise.
10890 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10891 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10892 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10893 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10894 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10895 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10896 (dump_case_nodes, try_switch_expansion): Likewise.
10897 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10898 (vect_do_peeling): Likewise.
10899 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10900 * tree-vect-stmts.c (vectorizable_load): Likewise.
10901 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10902 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10903 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10904 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10905 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10906 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10907 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10908 (evrp_dom_walker::before_dom_children): Likewise.
10909 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10910 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10911 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10912 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10913 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10914 (get_range_pos_neg): Likewise.
10915 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10916 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10917 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10918 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10919 * config/avr/avr.c (avr_fold_builtin): Likewise.
10920 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10921 * config/msp430/msp430.c (msp430_attr): Likewise.
10922 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10923 * config/powerpcspe/powerpcspe-c.c
10924 (altivec_resolve_overloaded_builtin): Likewise.
10925 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10926 (rs6000_expand_ternop_builtin): Likewise.
10927 * config/rs6000/rs6000-c.c
10928 (altivec_resolve_overloaded_builtin): Likewise.
10929 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10930 (rs6000_expand_ternop_builtin): Likewise.
10931 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10933 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10935 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10936 when copying loop nest with only one inner loop.
10938 2017-10-10 Richard Biener <rguenther@suse.de>
10940 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10941 blocks if SCEV is active.
10942 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10944 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10945 (scev_initialize): Assert we are not yet initialized.
10947 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10949 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10950 inner loop's exit stmt by making it always exit the loop, otherwise
10951 we would generate an infinite empty loop.
10953 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10955 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10956 renaming variables in new preheader if it's deleted.
10958 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10960 * tree-loop-distribution.c (struct partition): Remove unused field
10961 loops of the structure.
10962 (partition_alloc, partition_free): Ditto.
10963 (build_rdg_partition_for_vertex): Ditto.
10965 2017-10-09 Jeff Law <law@redhat.com>
10967 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10968 return type to match prototype and documentation.
10970 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10972 * config/rs6000/rs6000.c (processor_costs): Move to ...
10973 * config/rs6000/rs6000.h: ... here.
10974 (rs6000_cost): Declare.
10976 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10978 * except.c (setjmp_fn): New global variable.
10979 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10980 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10981 if DONT_USE_BUILTIN_SETJMP is defined.
10983 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10985 * target.def (insn_cost): New hook.
10986 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10987 * doc/tm.texi: Regenerate.
10988 * rtlanal.c (insn_cost): Use the new hook.
10990 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10992 * combine.c (combine_validate_cost): Compute the new insn_cost,
10993 not just pattern_cost.
10994 (try_combine): Adjust comment.
10996 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10998 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11000 * combine.c (uid_insn_cost): Adjust comment.
11001 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11003 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11004 * dse.c (find_shift_sequence): Ditto.
11005 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11006 (bb_valid_for_noce_process_p): Use pattern_cost.
11007 * rtl.h (insn_rtx_cost): Delete.
11008 (pattern_cost): New prototype.
11009 (insn_cost): New prototype.
11010 * rtlanal.c (insn_rtx_cost): Rename to...
11011 (pattern_cost): ... this.
11014 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11016 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11017 (*jcc<mode>_0_r_i387): Ditto.
11018 (*jccxf_r_i387): Ditto.
11019 (*jcc<mode>_r_i387): Ditto.
11020 (*jccu<mode>_r_i387): Ditto.
11021 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11022 (*jcc): Rename from *jcc_1.
11024 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11026 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11027 deferred rescans after the lvx/stvx recombination pre-pass.
11029 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11031 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11032 memory operation instruction support.
11033 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11034 (rs6000-ibm-aix[789]*): Likewise.
11035 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11036 Document new functions.
11038 2017-10-09 Richard Biener <rguenther@suse.de>
11040 PR tree-optimization/82397
11041 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11042 equality only for semantically equal trees.
11044 2017-10-09 Richard Biener <rguenther@suse.de>
11046 PR tree-optimization/82449
11047 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11048 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11049 allow constant addresses.
11050 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11053 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11055 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11058 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11061 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11064 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11067 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11069 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11072 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11073 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11075 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11077 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11078 (WI_BINARY_PREDICATE_RESULT): Likewise.
11079 (wi::binary_traits::operator_result): New type.
11080 (wi::binary_traits::predicate_result): Likewise.
11081 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11082 (generic_wide_int::operator==, generic_wide_int::operator!=)
11083 (generic_wide_int::operator&, generic_wide_int::and_not)
11084 (generic_wide_int::operator|, generic_wide_int::or_not)
11085 (generic_wide_int::operator^, generic_wide_int::operator+
11086 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11088 (operator~, unary operator-, operator==, operator!=, operator&)
11089 (operator|, operator^, operator+, binary operator-, operator*): New
11091 * expr.c (get_inner_reference): Use wi::bit_and_not.
11092 * fold-const.c (fold_binary_loc): Likewise.
11093 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11094 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11095 (bit_value_binop): Likewise.
11096 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11097 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11098 (extract_range_from_binary_expr_1): Likewise.
11099 (masked_increment): Likewise.
11100 (simplify_bit_ops_using_ranges): Likewise.
11102 2017-10-09 Martin Jambor <mjambor@suse.cz>
11105 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11106 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11107 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11109 (hsa_fixup_mov_insn_type): New function.
11110 (hsa_op_with_type::get_in_type): Use it.
11111 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11112 immediates in an assert.
11113 (hsa_op_with_type::extend_int_to_32bit): New method.
11114 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11115 types. Convert to dest type if necessary.
11116 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11117 (reg_for_gimple_ssa): Pass false as min32int to
11118 hsa_type_for_scalar_tree_type.
11119 (gen_hsa_addr): Fixup type when creating addresable temporary.
11120 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11121 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11122 necessary. Call hsa_fixup_mov_insn_type.
11123 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11124 extend operands and convert to dest type if necessary.
11125 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11126 to dest type if necessary.
11127 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11128 if conversion nt necessary and size matches.
11129 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11130 to dest type if necessary.
11131 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11132 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11134 (gen_hsa_clrsb): Likewise.
11135 (gen_hsa_ffs): Likewise.
11136 (gen_hsa_divmod): Extend operands and convert to dest type if
11138 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11140 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11142 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11143 Remove empty default arguments. Use a brace block as output
11145 (conditional return): Ditto.
11147 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11148 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11150 (group_ending_nop): Ditto.
11151 (doloop_end): Ditto.
11152 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11153 (splitters for those): Ditto.
11155 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11157 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11158 a conditional jump (and the compare for it) so that pc_rtx is the
11160 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11161 for the deleted and renamed ctr<mode>_internal[234] patterns.
11162 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11163 Delete second conditional return pattern.
11164 (ctr<mode>_internal2): Delete this second bdnz pattern.
11165 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11166 (ctr<mode>_internal4): Delete this second bdz pattern.
11168 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11170 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11171 (always_initialized_rtx_for_ssa_name_p): New predicate.
11172 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11173 (finish_out_of_ssa): Free new field of SA.
11174 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11175 * tree-ssa-coalesce.c: Include tree-ssa.h.
11176 (get_parm_default_def_partitions): Remove extern keyword.
11177 (get_undefined_value_partitions): New function.
11178 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11179 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11180 uninitialized bits.
11181 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11183 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11185 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11187 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11189 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11190 for targets that preffer 128bit.
11192 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11194 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11196 2017-10-08 Olivier Hainque <hainque@adacore.com>
11198 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11199 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11201 (thumb_set_return_address): Likewise.
11203 2017-10-08 Olivier Hainque <hainque@adacore.com>
11205 * common/config/arm/arm-common.c (arm_except_unwind_info):
11206 Handle DWARF2_UNWIND_INFO.
11208 2017-10-07 Michael Collison <michael.collison@arm.com>
11210 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11213 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11215 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11216 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11217 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11218 defined, force the creation of a new block for a dispatch label.
11220 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11222 * invoke.texi (Wsuggest-attribute=cold): Document.
11223 * common.opt (Wsuggest-attribute=cold): New
11224 * ipa-pure-const.c (warn_function_cold): New function.
11225 * predict.c (compute_function_frequency): Use it.
11226 * predict.h (warn_function_cold): Declare.
11228 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11230 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11233 2017-10-06 Martin Liska <mliska@suse.cz>
11235 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11236 keyword for member functions.
11237 (struct sanopt_tree_couple): New struct.
11238 (struct sanopt_tree_couple_hash): New function.
11239 (struct sanopt_ctx): Add new hash_map.
11240 (has_dominating_ubsan_ptr_check): New function.
11241 (record_ubsan_ptr_check_stmt): Likewise.
11242 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11243 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11244 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11246 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11249 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11250 aarch64_simd_valid_immediate on CONST_VECTORs.
11251 (aarch64_reg_or_bic_imm): Likewise.
11253 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11255 PR rtl-optimization/82396
11256 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11258 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11260 * graphite-dependences.c (scop_get_reads): Move code to...
11261 (scop_get_must_writes): Move code to...
11262 (scop_get_may_writes): Move code to...
11263 (scop_get_reads_and_writes): ... here.
11264 (scop_get_dependences): Call scop_get_reads_and_writes.
11266 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11268 PR tree-optimization/82434
11269 * fold-const.h (can_native_encode_type_p,
11270 can_native_encode_string_p): Remove.
11271 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11272 don't encode anything, just return what would be otherwise returned.
11273 (native_encode_fixed, native_encode_complex, native_encode_vector):
11275 (native_encode_string): Likewise. Inline by hand
11276 can_native_encode_string_p.
11277 (can_native_encode_type_p): Remove.
11278 (can_native_encode_string_p): Remove.
11279 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11280 STRING_CSTs using can_native_encode_string_p, test all
11281 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11282 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11283 argument from native_encode_expr.
11284 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11285 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11288 2017-10-06 Richard Biener <rguenther@suse.de>
11290 PR tree-optimization/82397
11291 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11292 operand_equal_p but rely on data_ref_compare_tree for detecting
11294 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11295 to match up with dr_group_sort_cmp.
11297 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11300 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11302 * config/s390/s390-builtin-types.def: Regenerate.
11304 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11307 * config/s390/s390-builtin-types.def: Regenerate.
11308 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11309 Change flag from B_VXE to B_VX.
11310 (s390_vec_min_dbl): Remove B_VXE flag.
11312 2017-10-06 Richard Biener <rguenther@suse.de>
11314 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11315 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11316 translate_isl_ast_to_gimple::is_valid_rename,
11317 translate_isl_ast_to_gimple::get_rename,
11318 translate_isl_ast_to_gimple::get_def_bb_for_const,
11319 translate_isl_ast_to_gimple::get_new_name,
11320 translate_isl_ast_to_gimple::collect_all_ssa_names,
11321 translate_isl_ast_to_gimple::copy_loop_phi_args,
11322 translate_isl_ast_to_gimple::collect_all_ssa_names,
11323 translate_isl_ast_to_gimple::copy_loop_phi_args,
11324 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11325 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11326 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11327 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11328 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11329 translate_isl_ast_to_gimple::copy_cond_phi_args,
11330 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11331 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11332 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11333 translate_isl_ast_to_gimple::rename_uses,
11334 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11335 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11336 (set_rename_for_each_def): Likewise.
11337 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11338 here. Handle rewriting SCEV analyzable uses here.
11339 (copy_bb_and_scalar_dependences): Generate code for PHI
11341 (graphite_regenerate_ast_isl): Adjust.
11342 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11343 (add_write, add_read): New functions.
11344 (build_cross_bb_scalars_def): Use it and simplify.
11345 (build_cross_bb_scalars_use): Likewise.
11346 (graphite_find_cross_bb_scalar_vars): Inline into...
11347 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11348 simulating out-of-SSA. Compute liveout and add dependencies.
11349 (build_scops): Force an empty entry block.
11350 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11352 (sese_build_liveouts): Declare.
11353 (sese_trivially_empty_bb_p): Likewise.
11354 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11355 compute liveout and debug_liveout.
11356 (sese_bad_liveouts_use): Remove.
11357 (sese_reset_debug_liveouts_bb): Likewise.
11358 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11359 (sese_build_liveouts): Build liveout and debug_liveout and store
11361 (new_sese_info): Adjust.
11362 (free_sese_info): Likewise.
11363 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11364 do not build liveout here.
11365 (move_sese_in_condition): Adjust region entry.
11366 (scev_analyzable_p): Match up with chrec_apply requirements.
11367 (sese_trivially_empty_bb_p): New.
11368 * tree-into-ssa.c (get_reaching_def): Properly support generating
11369 default-defs for incremental rewrite of anonymous names.
11371 2017-10-06 Richard Biener <rguenther@suse.de>
11373 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11374 precision do not perform modulo reduction.
11376 2017-10-06 Richard Biener <rguenther@suse.de>
11378 PR tree-optimization/82436
11379 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11380 conservatively choose the vectorization factor when checking
11381 whether we can perform the required load permutation.
11382 (vect_transform_slp_perm_load): Assert when we may not fail.
11384 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11386 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11387 message for incompatible -msdata=* and -mcall-* options.
11389 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11391 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11392 rate for post-reload scheduling.
11394 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11396 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11398 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11400 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11401 to improve monte carlo in scimark.
11403 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11405 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11406 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11407 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11408 pentium4_cost, nocona_cost): Set reassociation width to 1.
11409 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11410 width to 2 for fp operations and 1 otherwise.
11411 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11413 (atom_cost): Set reassociation width to 2.
11414 (slm_cost, generic_cost): Set fp reassociation width
11415 to 2 and 1 otherwise.
11416 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11417 (core_cost): Set fp reassociation width to 4 and vector to 2.
11418 (ix86_reassociation_width): Rewrite using cost table; special case
11419 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11420 and TARGET_AVX128_OPTIMAL.
11421 * config/i386/i386.h (processor_costs): Add
11422 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11423 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11424 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11425 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11426 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11427 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11429 2017-10-05 Nathan Sidwell <nathan@acm.org>
11431 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11433 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11435 * config/arm/arm.c (arm_test_fpu_data): New.
11436 (arm_run_selftests): Call arm_test_fpu_data.
11438 2017-10-04 Nathan Sidwell <nathan@acm.org>
11440 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11442 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11444 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11446 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11447 check type for aarch64_simd_valid_immediate.
11448 (aarch64_output_simd_mov_immediate): Update prototype.
11449 (aarch64_simd_valid_immediate): Update prototype.
11450 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11451 support for ORR-immediate.
11452 (and<mode>3): modified pattern to add support for BIC-immediate.
11453 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11454 now checks for valid immediate for BIC and ORR based on new enum
11456 (aarch64_output_simd_mov_immediate): Function now used to output
11457 BIC/ORR imm as well based on new enum argument.
11458 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11460 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11461 (aarch64_reg_or_bic_imm): Likewise.
11463 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11465 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11466 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11468 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11472 PR rtl-optimization/82396
11473 * haifa-sched.c (autopref_multipass_init): Simplify
11475 (autopref_rank_data): Simplify sort order.
11476 * sched-int.h (autopref_multipass_data_): Remove
11477 multi_mem_insn_p, min_offset and max_offset.
11479 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11481 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11483 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11485 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11486 vect_doubleint_cvt.
11488 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11490 * doc/sourcebuild.texi: Document vect_long_mult.
11492 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11494 PR tree-optimization/82413
11495 * fold-const.c (build_range_check): Use widest_int when comparing
11496 the maximum ETYPE value with HIGH.
11498 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11500 PR rtl-optimization/82396
11501 * haifa-sched.c (autopref_multipass_init): Simplify
11503 (autopref_rank_data): Simplify sort order.
11504 * sched-int.h (autopref_multipass_data_): Remove
11505 multi_mem_insn_p, min_offset and max_offset.
11507 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11509 PR tree-optimization/82381
11510 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11511 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11514 PR tree-optimization/82374
11515 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11516 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11517 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11518 current_function_decl to the new decl.
11520 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11522 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11523 helper macro for IEEE float128 hardware built-in functions.
11524 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11526 (TRUNCF128_ODD): Likewise.
11527 (ADDF128_ODD): Likewise.
11528 (SUBF128_ODD): Likewise.
11529 (MULF128_ODD): Likewise.
11530 (DIVF128_ODD): Likewise.
11531 (FMAF128_ODD): Likewise.
11532 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11533 UNSPEC_TRUNC_ROUND_TO_ODD.
11534 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11535 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11536 floating point round to odd instructions.
11537 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11538 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11539 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11540 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11541 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11542 (trunc<mode>sf2_hw): Change the truncate with round to odd
11543 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11544 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11545 to odd hardware instructions.
11546 (sub<mode>3_odd): Likewise.
11547 (mul<mode>3_odd): Likewise.
11548 (div<mode>3_odd): Likewise.
11549 (sqrt<mode>2_odd): Likewise.
11550 (fma<mode>4_odd): Likewise.
11551 (fms<mode>4_odd): Likewise.
11552 (nfma<mode>4_odd): Likewise.
11553 (nfms<mode>4_odd): Likewise.
11554 (trunc<mode>df2_odd): Change the truncate with round to odd
11555 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11557 * doc/extend.texi (PowerPC built-in functions): Update documentation
11558 for existing IEEE float128-bit built-in functions. Add built-in
11559 functions that generate the IEEE 128-bit floating point round to
11562 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11564 PR rtl-optimization/77729
11565 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11566 to (X&(C1&~C2))|C2 transformations.
11568 2017-10-03 Martin Jambor <mjambor@suse.cz>
11570 PR tree-optimization/82363
11571 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11572 mismatch, mark lacc written regardless of racc.
11574 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11576 PR tree-optimization/82381
11577 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11578 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11579 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11580 fallthrough into reassoc_stmt_dominates_stmt_p.
11583 * combine.c (combine_instructions): Don't combine in unreachable
11586 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11590 function to not use the have_cpu variable. Do not set cpu_index,
11591 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11592 or the default cpu.
11593 (rs6000_valid_attribute_p): Remove duplicate initializations of
11594 old_optimize and func_optimize.
11595 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11596 (rs6000_activate_target_options): Make global.
11597 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11600 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11602 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11603 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11604 Set *poffset to 0 and *psize and *pmax_size to -1 if
11605 *poffset + *psize overflows in HOST_WIDE_INT.
11607 PR tree-optimization/82387
11608 PR tree-optimization/82388
11609 PR tree-optimization/82389
11610 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11611 instead of live_bytes non-NULL.
11613 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11616 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11619 2017-10-02 Richard Biener <rguenther@suse.de>
11621 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11622 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11623 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11625 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11627 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11628 requested precision matches the type's.
11629 * calls.c (alloc_max_size): Calculate the new candidate size as
11630 a widest_int and use wi::to_widest when comparing it with the
11631 current candidate size.
11632 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11633 zero rather than integer_zero_node.
11634 * match.pd: Check for a no-op conversion before using wi::add
11635 rather than after. Use tree_to_uhwi when summing small shift
11636 counts into an unsigned int.
11638 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11639 Alan Hayward <alan.hayward@arm.com>
11640 David Sherwood <david.sherwood@arm.com>
11643 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11644 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11645 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11646 POINTER_AND_FP_REGS.
11648 2017-10-02 Richard Biener <rguenther@suse.de>
11650 PR tree-optimization/82355
11651 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11652 a mapping for the enclosing loop but avoid generating one for
11653 the loop tree root.
11654 (copy_bb_and_scalar_dependences): Remove premature codegen
11655 error on PHIs in blocks duplicated into multiple places.
11656 * graphite-scop-detection.c
11657 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11658 in the region use it as loop and nest to analyze the DR in.
11659 (try_generate_gimple_bb): Likewise.
11660 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11661 (add_loop_constraints): For blocks in a loop not in the region
11662 create a dimension with a single iteration.
11663 * sese.h (gbb_loop_at_index): Remove assert.
11665 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11667 * omp-expand.c (adjust_context_scope): New function.
11668 (expand_parallel_call): Call adjust_context_scope.
11670 2017-10-01 Jeff Law <law@redhat.com>
11672 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11673 dom_opt_dom_walker class with direct access to private members.
11674 Add comments. Call test_for_singularity.
11675 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11676 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11677 m_dummy_cond anymore.
11678 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11680 (pass_dominator:execute): Build the dummy_cond here and pass it
11681 to the dom_opt_dom_walker ctor.
11682 (test_for_singularity): New function.
11684 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11685 Maya Rashish <coypu@sdf.org>
11687 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11688 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11689 (alpha*-*-netbsd*) Use nbsd_tm_file.
11690 (arm*-*-netbsdelf*) Likewise.
11691 (i[34567]86-*-netbsdelf*) Likewise.
11692 (x86_64-*-netbsd*) Likewise.
11693 (mips*-*-netbsd*) Likewise.
11694 (powerpc-*-netbsd*) Likewise.
11695 (sh*-*-netbsd*) Likewise.
11696 (sparc-*-netbsdelf*) Likewise.
11697 (sparc64-*-netbsd*) Likewise.
11698 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
11700 (vax-*-netbsdelf*) Likewise.
11701 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
11702 (UINT_FAST8_TYPE) Likewise.
11703 (INT_FAST16_TYPE) Check CHAR_FAST16.
11704 (UINT_FAST16_TYPE) Likewise.
11706 2017-09-30 Jakub Jelinek <jakub@redhat.com>
11709 * config/i386/i386.md
11710 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
11711 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
11712 *divmodsi4_zext_2): New define_insn_and_split.
11713 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
11714 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
11715 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
11716 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
11717 New define_insn_and_split.
11718 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11719 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11720 operands[1] having DImode when mode is SImode.
11722 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11723 always SImode for DIV and MOD in REG_EQUAL notes.
11725 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11727 PR middle-end/82319
11728 * match.pd: Fix handling of NaNs in pattern.
11730 2017-09-29 Jeff Law <law@redhat.com>
11732 * sbitmap.c (bitmap_bit_in_range_p): New function.
11733 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11734 * tree-ssa-dse.c (live_bytes_read): New function.
11735 (dse_classify_store): Ignore reads of dead bytes.
11737 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11738 typos and whitespace errors.
11739 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11740 * config/s390/s390.c (s390_emit_prologue): Likewise.
11742 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11745 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11746 with a symbol for LRA.
11748 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11750 PR rtl-optimization/82338
11751 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11753 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11755 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11756 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11757 (qsort_chk): Declare.
11758 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11759 (qsort_chk): New function.
11761 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11763 PR tree-optimization/82337
11764 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11765 phi definition if the PHI result appears in an abnormal PHI.
11766 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11767 basis appears in an abnormal PHI.
11769 2017-09-29 Richard Biener <rguenther@suse.de>
11771 * graphite-isl-ast-to-gimple.c
11772 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11773 (binary_op_to_tree): Use it.
11774 (get_rename_from_scev): Likewise.
11775 (copy_loop_phi_nodes): Likewise.
11776 (copy_bb_and_scalar_dependences): Likewise.
11777 (translate_pending_phi_nodes): Likewise.
11779 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11782 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11783 for movabsq $(i32 << shift), r64.
11785 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11787 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11788 index when encoding %esp as %rsp to avoid 0x67 prefix.
11790 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11792 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11793 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11795 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11797 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11798 Extensions with more than 16 double VFP registers.
11799 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11800 to_clear_mask and all code related to it. Replace the remaining
11801 entry by a sbitmap and adapt code accordingly.
11803 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11805 * brig-builtins.def: Change pure attributes to const.
11807 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11809 * config.gcc (default_gnu_indirect_function): Default to yes for
11810 sparc*-*-linux* with glibc.
11812 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11814 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11815 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11816 when creating .init_array and .fini_array sections with priority
11819 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11822 * config/aarch64/aarch64.c
11823 (aarch64_builtin_support_vector_misalignment): Always return false
11824 when misalignment is unknown.
11826 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11828 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11829 this function to return false if the definition used by the swap
11830 instruction is artificial, or if the memory address from which the
11831 constant value is loaded is not represented by a base address held
11832 in a register or if the base address register is a frame or stack
11833 pointer. Additionally, return false if the base address of the
11834 loaded constant is a SYMBOL_REF but is not considered to be a
11836 (replace_swapped_load_constant): New function.
11837 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11838 loaded constant vector with a load of a swapped constant vector.
11840 2017-09-27 Carl Love <cel@us.ibm.com>
11842 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11843 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11844 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11847 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11849 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11850 first, always call autopref_rank_data otherwise.
11852 2017-09-27 Richard Biener <rguenther@suse.de>
11854 * graphite-scop-detection.c (find_scop_parameters): Move
11855 loop bound handling ...
11856 (gather_bbs::before_dom_children): ... here, avoiding the need
11857 to build scop_info->loop_nest.
11858 (record_loop_in_sese): Remove.
11859 * sese.h (sese_info_t::loop_nest): Remove.
11860 * sese.c (new_sese_info): Do not allocate loop_nest.
11861 (free_sese_info): Do not free loop_nest.
11863 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11866 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11867 lhs from calls if the lhs has addressable type.
11869 2017-09-27 Richard Biener <rguenther@suse.de>
11871 * graphite.h (scop::max_alias_set): New member.
11872 * graphite-scop-detection.c: Remove references to non-existing
11873 --param in comments.
11874 (build_alias_sets): Record the maximum alias set used for drs.
11875 (build_scops): Support zero as unlimited for
11876 --param graphite-max-arrays-per-scop.
11877 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11878 and inline into ...
11879 (build_poly_sr_1): ... here. Compute alias set based on the
11880 maximum alias set used for drs rather than
11881 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11883 2017-09-27 Richard Biener <rguenther@suse.de>
11885 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11886 --param loop-block-tile-size=0 to disable tiling.
11888 2017-09-27 Richard Biener <rguenther@suse.de>
11890 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11891 (graphite-max-nb-scop-params): Document special value zero.
11892 * domwalk.h (dom_walker::STOP): New symbolical constant.
11893 (dom_walker::dom_walker): Add optional parameter for bb to
11895 (dom_walker::~dom_walker): Declare.
11896 (dom_walker::before_dom_children): Document STOP return value.
11897 (dom_walker::m_user_bb_to_rpo): New member.
11898 (dom_walker::m_bb_to_rpo): Likewise.
11899 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11900 mapping here if not provided by the user.
11901 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11902 provided by the user.
11903 (dom_walker::STOP): Define.
11904 (dom_walker::walk): Do not compute bb to RPO mapping here.
11905 Support STOP return value from before_dom_children to stop
11907 * graphite-optimize-isl.c (optimize_isl): If the schedule
11908 is the same still generate code if -fgraphite-identity
11909 or -floop-parallelize-all are given.
11910 * graphite-scop-detection.c: Include cfganal.h.
11911 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11913 (gather_bbs::before_dom_children): Return STOP for BBs
11915 (build_scops): Compute bb to RPO mapping and pass it to
11916 the domwalk. Treat --param graphite-max-nb-scop-params=0
11917 as not limiting the number of params.
11918 * graphite.c (graphite_initialize): Remove limit on the
11919 number of basic-blocks in a function.
11920 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11921 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11922 default value of 10.
11924 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11926 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11927 Adjust code to eliminate needing to do the shift right 32-bits
11928 operation after XSCVDPSPN.
11930 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11932 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11933 ((X / Y) != 0 -> X >= Y): Likewise.
11935 2017-09-26 Carl Love <cel@us.ibm.com>
11937 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11938 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11939 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11940 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11941 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11942 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11943 definitions and overloading.
11944 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11945 statement for P9V_BUILTIN_XST_LEN_R.
11946 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11947 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11948 define_expand and define_insn for the instructions and builtins.
11949 * doc/extend.texi: Update the built-in documentation file for the new
11950 built-in functions.
11951 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11952 define_insn for the instructions
11954 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11957 * gcc/config/netbsd-protos.h: New file.
11958 * gcc/config/netbsd.c: New file.
11959 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11960 * gcc/config/t-netbsd: New file.
11961 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11962 (tmake_file) Add t-netbsd.
11963 (extra_objs) Add netbsd.o.
11965 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11969 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11971 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11973 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11974 sign extension from a vector register to a GPR by doing a 32-bit
11975 direct move and then an EXTSW.
11976 (extendsi<mode>2 splitter): Likewise.
11977 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11978 right or vector extract after doing XSCVDPSPN. Use
11979 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11981 (movdi_from_sf_zero_ext): Likewise.
11982 (reload_gpr_from_vsxsf): Likewise.
11983 (p8_mfvsrd_4_disf): Delete, no longer used.
11984 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11985 then needing to move the SFmode to a GPR to use the XSCVDPSP
11986 instruction instead of FRSP and XSCVDPSPN.
11987 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11988 it is adjacent to the other XSCVSPDP insns.
11989 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11990 SFmode to be in traditional Altivec registers.
11991 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11992 (vsx_xscvspdpn): Likewise.
11993 (vsx_xscvspdpn_scalar): Likewise.
11995 2017-09-26 Martin Jambor <mjambor@suse.cz>
11997 * tree-sra.c (compare_access_positions): Put integral types first,
11998 stabilize sorting of integral types, remove conditions putting
11999 non-full-precision integers last.
12000 (sort_and_splice_var_accesses): Disable scalarization if a
12001 non-integert would be represented by a non-full-precision integer.
12003 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12005 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12006 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12007 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12008 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12009 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12010 conditionals inside the function instead of around it. Call
12011 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12012 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12014 2017-09-26 Richard Biener <rguenther@suse.de>
12016 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12018 (scop_detection::build_scop_breadth): ... this. Removed.
12019 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12020 (scop_detection::harmful_stmt_in_bb): Likewise.
12021 (scop_detection::graphite_can_represent_stmt): Likewise.
12022 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12023 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12024 (scop_detection::can_represent_loop_1): ... this. Removed.
12025 (scop_detection::harmful_loop_in_region): Simplify after inlining
12026 the above and remove more quadraticness.
12027 (build_scops): Adjust.
12028 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12031 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12034 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12035 REGNO (base) == SP_REG if base is a REG.
12037 PR middle-end/35691
12038 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12039 if it is different SSA_NAME.
12040 (optimize_range_tests_cmp_bitwise): New function.
12041 (optimize_range_tests): Call it.
12043 2017-09-26 Richard Biener <rguenther@suse.de>
12045 PR tree-optimization/82321
12046 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12047 for the def being inside the loop.
12049 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12051 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12053 * config/s390/s390-builtins.def: Fix constraint on op4.
12055 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12057 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12058 independent expanders.
12059 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12060 ("vec_ordered", "vec_unordered"): New expanders.
12062 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12064 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12067 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12069 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12070 vec_unpacks_lo_v16qi.
12071 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12073 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12075 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12076 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12077 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12079 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12081 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12083 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12085 ("*vec_slb<mode>"): New insn pattern.
12086 ("vec_shr_<mode>"): New expander.
12087 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12088 and force the shift count operand to V16QImode.
12089 ("vec_srb<mode>"): Set shift count mode to V16QI.
12091 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12093 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12094 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12095 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12097 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12100 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12102 2017-09-26 Richard Biener <rguenther@suse.de>
12104 PR tree-optimization/82320
12105 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12108 2017-09-25 Jeff Law <law@redhat.com>
12110 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12111 prototype for new argument.
12112 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12113 mostly extracted from rs6000_emit_allocate_stack.
12114 (rs6000_emit_probe_stack_range_stack_clash): New function.
12115 (rs6000_emit_allocate_stack): Call
12116 rs6000_emit_probe_stack_range_stack_clash as needed.
12117 (rs6000_emit_probe_stack_range): Add additional argument
12118 to call to gen_probe_stack_range{si,di}.
12119 (output_probe_stack_range): New.
12120 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12121 (output_probe_stack_range_stack_clash): New.
12122 (rs6000_emit_prologue): Emit notes into dump file as requested.
12123 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12124 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12125 Add additional operand and pass it to output_probe_stack_range.
12127 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12129 PR tree-optimization/82163
12130 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12131 (checking_verify_loop_closed_ssa): New parameter.
12132 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12133 (check_loop_closed_ssa_stmt): Delete.
12134 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12135 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12136 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12139 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12141 * brig-builtins.def: Treat HSAIL barrier builtins as
12142 setjmp/longjump style functions.
12144 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12146 * target.def (constant_alignment): New hook.
12147 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12148 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12149 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12150 * doc/tm.texi: Regenerate.
12151 * targhooks.h (default_constant_alignment): Declare.
12152 (constant_alignment_word_strings): Likewise.
12153 * targhooks.c (default_constant_alignment): New function.
12154 (constant_alignment_word_strings): Likewise.
12155 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12156 instead of CONSTANT_ALIGNMENT.
12157 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12158 (force_const_mem): Likewise.
12159 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12160 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12161 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12162 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12163 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12165 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12166 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12167 constant_alignment_word_strings.
12168 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12169 (CONSTANT_ALIGNMENT): Likewise.
12170 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12171 (arm_constant_alignment): New function.
12172 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12173 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12174 constant_alignment_word_strings.
12175 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12176 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12177 constant_alignment_word_strings.
12178 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12179 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12180 (cris_constant_alignment): New function.
12181 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12182 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12183 (epiphany_constant_alignment): New function.
12184 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12185 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12186 constant_alignment_word_strings.
12187 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12188 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12189 constant_alignment_word_strings.
12190 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12191 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12192 constant_alignment_word_strings.
12193 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12194 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12195 * config/i386/i386.c (ix86_constant_alignment): Make static.
12196 Use the same interface as the target hook.
12197 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12198 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12199 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12200 constant_alignment_word_strings.
12201 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12202 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12203 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12204 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12205 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12206 constant_alignment_word_strings.
12207 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12208 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12209 constant_alignment_word_strings.
12210 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12211 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12212 constant_alignment_word_strings.
12213 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12214 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12216 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12217 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12218 * config/mips/mips.c (mips_constant_alignment): New function.
12219 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12220 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12221 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12222 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12223 (mmix_constant_alignment): Make static. Use the same interface
12224 as the target hook.
12225 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12226 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12227 constant_alignment_word_strings.
12228 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12229 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12230 constant_alignment_word_strings.
12231 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12232 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12233 constant_alignment_word_strings.
12234 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12235 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12236 (rs6000_constant_alignment): New function.
12237 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12238 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12239 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12240 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12241 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12242 (rs6000_constant_alignment): New function.
12243 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12244 * config/s390/s390.c (s390_constant_alignment): New function.
12245 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12246 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12247 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12248 constant_alignment_word_strings.
12249 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12250 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12251 (sparc_constant_alignment): New function.
12252 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12253 * config/spu/spu.c (spu_constant_alignment): New function.
12254 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12255 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12256 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12257 constant_alignment_word_strings.
12258 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12259 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12260 constant_alignment_word_strings.
12261 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12262 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12263 constant_alignment_word_strings.
12264 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12265 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12266 (visium_constant_alignment): New function.
12267 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12268 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12269 (xtensa_constant_alignment): New function.
12270 * system.h (CONSTANT_ALIGNMENT): Poison.
12272 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12274 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12275 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12276 (rs6000_builtin_valid_without_lhs): New helper function.
12277 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12278 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12280 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12282 * target.h (vec_perm_indices): Use unsigned short rather than
12284 (auto_vec_perm_indices): Likewise.
12285 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12286 Use unsigned int rather than unsigned char.
12287 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12289 2017-09-25 Richard Biener <rguenther@suse.de>
12291 * cfgloop.h (sort_sibling_loops): Declare.
12292 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12293 (sort_sibling_loops): New function sorting the sibling loop list
12295 * graphite.c (graphite_transform_loops): Sort sibling loops.
12297 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12299 * target.def (vec_perm_const_ok): Change sel parameter to
12301 * optabs-query.c (can_vec_perm_p): Update accordingly.
12302 * doc/tm.texi: Regenerate.
12303 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12304 auto_vec_perm_indices and remove separate nelt field.
12305 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12306 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12307 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12308 (aarch64_expand_vec_perm_const): Update accordingly.
12309 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12310 to vec_perm_indices.
12311 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12312 auto_vec_perm_indices and remove separate nelt field.
12313 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12314 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12315 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12317 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12318 to vec_perm_indices.
12319 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12320 sel to vec_perm_indices.
12321 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12322 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12323 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12325 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12328 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12331 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12332 on the FUNCTION_DECL function context if it has a DIE that is a
12335 2017-09-25 Richard Biener <rguenther@suse.de>
12337 PR tree-optimization/82285
12338 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12341 2017-09-25 Tom de Vries <tom@codesourcery.com>
12345 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12348 2017-09-25 Richard Biener <rguenther@suse.de>
12350 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12351 ISL errors other than isl_error_quota happen. Dump if the
12352 schedule is the same.
12353 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12354 errors instead of aborting inside ISL.
12356 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12359 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12360 of libgcc_eh for m64.
12361 * config/i386/darwin64.h: Likewise.
12363 2017-09-25 Richard Biener <rguenther@suse.de>
12365 PR middle-end/82144
12366 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12367 attribute for incomplete types nor twice for complete ones.
12369 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12372 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12373 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12376 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12379 * config/i386/i386.opt (mprefer-avx256): Use
12380 ix86_target_flags variable.
12381 * config/i386/i386.c (ix86_target_string): Move
12382 -mprefer-avx256 to flag2_opts.
12384 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12386 PR middle-end/35691
12387 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12388 and x != -1 | y != -1 into (x & y) != -1.
12390 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12392 * config.gcc: Add new case statement to set
12393 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12394 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12395 s390x-*-linux* case statements. Added aarch64 to the list of
12396 supported architectures.
12398 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12400 PR tree-optimization/82289
12401 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12402 STMT_VINFO_RELEVANT_P.
12404 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12405 Alan Hayward <alan.hayward@arm.com>
12406 David Sherwood <david.sherwood@arm.com>
12408 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12409 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12410 (extract_range_from_binary_expr_1): Don't call
12411 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12413 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12414 Alan Hayward <alan.hayward@arm.com>
12415 David Sherwood <david.sherwood@arm.com>
12417 * target.def (preferred_vector_alignment): New hook.
12418 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12420 * doc/tm.texi: Regenerate.
12421 * targhooks.h (default_preferred_vector_alignment): Declare.
12422 * targhooks.c (default_preferred_vector_alignment): New function.
12423 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12425 (DR_TARGET_ALIGNMENT): New macro.
12426 (aligned_access_p): Update commentary.
12427 (vect_known_alignment_in_bytes): New function.
12428 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12430 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12431 Calculate the misalignment based on the target alignment rather than
12433 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12434 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12435 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12436 the target alignment, rather than masking the element misalignment
12437 with the number of elements in a vector. Also use the target
12438 alignment when calculating the maximum number of peels.
12439 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12440 instead of TYPE_ALIGN_UNIT.
12441 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12442 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12443 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12444 (vect_create_data_ref_ptr): Likewise.
12445 (vect_setup_realignment): Realign by ANDing with
12446 -DR_TARGET_MISALIGNMENT.
12447 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12448 the number of peels based on DR_TARGET_ALIGNMENT.
12449 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12450 with the guaranteed alignment boundary when deciding whether
12452 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12453 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12454 (ensure_base_align): Remove stmt_info parameter. Get the
12455 target base alignment from DR_TARGET_ALIGNMENT.
12456 (vectorizable_store): Update call accordingly. Interpret
12457 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12459 (vectorizable_load): Likewise.
12461 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12462 Alan Hayward <alan.hayward@arm.com>
12463 David Sherwood <david.sherwood@arm.com>
12465 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12466 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12467 (vect_enhance_data_refs_alignment): Likewise.
12469 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12471 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12472 error. Only quit immediately if parsing is complete.
12473 (BEGIN): Initialize fatal_err and parse_done.
12474 (begin fpu, end fpu): Check number of arguments.
12475 (begin arch, end arch): Likewise.
12476 (begin cpu, end cpu): Likewise.
12477 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12478 (optalias): Likewise.
12480 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12482 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12483 * config/arm/arm-isa.h: Delete. Move definitions to ...
12484 * arm-cpus.in: ... here. Use new feature and fgroup values.
12485 * config/arm/arm.c (arm_option_override): Use lower case for feature
12487 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12488 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12489 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12491 (print_isa_bits_for): New function.
12492 (gen_isa): New function.
12493 (gen_comm_data): Use print_isa_bits_for.
12494 (define feature): New keyword.
12495 (define fgroup): New keyword.
12496 * config/arm/t-arm (TM_H): Remove.
12497 (GTM_H): Add arm-isa.h.
12498 (arm-isa.h): Add rule to generate file.
12499 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12500 case for feature bit names.
12502 2017-09-22 Richard Biener <rguenther@suse.de>
12504 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12506 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12507 print of no dependency loops ...
12508 * graphite.c (graphite_transform_loops): ... here.
12509 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12511 (same_close_phi_node, remove_duplicate_close_phi,
12512 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12513 (canonicalize_loop_closed_ssa): ... here and simplify.
12514 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12515 (optimize_isl): Use dump_printf_loc to tell when we stopped
12516 optimizing because of an ISL timeout.
12518 2017-09-22 Richard Biener <rguenther@suse.de>
12520 PR tree-optimization/82291
12521 * tree-if-conv.c (predicate_mem_writes): Make sure to
12522 remove writes in blocks predicated with false.
12524 2017-09-22 Richard Biener <rguenther@suse.de>
12526 * sese.c: Include cfganal.h.
12527 (if_region_set_false_region): Remove.
12528 (create_if_region_on_edge): Likewise.
12529 (move_sese_in_condition): Re-implement without destroying
12532 2017-09-22 Richard Biener <rguenther@suse.de>
12534 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12535 Verify both BBs contain loop PHI nodes before dispatching to
12536 copy_loop_phi_args.
12537 (graphite_regenerate_ast_isl): Do not recompute dominators,
12538 do not verify three times. Restructure for clarity.
12539 * graphite-scop-detection.c (same_close_phi_node,
12540 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12541 defined_in_loop_p, canonicalize_loop_closed_ssa,
12542 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12543 checking and SSA rewrite, move to ...
12544 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12545 (graphite_initialize): Do not pass in ctx, do not reset the
12546 SCEV cache, compute only dominators.
12547 (graphite_transform_loops): Allocate ISL ctx after
12548 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12549 Maintain post-dominators only around build_scops.
12550 * sese.c (if_region_set_false_region): Make static. Free
12551 and recompute dominators.
12552 (move_sese_in_condition): Assert we don't get called with
12553 post-dominators computed.
12554 * sese.h (if_region_set_false_region): Remove.
12556 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12558 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12559 mode attribute for TARGET_AVX512VL.
12561 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12563 * config/i386/i386.opt (mprefer-avx256): New option.
12564 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12566 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12567 for TARGET_PREFER_AVX256.
12568 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12570 2017-09-21 Jeff Law <law@redhat.com>
12572 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12573 Fix dump output if the only stack space is for pushed registers.
12575 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12577 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12580 2017-09-21 Martin Sebor <msebor@redhat.com>
12583 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12584 code (in C++) or code that triggers warnings.
12586 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12588 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12590 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12592 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12593 * sched-rgn.c: Ditto.
12594 * sel-sched-ir.c: Ditto.
12596 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12598 * toplev.h (set_random_seed): Adjust return type.
12599 * toplev.c (init_local_tick): Move eager initialization of random_seed
12600 to get_random_seed. Adjust comment.
12601 (init_random_seed): Inline to get_random_seed, delete.
12602 (get_random_seed): Initialize random_seed lazily.
12603 (set_random_seed): Do not return previous value.
12604 (print_switch_value): Do not call get_random_seed.
12606 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12608 * cgraph.c (delete_function_version): New, broken out from...
12609 (cgraph_node::delete_function_version): ...here. Rename to
12610 cgraph_node::delete_function_version_by_decl. Update all uses.
12611 (cgraph_node::remove): Call delete_function_version.
12613 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12616 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12617 VAR_DECLs to which addressable non-volatile parameters are mapped
12618 and for id->retvar after the return value assignment. Clear
12619 id->retval and id->retbnd after inlining.
12621 2017-09-21 Richard Biener <rguenther@suse.de>
12623 PR tree-optimization/82276
12624 PR tree-optimization/82244
12625 * tree-vrp.c (build_assert_expr_for): Set
12626 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12628 (remove_range_assertions): Revert earlier change.
12630 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12633 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12635 2017-09-21 Richard Biener <rguenther@suse.de>
12637 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12638 Restore valid IL after code generation errors.
12639 * graphite.c (graphite_transform_loops): Diagnose code
12640 generation issues as MSG_MISSED_OPTIMIZATION and continue
12641 with processing SCOPs.
12643 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12644 Alan Hayward <alan.hayward@arm.com>
12645 David Sherwood <david.sherwood@arm.com>
12647 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12648 rather than choosing between plus_constant and gen_rtx_<CODE>.
12649 * expr.c (emit_push_insn): Likewise.
12650 (expand_expr_real_2): Likewise.
12652 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12653 Alan Hayward <alan.hayward@arm.com>
12654 David Sherwood <david.sherwood@arm.com>
12656 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12658 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12659 Alan Hayward <alan.hayward@arm.com>
12660 David Sherwood <david.sherwood@arm.com>
12662 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12663 calling tree_to_uhwi.
12665 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12666 Alan Hayward <alan.hayward@arm.com>
12667 David Sherwood <david.sherwood@arm.com>
12669 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12670 INTEGER_CST rather than a negative test for ADDR_EXPR.
12672 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12673 Alan Hayward <alan.hayward@arm.com>
12674 David Sherwood <david.sherwood@arm.com>
12676 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12677 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12679 2017-09-21 Richard Biener <rguenther@suse.de>
12681 PR tree-optimization/71351
12682 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12683 graphite_create_new_loop_guard): Remove, fold remaining parts
12685 (translate_isl_ast_node_for): ... here and simplify.
12687 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12690 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12691 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12692 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12693 alternative always use QI mode, for -Os imov (=R,R) alternative
12694 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
12697 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12698 Jeff Law <law@redhat.com>
12700 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
12701 (allocate_stack_space): New function, partially extracted from
12702 s390_emit_prologue.
12703 (s390_emit_prologue): Track offset to most recent stack probe.
12704 Code to allocate space moved into allocate_stack_space.
12705 Dump actions when no stack is allocated.
12706 (s390_prologue_plus_offset): New function.
12707 (s390_emit_stack_probe): Likewise.
12709 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
12711 * common.opt (Wa, Wl, Wp, g, gz=): Add
12713 (gno-column-info): Remove.
12714 (gcolumn-info): Drop RejectNegative.
12715 (gno-): New prefix.
12716 (gno-record-gcc-switches): Remove.
12717 (grecord-gcc-switches): Drop RejectNegative.
12718 (gno-split-dwarf): Remove.
12719 (gsplit-dwarf): Drop RejectNegative.
12720 (gno-strict-dwarf): Remove.
12721 (gstrict-dwarf): Drop RejectNegative.
12722 * config/darwin.opt (gfull, gused): Add RejectNegative.
12723 * dwarf2out.c (gen_producer_string): Drop
12724 gno-record-gcc-switches handler.
12725 * optc-gen.awk: Add g to prefixes with negative forms.
12726 * opts-common.c (remapping_prefix_p): New.
12727 (find_opt): Check it.
12728 (generate_canonical_option): Test g prefix.
12729 (option_map): Add -gno- mapping.
12730 (add_misspelling_candidates): Check remapping_prefix_p.
12732 2017-09-20 Jeff Law <law@redhat.com>
12734 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12735 thinko in stack clash protection support.
12737 * explow.c (compute_stack_clash_protection_loop_data): Use
12738 CONST_INT_P instead of explicit test. Verify object is a
12739 CONST_INT_P before looking at INTVAL.
12740 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12741 instead of explicit test.
12743 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12746 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12747 address instead of to r1 and r11.
12749 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12751 * config.gcc: Support "knm".
12752 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12753 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12755 * config/i386/i386.c (m_KNM): Define.
12756 (processor_target_table): Add "knm".
12758 (ix86_option_override_internal): Add "knm".
12759 (ix86_issue_rate): Add PROCESSOR_KNM.
12760 (ix86_adjust_cost): Ditto.
12761 (ia32_multipass_dfa_lookahead): Ditto.
12762 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12763 (fold_builtin_cpu): Add M_INTEL_KNM.
12764 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12765 (processor_type): Add PROCESSOR_KNM.
12766 * config/i386/x86-tune.def: Add m_KNM.
12767 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12769 2017-09-20 Richard Biener <rguenther@suse.de>
12771 PR tree-optimization/80213
12772 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12773 are allowed in empty BBs as well.
12774 (canonicalize_loop_closed_ssa): Also look for other complex
12776 (scop_detection::get_sese): Include the loop-closed PHI block
12778 (scop_detection::merge_sese): Remove code adding extra blocks.
12779 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12780 (build_scops): Assert the final returned scop is invalid.
12782 2017-09-20 Richard Biener <rguenther@suse.de>
12784 PR tree-optimization/82264
12785 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12787 (vn_phi_lookup): Likewise.
12788 (vn_phi_insert): Likewise.
12790 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12792 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12793 that fits into uhwi or shwi, add DW_AT_const_value regardless
12794 of early_dwarf without going through RTL, using add_AT_unsigned
12797 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12798 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12799 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12800 (DEBUG_MACRO_SECTION): Likewise.
12801 (DEBUG_LTO_MACRO_SECTION): Likewise.
12802 (DEBUG_STR_DWO_SECTION): Likewise.
12803 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12804 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12805 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12806 (DEBUG_LTO_LINE_STR_SECTION): Define.
12807 (init_sections_and_labels): Initialize debug_line_str_section
12808 variable. Initialize debug_loc_section for -gdwarf-5 to
12809 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12811 2017-09-20 Richard Biener <rguenther@suse.de>
12813 * graphite-sese-to-poly.c (extract_affine): Properly handle
12814 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12816 2017-09-20 Richard Biener <rguenther@suse.de>
12818 PR tree-optimization/81373
12819 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12820 Force SESE live-out defs to be handled even if they are
12823 2017-09-19 Jeff Law <law@redhat.com>
12825 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12826 nothing for stack adjustments with REG_STACK_CHECK.
12827 * sched-deps.c (parse_add_or_inc): Reject insns with
12828 REG_STACK_CHECK from dependency breaking.
12829 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12830 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12831 * reg-notes.def (STACK_CHECK): New note.
12833 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12834 (ix86_expand_prologue): Dump stack clash info as needed.
12835 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12837 * function.c (dump_stack_clash_frame_info): New function.
12838 * function.h (dump_stack_clash_frame_info): Prototype.
12839 (enum stack_clash_probes): New enum.
12841 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12842 flag_stack_clash_protection.
12843 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12844 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12845 (arm_frame_pointer_required): Likewise.
12846 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12847 (ia64_expand_prologue): Likewise.
12848 * config/mips/mips.c (mips_expand_prologue): Likewise.
12849 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12850 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12851 (sparc_flat_expand_prologue): Likewise.
12852 * config/spu/spu.c (spu_expand_prologue): Likewise.
12854 * explow.c: Include "params.h".
12855 (anti_adjust_stack_and_probe_stack_clash): New function.
12856 (get_stack_check_protect): Likewise.
12857 (compute_stack_clash_protection_loop_data): Likewise.
12858 (emit_stack_clash_protection_loop_start): Likewise.
12859 (emit_stack_clash_protection_loop_end): Likewise.
12860 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12861 Use anti_adjust_stack_and_probe_stack_clash.
12862 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12863 (emit_stack_clash_protection_loop_start): Likewise.
12864 (emit_stack_clash_protection_loop_end): Likewise.
12865 * rtl.h (get_stack_check_protect): Prototype.
12866 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12867 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12868 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12870 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12872 * doc/tm.texi: Rebuilt.
12873 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12874 get_stack_check_protect.
12875 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12876 * config/arm/arm.c (arm_expand_prologue): Likewise.
12877 (arm_frame_pointer_required): Likewise.
12878 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12879 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12880 * config/mips/mips.c (mips_expand_prologue): Likewise.
12881 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12882 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12883 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12884 (sparc_flat_expand_prologue): Likewise.
12886 * common.opt (-fstack-clash-protection): New option.
12887 * flag-types.h (enum stack_check_type): Note difference between
12888 -fstack-check= and -fstack-clash-protection.
12889 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12890 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12891 * toplev.c (process_options): Issue warnings/errors for cases
12892 not handled with -fstack-clash-protection.
12893 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12894 (-fstack-check): Note additional problem with -fstack-check=generic.
12895 Note that -fstack-check is primarily for Ada and refer users
12896 to -fstack-clash-protection for stack-clash-protection.
12897 Document new params for stack clash protection.
12899 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12901 * config/i386/i386.c (ix86_split_long_move): Do not handle
12902 address used for LEA in a special way.
12904 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12906 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12908 2017-09-19 Martin Sebor <msebor@redhat.com>
12911 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12912 of incompatible types.
12914 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12917 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12918 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12919 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12921 2017-09-19 Richard Biener <rguenther@suse.de>
12923 PR tree-optimization/82244
12924 * tree-vrp.c (remove_range_assertions): Do not propagate
12925 a constant to abnormals but replace the assert with a copy.
12927 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12929 PR rtl-optimization/57878
12930 PR rtl-optimization/68988
12931 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12932 avoidance test involving non_reload_pseudos. Move frequency test
12933 below the general fragmentation avoidance test.
12935 2017-09-19 Richard Biener <rguenther@suse.de>
12937 PR tree-optimization/69728
12938 * graphite-sese-to-poly.c (schedule_error): New global.
12939 (add_loop_schedule): Handle empty domain by failing the
12941 (build_original_schedule): Handle schedule_error.
12943 2017-09-19 Richard Biener <rguenther@suse.de>
12945 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12946 Do not iterate to sibling loops but only to siblings of inner
12949 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12952 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12953 that the registers are different.
12955 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12957 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12958 to processor_model and "amdfam17h" to arch_names_table.
12959 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12961 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12964 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12966 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12967 Alan Hayward <alan.hayward@arm.com>
12968 David Sherwood <david.sherwood@arm.com>
12970 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12972 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12974 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12975 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12976 vect_schedule_slp_instance.
12977 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12978 Update call to vect_slp_analyze_node_operations. Simplify return
12980 (vect_slp_analyze_bb_1): Update call accordingly.
12981 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12982 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12983 (vect_schedule_slp): Update call accordingly.
12985 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12986 Alan Hayward <alan.hayward@arm.com>
12987 David Sherwood <david.sherwood@arm.com>
12989 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12990 with types that aren't in fact scalar.
12992 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12994 * tree-vect-slp.c (vect_record_max_nunits): New function,
12996 (vect_build_slp_tree_1): ...here.
12997 (vect_build_slp_tree_2): Call it for phis too.
12999 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13001 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13002 to vect_get_vec_def_for_operand when getting the mask operand.
13004 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13005 Alan Hayward <alan.hayward@arm.com>
13006 David Sherwood <david.sherwood@arm.com>
13008 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13011 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13012 Alan Hayward <alan.hayward@arm.com>
13013 David Sherwood <david.sherwood@arm.com>
13015 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13016 calculation for vector booleans.
13018 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13019 Alan Hayward <alan.hayward@arm.com>
13020 David Sherwood <david.sherwood@arm.com>
13022 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13024 (vect_transform_stmt): ...here.
13025 (vect_analyze_stmt): Use it instead of calling
13026 vectorizable_live_operation directly.
13028 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13030 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13031 non-SIMT targets in acc vector loops.
13033 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13035 * configure.ac: Add arc and check if assembler supports gdwarf2.
13036 * configure: Regenerate.
13038 2017-09-18 Richard Biener <rguenther@suse.de>
13040 PR tree-optimization/82220
13041 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13042 epilogue niters from the min_profitable_iters compute.
13044 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13047 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13048 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13049 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13051 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13054 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13055 switching to a new text section.
13057 2017-09-18 Richard Biener <rguenther@suse.de>
13059 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13061 (build_alias_set): Reject aliases with no access function.
13063 2017-09-18 Richard Biener <rguenther@suse.de>
13065 PR tree-optimization/79622
13066 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13068 (build_cross_bb_scalars_use): Likewise.
13070 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13072 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13074 2017-09-18 Alan Modra <amodra@gmail.com>
13077 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13078 stack_pointer_rtx for count 0. Update comments. Break up
13079 large rtl expression.
13081 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13083 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13084 Increase to 20 bytes.
13085 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13086 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13087 or avx version of the stub.
13089 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13092 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13093 compute the minimum stack alignment. Also update preferred stack
13094 boundary for leaf functions.
13096 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13098 PR tree-optimization/82228
13099 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13102 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13104 * common/config/nds32/nds32-common.c
13105 (nds32_option_optimization_table): Refine formatting.
13106 (nds32_option_optimization_table): Use -fsched-pressure and
13107 -fomit-frame-pointer for specific optimization level.
13109 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13111 * config/nds32/nds32.c: Refine formatting and comments.
13112 * config/nds32/nds32.h: Likewise.
13113 * config/nds32/nds32.md: Likewise.
13114 * config/nds32/nds32-cost.c: Likewise.
13115 * config/nds32/nds32-isr.c: Likewise.
13116 * config/nds32/nds32-md-auxiliary.c: Likewise.
13117 * config/nds32/nds32-multiple.md: Likewise.
13118 * config/nds32/nds32-predicates.c: Likewise.
13120 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13121 Jakub Jelinek <jakub@redhat.com>
13123 Add support for -std=c++2a.
13124 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13126 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13128 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13131 * doc/extend.texi (Common Function Attributes): Add
13132 references to ARM, AArch64, and S/390 specific attributes.
13133 (Function Specific Option Pragmas): Add AArch64 and S/390
13134 to list of back ends that support the target pragma.
13136 2017-09-15 Nathan Sidwell <nathan@acm.org>
13138 * doc/standards.texi: Fix C++17 description. Update URLs for
13141 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13143 * common.opt (Wcast-align=strict): New warning option.
13144 * doc/invoke.texi: Document -Wcast-align=strict.
13146 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13148 * cgraph.h (cgraph_thunk_info): Add comments.
13149 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13150 assert for VIRTUAL_* arguments stricter.
13152 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13154 PR tree-optimization/71026
13155 * match.pd: Move RDIV patterns from fold-const.c
13156 * fold-const.c (distribute_real_division): Removed.
13157 (fold_binary_loc): Remove calls to distribute_real_divison.
13159 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13161 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13162 c++1z and gnu++1z as deprecated. Change other references to
13163 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13164 Change -Wc++1z-compat to -Wc++17-compat.
13165 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13166 * dwarf2out.c (highest_c_language): Handle C++17.
13167 (gen_compile_unit_die): Likewise.
13169 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13171 PR rtl-optimization/82192
13172 * combine.c (make_extraction): Don't look through non-paradoxical
13173 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13176 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13177 Alan Hayward <alan.hayward@arm.com>
13178 David Sherwood <david.sherwood@arm.com>
13180 * target.def (function_arg_offset): New hook.
13181 * targhooks.h (default_function_arg_offset): Declare.
13182 * targhooks.c (default_function_arg_offset): New function.
13183 * function.c (locate_and_pad_parm): Use
13184 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13185 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13186 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13187 * doc/tm.texi: Regenerate.
13188 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13189 * config/spu/spu.c (spu_function_arg_offset): New function.
13190 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13191 * system.h (FUNCTION_ARG_OFFSET): Poison.
13193 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13194 Alan Hayard <alan.hayward@arm.com>
13195 David Sherwood <david.sherwood@arm.com>
13197 * target.def (truly_noop_truncation): New hook.
13198 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13199 than TRULY_NOOP_TRUNCATION.
13200 * hooks.h (hook_bool_uint_uint_true): Declare.
13201 * hooks.c (hook_bool_uint_uint_true): New function.
13202 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13203 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13204 * doc/tm.texi: Regenerate.
13205 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13206 rather than TRULY_NOOP_TRUNCATION in comments.
13207 (simplify_comparison): Likewise.
13208 (record_truncated_value): Likewise.
13209 * expmed.c (extract_bit_field_1): Likewise.
13210 (extract_split_bit_field): Likewise.
13211 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13212 instead of TRULY_NOOP_TRUNCATION.
13213 * function.c (assign_parm_setup_block): Likewise.
13214 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13215 * rtlhooks.c: Include target.h.
13216 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13217 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13218 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13219 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13220 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13221 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13222 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13223 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13224 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13225 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13226 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13227 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13228 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13229 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13230 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13231 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13232 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13233 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13234 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13235 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13236 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13237 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13238 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13239 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13240 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13241 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13242 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13243 rather than TRULY_NOOP_TRUNCATION in comments.
13244 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13245 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13246 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13247 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13248 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13249 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13250 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13251 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13252 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13253 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13254 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13255 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13256 rather than TRULY_NOOP_TRUNCATION in comments.
13257 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13258 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13259 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13260 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13261 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13262 TRULY_NOOP_TRUNCATION condition.
13263 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13264 (TRULY_NOOP_TRUNCATION): Delete.
13265 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13266 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13267 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13268 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13269 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13270 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13271 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13272 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13273 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13274 rather than TRULY_NOOP_TRUNCATION in comments.
13275 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13276 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13277 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13278 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13279 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13280 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13282 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13285 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13286 (*cmp_ior): Likewise.
13287 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13288 (*ior_scc_scc_cmp): Likewise.
13289 (*and_scc_scc): Likewise.
13290 (*and_scc_scc_cmp): Likewise.
13292 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13293 Alan Hayard <alan.hayward@arm.com>
13294 David Sherwood <david.sherwood@arm.com>
13296 * target.def (can_change_mode_class): New hook.
13297 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13298 (hard_regno_nregs): Likewise.
13299 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13300 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13301 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13302 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13303 (LOAD_EXTEND_OP): Update accordingly.
13304 * doc/tm.texi: Regenerate.
13305 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13306 CANNOT_CHANGE_MODE_CLASS.
13307 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13308 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13309 * combine.c (simplify_set): Update accordingly.
13310 * emit-rtl.c (validate_subreg): Likewise.
13311 * recog.c (general_operand): Likewise.
13312 * regcprop.c (mode_change_ok): Likewise.
13313 * reload1.c (choose_reload_regs): Likewise.
13314 (inherit_piecemeal_p): Likewise.
13315 * rtlanal.c (simplify_subreg_regno): Likewise.
13316 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13317 instead of CANNOT_CHANGE_MODE_CLASS.
13318 (reload_cse_simplify_operands): Likewise.
13319 * reload.c (push_reload): Use targetm.can_change_mode_class
13320 instead of CANNOT_CHANGE_MODE_CLASS.
13321 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13322 REG_CANNOT_CHANGE_MODE_P.
13323 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13324 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13325 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13326 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13327 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13328 (arm_can_change_mode_class): New function.
13329 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13330 than CANNOT_CHANGE_MODE_CLASS in comments.
13331 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13332 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13333 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13334 (ix86_can_change_mode_class): ...this new function, inverting the
13335 sense of the return value.
13336 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13337 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13338 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13339 (ia64_can_change_mode_class): New function.
13340 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13341 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13342 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13343 (m32c_can_change_mode_class): ...this new function, inverting the
13344 sense of the return value.
13345 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13346 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13347 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13348 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13349 (mips_can_change_mode_class): ...this new function, inverting the
13350 sense of the return value.
13351 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13352 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13353 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13354 (msp430_can_change_mode_class): New function.
13355 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13356 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13357 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13358 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13359 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13360 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13361 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13362 (pa_cannot_change_mode_class): Replace with...
13363 (pa_can_change_mode_class): ...this new function, inverting the
13364 sense of the return value.
13365 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13366 than CANNOT_CHANGE_MODE_CLASS in comments.
13367 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13368 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13369 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13370 (pdp11_cannot_change_mode_class): Replace with...
13371 (pdp11_can_change_mode_class): ...this new function, inverting the
13372 sense of the return value.
13373 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13374 * config/powerpcspe/powerpcspe-protos.h
13375 (rs6000_cannot_change_mode_class_ptr): Delete.
13376 * config/powerpcspe/powerpcspe.c
13377 (rs6000_cannot_change_mode_class_ptr): Delete.
13378 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13379 (rs6000_option_override_internal): Assign to
13380 targetm.can_change_mode_class instead of
13381 rs6000_cannot_change_mode_class_ptr.
13382 (rs6000_cannot_change_mode_class): Replace with...
13383 (rs6000_can_change_mode_class): ...this new function, inverting the
13384 sense of the return value.
13385 (rs6000_debug_cannot_change_mode_class): Replace with...
13386 (rs6000_debug_can_change_mode_class): ...this new function.
13387 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13388 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13389 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13390 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13391 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13393 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13394 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13395 (rs6000_option_override_internal): Assign to
13396 targetm.can_change_mode_class instead of
13397 rs6000_cannot_change_mode_class_ptr.
13398 (rs6000_cannot_change_mode_class): Replace with...
13399 (rs6000_can_change_mode_class): ...this new function, inverting the
13400 sense of the return value.
13401 (rs6000_debug_cannot_change_mode_class): Replace with...
13402 (rs6000_debug_can_change_mode_class): ...this new function.
13403 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13404 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13405 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13406 (s390_can_change_mode_class): ...this new function, inverting the
13407 sense of the return value.
13408 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13409 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13410 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13411 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13412 (sh_cannot_change_mode_class): Replace with...
13413 (sh_can_change_mode_class): ...this new function, inverting the
13414 sense of the return value.
13415 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13416 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13417 (sparc_can_change_mode_class): New function.
13418 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13419 * config/spu/spu.c (spu_can_change_mode_class): New function.
13420 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13421 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13422 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13423 (visium_can_change_mode_class): New function.
13424 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13426 2017-09-15 Richard Biener <rguenther@suse.de>
13428 PR tree-optimization/82217
13429 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13430 but not undefined case.
13432 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13435 * postreload.c (reload_cse_simplify_operands): Skip
13436 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13438 2017-09-15 Richard Biener <rguenther@suse.de>
13440 PR tree-optimization/68823
13441 * graphite-scop-detection.c (build_alias_set): If we have a
13442 possible dependence check whether we can handle them by just
13443 looking at the DRs DR_ACCESS_FNs.
13444 (build_scops): If build_alias_set fails, fail the SCOP.
13446 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13448 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13449 to support float128 built-in functions that require the ISA 3.0
13451 (BU_FLOAT128_3_HW): Likewise.
13452 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13453 built-in functions.
13454 (FMAF128): Likewise.
13456 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13457 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13458 floating point instructions.
13459 (rs6000_invalid_builtin): Likewise.
13460 (rs6000_builtin_mask_names): Likewise.
13461 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13462 (RS6000_BTM_FLOAT128_HW): Likewise.
13463 (RS6000_BTM_COMMON): Likewise.
13464 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13466 * doc/extend.texi (RS/6000 built-in functions): Document the
13467 IEEE 128-bit floating point square root and fused multiply-add
13468 built-in functions.
13470 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13472 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13473 reg (r2) isn't in the set of registers defined in the prologue.
13475 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13476 Alan Hayward <alan.hayward@arm.com>
13477 David Sherwood <david.sherwood@arm.com>
13479 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13480 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13481 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13482 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13483 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13485 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13486 max_vectorization_factor.
13487 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13489 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13490 Alan Hayward <alan.hayward@arm.com>
13491 David Sherwood <david.sherwood@arm.com>
13493 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13494 (vect_worthwhile_without_simd_p): Declare.
13495 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13496 (vectorizable_reduction): Use it.
13497 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13498 (vectorizable_operation): Likewise.
13500 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13501 Alan Hayward <alan.hayward@arm.com>
13502 David Sherwood <david.sherwood@arm.com>
13504 * tree-vectorizer.h (vect_get_num_copies): New function.
13505 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13506 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13507 (vectorizable_induction): Likewise.
13508 (vectorizable_live_operation): Likewise.
13509 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13510 (vectorizable_bswap): Likewise.
13511 (vectorizable_call): Likewise.
13512 (vectorizable_conversion): Likewise.
13513 (vectorizable_assignment): Likewise.
13514 (vectorizable_shift): Likewise.
13515 (vectorizable_operation): Likewise.
13516 (vectorizable_store): Likewise.
13517 (vectorizable_load): Likewise.
13518 (vectorizable_condition): Likewise.
13519 (vectorizable_comparison): Likewise.
13520 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13522 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13523 Alan Hayward <alan.hayward@arm.com>
13524 David Sherwood <david.sherwood@arm.com>
13526 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13527 of vect_init_vector.
13529 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13530 Alan Hayward <alan.hayward@arm.com>
13531 David Sherwood <david.sherwood@arm.com>
13533 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13534 an inline wrapper that provides a location.
13535 (gimple_build_vector): Likewise.
13536 * gimple-fold.c (gimple_build_vector_from_val): New function.
13537 (gimple_build_vector): Likewise.
13538 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13539 functions to build the initial value. Always return a gimple value.
13540 (get_initial_defs_for_reduction): Likewise. Only compute
13542 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13543 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13544 (vectorizable_induction): Use gimple_build_vector rather than
13547 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13548 Alan Hayward <alan.hayward@arm.com>
13549 David Sherwood <david.sherwood@arm.com>
13551 * target.h (vec_perm_indices): New typedef.
13552 (auto_vec_perm_indices): Likewise.
13553 * optabs-query.h: Include target.h
13554 (can_vec_perm_p): Take a vec_perm_indices *.
13555 * optabs-query.c (can_vec_perm_p): Likewise.
13556 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13557 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13558 * tree-vect-generic.c (lower_vec_perm): Likewise.
13559 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13560 (vect_grouped_load_supported): Likewise.
13561 (vect_shift_permute_load_chain): Likewise.
13562 (vect_permute_store_chain): Use auto_vec_perm_indices.
13563 (vect_permute_load_chain): Likewise.
13564 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13565 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13566 Update uses of can_vec_perm_p.
13567 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13568 mode with a number of elements. Take a vec_perm_indices *.
13569 (vect_create_epilog_for_reduction): Update accordingly.
13570 Use auto_vec_perm_indices.
13571 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13572 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13573 (vect_transform_slp_perm_load): Likewise.
13574 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13575 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13576 (vect_gen_perm_mask_checked): Likewise.
13577 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13578 (vect_gen_perm_mask_checked): Likewise.
13579 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13580 (vectorizable_store): Likewise.
13581 (vectorizable_load): Likewise.
13582 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13583 (vectorizable_bswap): Likewise.
13585 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13586 Alan Hayward <alan.hayward@arm.com>
13587 David Sherwood <david.sherwood@arm.com>
13589 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13590 * tree.c (build_vector): Likewise.
13591 (build_vector_from_ctor): Update accordingly.
13592 (build_vector_from_val): Likewise.
13593 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13594 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13595 * tree-vect-generic.c (add_rshift): Likewise.
13596 (expand_vector_divmod): Likewise.
13597 (optimize_vector_constructor): Likewise.
13598 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13599 (vect_transform_slp_perm_load): Likewise.
13600 (vect_schedule_slp_instance): Likewise.
13601 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13602 (vectorizable_call): Likewise.
13603 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13604 * expmed.c (make_tree): Likewise.
13605 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13606 a vector passed to build_vector.
13607 (fold_convert_const): Likewise.
13608 (exact_inverse): Likewise.
13609 (fold_ternary_loc): Likewise.
13610 (fold_relational_const): Likewise.
13611 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13612 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13613 (const_unop): Likewise. Store the reduction accumulator in a
13614 variable rather than an array.
13615 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13616 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13617 the new vector, rather than constructing it after the input arrays.
13618 (native_interpret_vector): Use auto_vec<tree> when building
13619 a vector passed to build_vector. Add elements in order.
13620 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13621 auto_vec<tree> when building a vector passed to build_vector.
13622 (vect_create_epilog_for_reduction): Likewise.
13623 (vectorizable_induction): Likewise.
13624 (get_initial_def_for_reduction): Likewise. Fix indentation of
13626 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13628 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13629 passed to build_vector.
13631 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13632 Alan Hayward <alan.hayward@arm.com>
13633 David Sherwood <david.sherwood@arm.com>
13635 * tree-core.h (tree_base::u): Add an "nelts" field.
13636 (tree_vector): Use VECTOR_CST_NELTS as the length.
13637 * tree.c (tree_size): Likewise.
13638 (make_vector): Initialize VECTOR_CST_NELTS.
13639 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13640 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13641 TYPE_VECTOR_SUBPARTS.
13642 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13643 as the number of units, setting it from VECTOR_CST_NELTS.
13644 (const_vector_from_tree): Likewise.
13645 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13646 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13647 (fold_negate_expr_1): Likewise.
13648 (fold_convert_const): Likewise.
13649 (const_binop): Likewise. Differentiate the number of output and
13651 (const_unop): Likewise.
13652 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13653 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13654 in cases that did the opposite.
13656 2017-09-14 Richard Biener <rguenther@suse.de>
13658 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13661 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13663 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13665 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13668 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13669 if and where to split a bb, except for splitting before debug insn
13670 sequences followed by non-label real insn. Delete debug insns
13671 in between basic blocks.
13673 * combine.c (make_compound_operation_int): Formatting fixes.
13675 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13676 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13677 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13678 * config/sol2.h (LINK_EH_SPEC): Likewise.
13679 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13680 * config/s390/linux.h (LINK_SPEC): Likewise.
13681 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13682 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13683 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13684 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13685 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13686 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13687 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13688 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13690 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13692 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13694 (ENDFILE_LINUX_SPEC): Likewise.
13695 (LINK_EH_SPEC): Likewise.
13696 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
13697 (LINK_OS_LINUX_SPEC32): Likewise.
13698 (LINK_OS_LINUX_SPEC64): Likewise.
13699 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
13700 (LINK_OS_LINUX_SPEC): Likewise.
13702 2017-09-13 Martin Liska <mliska@suse.cz>
13704 PR middle-end/82154
13705 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
13706 CASE_HIGH is NULL_TREE.
13708 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13709 Alan Hayward <alan.hayward@arm.com>
13710 David Sherwood <david.sherwood@arm.com>
13712 * target.def (secondary_memory_needed): New hook.
13713 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13714 instead of SECONDARY_MEMORY_NEEDED.
13715 (secondary_memory_needed_mode): Likewise.
13716 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
13717 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13718 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13719 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13720 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13721 * doc/tm.texi: Regenerate.
13722 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13723 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13724 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13725 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13726 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13727 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13728 mode argument first and change the reg_class arguments to reg_class_t.
13729 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13730 Make static. Update the call to inline_secondary_memory_needed.
13731 (ix86_register_move_cost): Update the call to
13732 inline_secondary_memory_needed.
13733 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13734 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13736 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13737 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13739 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13740 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13741 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13742 and match hook interface. Add comment from mips.h.
13743 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13744 * config/mmix/mmix.md (truncdfsf2): Refer to
13745 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13747 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13748 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13749 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13750 (pa_secondary_memory_needed): New function.
13751 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13752 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13753 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13754 (pdp11_secondary_memory_needed): Make static and match hook interface.
13755 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13756 * config/powerpcspe/powerpcspe-protos.h
13757 (rs6000_secondary_memory_needed_ptr): Delete.
13758 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13760 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13761 (rs6000_option_override_internal): Assign to
13762 targetm.secondary_memory_needed rather than
13763 rs6000_secondary_memory_needed_ptr.
13764 (rs6000_secondary_memory_needed): Match hook interface.
13765 (rs6000_debug_secondary_memory_needed): Likewise.
13766 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13767 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13768 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13769 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13770 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13771 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13773 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13774 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13775 (rs6000_option_override_internal): Assign to
13776 targetm.secondary_memory_needed rather than
13777 rs6000_secondary_memory_needed_ptr.
13778 (rs6000_secondary_memory_needed): Match hook interface.
13779 (rs6000_debug_secondary_memory_needed): Likewise.
13780 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13781 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13782 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13783 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13784 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13785 (sparc_secondary_memory_needed): New function.
13786 * lra-constraints.c (check_and_process_move): Refer to
13787 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13789 (curr_insn_transform): Likewise.
13790 (process_alt_operands): Use targetm.secondary_memory_needed
13791 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13792 (check_secondary_memory_needed_p): Likewise.
13793 (choose_split_class): Likewise.
13794 * reload.c: Unconditionally include code that was previously
13795 conditional on SECONDARY_MEMORY_NEEDED.
13796 (push_secondary_reload): Use targetm.secondary_memory_needed
13797 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13798 (push_reload): Likewise.
13799 * reload1.c: Unconditionally include code that was previously
13800 conditional on SECONDARY_MEMORY_NEEDED.
13801 (choose_reload_regs): Use targetm.secondary_memory_needed
13802 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13803 (gen_reload): Likewise.
13804 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13806 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13807 Alan Hayward <alan.hayward@arm.com>
13808 David Sherwood <david.sherwood@arm.com>
13810 * target.def (secondary_memory_needed_mode): New hook:
13811 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13812 * targhooks.h (default_secondary_memory_needed_mode): New function.
13813 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13814 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13815 * doc/tm.texi: Regenerate.
13816 * lra-constraints.c (check_and_process_move): Use
13817 targetm.secondary_memory_needed_mode instead of
13818 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13819 (curr_insn_transform): Likewise.
13820 * reload.c (get_secondary_mem): Likewise.
13821 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13822 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13824 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13825 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13826 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13827 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13828 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13830 * config/powerpcspe/powerpcspe-protos.h
13831 (rs6000_secondary_memory_needed_mode): Delete.
13832 * config/powerpcspe/powerpcspe.c
13833 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13834 (rs6000_secondary_memory_needed_mode): Make static.
13835 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13836 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13838 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13840 (rs6000_secondary_memory_needed_mode): Make static.
13841 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13842 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13843 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13844 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13845 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13847 (sparc_secondary_memory_needed_mode): New function.
13848 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13850 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13852 * config/aarch64/constraints.md (Umq): New constraint.
13853 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13855 (mov<mode>): Update condition.
13857 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13859 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13860 when bitposition is the same.
13862 2017-09-13 Richard Biener <rguenther@suse.de>
13864 * dwarf2out.c (output_die_symbol): Remove.
13865 (output_die): Do not output a DIEs symbol.
13867 2017-09-13 Richard Biener <rguenther@suse.de>
13869 PR middle-end/82128
13870 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13871 default-def to avoid breaking iterator update with the weird
13872 interaction with cgraph_update_edges_for_call_stmt_node.
13874 2017-09-13 Richard Biener <rguenther@suse.de>
13876 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13877 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13878 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13879 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13881 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13883 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13884 Disable pc relative literal load irrespective of
13885 TARGET_FIX_ERR_A53_84341 for default.
13887 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13889 * config/sparc/sparc.c (output_return): Output the source location of
13890 the insn in the delay slot, if any.
13891 (output_sibcall): Likewise.
13893 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13896 * common.opt (-static-pie): New alias.
13897 (shared): Negate static-pie.
13898 (-no-pie): Update help text.
13900 (static-pie): New option.
13901 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13902 -static-pie support.
13903 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13904 (LINK_EH_SPEC): Likewise.
13905 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13906 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13907 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13908 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13909 (init_gcc_specs): Likewise.
13910 (init_spec): Likewise.
13911 (display_help): Update help message for -pie.
13912 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13915 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13917 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13918 (movdi_aarch64): Likewise.
13919 (movti_aarch64): Likewise.
13921 2017-09-12 Simon Wright <simon@pushface.org>
13924 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13925 calculation of the minor version, always output as 0.
13927 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13930 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13931 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13932 on it early, rather than manual conversion late. For
13933 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13934 instead of performing manual conversion.
13936 2017-09-12 Carl Love <cel@us.ibm.com>
13938 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13939 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13941 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13942 VMULOSW): Add definitions.
13943 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13944 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13945 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13947 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13949 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13951 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13953 (movti_aarch64): Likewise.
13954 (movdf_aarch64): Likewise.
13955 (movtf_aarch64): Likewise.
13956 (load_pairdi): Likewise.
13957 (store_pairdi): Likewise.
13958 (load_pairdf): Likewise.
13959 (store_pairdf): Likewise.
13960 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13961 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13962 (ldr_got_small_<mode>): Likewise.
13963 (ldr_got_small_28k_<mode>): Likewise.
13964 (ldr_got_tiny): Likewise.
13965 * config/aarch64/iterators.md (ldst_sz): New.
13966 (ldpstp_sz): Likewise.
13967 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13969 (thunderx_load): Split load_8 to load_16.
13970 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13972 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13973 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13974 (xgene1_store_pair): Split store_8 to store_16.
13975 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13976 (falkor_st_0_st_sd): Split store_8 to store_16.
13978 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13980 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13981 and store1/2/3/4 to store_4/8/12/16.
13982 * config/aarch64/aarch64.md: Update for rename.
13983 * config/arm/arm.md: Likewise.: Likewise.
13984 * config/arm/arm.c: Likewise.
13985 * config/arm/thumb1.md: Likewise.
13986 * config/arm/thumb2.md: Likewise.
13987 * config/arm/vfp.md: Likewise.
13988 * config/arm/arm-generic.md: Likewise.
13989 * config/arm/arm1020e.md: Likewise.
13990 * config/arm/arm1026ejs.md: Likewise.
13991 * config/arm/arm1136jfs.md: Likewise.
13992 * config/arm/arm926ejs.md: Likewise.
13993 * config/arm/cortex-a15.md: Likewise.
13994 * config/arm/cortex-a17.md: Likewise.
13995 * config/arm/cortex-a5.md: Likewise.
13996 * config/arm/cortex-a53.md: Likewise.
13997 * config/arm/cortex-a57.md: Likewise.
13998 * config/arm/cortex-a7.md: Likewise.
13999 * config/arm/cortex-a8.md: Likewise.
14000 * config/arm/cortex-a9.md: Likewise.
14001 * config/arm/cortex-m4.md: Likewise.
14002 * config/arm/cortex-m7.md: Likewise.
14003 * config/arm/cortex-r4.md: Likewise.
14004 * config/arm/exynos-m1.md: Likewise.
14005 * config/arm/fa526.md: Likewise.
14006 * config/arm/fa606te.md: Likewise.
14007 * config/arm/fa626te.md: Likewise.
14008 * config/arm/fa726te.md: Likewise.
14009 * config/arm/fmp626.md: Likewise.
14010 * config/arm/iwmmxt.md: Likewise.
14011 * config/arm/ldmstm.md: Likewise.
14012 * config/arm/marvell-pj4.md: Likewise.
14013 * config/arm/xgene1.md: Likewise.
14014 * config/aarch64/thunderx.md: Likewise.
14015 * config/aarch64/thunderx2t99.md: Likewise.
14016 * config/aarch64/falkor.md: Likewise.
14018 2017-09-12 Martin Liska <mliska@suse.cz>
14020 * attribs.c (private_lookup_attribute): New function.
14021 * attribs.h (private_lookup_attribute): Declared here.
14022 (lookup_attribute): Called from this place.
14024 2017-09-12 Richard Biener <rguenther@suse.de>
14026 PR tree-optimization/82157
14027 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14028 stmts with side-effects.
14030 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14031 Alan Hayward <alan.hayward@arm.com>
14032 David Sherwood <david.sherwood@arm.com>
14034 * target.def (hard_regno_nregs): New hook.
14035 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14036 * targhooks.h (default_hard_regno_nregs): Declare.
14037 * targhooks.c (default_hard_regno_nregs): New function.
14038 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14039 (TARGET_HARD_REGNO_NREGS): ...this hook.
14040 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14041 (CLASS_MAX_NREGS): Likewise.
14042 * doc/tm.texi: Regenerate.
14043 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14044 instead of HARD_REGNO_NREGS.
14045 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14046 HARD_REGNO_NREGS in the comment.
14047 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14048 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14049 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14050 Return an unsigned int.
14051 (TARGET_HARD_REGNO_NREGS): Redefine.
14052 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14053 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14054 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14055 (arc_hard_regno_nregs): New function.
14056 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14057 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14058 (arm_hard_regno_nregs): New function.
14059 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14060 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14061 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14062 (TARGET_HARD_REGNO_NREGS): Redefine.
14063 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14064 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14065 (HARD_REGNO_NREGS): Delete.
14066 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14067 (cr16_hard_regno_nregs): New function.
14068 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14069 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14070 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14071 (cris_hard_regno_nregs): New function.
14072 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14073 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14074 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14075 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14076 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14077 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14078 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14079 (frv_hard_regno_nregs): Make static. Take and return an
14081 (frv_class_max_nregs): Remove outdated copy of documentation.
14082 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14083 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14084 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14085 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14086 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14087 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14088 (TARGET_HARD_REGNO_NREGS): Redefine.
14089 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14090 (CLASS_MAX_NREGS): Update comment.
14091 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14092 (ia64_hard_regno_nregs): New function.
14093 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14094 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14095 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14096 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14097 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14099 (m32c_hard_regno_nregs): Likewise. Make static.
14100 (TARGET_HARD_REGNO_NREGS): Redefine.
14101 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14102 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14103 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14104 (m68k_hard_regno_nregs): New function.
14105 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14106 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14107 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14108 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14109 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14110 Take and return an unsigned int.
14111 (TARGET_HARD_REGNO_NREGS): Redefine.
14112 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14113 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14114 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14115 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14116 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14117 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14118 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14119 (msp430_hard_regno_nregs): Make static. Take and return an
14121 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14122 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14123 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14124 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14125 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14126 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14127 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14128 (TARGET_HARD_REGNO_NREGS): Redefine.
14129 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14130 (PA_HARD_REGNO_NREGS): ...this.
14131 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14132 (PA_HARD_REGNO_NREGS): ...this.
14133 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14134 (pa_hard_regno_nregs): New function.
14135 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14136 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14137 (pdp11_hard_regno_nregs): New function.
14138 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14139 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14140 (rs6000_hard_regno_nregs_hook): New function.
14141 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14142 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14143 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14144 Take and return an unsigned int. Move earlier in file.
14145 (TARGET_HARD_REGNO_NREGS): Redefine.
14146 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14147 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14148 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14149 (rl78_hard_regno_nregs): Make static. Take and return an
14151 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14152 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14153 (rs6000_hard_regno_nregs_hook): New function.
14154 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14155 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14156 (TARGET_HARD_REGNO_NREGS): Redefine.
14157 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14158 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14159 instead of HARD_REGNO_NREGS.
14160 (s390_hard_regno_nregs): New function.
14161 (s390_hard_regno_mode_ok): Add comment from s390.h.
14162 (TARGET_HARD_REGNO_NREGS): Redefine.
14163 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14164 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14165 (sh_hard_regno_nregs): New function.
14166 (sh_pass_in_reg_p): Use it.
14167 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14168 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14169 (sparc_hard_regno_nregs): New function.
14170 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14171 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14172 (spu_function_arg_advance): Use it, supplying a valid register number.
14173 (TARGET_HARD_REGNO_NREGS): Redefine.
14174 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14175 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14176 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14177 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14178 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14179 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14180 (CLASS_MAX_NREGS): Remove copy of old documentation.
14181 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14182 (visium_hard_regno_nregs): New function.
14183 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14184 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14185 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14186 xtensa_hard_regno_nregs): New function.
14187 * system.h (HARD_REGNO_NREGS): Poison.
14189 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14191 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14192 hard_regno_nregs instead of HARD_REGNO_NREGS.
14193 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14194 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14195 (c6x_expand_epilogue): Likewise.
14196 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14197 (frv_read_iacc_argument): Likewise.
14198 * config/sh/sh.c: Include regs.h.
14199 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14200 (regs_used): Likewise.
14201 (output_stack_adjust): Likewise.
14202 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14203 * expmed.c: Include regs.h.
14204 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14205 * ree.c: Include regs.h.
14206 (combine_reaching_defs): Use hard_regno_nregs instead of
14208 (add_removable_extension): Likewise.
14210 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14212 * regs.h (hard_regno_nregs): Turn into a function.
14213 (end_hard_regno): Update accordingly.
14214 * caller-save.c (setup_save_areas): Likewise.
14215 (save_call_clobbered_regs): Likewise.
14216 (replace_reg_with_saved_mem): Likewise.
14217 (insert_restore): Likewise.
14218 (insert_save): Likewise.
14219 * combine.c (can_change_dest_mode): Likewise.
14220 (move_deaths): Likewise.
14221 (distribute_notes): Likewise.
14222 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14223 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14224 (rs6000_split_multireg_move): Likewise.
14225 (rs6000_register_move_cost): Likewise.
14226 (rs6000_memory_move_cost): Likewise.
14227 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14228 (rs6000_split_multireg_move): Likewise.
14229 (rs6000_register_move_cost): Likewise.
14230 (rs6000_memory_move_cost): Likewise.
14231 * cselib.c (cselib_reset_table): Likewise.
14232 (cselib_lookup_1): Likewise.
14233 * emit-rtl.c (set_mode_and_regno): Likewise.
14234 * function.c (aggregate_value_p): Likewise.
14235 * ira-color.c (setup_profitable_hard_regs): Likewise.
14236 (check_hard_reg_p): Likewise.
14237 (calculate_saved_nregs): Likewise.
14238 (assign_hard_reg): Likewise.
14239 (improve_allocation): Likewise.
14240 (calculate_spill_cost): Likewise.
14241 * ira-emit.c (modify_move_list): Likewise.
14242 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14243 (ira_hard_reg_in_set_p): Likewise.
14244 * ira.c (setup_reg_mode_hard_regset): Likewise.
14245 (clarify_prohibited_class_mode_regs): Likewise.
14246 (check_allocation): Likewise.
14247 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14248 (lra_setup_reg_renumber): Likewise.
14249 (setup_try_hard_regno_pseudos): Likewise.
14250 (spill_for): Likewise.
14251 (assign_hard_regno): Likewise.
14252 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14253 * lra-constraints.c (in_class_p): Likewise.
14254 (lra_constraint_offset): Likewise.
14255 (simplify_operand_subreg): Likewise.
14256 (lra_constraints): Likewise.
14257 (split_reg): Likewise.
14258 (split_if_necessary): Likewise.
14259 (invariant_p): Likewise.
14260 (inherit_in_ebb): Likewise.
14261 * lra-lives.c (process_bb_lives): Likewise.
14262 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14263 (get_hard_regs): Likewise.
14264 (do_remat): Likewise.
14265 * lra-spills.c (assign_spill_hard_regs): Likewise.
14266 * mode-switching.c (create_pre_exit): Likewise.
14267 * postreload.c (reload_combine_recognize_pattern): Likewise.
14268 * recog.c (peep2_find_free_register): Likewise.
14269 * regcprop.c (kill_value_regno): Likewise.
14270 (set_value_regno): Likewise.
14271 (copy_value): Likewise.
14272 (maybe_mode_change): Likewise.
14273 (find_oldest_value_reg): Likewise.
14274 (copyprop_hardreg_forward_1): Likewise.
14275 * regrename.c (check_new_reg_p): Likewise.
14276 (regrename_do_replace): Likewise.
14277 * reload.c (push_reload): Likewise.
14278 (combine_reloads): Likewise.
14279 (find_dummy_reload): Likewise.
14280 (operands_match_p): Likewise.
14281 (find_reloads): Likewise.
14282 (find_equiv_reg): Likewise.
14283 (reload_adjust_reg_for_mode): Likewise.
14284 * reload1.c (count_pseudo): Likewise.
14285 (count_spilled_pseudo): Likewise.
14286 (find_reg): Likewise.
14287 (clear_reload_reg_in_use): Likewise.
14288 (free_for_value_p): Likewise.
14289 (allocate_reload_reg): Likewise.
14290 (choose_reload_regs): Likewise.
14291 (reload_adjust_reg_for_temp): Likewise.
14292 (emit_reload_insns): Likewise.
14293 (delete_output_reload): Likewise.
14294 * rtlanal.c (subreg_get_info): Likewise.
14295 * sched-deps.c (sched_analyze_reg): Likewise.
14296 * sel-sched.c (init_regs_for_mode): Likewise.
14297 (mark_unavailable_hard_regs): Likewise.
14298 (choose_best_reg_1): Likewise.
14299 (verify_target_availability): Likewise.
14300 * valtrack.c (dead_debug_insert_temp): Likewise.
14301 * var-tracking.c (track_loc_p): Likewise.
14302 (emit_note_insn_var_location): Likewise.
14303 * varasm.c (make_decl_rtl): Likewise.
14304 * reginfo.c (choose_hard_reg_mode): Likewise.
14305 (init_reg_modes_target): Refer directly to
14306 this_target_regs->x_hard_regno_nregs.
14308 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14310 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14311 instead of hard_regno_nregs.
14313 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14315 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14316 end_hard_regno instead of hard_regno_nregs.
14317 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14318 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14319 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14320 * ira-color.c (improve_allocation): Likewise.
14321 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14322 * lra-lives.c (mark_regno_live): Likewise.
14323 (mark_regno_dead): Likewise.
14324 * lra-remat.c (operand_to_remat): Likewise.
14325 * lra.c (collect_non_operand_hard_regs): Likewise.
14326 * postreload.c (reload_combine_note_store): Likewise.
14327 (move2add_valid_value_p): Likewise.
14328 * reload.c (regno_clobbered_p): Likewise.
14330 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14332 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14334 * config/v850/v850.c (v850_reorg): Likewise.
14335 * reload.c (refers_to_regno_for_reload_p): Likewise.
14336 (find_equiv_reg): Likewise.
14337 * reload1.c (reload_reg_reaches_end_p): Likewise.
14339 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14341 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14343 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14344 * config/arm/arm.c (output_move_neon): Likewise.
14345 (arm_attr_length_move_neon): Likewise.
14346 (neon_split_vcombine): Likewise.
14347 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14348 (c6x_mark_reg_written): Likewise.
14349 (c6x_dwarf_register_span): Likewise.
14350 * config/i386/i386.c (ix86_save_reg): Likewise.
14351 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14352 (rws_access_reg): Likewise.
14353 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14354 * mode-switching.c (create_pre_exit): Likewise.
14355 * ree.c (combine_reaching_defs): Likewise.
14356 (add_removable_extension): Likewise.
14357 * regcprop.c (find_oldest_value_reg): Likewise.
14358 (copyprop_hardreg_forward_1): Likewise.
14359 * reload.c (reload_inner_reg_of_subreg): Likewise.
14360 (push_reload): Likewise.
14361 (combine_reloads): Likewise.
14362 (find_dummy_reload): Likewise.
14363 (reload_adjust_reg_for_mode): Likewise.
14364 * reload1.c (find_reload_regs): Likewise.
14365 (forget_old_reloads_1): Likewise.
14366 (reload_reg_free_for_value_p): Likewise.
14367 (reload_adjust_reg_for_temp): Likewise.
14368 (emit_reload_insns): Likewise.
14369 (delete_output_reload): Likewise.
14370 * sel-sched.c (choose_best_reg_1): Likewise.
14371 (choose_best_pseudo_reg): Likewise.
14373 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14374 Alan Hayward <alan.hayward@arm.com>
14375 David Sherwood <david.sherwood@arm.com>
14377 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14378 * target.def (slow_unaligned_access): New hook.
14379 * targhooks.h (default_slow_unaligned_access): Declare.
14380 * targhooks.c (default_slow_unaligned_access): New function.
14381 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14382 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14383 * doc/tm.texi: Regenerate.
14384 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14385 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14386 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14388 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14389 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14391 (rs6000_slow_unaligned_access): New function.
14392 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14393 (expand_block_compare): Likewise.
14394 (expand_strn_compare): Likewise.
14395 (rs6000_rtx_costs): Likewise.
14396 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14397 (riscv_slow_unaligned_access): Likewise.
14398 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14399 (riscv_slow_unaligned_access_p): ...this and make static.
14400 (riscv_option_override): Update accordingly.
14401 (riscv_slow_unaligned_access): New function.
14402 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14403 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14404 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14405 (rs6000_slow_unaligned_access): New function.
14406 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14407 (rs6000_rtx_costs): Likewise.
14408 * config/rs6000/rs6000-string.c (expand_block_compare)
14409 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14410 of SLOW_UNALIGNED_ACCESS.
14411 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14412 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14413 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14414 of SLOW_UNALIGNED_ACCESS.
14415 * expmed.c (simple_mem_bitfield_p): Likewise.
14416 * expr.c (alignment_for_piecewise_move): Likewise.
14417 (emit_group_load_1): Likewise.
14418 (emit_group_store): Likewise.
14419 (copy_blkmode_from_reg): Likewise.
14420 (emit_push_insn): Likewise.
14421 (expand_assignment): Likewise.
14422 (store_field): Likewise.
14423 (expand_expr_real_1): Likewise.
14424 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14425 * lra-constraints.c (simplify_operand_subreg): Likewise.
14426 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14427 * gimple-ssa-store-merging.c: Likewise in block comment at start
14429 * tree-ssa-strlen.c: Include target.h.
14430 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14431 of SLOW_UNALIGNED_ACCESS.
14432 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14434 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14436 PR rtl-optimization/82185
14437 * expmed.c (emit_store_flag_int): Only test tem if it has been
14440 2017-09-12 Richard Biener <rguenther@suse.de>
14442 PR middle-end/82149
14443 * match.pd ((FTYPE) N CMP CST): Fix typo.
14445 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14447 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14449 (mips_near_type_p): Add 'short_call' attribute as a synonym
14451 * doc/extend.texi (short_call): Document new function attribute.
14453 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14456 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14457 assertion check that in the condition.
14458 (get_atomic_generic_size): Likewise. Before testing if parameter
14459 has pointer type, if it has array type, call for C++
14460 default_conversion to perform array-to-pointer conversion.
14462 2017-09-12 Richard Biener <rguenther@suse.de>
14464 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14465 for operations we cannot scalarize.
14467 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14469 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14470 vectors heap vectors. Clean up comments.
14471 Make visited_bbs a reference.
14472 (profitable_jump_thread_path): Make GC
14473 vectors heap vectors. Clean up comments.
14475 (convert_and_register_jump_thread_path): Make GC vectors heap
14477 (check_subpath_and_update_thread_path): Same. Clean up comments.
14478 Make visited_bbs a reference.
14479 (handle_phi): Abstract common code to to
14480 register_jump_thread_path_if_profitable.
14481 Rename VAR_BB to DEF_BB.
14483 Make GC vectors heap vectors.
14484 Make visited_bbs a reference.
14485 (handle_assignment): Same.
14486 (register_jump_thread_path_if_profitable): New.
14487 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14489 Make GC vectors heap vectors. Clean up comments.
14490 Make visited_bbs a reference.
14491 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14492 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14495 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14498 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14499 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14501 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14503 Revert r251800 and r251799.
14505 2017-09-11 Martin Jambor <mjambor@suse.cz>
14508 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14509 arguments in advance.
14510 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14511 use it to find predecessor edges.
14512 (naive_outof_ssa): Collect vector of predecessors.
14514 2017-09-08 Jason Merrill <jason@redhat.com>
14516 PR c++/70029 - ICE with ref-qualifier and -flto
14517 * langhooks.h (struct lang_hooks_for_types): Add
14518 copy_lang_qualifiers.
14519 * attribs.c (build_type_attribute_qual_variant): Use it.
14520 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14522 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14523 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14525 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14528 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14529 (*mulsi3_sp64): New instruction.
14530 (mulsi3): New expander.
14532 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14534 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14536 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14538 * sancov.c: Include memmodel.h.
14540 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14543 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14546 2017-09-07 Carl Love <cel@us.ibm.com>
14548 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14549 the sldi instruction.
14551 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14553 * sancov.c: Include tm_p.h.
14555 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14558 * output.h (switch_to_other_text_partition): New declaration.
14559 * varasm.c (switch_to_other_text_partition): New function.
14560 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14561 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14562 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14563 to the other text partition before emitting LCL label and switch back
14564 after emitting the word after it.
14566 2017-09-07 Richard Biener <rguenther@suse.de>
14568 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14569 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14570 critical edges here, after loop init.
14571 (pass_data_pre): Remove PROP_no_crit_edges flags.
14572 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14573 for valueization of call args to avoid leaking VN_TOP.
14574 (visit_use): Assert we do not visit default defs.
14575 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14576 Use error_mark_node to more easily detect leaking VN_TOP.
14577 All default-defs are varying, not VN_TOP. Mark them visited.
14578 (run_scc_vn): Make code match comment.
14580 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14582 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14583 OPTION_MASK_FLOAT128_KEYWORD.
14584 (POWERPC_MASKS): Likewise.
14585 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14586 support for the -mfloat128-type option, and make -mfloat128
14587 default on PowerPC Linux systems. Define or undefine
14588 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14589 Define __float128 to be __ieee128 if IEEE 128-bit support is
14590 enabled, or undefine it.
14591 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14592 Delete defining __FLOAT128_TYPE__.
14593 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14594 -mfloat128-type option and make -mfloat128 default on PowerPC
14596 (TARGET_FLOAT128_TYPE): Likewise.
14597 (-mfloat128-type): Likewise.
14598 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14599 Delete the -mfloat128-type option and make -mfloat128 default on
14600 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14601 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14602 128-bit floating point is enabled. Change tests from using
14603 -mfloat128-type to -mfloat128.
14604 (rs6000_mangle_type): Use the correct mangling for the __float128
14605 type even if normal long double is restricted to 64-bits.
14606 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14608 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14609 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14610 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14612 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14613 documentation for -mfloat128.
14615 2017-09-06 Olivier Hainque <hainque@adacore.com>
14617 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14619 2017-09-06 Wish Wu <wishwu007@gmail.com>
14620 Jakub Jelinek <jakub@redhat.com>
14622 * asan.c (initialize_sanitizer_builtins): Add
14623 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14624 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14625 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14626 BT_FN_VOID_UINT64_PTR variables.
14627 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14628 (BT_FN_VOID_UINT16_UINT16): Likewise.
14629 (BT_FN_VOID_UINT32_UINT32): Likewise.
14630 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14631 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14632 (BT_FN_VOID_UINT64_PTR): Likewise.
14633 * common.opt (flag_sanitize_coverage): New variable.
14634 (fsanitize-coverage=trace-pc): Remove.
14635 (fsanitize-coverage=): Add.
14636 * flag-types.h (enum sanitize_coverage_code): New enum.
14637 * fold-const.c (fold_range_test): Disable non-short-circuit
14638 optimization if flag_sanitize_coverage.
14639 (fold_truth_andor): Likewise.
14640 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14641 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14642 (coverage_sanitizer_opts): New array.
14643 (get_closest_sanitizer_option): Add OPTS argument, handle also
14644 OPT_fsanitize_coverage_.
14645 (parse_sanitizer_options): Adjusted to also handle
14646 OPT_fsanitize_coverage_.
14647 (common_handle_option): Add OPT_fsanitize_coverage_.
14648 * sancov.c (instrument_comparison, instrument_switch): New function.
14649 (sancov_pass): Add trace-cmp support.
14650 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14651 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14652 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14653 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14654 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14655 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14656 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14657 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14658 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14659 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14661 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14663 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14664 error. Only quit immediately if parsing is complete.
14665 (BEGIN): Initialize fatal_err and parse_done.
14666 (begin fpu, end fpu): Check number of arguments.
14667 (begin arch, end arch): Likewise.
14668 (begin cpu, end cpu): Likewise.
14669 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14670 (optalias): Likewise.
14672 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14674 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14675 * config/arm/arm-isa.h: Delete. Move definitions to ...
14676 * arm-cpus.in: ... here. Use new feature and fgroup values.
14677 * config/arm/arm.c (arm_option_override): Use lower case for feature
14679 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14680 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14681 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14683 (print_isa_bits_for): New function.
14684 (gen_isa): New function.
14685 (gen_comm_data): Use print_isa_bits_for.
14686 (define feature): New keyword.
14687 (define fgroup): New keyword.
14688 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14689 (arm-isa.h): Add rule to generate file.
14690 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14691 case for feature bit names.
14693 2017-09-06 Richard Biener <rguenther@suse.de>
14695 * tree-ssa-pre.c (NECESSARY): Remove.
14696 (create_expression_by_pieces): Do not touch pass-local flags.
14697 (insert_into_preds_of_block): Likewise.
14698 (do_pre_regular_insertion): Likewise.
14699 (eliminate_insert): Likewise.
14700 (eliminate_dom_walker::before_dom_children): Likewise.
14701 (fini_eliminate): Do not look at inserted_exprs.
14702 (mark_operand_necessary): Remove.
14703 (remove_dead_inserted_code): Replace with simple work-list
14704 algorithm based on inserted_exprs and SSA uses.
14705 (pass_pre::execute): Re-order fini_eliminate and
14706 remove_dead_inserted_code.
14708 2017-09-06 Olivier Hainque <hainque@adacore.com>
14710 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
14711 for VxWorks 7. Adjust surrounding comments.
14713 2017-09-06 Richard Biener <rguenther@suse.de>
14715 * gimple-ssa-strength-reduction.c
14716 (find_candidates_dom_walker::before_dom_children): Also allow
14719 2017-09-06 Richard Biener <rguenther@suse.de>
14721 PR tree-optimization/82108
14722 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14723 for gap in the non-permutation SLP case.
14725 2017-09-06 Martin Jambor <mjambor@suse.cz>
14727 PR tree-optimization/82078
14728 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14729 add_access_to_work_queue...
14730 (build_accesses_from_assign): ...here.
14731 (propagate_all_subaccesses): Make sure racc is the group
14732 representative, if there is one.
14734 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14736 PR middle-end/82095
14737 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14740 2017-09-06 Richard Biener <rguenther@suse.de>
14742 * gimple-ssa-strength-reduction.c
14743 (find_candidates_dom_walker::before_doom_children): Use a
14744 type and not a mode check.
14746 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14749 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14751 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14753 2017-09-05 Jeff Law <law@redhat.com>
14755 PR tree-optimization/64910
14756 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14757 cases where we have 3 or more operands.
14759 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14761 PR middle-end/81768
14762 * omp-low.c (lower_omp_for): Recompute tree invariant if
14763 gimple_omp_for_initial/final is ADDR_EXPR.
14765 PR middle-end/81768
14766 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14767 into gimple val before gimplification fo the COND_EXPR.
14769 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14771 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14772 REGION_COPY argument.
14773 (thread_through_all_blocks): Remove unused argument to
14774 duplicate_thread_path.
14776 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14777 Alan Hayward <alan.hayward@arm.com>
14778 David Sherwood <david.sherwood@arm.com>
14780 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14781 Take a scalar_mode rather than a machine_mode.
14782 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14783 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14784 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14785 (aarch64_gen_adjusted_ldpstp): Likewise.
14786 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14788 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14789 Alan Hayward <alan.hayward@arm.com>
14790 David Sherwood <david.sherwood@arm.com>
14792 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14793 Take a scalar_int_mode instead of a machine_mode.
14794 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14795 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14796 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14797 (aarch64_simd_attr_length_rglist): Delete.
14798 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14799 a scalar_int_mode instead of a machine_mode.
14800 (aarch64_add_offset): Likewise.
14801 (aarch64_internal_mov_immediate): Likewise
14802 (aarch64_add_constant_internal): Likewise.
14803 (aarch64_add_constant): Likewise.
14804 (aarch64_movw_imm): Likewise.
14805 (aarch64_rtx_arith_op_extract_p): Likewise.
14806 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14807 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14808 Remove assert that the mode isn't a vector.
14809 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14810 (aarch64_expand_mov_immediate): Update calls after above changes.
14811 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14812 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14813 (aarch64_move_imm): Likewise.
14814 (aarch64_can_const_movi_rtx_p): Likewise.
14815 (aarch64_strip_extend): Likewise.
14816 (aarch64_extr_rtx_p): Likewise.
14817 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14818 a CONST_INT when the mode parameter is VOIDmode.
14819 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14821 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14823 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14824 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14825 (bitwise_type_for_mode): Update accordingly.
14827 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14829 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14830 * stor-layout.c (mode_for_size_tree): Likewise.
14831 (mode_for_array): Update accordingly.
14832 (layout_decl): Likewise.
14833 (compute_record_mode): Likewise. Only set the mode once.
14835 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14837 * target.def (get_mask_mode): Change return type to opt_mode.
14839 * doc/tm.texi: Regenerate.
14840 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14841 * targhooks.c (default_get_mask_mode): Likewise.
14842 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14843 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14844 targetm.get_mask_mode.
14845 * tree.c (build_truth_vector_type): Likewise.
14847 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14849 * machmode.h (mode_for_vector): Return an opt_mode.
14850 * stor-layout.c (mode_for_vector): Likewise.
14851 (mode_for_int_vector): Update accordingly.
14852 (layout_type): Likewise.
14853 * config/i386/i386.c (emit_memmov): Likewise.
14854 (ix86_expand_set_or_movmem): Likewise.
14855 (ix86_expand_vector_init): Likewise.
14856 (ix86_get_mask_mode): Likewise.
14857 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14859 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14860 * expmed.c (extract_bit_field_1): Likewise.
14861 * expr.c (expand_expr_real_2): Likewise.
14862 * optabs-query.c (can_vec_perm_p): Likewise.
14863 (can_vec_mask_load_store_p): Likewise.
14864 * optabs.c (expand_vec_perm): Likewise.
14865 * targhooks.c (default_get_mask_mode): Likewise.
14866 * tree-vect-stmts.c (vectorizable_store): Likewise.
14867 (vectorizable_load): Likewise.
14868 (get_vectype_for_scalar_type_and_size): Likewise.
14870 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14872 * machmode.h (mode_for_int_vector): New function.
14873 * stor-layout.c (mode_for_int_vector): Likewise.
14874 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14875 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14876 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14877 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14878 (s390_expand_vcond): Likewise.
14880 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14882 * machmode.h (opt_machine_mode): New type.
14883 (opt_mode<T>): Allow construction from anything that can be
14885 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14886 (mode_for_size): Return an opt_machine_mode.
14887 * stor-layout.c (mode_for_size): Likewise.
14888 (mode_for_size_tree): Update call accordingly.
14889 (bitwise_mode_for_mode): Likewise.
14890 (make_fract_type): Likewise.
14891 (make_accum_type): Likewise.
14892 * caller-save.c (replace_reg_with_saved_mem): Update call
14894 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14895 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14896 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14897 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14898 * expmed.c (extract_bit_field_1): Likewise.
14899 * reload.c (get_secondary_mem): Likewise.
14900 * varasm.c (assemble_integer): Likewise.
14901 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14904 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14906 * machmode.h (decimal_float_mode_for_size): New function.
14907 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14908 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14909 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14910 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14911 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14912 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14914 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14916 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14917 (get_builtin_sync_mode): Likewise.
14918 (expand_ifn_atomic_compare_exchange): Likewise.
14919 (expand_builtin_atomic_clear): Likewise.
14920 (expand_builtin_atomic_test_and_set): Likewise.
14921 (fold_builtin_atomic_always_lock_free): Likewise.
14922 * calls.c (compute_argument_addresses): Likewise.
14923 (emit_library_call_value_1): Likewise.
14924 (store_one_arg): Likewise.
14925 * combine.c (combine_instructions): Likewise.
14926 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14927 * config/arm/arm.c (arm_function_value): Likewise.
14928 (aapcs_allocate_return_reg): Likewise.
14929 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14930 * config/i386/i386.c (construct_container): Likewise.
14931 (ix86_gimplify_va_arg): Likewise.
14932 (ix86_expand_sse_cmp): Likewise.
14933 (emit_memmov): Likewise.
14934 (emit_memset): Likewise.
14935 (expand_small_movmem_or_setmem): Likewise.
14936 (ix86_expand_pextr): Likewise.
14937 (ix86_expand_pinsr): Likewise.
14938 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14939 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14941 * config/mips/mips.c (mips_function_value_1) Likewise.
14942 (mips_block_move_straight): Likewise.
14943 (mips_expand_ins_as_unaligned_store): Likewise.
14944 * config/powerpcspe/powerpcspe.c
14945 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14946 (rs6000_darwin64_record_arg_flush): Likewise.
14947 * config/rs6000/rs6000.c
14948 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14949 (rs6000_darwin64_record_arg_flush): Likewise.
14950 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14951 (sparc_function_value_1): Likewise.
14952 * config/spu/spu.c (adjust_operand): Likewise.
14953 (spu_emit_branch_or_set): Likewise.
14954 (arith_immediate_p): Likewise.
14955 * emit-rtl.c (gen_lowpart_common): Likewise.
14956 * expr.c (expand_expr_real_1): Likewise.
14957 * function.c (assign_parm_setup_block): Likewise.
14958 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14959 * reload1.c (alter_reg): Likewise.
14960 * stor-layout.c (mode_for_vector): Likewise.
14961 (layout_type): Likewise.
14963 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14965 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14966 (spu_convert_move): Likewise.
14967 * lower-subreg.c (resolve_simple_move): Likewise.
14969 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14972 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14973 define_insn to a define_expand.
14974 (altivec_vsum2sws_direct): New define_insn.
14975 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14977 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14979 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14982 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14987 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14988 realign stack if stack alignment needed is less than incoming
14991 2017-09-05 Marek Polacek <polacek@redhat.com>
14994 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14997 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14999 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15001 2017-09-05 Richard Biener <rguenther@suse.de>
15003 PR tree-optimization/82084
15004 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15006 2017-09-05 Richard Biener <rguenther@suse.de>
15008 PR tree-optimization/82102
15009 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15011 2017-09-05 Martin Liska <mliska@suse.cz>
15013 PR tree-optimization/82032
15014 * tree-cfg.c (generate_range_test): New function.
15015 * tree-cfg.h (generate_range_test): Declared here.
15016 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15017 (cleanup_control_expr_graph): Use it.
15018 * tree-switch-conversion.c (try_switch_expansion): Remove
15020 (emit_case_nodes): Use generate_range_test.
15022 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15025 * config/i386/i386.md (*<btsc><mode>_mask): Add
15026 TARGET_USE_BT to insn constraint.
15027 (*btr<mode>_mask): Ditto.
15029 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15031 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15032 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15034 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15037 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15038 TARGET_NEON and TARGET_IWMMXT.
15039 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15040 TARGET_NEON and TARGET_IWMMXT.
15041 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15043 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15045 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15046 (ix86_rewrite_tls_address): Ditto.
15047 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15048 (ix86_rewrite_tls_address_1): Ditto.
15049 (ix86_rewrite_tls_address): Ditto.
15050 * config/i386/predicates.md (tls_address_pattern): New predicate.
15051 * config/i386/i386.md (TLS address splitter): New splitter.
15053 2017-09-04 Richard Biener <rguenther@suse.de>
15055 PR tree-optimization/82084
15056 * fold-const.h (can_native_encode_string_p): Declare.
15057 * fold-const.c (can_native_encode_string_p): Factor out from ...
15058 (native_encode_string): ... here.
15059 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15060 vectorizing stores from constants we later cannot handle.
15062 2017-09-04 Marek Polacek <polacek@redhat.com>
15065 * doc/invoke.texi: Update -Wtautological-compare documentation.
15067 2017-09-04 Jeff Law <law@redhat.com>
15069 PR tree-optimization/64910
15070 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15071 swap the first and last operand if the last is a constant.
15073 2017-09-04 Marek Polacek <polacek@redhat.com>
15076 * convert.c (do_narrow): When sanitizing signed integer overflows,
15077 bail out for signed types.
15078 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15080 2017-09-04 Richard Biener <rguenther@suse.de>
15082 PR tree-optimization/82060
15083 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15084 Move devirtualization after stmt folding and before EH/AB/noreturn
15085 cleanup to get the stmt refs canonicalized. Use a bool instead
15086 of gimple_modified_p since that doesn't work for NOPs. Schedule
15087 NOPs generated by folding for removal.
15089 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15090 Alan Hayward <alan.hayward@arm.com>
15091 David Sherwood <david.sherwood@arm.com>
15093 * coretypes.h (pad_direction): New enum.
15094 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15095 (FUNCTION_ARG_PADDING): Likewise.
15096 * target.def (function_arg_padding): New hook.
15097 * targhooks.h (default_function_arg_padding): Declare.
15098 * targhooks.c (default_function_arg_padding): New function.
15099 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15100 (TARGET_FUNCTION_ARG_PADDING): ...this.
15101 * doc/tm.texi: Regenerate.
15102 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15103 instead of direction.
15104 (compute_argument_addresses): Likewise.
15105 (load_register_parameters): Likewise.
15106 (emit_library_call_value_1): Likewise.
15107 (store_one_arg): Use targetm.calls.function_arg_padding instead
15108 of FUNCTION_ARG_PADDING.
15109 (must_pass_in_stack_var_size_or_pad): Likewise.
15110 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15111 (emit_group_store): Likewise.
15112 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15113 instead of FUNCTION_ARG_PADDING.
15114 (emit_push_insn): Likewise, and propagate enum change throughout
15116 * function.h (direction): Delete.
15117 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15119 * function.c (assign_parm_find_stack_rtl): Likewise.
15120 (assign_parm_setup_block_p): Likewise.
15121 (assign_parm_setup_block): Likewise.
15122 (gimplify_parameters): Likewise.
15123 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15124 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15126 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15127 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15128 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15129 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15130 (aarch64_function_arg_padding): ...this new function.
15131 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15132 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15133 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15134 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15135 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15136 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15137 (arm_pad_arg_upward): Replace with...
15138 (arm_function_arg_padding): ...this new function.
15139 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15141 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15142 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15143 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15144 (ia64_hpux_function_arg_padding): Replace with...
15145 (ia64_function_arg_padding): ...this new function. Use pad_direction
15146 instead of direction. Check for TARGET_HPUX.
15147 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15148 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15149 (iq2000_function_arg_padding): New function.
15150 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15151 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15152 (mips_function_arg_padding): ...this new function.
15153 (mips_pad_reg_upward): Update accordingly.
15154 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15155 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15156 targetm.calls.function_arg_padding.
15157 (FUNCTION_ARG_PADDING): Delete.
15158 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15159 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15160 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15161 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15162 (nios2_block_reg_padding): Return pad_direction instead of direction.
15163 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15164 instead of direction.
15165 (nios2_function_arg_padding): Likewise. Make static.
15166 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15167 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15168 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15169 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15170 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15171 (pa_function_arg_padding): Make static. Return pad_direction instead
15173 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15174 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15175 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15176 instead of direction. Use targetm.calls.function_arg_padding.
15177 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15178 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15179 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15180 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15181 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15183 (function_arg_padding): Rename to...
15184 (rs6000_function_arg_padding): ...this. Make static. Return
15185 pad_direction instead of direction.
15186 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15187 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15188 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15189 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15190 instead of direction. Use targetm.calls.function_arg_padding.
15191 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15192 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15193 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15194 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15195 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15196 (function_arg_padding): Rename to...
15197 (rs6000_function_arg_padding): ...this. Make static. Return
15198 pad_direction instead of direction.
15199 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15200 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15201 * config/s390/s390.c (s390_function_arg_padding): New function.
15202 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15203 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15204 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15205 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15206 (function_arg_padding): Rename to...
15207 (sparc_function_arg_padding): ...this. Make static. Return
15208 pad_direction instead of direction.
15209 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15210 * config/spu/spu.c (spu_function_arg_padding): New function.
15211 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15212 * system.h (FUNCTION_ARG_PADDING): Poison.
15214 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15215 Alan Hayward <alan.hayward@arm.com>
15216 David Sherwood <david.sherwood@arm.com>
15218 * target.def (modes_tieable_p): New hook.
15219 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15220 (TARGET_MODES_TIEABLE_P): ...this.
15221 * doc/tm.texi.in: Regenerate.
15222 * hooks.h (hook_bool_mode_mode_true): Declare.
15223 * hooks.c (hook_bool_mode_mode_true): New function.
15224 * combine.c (subst): Use targetm.modes_tieable_p instead of
15226 * dse.c (find_shift_sequence): Likewise.
15227 * expmed.c (extract_low_bits): Likewise.
15228 * lower-subreg.c: Include target.h.
15229 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15231 * rtlanal.c (rtx_cost): Likewise.
15232 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15233 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15234 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15235 (TARGET_MODES_TIEABLE_P): Redefine.
15236 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15237 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15238 (TARGET_MODES_TIEABLE_P): Redefine.
15239 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15240 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15241 (arc_modes_tieable_p): New function.
15242 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15243 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15244 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15245 (arm_modes_tieable_p): Make static.
15246 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15247 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15248 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15249 (TARGET_MODES_TIEABLE_P): Redefine.
15250 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15251 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15252 (TARGET_MODES_TIEABLE_P): Redefine.
15253 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15254 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15255 (cr16_modes_tieable_p): New function.
15256 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15257 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15258 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15259 (TRULY_NOOP_TRUNCATION): Update comment.
15260 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15261 (TRULY_NOOP_TRUNCATION): Update comment.
15262 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15263 (frv_modes_tieable_p): New function.
15264 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15265 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15266 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15267 (TARGET_MODES_TIEABLE_P): Redefine.
15268 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15269 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15270 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15271 (TARGET_MODES_TIEABLE_P): Redefine.
15272 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15273 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15274 (ia64_modes_tieable_p): New function.
15275 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15276 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15277 (iq2000_modes_tieable_p): New function.
15278 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15279 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15280 (lm32_modes_tieable_p): New function.
15281 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15282 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15283 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15284 (TARGET_MODES_TIEABLE_P): Redefine.
15285 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15286 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15287 (m32r_modes_tieable_p): New function.
15288 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15289 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15290 (m68k_modes_tieable_p): New function.
15291 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15292 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15293 (mcore_modes_tieable_p): New function.
15294 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15295 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15297 (TARGET_MODES_TIEABLE_P): Redefine.
15298 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15299 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15300 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15301 (TARGET_MODES_TIEABLE_P): Redefine.
15302 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15303 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15304 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15305 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15306 (mn10300_modes_tieable_p): ...this and make static.
15307 (TARGET_MODES_TIEABLE_P): Redefine.
15308 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15309 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15310 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15311 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15312 (msp430_modes_tieable_p): Make static.
15313 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15314 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15315 (TARGET_MODES_TIEABLE_P): Redefine.
15316 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15317 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15318 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15319 (TARGET_MODES_TIEABLE_P): Redefine.
15320 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15321 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15322 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15323 (TARGET_MODES_TIEABLE_P): Redefine.
15324 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15325 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15326 (pdp11_modes_tieable_p): New function.
15327 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15328 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15329 (rs6000_modes_tieable_p): New function.
15330 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15331 * config/powerpcspe/powerpcspe.md: Update comment.
15332 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15333 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15334 (TARGET_MODES_TIEABLE_P): Redefine.
15335 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15336 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15337 (rl78_modes_tieable_p): New function.
15338 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15339 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15340 (rs6000_modes_tieable_p): New function.
15341 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15342 * config/rs6000/rs6000.md: Update comment.
15343 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15344 * config/rx/rx.c (rx_modes_tieable_p): New function.
15345 (TARGET_MODES_TIEABLE_P): Redefine.
15346 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15347 * config/s390/s390.c (s390_modes_tieable_p): New function.
15348 (TARGET_MODES_TIEABLE_P): Redefine.
15349 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15350 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15351 (sh_modes_tieable_p): New function.
15352 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15353 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15354 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15355 (sparc_modes_tieable_p): Make static.
15356 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15357 * config/spu/spu.c (spu_modes_tieable_p): New function.
15358 (TARGET_MODES_TIEABLE_P): Redefine.
15359 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15360 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15361 (TARGET_MODES_TIEABLE_P): Redefine.
15362 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15363 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15364 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15365 * config/v850/v850.c (v850_modes_tieable_p): New function.
15366 (TARGET_MODES_TIEABLE_P): Redefine.
15367 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15368 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15369 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15370 (visium_modes_tieable_p): New function.
15371 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15372 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15373 (xtensa_modes_tieable_p): New function.
15374 * system.h (MODES_TIEABLE_P): Poison.
15376 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15377 Alan Hayward <alan.hayward@arm.com>
15378 David Sherwood <david.sherwood@arm.com>
15380 * target.def (hard_regno_mode_ok): New hook.
15381 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15382 (TARGET_HARD_REGNO_MODE_OK): ...this.
15383 * doc/tm.texi.in: Regenerate.
15384 * hooks.h (hook_bool_uint_mode_true): Declare.
15385 * hooks.c (hook_bool_uint_mode_true): New function.
15386 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15387 HARD_REGNO_MODE_OK.
15388 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15389 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15390 instead of HARD_REGNO_MODE_OK.
15391 * caller-save.c: Include target.h.
15392 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15393 HARD_REGNO_MODE_OK.
15394 * combine.c (can_combine_p): Likewise.
15395 (combinable_i3pat): Likewise.
15396 (can_change_dest_mode): Likewise.
15397 * expr.c (init_expr_target): Likewise.
15398 (convert_move): Likewise.
15399 (convert_modes): Likewise.
15400 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15401 (setup_prohibited_mode_move_regs): Likewise.
15402 * ira.h (target_ira): Likewise.
15403 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15404 * lra-constraints.c (process_alt_operands): Likewise.
15405 (split_reg): Likewise.
15406 * recog.c (peep2_find_free_register): Likewise.
15407 * ree.c (combine_reaching_defs): Likewise.
15408 * regcprop.c (maybe_mode_change): Likewise.
15409 * reginfo.c (init_reg_sets_1): Likewise.
15410 (choose_hard_reg_mode): Likewise.
15411 (simplifiable_subregs): Likewise.
15412 * regrename.c (check_new_reg_p): Likewise.
15413 * reload.c (find_valid_class): Likewise.
15414 (find_valid_class_1): Likewise.
15415 (reload_inner_reg_of_subreg): Likewise.
15416 (push_reload): Likewise.
15417 (combine_reloads): Likewise.
15418 (find_dummy_reload): Likewise.
15419 (find_reloads): Likewise.
15420 * reload1.c (find_reg): Likewise.
15421 (set_reload_reg): Likewise.
15422 (allocate_reload_reg): Likewise.
15423 (choose_reload_regs): Likewise.
15424 (reload_adjust_reg_for_temp): Likewise.
15425 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15426 (simplify_subreg_regno): Likewise.
15427 * sel-sched.c (init_regs_for_mode): Likewise.
15428 * varasm.c (make_decl_rtl): Likewise.
15429 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15430 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15431 HARD_REGNO_MODE_OK.
15432 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15433 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15434 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15435 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15436 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15437 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15438 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15439 (arc_mode_class): Delete.
15440 (HARD_REGNO_MODE_OK): Delete.
15441 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15442 (arc_hard_regno_mode_ok): Rename old array to...
15443 (arc_hard_regno_mode_ok_modes): ...this.
15444 (arc_conditional_register_usage): Update accordingly.
15445 (arc_mode_class): Make static.
15446 (arc_hard_regno_mode_ok): New function.
15447 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15448 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15449 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15450 (arm_hard_regno_mode_ok): Make static.
15451 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15452 HARD_REGNO_MODE_OK.
15453 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15454 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15455 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15457 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15458 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15459 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15460 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15461 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15462 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15463 * config/bfin/predicates.md (valid_reg_operand): Use
15464 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15465 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15466 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15467 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15468 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15469 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15470 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15471 (cr16_hard_regno_mode_ok): Make static and return a bool.
15472 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15473 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15474 (cris_hard_regno_mode_ok): New function.
15475 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15476 (epiphany_mode_class): Delete.
15477 (HARD_REGNO_MODE_OK): Delete.
15478 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15479 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15480 (hard_regno_mode_ok): Rename to...
15481 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15482 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15483 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15484 HARD_REGNO_MODE_OK.
15485 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15486 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15487 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15488 (frv_hard_regno_mode_ok): Make static and return a bool.
15489 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15490 HARD_REGNO_MODE_OK.
15491 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15492 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15493 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15494 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15496 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15497 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15498 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15499 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15502 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15503 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15504 (ia64_hard_regno_mode_ok): New function.
15505 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15506 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15507 (iq2000_hard_regno_mode_ok): New function.
15508 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15509 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15510 (lm32_hard_regno_mode_ok): New function.
15511 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15512 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15513 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15514 instead of HARD_REGNO_MODE_OK.
15515 (m32c_hard_regno_ok): Rename to...
15516 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15517 (m32c_cannot_change_mode_class): Update accordingly.
15518 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15519 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15520 (m32r_mode_class): Delete.
15521 (HARD_REGNO_MODE_OK): Delete.
15522 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15523 (m32r_hard_regno_mode_ok): Rename to...
15524 (m32r_hard_regno_modes): ...this.
15525 (m32r_mode_class): Make static.
15526 (m32r_hard_regno_mode_ok): New function.
15527 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15528 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15529 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15530 (m68k_hard_regno_mode_ok): Make static.
15531 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15532 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15533 (mcore_hard_regno_mode_ok): New function.
15534 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15535 (HARD_REGNO_MODE_OK): Delete.
15536 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15538 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15539 (microblaze_hard_regno_mode_ok): New function.
15540 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15541 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15542 (mips_hard_regno_mode_ok): Delete.
15543 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15544 (mips_hard_regno_mode_ok_p): ...this and make static.
15545 (mips_hard_regno_mode_ok_p): Rename to...
15546 (mips_hard_regno_mode_ok_uncached): ...this.
15547 (mips_hard_regno_mode_ok): New function.
15548 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15549 of HARD_REGNO_MODE_OK.
15550 (mips_option_override): Update after above name changes.
15551 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15552 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15553 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15554 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15555 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15556 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15557 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15558 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15559 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15560 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15561 (msp430_hard_regno_mode_ok): Make static and return a bool.
15562 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15563 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15564 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15566 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15567 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15568 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15569 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15570 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15571 (PA_HARD_REGNO_MODE_OK): ...this
15572 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15573 (PA_HARD_REGNO_MODE_OK): ...this.
15574 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15575 (pa_hard_regno_mode_ok): New function.
15576 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15577 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15578 (pdp11_hard_regno_mode_ok): New function.
15579 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15580 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15582 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15585 (rs6000_hard_regno_mode_ok): Rename to...
15586 (rs6000_hard_regno_mode_ok_uncached): ...this.
15587 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15588 (rs6000_hard_regno_mode_ok): New function.
15589 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15590 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15591 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15592 (riscv_hard_regno_mode_ok): ...this and make static.
15593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15594 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15595 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15596 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15597 (rl78_hard_regno_mode_ok): Make static and return bool.
15598 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15599 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15601 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15602 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15603 (rs6000_hard_regno_mode_ok): Rename to...
15604 (rs6000_hard_regno_mode_ok_uncached): ...this.
15605 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15606 (rs6000_hard_regno_mode_ok): New function.
15607 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15608 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15609 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15610 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15611 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15612 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15613 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15614 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15615 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15616 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15617 (sh_hard_regno_mode_ok): Make static.
15618 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15619 instead of HARD_REGNO_MODE_OK.
15620 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15621 (sparc_mode_class): Delete.
15622 (HARD_REGNO_MODE_OK): Delete.
15623 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15624 (hard_regno_mode_classes): Make static.
15625 (sparc_mode_class): Likewise.
15626 (sparc_hard_regno_mode_ok): New function.
15627 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15628 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15629 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15632 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15633 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15634 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15635 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15636 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15637 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15638 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15639 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15640 (visium_hard_regno_mode_ok): New function.
15641 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15642 instead of HARD_REGNO_MODE_OK.
15643 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15644 (HARD_REGNO_MODE_OK): Delete.
15645 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15646 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15647 (xtensa_option_override): Update accordingly.
15648 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15649 (xtensa_hard_regno_mode_ok): New function.
15650 * system.h (HARD_REGNO_MODE_OK): Poison.
15652 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15653 Alan Hayward <alan.hayward@arm.com>
15654 David Sherwood <david.sherwood@arm.com>
15656 * target.def (hard_regno_call_part_clobbered): New hook.
15657 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15658 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15659 * doc/tm.texi: Regenerate.
15660 * hooks.h (hook_bool_uint_mode_false): Declare.
15661 * hooks.c (hook_bool_uint_mode_false): New function.
15662 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15663 * cselib.c (cselib_process_insn): Use
15664 targetm.hard_regno_call_part_clobbered instead of
15665 HARD_REGNO_CALL_PART_CLOBBERED.
15666 * ira-conflicts.c (ira_build_conflicts): Likewise.
15667 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15668 * lra-constraints.c (need_for_call_save_p): Likewise.
15669 * lra-lives.c: Include target.h.
15670 (check_pseudos_live_through_calls): Use
15671 targetm.hard_regno_call_part_clobbered instead of
15672 HARD_REGNO_CALL_PART_CLOBBERED.
15673 * regcprop.c: Include target.h.
15674 (copyprop_hardreg_forward_1): Use
15675 targetm.hard_regno_call_part_clobbered instead of
15676 HARD_REGNO_CALL_PART_CLOBBERED.
15677 * reginfo.c (choose_hard_reg_mode): Likewise.
15678 * regrename.c (check_new_reg_p): Likewise.
15679 * reload.c (find_equiv_reg): Likewise.
15680 * reload1.c (emit_reload_insns): Likewise.
15681 * sched-deps.c (deps_analyze_insn): Likewise.
15682 * sel-sched.c (init_regs_for_mode): Likewise.
15683 (mark_unavailable_hard_regs): Likewise.
15684 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15685 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15686 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15688 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15689 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15690 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15692 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15694 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15695 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15696 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
15698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15699 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15700 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
15702 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15703 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
15705 * config/powerpcspe/powerpcspe.c
15706 (rs6000_hard_regno_call_part_clobbered): New function.
15707 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15708 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15709 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
15711 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15712 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15713 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
15715 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15716 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15717 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15719 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15720 Alan Hayward <alan.hayward@arm.com>
15721 David Sherwood <david.sherwood@arm.com>
15723 * rtl.h (subreg_memory_offset): Declare.
15724 * emit-rtl.c (subreg_memory_offset): New function.
15725 * expmed.c (store_bit_field_1): Use it.
15726 * expr.c (undefined_operand_subword_p): Likewise.
15727 * simplify-rtx.c (simplify_subreg): Likewise.
15729 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15731 PR rtl-optimization/57448
15734 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15735 using atomic_load pattern.
15736 (expand_atomic_store): Likewise.
15738 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15741 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15742 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15743 handling. Use replace_call_with_value with NULL instead of
15744 gsi_replace, unlink_stmt_vdef and release_defs.
15746 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15749 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15751 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15754 * rtl.h (emit_library_call_value_1): Declare.
15755 (emit_library_call): Replace declaration with a series of overloads.
15756 Remove the parameter count argument.
15757 (emit_library_call_value): Likewise.
15758 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15759 with an "rtx_mode_t *".
15760 (emit_library_call_value): Delete.
15761 (emit_library_call): Likewise.
15762 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15763 (asan_emit_allocas_unpoison): Likewise.
15764 * builtins.c (expand_builtin_powi): Likewise.
15765 (expand_asan_emit_allocas_unpoison): Likewise.
15766 * cfgexpand.c (expand_main_function): Likewise.
15767 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15768 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15769 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15770 * config/arm/arm.c (arm_trampoline_init): Likewise.
15771 (arm_call_tls_get_addr): Likewise.
15772 (arm_expand_divmod_libfunc): Likewise.
15773 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15774 (smulsi3_highpart): Likewise.
15775 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15776 (c6x_expand_compare): Likewise.
15777 (c6x_expand_movmem): Likewise.
15778 * config/frv/frv.c (frv_trampoline_init): Likewise.
15779 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15780 (ix86_expand_divmod_libfunc): Likewise.
15781 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15782 (ia64_expand_compare): Likewise.
15783 (ia64_profile_hook): Likewise.
15784 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15785 (nonlocal_goto): Likewise.
15786 (restore_stack_nonlocal): Likewise.
15787 * config/m32r/m32r.c (block_move_call): Likewise.
15788 (m32r_trampoline_init): Likewise.
15789 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15790 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15791 (m68k_call_m68k_read_tp): Likewise.
15792 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15793 (microblaze_expand_divide): Likewise.
15794 * config/mips/mips.h (mips_args): Likewise.
15795 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15796 (MIPS_ICACHE_SYNC): Likewise.
15797 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15798 (nios2_trampoline_init): Likewise.
15799 * config/pa/pa.c (hppa_tls_call): Likewise.
15800 (pa_trampoline_init): Likewise.
15801 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15802 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15803 (expand_strn_compare): Likewise.
15804 (rs6000_generate_compare): Likewise.
15805 (rs6000_expand_float128_convert): Likewise.
15806 (output_profile_hook): Likewise.
15807 (rs6000_trampoline_init): Likewise.
15808 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15809 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15810 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15811 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15812 (rs6000_generate_compare): Likewise.
15813 (rs6000_expand_float128_convert): Likewise.
15814 (output_profile_hook): Likewise.
15815 (rs6000_trampoline_init): Likewise.
15816 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15817 * config/sh/sh.c (sh_trampoline_init): Likewise.
15818 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15819 (sparc_emit_float_lib_cmp): Likewise.
15820 (sparc32_initialize_trampoline): Likewise.
15821 (sparc64_initialize_trampoline): Likewise.
15822 (sparc_profile_hook): Likewise.
15823 * config/spu/spu.c (ea_load_store): Likewise.
15824 * config/spu/spu.md (floatunssidf2): Likewise.
15825 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15826 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15827 * config/visium/visium.c (expand_block_move_4): Likewise.
15828 (expand_block_move_2): Likewise.
15829 (expand_block_move_1): Likewise.
15830 (expand_block_set_4): Likewise.
15831 (expand_block_set_2): Likewise.
15832 (expand_block_set_1): Likewise.
15833 (visium_trampoline_init): Likewise.
15834 (visium_profile_hook): Likewise.
15835 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15836 (xtensa_setup_frame_addresses): Likewise.
15837 (xtensa_trampoline_init): Likewise.
15838 * except.c (sjlj_emit_function_enter): Likewise.
15839 (sjlj_emit_function_exit): Likewise.
15840 * explow.c (allocate_dynamic_stack_space): Likewise.
15841 (probe_stack_range): Likewise.
15842 * expr.c (convert_mode_scalar): Likewise.
15843 * optabs.c (expand_binop): Likewise.
15844 (expand_twoval_binop_libfunc): Likewise.
15845 (expand_unop): Likewise.
15846 (prepare_cmp_insn): Likewise.
15847 (prepare_float_lib_cmp): Likewise.
15848 (expand_float): Likewise.
15849 (expand_fix): Likewise.
15850 (expand_fixed_convert): Likewise.
15851 (maybe_emit_sync_lock_test_and_set): Likewise.
15852 (expand_atomic_compare_and_swap): Likewise.
15853 (expand_mem_thread_fence): Likewise.
15854 (expand_atomic_fetch_op): Likewise.
15856 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15858 * doc/generic.texi (OpenACC): Adjust URL.
15859 * doc/invoke.texi (C Dialect Options): Ditto.
15861 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15863 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15864 predicate for operand 1. Add (m,<S>) constraint.
15865 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15866 Prevent memory operand 1 with register operand 2.
15868 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15870 PR rtl-optimization/82024
15871 * combine.c (try_combine): If the combination result is a PARALLEL,
15872 and we only need to retain the SET in there that would be placed
15873 at I2, check that we can place that at I3 instead, before doing so.
15875 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15878 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15880 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15881 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15884 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15885 Jeff Law <law@redhat.com>
15887 * varasm.c (bss_initializer_p): Do not put constants into .bss
15888 (categorize_decl_for_section): Handle bss_initializer_p returning
15889 false when DECL_INITIAL is NULL.
15891 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15894 * config/s390/s390.c (s390_can_inline_p): New function.
15896 2017-09-01 Jeff Law <law@redhat.com>
15898 PR tree-optimization/82052
15899 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15900 Always initialize the returned slot after a hash table miss
15901 when INSERT is true.
15903 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15905 * config/s390/s390.md (mem_signal_fence): Remove.
15906 * doc/md.texi (mem_signal_fence): Remove.
15907 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15909 * target-insns.def (mem_signal_fence): Remove.
15911 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15914 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15917 * asan.c (create_odr_indicator): Strip name encoding from assembler
15918 name before appending it after __odr_asan_.
15920 2017-09-01 Martin Liska <mliska@suse.cz>
15922 PR tree-optimization/82059
15923 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15924 frequency only when an edge is redirected.
15926 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15928 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15929 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15930 (arc_conditional_register_usage): Remove ARC600 lp_count
15932 (arc_file_start): Emit Tag_ARC_CPU_variation.
15933 (arc_can_use_doloop_p): New conditions to use ZOLs.
15934 (hwloop_fail): New function.
15935 (hwloop_optimize): Likewise.
15936 (hwloop_pattern_reg): Likewise.
15937 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15938 (arc_reorg_loops): New function, calls reorg_loops.
15939 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15940 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15942 (arc_loop_hazard): Remove function, functionality moved into
15944 (arc_hazard): Remove arc_loop_hazard call.
15945 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15946 into hwloop_optimize.
15947 (arc_label_align): Remove ZOL handling.
15948 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15949 * config/arc/arc.md (doloop_begin): Remove pattern.
15950 (doloop_begin_i): Likewise.
15951 (doloop_end_i): Likewise.
15952 (doloop_fallback): Likewise.
15953 (doloop_fallback_m): Likewise.
15954 (doloop_end): Reimplement expand.
15955 (arc_lp): New pattern for LP instruction.
15956 (loop_end): New pattern.
15957 (loop_fail): Likewise.
15958 (decrement_and_branch_until_zero): Likewise.
15959 * config/arc/arc.opt (mlpc-width): New option.
15960 * doc/invoke.texi (mlpc-width): Document option.
15962 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15964 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15965 (arc_ccfsm_advance): Fix checking for delay slots.
15966 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15968 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15970 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15972 (movhi_insn): Update store instruction constraint which are saving
15973 6-bit short immediates.
15974 (movsi_insn): Consider also short scaled load operations.
15975 (zero_extendhisi2_i): Use Usd constraint instead of T.
15976 (extendhisi2_i): Add q constraint.
15977 (arc_clzsi2): Add type and length attributes.
15978 (arc_ctzsi2): Likewise.
15979 * config/arc/constraints.md (Usc): Update constraint, the
15980 assembler can parse two relocations for a single instruction.
15982 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15984 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15985 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15987 2017-08-31 Olivier Hainque <hainque@adacore.com>
15989 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15990 match as powerpc-wrs-vxworks*.
15992 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15994 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15995 register constraint for by-element operand.
15996 (aarch64_mls_elt_merge<mode>): Likewise.
15998 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16000 * config/arc/arc.c (arc_can_follow_jump): Check for short
16003 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16005 * config.gcc: Use g.opt for arc.
16006 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16007 functionality moved to ...
16008 (legitimate_scaled_address_p): New function, ...here.
16009 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16010 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16011 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16013 (arc_override_options): Handle G option.
16014 (arc_output_pic_addr_const): Correct function definition.
16015 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16016 (arc_decl_anon_ns_mem_p): Delete.
16017 (arc_in_small_data_p): Overhaul this function to take into
16018 consideration the value given via G option.
16019 (arc_rewrite_small_data_1): Renamed and corrected old
16020 arc_rewrite_small_data function.
16021 (arc_rewrite_small_data): New function.
16022 (small_data_pattern): Don't use pic_offset_table_rtx.
16023 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16024 * config/arc/simdext.md (movmisalignv2hi): Use
16025 prepare_move_operands function.
16027 (movmisalign*): Likewise.
16028 * doc/invoke.texi (ARC options): Document -G option.
16030 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16032 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16034 * config/arc/arc.c (arc_print_operand): Output scalled address for
16035 sdata whenever is possible.
16036 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16037 load/stores are available.
16038 (compact_sda_memory_operand): Check for the alignment required by
16039 code density instructions.
16040 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16042 * config/arc/constraints.md (Usd): Update constraint.
16043 (Us0): New constraint.
16044 (Usc): Update constraint.
16046 2017-08-31 Richard Biener <rguenther@suse.de>
16048 PR middle-end/82054
16049 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16050 function only once.
16052 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16054 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16055 Resize type_signature.
16057 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16058 Alan Hayward <alan.hayward@arm.com>
16059 David Sherwood <david.sherwood@arm.com>
16061 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16062 subregs whose inner modes can be stored in GPRs.
16063 (aarch64_classify_index): Likewise.
16065 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16066 Alan Hayward <alan.hayward@arm.com>
16067 David Sherwood <david.sherwood@arm.com>
16069 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16070 (V_INT_EQUIV): ...this.
16071 (v_cmp_result): Rename to...
16072 (v_int_equiv): ...this.
16073 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16074 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16075 (copysign<mode>3): Likewise.
16076 (aarch64_simd_bsl<mode>_internal): Likewise.
16077 (aarch64_simd_bsl<mode>): Likewise.
16078 (vec_cmp<mode><mode>): Likewise.
16079 (vcond<mode><mode>): Likewise.
16080 (vcond<v_cmp_mixed><mode>): Likewise.
16081 (vcondu<mode><v_cmp_mixed>): Likewise.
16082 (aarch64_cm<optab><mode>): Likewise.
16083 (aarch64_cmtst<mode>): Likewise.
16084 (aarch64_fac<optab><mode>): Likewise.
16085 (vec_perm_const<mode>): Likewise.
16086 (vcond_mask_<mode><v_cmp_result>): Rename to...
16087 (vcond_mask_<mode><v_int_equiv>): ...this.
16088 (vec_cmp<mode><v_cmp_result>): Rename to...
16089 (vec_cmp<mode><v_int_equiv>): ...this.
16091 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16092 Alan Hayward <alan.hayward@arm.com>
16093 David Sherwood <david.sherwood@arm.com>
16095 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16097 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16098 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16099 (UNSPEC_LD4_DREG): New unspecs.
16100 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16101 (aarch64_ld2<mode>_dreg_be): Replace with...
16102 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16104 (aarch64_ld3<mode>_dreg_le)
16105 (aarch64_ld3<mode>_dreg_be): Replace with...
16106 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16108 (aarch64_ld4<mode>_dreg_le)
16109 (aarch64_ld4<mode>_dreg_be): Replace with...
16110 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16113 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16115 PR tree-optimization/81987
16116 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16117 insert an initializer in a location not dominated by the stride
16120 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16122 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16123 on the entire header of the finally block in the fallthru case.
16125 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16127 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16129 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16131 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16132 rs6000_emit_move_from_cr and call renamed function.
16133 (rs6000_emit_prologue): Call renamed functions.
16134 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16135 movesi_from_cr, remove volatile CRs.
16137 2017-08-30 Jon Beniston <jon@beniston.com>
16138 Richard Biener <rguenther@suse.de>
16140 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16141 of VECTOR_MODE_P check.
16142 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16143 element vector types.
16145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16147 * df.h (df_read_modify_subreg_p): Remove in favor of...
16148 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16149 const_rtx instead of an rtx.
16150 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16151 * df-problems.c (df_word_lr_mark_ref): Likewise.
16152 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16153 (mark_pseudo_reg_dead): Likewise.
16154 (mark_ref_dead): Likewise.
16155 * reginfo.c (init_subregs_of_mode): Likewise.
16156 * sched-deps.c (sched_analyze_1): Likewise.
16157 * df-scan.c (df_def_record_1): Likewise.
16158 (df_uses_record): Likewise.
16159 (df_read_modify_subreg_p): Remove in favor of...
16160 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16161 const_rtx instead of an rtx.
16163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16164 Alan Hayward <alan.hayward@arm.com>
16165 David Sherwood <david.sherwood@arm.com>
16167 * rtl.h (partial_subreg_p): New function.
16168 * caller-save.c (save_call_clobbered_regs): Use it.
16169 * calls.c (expand_call): Likewise.
16170 * combine.c (combinable_i3pat): Likewise.
16171 (simplify_set): Likewise.
16172 (make_extraction): Likewise.
16173 (make_compound_operation_int): Likewise.
16174 (gen_lowpart_or_truncate): Likewise.
16175 (force_to_mode): Likewise.
16176 (make_field_assignment): Likewise.
16177 (reg_truncated_to_mode): Likewise.
16178 (record_truncated_value): Likewise.
16179 (move_deaths): Likewise.
16180 * cse.c (record_jump_cond): Likewise.
16181 (cse_insn): Likewise.
16182 * cselib.c (cselib_lookup_1): Likewise.
16183 * expmed.c (extract_bit_field_using_extv): Likewise.
16184 * function.c (assign_parm_setup_reg): Likewise.
16185 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16186 * ira-build.c (create_insn_allocnos): Likewise.
16187 * lra-coalesce.c (merge_pseudos): Likewise.
16188 * lra-constraints.c (match_reload): Likewise.
16189 (simplify_operand_subreg): Likewise.
16190 (curr_insn_transform): Likewise.
16191 * lra-lives.c (process_bb_lives): Likewise.
16192 * lra.c (new_insn_reg): Likewise.
16193 (lra_substitute_pseudo): Likewise.
16194 * regcprop.c (mode_change_ok): Likewise.
16195 (maybe_mode_change): Likewise.
16196 (copyprop_hardreg_forward_1): Likewise.
16197 * reload.c (push_reload): Likewise.
16198 (find_reloads): Likewise.
16199 (find_reloads_subreg_address): Likewise.
16200 * reload1.c (alter_reg): Likewise.
16201 (eliminate_regs_1): Likewise.
16202 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16204 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16206 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16207 back to if statements, including unpack.
16209 2017-08-30 Martin Liska <mliska@suse.cz>
16211 PR inline-asm/82001
16212 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16214 (func_checker::compare_asm_inputs_outputs): ... this function.
16215 (func_checker::compare_gimple_asm): Use the function to compare
16216 also ASM constrains.
16217 * ipa-icf-gimple.h: Rename the function.
16219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16220 Alan Hayward <alan.hayward@arm.com>
16221 David Sherwood <david.sherwood@arm.com>
16223 * coretypes.h (complex_mode): New type.
16224 * gdbhooks.py (build_pretty_printer): Handle it.
16225 * machmode.h (complex_mode): New class.
16226 (complex_mode::includes_p): New function.
16227 (is_complex_int_mode): Likewise.
16228 (is_complex_float_mode): Likewise.
16229 * genmodes.c (get_mode_class): Handle complex mode classes.
16230 * function.c (expand_function_end): Use is_complex_int_mode.
16232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16233 Alan Hayward <alan.hayward@arm.com>
16234 David Sherwood <david.sherwood@arm.com>
16236 * coretypes.h (scalar_mode_pod): New typedef.
16237 * gdbhooks.py (build_pretty_printer): Handle it.
16238 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16239 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16240 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16241 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16242 as_a <scalar_mode>.
16244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16245 Alan Hayward <alan.hayward@arm.com>
16246 David Sherwood <david.sherwood@arm.com>
16248 * machmode.h (mode_for_vector): Take a scalar_mode instead
16250 * stor-layout.c (mode_for_vector): Likewise.
16251 * explow.c (promote_mode): Use as_a <scalar_mode>.
16252 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16255 Alan Hayward <alan.hayward@arm.com>
16256 David Sherwood <david.sherwood@arm.com>
16258 * target.def (preferred_simd_mode): Take a scalar_mode
16259 instead of a machine_mode.
16260 * targhooks.h (default_preferred_simd_mode): Likewise.
16261 * targhooks.c (default_preferred_simd_mode): Likewise.
16262 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16263 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16264 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16265 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16266 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16267 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16268 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16269 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16271 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16272 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16273 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16274 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16275 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16276 * doc/tm.texi: Regenerate.
16277 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16281 Alan Hayward <alan.hayward@arm.com>
16282 David Sherwood <david.sherwood@arm.com>
16284 * target.def (scalar_mode_supported_p): Take a scalar_mode
16285 instead of a machine_mode.
16286 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16287 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16288 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16289 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16290 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16291 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16292 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16293 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16294 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16295 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16296 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16297 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16298 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16299 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16301 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16302 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16303 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16304 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16305 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16307 * doc/tm.texi: Regenerate.
16309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16310 Alan Hayward <alan.hayward@arm.com>
16311 David Sherwood <david.sherwood@arm.com>
16313 * coretypes.h (opt_scalar_mode): New typedef.
16314 * gdbhooks.py (build_pretty_printers): Handle it.
16315 * machmode.h (mode_iterator::get_2xwider): Add overload for
16317 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16319 * expr.c (convert_mode_scalar): Likewise.
16320 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16321 * optabs.c (expand_float): Likewise.
16322 (expand_fix): Likewise.
16323 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16326 Alan Hayward <alan.hayward@arm.com>
16327 David Sherwood <david.sherwood@arm.com>
16329 * optabs.c (expand_float): Explicitly check for scalars before
16330 using a branching expansion.
16331 (expand_fix): Likewise.
16333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16334 Alan Hayward <alan.hayward@arm.com>
16335 David Sherwood <david.sherwood@arm.com>
16337 * expr.c (convert_mode): Split scalar handling out into...
16338 (convert_mode_scalar): ...this new function. Treat the modes
16341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16342 Alan Hayward <alan.hayward@arm.com>
16343 David Sherwood <david.sherwood@arm.com>
16345 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16347 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16350 Alan Hayward <alan.hayward@arm.com>
16351 David Sherwood <david.sherwood@arm.com>
16353 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16354 rather than a machine_mode.
16355 (fixed_from_string): Likewise.
16356 (fixed_convert): Likewise.
16357 (fixed_convert_from_int): Likewise.
16358 (fixed_convert_from_real): Likewise.
16359 (real_convert_from_fixed): Likewise.
16360 * fixed-value.c (fixed_from_double_int): Likewise.
16361 (fixed_from_string): Likewise.
16362 (fixed_convert): Likewise.
16363 (fixed_convert_from_int): Likewise.
16364 (fixed_convert_from_real): Likewise.
16365 (real_convert_from_fixed): Likewise.
16366 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16369 Alan Hayward <alan.hayward@arm.com>
16370 David Sherwood <david.sherwood@arm.com>
16372 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16373 of separate mode class checks. Do not allow vector modes here.
16374 (immed_wide_int_const): Use as_a <scalar_mode>.
16375 * explow.c (trunc_int_for_mode): Likewise.
16376 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16377 (wi::shwi): Likewise.
16378 (wi::min_value): Likewise.
16379 (wi::max_value): Likewise.
16380 * dwarf2out.c (loc_descriptor): Likewise.
16381 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16382 for CONST_WIDE_INT.
16384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16385 Alan Hayward <alan.hayward@arm.com>
16386 David Sherwood <david.sherwood@arm.com>
16388 * tree.h (SCALAR_TYPE_MODE): New macro.
16389 * expr.c (expand_expr_addr_expr_1): Use it.
16390 (expand_expr_real_2): Likewise.
16391 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16392 (fold_convert_const_fixed_from_int): Likewise.
16393 (fold_convert_const_fixed_from_real): Likewise.
16394 (native_encode_fixed): Likewise
16395 (native_encode_complex): Likewise
16396 (native_encode_vector): Likewise.
16397 (native_interpret_fixed): Likewise.
16398 (native_interpret_real): Likewise.
16399 (native_interpret_complex): Likewise.
16400 (native_interpret_vector): Likewise.
16401 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16402 (simd_clone_adjust_argument_types): Likewise.
16403 (simd_clone_init_simd_arrays): Likewise.
16404 (simd_clone_adjust): Likewise.
16405 * stor-layout.c (layout_type): Likewise.
16406 * tree.c (build_minus_one_cst): Likewise.
16407 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16408 * tree-inline.c (estimate_move_cost): Likewise.
16409 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16410 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16411 (vectorizable_reduction): Likewise.
16412 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16413 (vect_recog_mixed_size_cond_pattern): Likewise.
16414 (check_bool_pattern): Likewise.
16415 (adjust_bool_pattern): Likewise.
16416 (search_type_for_mask_1): Likewise.
16417 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16418 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16419 (vectorizable_load): Likewise.
16420 (vectorizable_store): Likewise.
16421 * ubsan.c (ubsan_encode_value): Likewise.
16422 * varasm.c (output_constant): Likewise.
16424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16425 Alan Hayward <alan.hayward@arm.com>
16426 David Sherwood <david.sherwood@arm.com>
16428 * coretypes.h (scalar_mode): New class.
16429 * machmode.h (scalar_mode): Likewise.
16430 (scalar_mode::includes_p): New function.
16431 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16432 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16433 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16434 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16435 * expmed.c (store_bit_field_1): Likewise.
16436 (extract_bit_field_1): Likewise.
16437 * expr.c (write_complex_part): Likewise.
16438 (read_complex_part): Likewise.
16439 (emit_move_complex_push): Likewise.
16440 (expand_expr_real_2): Likewise.
16441 * function.c (assign_parm_setup_reg): Likewise.
16442 (assign_parms_unsplit_complex): Likewise.
16443 * optabs.c (expand_binop): Likewise.
16444 * rtlanal.c (subreg_get_info): Likewise.
16445 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16446 * varasm.c (output_constant_pool_2): Likewise.
16448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16449 Alan Hayward <alan.hayward@arm.com>
16450 David Sherwood <david.sherwood@arm.com>
16452 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16454 (expmed_mult_highpart_optab): Likewise.
16455 (expmed_mult_highpart): Likewise.
16457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16458 Alan Hayward <alan.hayward@arm.com>
16459 David Sherwood <david.sherwood@arm.com>
16461 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16462 instead of a machine_mode.
16463 (builtin_memset_read_str): Likewise.
16464 * builtins.c (c_readstr): Likewise.
16465 (builtin_memcpy_read_str): Likewise.
16466 (builtin_strncpy_read_str): Likewise.
16467 (builtin_memset_read_str): Likewise.
16468 (builtin_memset_gen_str): Likewise.
16469 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16470 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16471 instead of a machine_mode.
16472 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16474 (make_extraction): Likewise.
16475 (try_widen_shift_mode): Take and return scalar_int_modes instead
16477 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16478 a scalar_int_mode instead of a machine_mode.
16479 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16480 (avr_addr_space_pointer_mode): Likewise.
16481 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16482 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16483 (msp430_unwind_word_mode): Likewise.
16484 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16485 (spu_addr_space_pointer_mode): Likewise.
16486 (spu_addr_space_address_mode): Likewise.
16487 (spu_libgcc_cmp_return_mode): Likewise.
16488 (spu_libgcc_shift_count_mode): Likewise.
16489 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16490 (rl78_addr_space_pointer_mode): Likewise.
16491 (fl78_unwind_word_mode): Likewise.
16492 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16494 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16495 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16496 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16497 (mips_valid_pointer_mode): Likewise.
16498 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16499 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16500 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16502 (ft32_addr_space_address_mode): Likewise.
16503 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16504 scalar_int_mode instead of a machine_mode.
16505 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16507 (m32c_addr_space_address_mode): Likewise.
16508 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16509 (rs6000_eh_return_filter_mode): Likewise.
16510 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16511 (rs6000_eh_return_filter_mode): Likewise.
16512 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16513 (s390_libgcc_shift_count_mode): Likewise.
16514 (s390_unwind_word_mode): Likewise.
16515 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16517 * target.def (mode_rep_extended): Likewise.
16518 (valid_pointer_mode): Likewise.
16519 (addr_space.valid_pointer_mode): Likewise.
16520 (eh_return_filter_mode): Return a scalar_int_mode rather than
16522 (libgcc_cmp_return_mode): Likewise.
16523 (libgcc_shift_count_mode): Likewise.
16524 (unwind_word_mode): Likewise.
16525 (addr_space.pointer_mode): Likewise.
16526 (addr_space.address_mode): Likewise.
16527 * doc/tm.texi: Regenerate.
16528 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16530 (do_jump): Use scalar_int_mode for local variables.
16531 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16532 rather than a machine_mode.
16533 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16534 (scompare_loc_descriptor_wide): Likewise.
16535 (scompare_loc_descriptor_narrow): Likewise.
16536 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16538 * except.c (sjlj_emit_dispatch_table): Likewise.
16539 (expand_builtin_eh_copy_values): Likewise.
16540 * explow.c (convert_memory_address_addr_space_1): Likewise.
16541 Take a scalar_int_mode rather than a machine_mode.
16542 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16543 than a machine_mode.
16544 (memory_address_addr_space): Use scalar_int_mode for local variables.
16545 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16546 rather than a machine_mode.
16547 * expmed.c (mask_rtx): Likewise.
16548 (init_expmed_one_conv): Likewise.
16549 (expand_mult_highpart_adjust): Likewise.
16550 (extract_high_half): Likewise.
16551 (expmed_mult_highpart_optab): Likewise.
16552 (expmed_mult_highpart): Likewise.
16553 (expand_smod_pow2): Likewise.
16554 (expand_sdiv_pow2): Likewise.
16555 (emit_store_flag_int): Likewise.
16556 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16558 (extract_low_bits): Likewise.
16559 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16561 * expr.c (pieces_addr::adjust): Likewise.
16562 (can_store_by_pieces): Likewise.
16563 (store_by_pieces): Likewise.
16564 (clear_by_pieces_1): Likewise.
16565 (expand_expr_addr_expr_1): Likewise.
16566 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16567 (expand_expr_real_1): Likewise.
16568 (try_casesi): Likewise.
16569 * final.c (shorten_branches): Likewise.
16570 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16571 type of "mode" to machine_mode.
16572 * internal-fn.c (expand_arith_overflow_result_store): Take a
16573 scalar_int_mode rather than a machine_mode.
16574 (expand_mul_overflow): Use scalar_int_mode for local variables.
16575 * loop-doloop.c (doloop_modify): Likewise.
16576 (doloop_optimize): Likewise.
16577 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16578 than a machine_mode.
16579 (expand_doubleword_shift_condmove): Likewise.
16580 (expand_doubleword_shift): Likewise.
16581 (expand_doubleword_clz): Likewise.
16582 (expand_doubleword_popcount): Likewise.
16583 (expand_doubleword_parity): Likewise.
16584 (expand_absneg_bit): Use scalar_int_mode for local variables.
16585 (prepare_float_lib_cmp): Likewise.
16586 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16587 rather than a machine_mode.
16588 (convert_memory_address_addr_space): Likewise.
16589 (get_mode_bounds): Likewise.
16590 (get_address_mode): Return a scalar_int_mode rather than a
16592 * rtlanal.c (get_address_mode): Likewise.
16593 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16594 than a machine_mode.
16595 * targhooks.c (default_mode_rep_extended): Likewise.
16596 (default_valid_pointer_mode): Likewise.
16597 (default_addr_space_valid_pointer_mode): Likewise.
16598 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16599 than a machine_mode.
16600 (default_libgcc_cmp_return_mode): Likewise.
16601 (default_libgcc_shift_count_mode): Likewise.
16602 (default_unwind_word_mode): Likewise.
16603 (default_addr_space_pointer_mode): Likewise.
16604 (default_addr_space_address_mode): Likewise.
16605 * targhooks.h (default_eh_return_filter_mode): Likewise.
16606 (default_libgcc_cmp_return_mode): Likewise.
16607 (default_libgcc_shift_count_mode): Likewise.
16608 (default_unwind_word_mode): Likewise.
16609 (default_addr_space_pointer_mode): Likewise.
16610 (default_addr_space_address_mode): Likewise.
16611 (default_mode_rep_extended): Take a scalar_int_mode rather than
16613 (default_valid_pointer_mode): Likewise.
16614 (default_addr_space_valid_pointer_mode): Likewise.
16615 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16617 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16618 rather than a machine_mode.
16619 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16620 for local variables.
16621 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16622 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16623 than a machine_mode.
16625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16626 Alan Hayward <alan.hayward@arm.com>
16627 David Sherwood <david.sherwood@arm.com>
16629 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16630 the mode argument to scalar_int_mode.
16631 (do_jump_by_parts_zero_rtx): Likewise.
16632 (do_jump_by_parts_equality_rtx): Likewise.
16633 (do_jump_by_parts_greater): Take a mode argument.
16634 (do_jump_by_parts_equality): Likewise.
16635 (do_jump_1): Update calls accordingly.
16637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16638 Alan Hayward <alan.hayward@arm.com>
16639 David Sherwood <david.sherwood@arm.com>
16641 * is-a.h (safe_dyn_cast): New function.
16642 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16643 (jump_table_for_label): Likewise.
16644 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16645 instead of an rtx_insn *.
16646 (shorten_branches): Use dyn_cast instead of LABEL_P and
16647 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16648 rtx_jump_table_data::get_data_mode.
16649 (final_scan_insn): Likewise.
16651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16652 Alan Hayward <alan.hayward@arm.com>
16653 David Sherwood <david.sherwood@arm.com>
16655 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16656 trying to combine a full-register integer set with a subreg
16659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16660 Alan Hayward <alan.hayward@arm.com>
16661 David Sherwood <david.sherwood@arm.com>
16663 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16664 that is always either address_mode or pointer_mode.
16666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16667 Alan Hayward <alan.hayward@arm.com>
16668 David Sherwood <david.sherwood@arm.com>
16670 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16671 when the two are known to be equal.
16673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16674 Alan Hayward <alan.hayward@arm.com>
16675 David Sherwood <david.sherwood@arm.com>
16677 * simplify-rtx.c (simplify_const_unary_operation): Use
16678 is_a <scalar_int_mode> instead of checking for a nonzero
16679 precision. Forcibly convert op_mode to a scalar_int_mode
16680 in that case. More clearly differentiate the operand and
16681 result modes and use the former when deciding what the value
16682 of a count-bits operation should be. Use is_int_mode instead
16683 of checking for a MODE_INT. Remove redundant check for whether
16684 this mode has a zero precision.
16686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16687 Alan Hayward <alan.hayward@arm.com>
16688 David Sherwood <david.sherwood@arm.com>
16690 * optabs.c (widen_leading): Change the type of the mode argument
16691 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16692 (widen_bswap): Likewise.
16693 (expand_parity): Likewise.
16694 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16695 (expand_ffs): Likewise.
16696 (epand_unop): Check for scalar integer modes before calling the
16699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16700 Alan Hayward <alan.hayward@arm.com>
16701 David Sherwood <david.sherwood@arm.com>
16703 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
16705 (expand_expr_real_1): Update call accordingly.
16707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16708 Alan Hayward <alan.hayward@arm.com>
16709 David Sherwood <david.sherwood@arm.com>
16711 * expmed.c (store_bit_field_using_insv): Add op0_mode and
16712 value_mode arguments. Use scalar_int_mode internally.
16713 (store_bit_field_1): Rename the new integer mode from imode
16714 to op0_mode and use it instead of GET_MODE (op0). Update calls
16715 to store_split_bit_field, store_bit_field_using_insv and
16716 store_fixed_bit_field.
16717 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16718 Use scalar_int_mode internally. Use a bit count rather than a mode
16719 when calculating the largest bit size for get_best_mode.
16720 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16721 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16722 Remove assertion that OP0 has a scalar integer mode.
16723 (store_split_bit_field): Add op0_mode and value_mode arguments.
16724 Update calls to extract_fixed_bit_field.
16725 (extract_bit_field_using_extv): Add an op0_mode argument.
16726 Use scalar_int_mode internally.
16727 (extract_bit_field_1): Rename the new integer mode from imode to
16728 op0_mode and use it instead of GET_MODE (op0). Update calls to
16729 extract_split_bit_field, extract_bit_field_using_extv and
16730 extract_fixed_bit_field.
16731 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16732 to extract_split_bit_field and extract_fixed_bit_field_1.
16733 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16734 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16735 on the target mode.
16736 (extract_split_bit_field): Add an op0_mode argument. Update call
16737 to extract_fixed_bit_field.
16739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16740 Alan Hayward <alan.hayward@arm.com>
16741 David Sherwood <david.sherwood@arm.com>
16743 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16744 * explow.c (hard_function_value): Likewise.
16745 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16746 convert_to_mode call outside the loop.
16747 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16748 for the mode iterator. Require the mode specified by max_pieces
16750 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16752 (copy_blkmode_to_reg): Likewise.
16753 (set_storage_via_setmem): Likewise.
16754 * optabs.c (prepare_cmp_insn): Likewise.
16755 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16756 * stor-layout.c (finish_bitfield_representative): Likewise.
16758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16759 Alan Hayward <alan.hayward@arm.com>
16760 David Sherwood <david.sherwood@arm.com>
16762 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16763 * expr.c (convert_move): Use them.
16764 (convert_modes): Likewise.
16765 (store_expr_with_bounds): Likewise.
16767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16768 Alan Hayward <alan.hayward@arm.com>
16769 David Sherwood <david.sherwood@arm.com>
16771 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16772 parameter for the mode of "x". Remove the "known_x", "known_mode"
16773 and "known_ret" arguments. Change the type of the mode argument
16774 to scalar_int_mode.
16775 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16776 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16777 (reg_num_sign_bit_copies_for_combine): Likewise.
16778 * rtlanal.c (nonzero_bits1): Likewise.
16779 (num_sign_bit_copies1): Likewise.
16780 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16781 (reg_num_sign_bit_copies_general): Likewise.
16782 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16783 (reg_nonzero_bits_general): Likewise.
16785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16786 Alan Hayward <alan.hayward@arm.com>
16787 David Sherwood <david.sherwood@arm.com>
16789 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16790 than in subroutines. Return 1 for non-integer modes.
16791 (cached_num_sign_bit_copies): Change the type of the mode parameter
16792 to scalar_int_mode.
16793 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16794 classes. Handle CONST_INT_P first and then check whether X also
16795 has a scalar integer mode. Check the same thing for inner registers
16796 of a SUBREG and for values that are being extended or truncated.
16798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16799 Alan Hayward <alan.hayward@arm.com>
16800 David Sherwood <david.sherwood@arm.com>
16802 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16803 in subroutines. Return the mode mask for non-integer modes.
16804 (cached_nonzero_bits): Change the type of the mode parameter
16805 to scalar_int_mode.
16806 (nonzero_bits1): Likewise. Remove early exit for other mode
16807 classes. Handle CONST_INT_P first and then check whether X
16808 also has a scalar integer mode.
16810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16811 Alan Hayward <alan.hayward@arm.com>
16812 David Sherwood <david.sherwood@arm.com>
16814 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16815 Return a scalar_int_mode and assert that the size is greater than
16817 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16818 (op_by_pieces_d::op_by_pieces_d): Likewise.
16819 (op_by_pieces_d::run): Likewise.
16820 (can_store_by_pieces): Likewise.
16822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16823 Alan Hayward <alan.hayward@arm.com>
16824 David Sherwood <david.sherwood@arm.com>
16826 * combine.c (extract_left_shift): Add a mode argument and update
16828 (make_compound_operation_int): Change the type of the mode parameter
16829 to scalar_int_mode and update the call to extract_left_shift.
16831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16832 Alan Hayward <alan.hayward@arm.com>
16833 David Sherwood <david.sherwood@arm.com>
16835 * combine.c (simplify_and_const_int): Change the type of the mode
16836 parameter to scalar_int_mode.
16837 (simplify_and_const_int_1): Likewise. Update recursive call.
16839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16840 Alan Hayward <alan.hayward@arm.com>
16841 David Sherwood <david.sherwood@arm.com>
16843 * combine.c (simplify_compare_const): Check that the mode is a
16844 scalar_int_mode (rather than VOIDmode) before testing its
16846 (simplify_comparison): Move COMPARISON_P handling out of the
16847 loop and restrict the latter part of the loop to scalar_int_modes.
16848 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16849 and when considering SUBREG_REGs. Use is_int_mode instead of
16850 checking GET_MODE_CLASS against MODE_INT.
16852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16853 Alan Hayward <alan.hayward@arm.com>
16854 David Sherwood <david.sherwood@arm.com>
16856 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16857 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16858 shift_unit_mode and for modes involved in scalar shifts.
16860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16861 Alan Hayward <alan.hayward@arm.com>
16862 David Sherwood <david.sherwood@arm.com>
16864 * combine.c (force_int_to_mode): New function, split out from...
16865 (force_to_mode): ...here. Keep xmode up-to-date and use it
16866 instead of GET_MODE (x).
16868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16869 Alan Hayward <alan.hayward@arm.com>
16870 David Sherwood <david.sherwood@arm.com>
16872 * optabs-query.h (extraction_insn::struct_mode): Change type to
16873 opt_scalar_int_mode and update comment.
16874 (extraction_insn::field_mode): Change type to scalar_int_mode.
16875 (extraction_insn::pos_mode): Likewise.
16876 * combine.c (make_extraction): Update accordingly.
16877 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16878 (get_optab_extraction_insn): Likewise.
16879 * recog.c (simplify_while_replacing): Likewise.
16880 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16881 parameter to opt_scalar_int_mode.
16883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16884 Alan Hayward <alan.hayward@arm.com>
16885 David Sherwood <david.sherwood@arm.com>
16887 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16888 to a scalar_int_mode instead of a machine_mode.
16889 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16890 (get_best_mode): Return a boolean and use a pointer argument to store
16891 the selected mode. Replace the limit mode parameter with a bit limit.
16892 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16893 for the values returned by bit_field_mode_iterator::next_mode.
16894 (store_bit_field): Update call to get_best_mode.
16895 (store_fixed_bit_field): Likewise.
16896 (extract_fixed_bit_field): Likewise.
16897 * expr.c (optimize_bitfield_assignment_op): Likewise.
16898 * fold-const.c (optimize_bit_field_compare): Likewise.
16899 (fold_truth_andor_1): Likewise.
16900 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16901 Update for new type of m_mode.
16902 (get_best_mode): As above.
16904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16905 Alan Hayward <alan.hayward@arm.com>
16906 David Sherwood <david.sherwood@arm.com>
16908 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16909 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16910 (store_bit_field): Check is_a <scalar_int_mode> before calling
16911 strict_volatile_bitfield_p.
16912 (extract_bit_field): Likewise.
16914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16915 Alan Hayward <alan.hayward@arm.com>
16916 David Sherwood <david.sherwood@arm.com>
16918 * target.def (cstore_mode): Return a scalar_int_mode.
16919 * doc/tm.texi: Regenerate.
16920 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16921 * targhooks.h (default_cstore_mode): Likewise.
16922 * targhooks.c (default_cstore_mode): Likewise, using a forced
16924 * expmed.c (emit_cstore): Expect the target of the cstore to be
16927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16928 Alan Hayward <alan.hayward@arm.com>
16929 David Sherwood <david.sherwood@arm.com>
16931 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16933 (niter_desc): Likewise mode.
16934 (iv_analyze): Add a mode parameter.
16936 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16937 and change its type to scalar_int_mode.
16938 * loop-iv.c: Update commentary at head of file.
16939 (iv_constant): Pass the mode paraeter before the rtx it describes
16940 and change its type to scalar_int_mode. Remove VOIDmode handling.
16941 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16942 (iv_extend): Likewise.
16943 (shorten_into_mode): Likewise.
16944 (iv_add): Use scalar_int_mode.
16945 (iv_mult): Likewise.
16946 (iv_shift): Likewise.
16947 (canonicalize_iv_subregs): Likewise.
16948 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16949 it describes and change its mode to scalar_int_mode. Also change
16950 the type of the returned inner_mode to scalar_int_mode.
16951 (get_biv_step): Likewise, turning outer_mode from a pointer
16952 into a direct parameter. Update call to get_biv_step_1.
16953 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16954 iv_constant and get_biv_step.
16955 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16956 and change its type to scalar_int_mode. Don't initialise iv->mode
16957 to VOIDmode and remove later checks for its still being VOIDmode.
16958 Update calls to iv_analyze_op and iv_analyze_expr. Check
16959 is_a <scalar_int_mode> when changing the mode under consideration.
16960 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16961 Update call to iv_analyze_expr.
16962 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16963 inner register is not also a scalar_int_mode. Update call to
16965 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16966 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16967 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16968 separate mode class checks. Update calls to iv_analyze. Remove
16969 fix-up of VOIDmodes after iv_analyze_biv.
16970 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16971 don't have a scalar_int_mode. Update call to biv_p.
16973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16974 Alan Hayward <alan.hayward@arm.com>
16975 David Sherwood <david.sherwood@arm.com>
16977 * cfgexpand.c (convert_debug_memory_address): Use
16978 as_a <scalar_int_mode>.
16979 * combine.c (expand_compound_operation): Likewise.
16980 (make_extraction): Likewise.
16981 (change_zero_ext): Likewise.
16982 (simplify_comparison): Likewise.
16983 * cse.c (cse_insn): Likewise.
16984 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16985 (mem_loc_descriptor): Likewise.
16986 (loc_descriptor): Likewise.
16987 * expmed.c (init_expmed_one_mode): Likewise.
16988 (synth_mult): Likewise.
16989 (emit_store_flag_1): Likewise.
16990 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16991 of a comparison with size.
16992 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16993 (reduce_to_bit_field_precision): Likewise.
16994 * function.c (expand_function_end): Likewise.
16995 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16996 * loop-doloop.c (doloop_modify): Likewise.
16997 * optabs.c (expand_binop): Likewise.
16998 (expand_unop): Likewise.
16999 (expand_copysign_absneg): Likewise.
17000 (prepare_cmp_insn): Likewise.
17001 (maybe_legitimize_operand): Likewise.
17002 * recog.c (const_scalar_int_operand): Likewise.
17003 * rtlanal.c (get_address_mode): Likewise.
17004 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17005 (simplify_cond_clz_ctz): Likewise.
17006 * tree-nested.c (get_nl_goto_field): Likewise.
17007 * tree.c (build_vector_type_for_mode): Likewise.
17008 * var-tracking.c (use_narrower_mode): Likewise.
17010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17011 Alan Hayward <alan.hayward@arm.com>
17012 David Sherwood <david.sherwood@arm.com>
17014 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17015 * builtins.c (expand_builtin_signbit): Use it.
17016 * cfgexpand.c (expand_debug_expr): Likewise.
17017 * dojump.c (do_jump): Likewise.
17018 (do_compare_and_jump): Likewise.
17019 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17020 * expmed.c (make_tree): Likewise.
17021 * expr.c (expand_expr_real_2): Likewise.
17022 (expand_expr_real_1): Likewise.
17023 (try_casesi): Likewise.
17024 * fold-const-call.c (fold_const_call_ss): Likewise.
17025 * fold-const.c (unextend): Likewise.
17026 (extract_muldiv_1): Likewise.
17027 (fold_single_bit_test): Likewise.
17028 (native_encode_int): Likewise.
17029 (native_encode_string): Likewise.
17030 (native_interpret_int): Likewise.
17031 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17032 * internal-fn.c (expand_addsub_overflow): Likewise.
17033 (expand_neg_overflow): Likewise.
17034 (expand_mul_overflow): Likewise.
17035 (expand_arith_overflow): Likewise.
17036 * match.pd: Likewise.
17037 * stor-layout.c (layout_type): Likewise.
17038 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17039 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17040 * tree-ssanames.c (get_range_info): Likewise.
17041 * tree-switch-conversion.c (array_value_type) Likewise.
17042 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17043 (vect_recog_divmod_pattern): Likewise.
17044 (vect_recog_mixed_size_cond_pattern): Likewise.
17045 * tree-vrp.c (extract_range_basic): Likewise.
17046 (simplify_float_conversion_using_ranges): Likewise.
17047 * tree.c (int_fits_type_p): Likewise.
17048 * ubsan.c (instrument_bool_enum_load): Likewise.
17049 * varasm.c (mergeable_string_section): Likewise.
17050 (narrowing_initializer_constant_valid_p): Likewise.
17051 (output_constant): Likewise.
17053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17054 Alan Hayward <alan.hayward@arm.com>
17055 David Sherwood <david.sherwood@arm.com>
17057 * machmode.h (NARROWEST_INT_MODE): New macro.
17058 * expr.c (alignment_for_piecewise_move): Use it instead of
17059 GET_CLASS_NARROWEST_MODE (MODE_INT).
17060 (push_block): Likewise.
17061 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17063 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17066 Alan Hayward <alan.hayward@arm.com>
17067 David Sherwood <david.sherwood@arm.com>
17069 * postreload.c (move2add_valid_value_p): Change the type of the
17070 mode parameter to scalar_int_mode.
17071 (move2add_use_add2_insn): Add a mode parameter and use it instead
17073 (move2add_use_add3_insn): Likewise.
17074 (reload_cse_move2add): Update accordingly.
17076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17077 Alan Hayward <alan.hayward@arm.com>
17078 David Sherwood <david.sherwood@arm.com>
17080 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17082 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17083 * optabs.c (expand_unop): Likewise.
17085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17086 Alan Hayward <alan.hayward@arm.com>
17087 David Sherwood <david.sherwood@arm.com>
17089 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17090 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17091 (popcount_loc_descriptor): Likewise.
17092 (bswap_loc_descriptor): Likewise.
17093 (rotate_loc_descriptor): Likewise.
17094 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17095 calling the functions above.
17097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17098 Alan Hayward <alan.hayward@arm.com>
17099 David Sherwood <david.sherwood@arm.com>
17101 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17103 (try_combine): Likewise.
17104 (simplify_if_then_else): Likewise.
17105 * cse.c (cse_insn): Likewise.
17106 * dwarf2out.c (mem_loc_descriptor): Likewise.
17107 * emit-rtl.c (gen_lowpart_common): Likewise.
17108 * simplify-rtx.c (simplify_truncation): Likewise.
17109 (simplify_binary_operation_1): Likewise.
17110 (simplify_const_relational_operation): Likewise.
17111 (simplify_ternary_operation): Likewise.
17112 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17115 Alan Hayward <alan.hayward@arm.com>
17116 David Sherwood <david.sherwood@arm.com>
17118 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17119 * reload.c (push_reload): Likewise.
17120 (find_reloads): Likewise.
17122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17123 Alan Hayward <alan.hayward@arm.com>
17124 David Sherwood <david.sherwood@arm.com>
17126 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17127 (make_compound_operation_int): Likewise.
17128 (change_zero_ext): Likewise.
17129 * expr.c (convert_move): Likewise.
17130 (convert_modes): Likewise.
17131 * fwprop.c (forward_propagate_subreg): Likewise.
17132 * loop-iv.c (get_biv_step_1): Likewise.
17133 * optabs.c (widen_operand): Likewise.
17134 * postreload.c (move2add_valid_value_p): Likewise.
17135 * recog.c (simplify_while_replacing): Likewise.
17136 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17137 (simplify_binary_operation_1): Likewise. Remove redundant
17138 mode equality check.
17140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17141 Alan Hayward <alan.hayward@arm.com>
17142 David Sherwood <david.sherwood@arm.com>
17144 * combine.c (combine_simplify_rtx): Add checks for
17145 is_a <scalar_int_mode>.
17146 (simplify_if_then_else): Likewise.
17147 (make_field_assignment): Likewise.
17148 (simplify_comparison): Likewise.
17149 * ifcvt.c (noce_try_bitop): Likewise.
17150 * loop-invariant.c (canonicalize_address_mult): Likewise.
17151 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17154 Alan Hayward <alan.hayward@arm.com>
17155 David Sherwood <david.sherwood@arm.com>
17157 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17158 is_a <scalar_int_mode> instead of != BLKmode.
17160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17161 Alan Hayward <alan.hayward@arm.com>
17162 David Sherwood <david.sherwood@arm.com>
17164 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17165 instead of != VOIDmode.
17166 * combine.c (if_then_else_cond): Likewise.
17167 (change_zero_ext): Likewise.
17168 * dwarf2out.c (mem_loc_descriptor): Likewise.
17169 (loc_descriptor): Likewise.
17170 * rtlanal.c (canonicalize_condition): Likewise.
17171 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17174 Alan Hayward <alan.hayward@arm.com>
17175 David Sherwood <david.sherwood@arm.com>
17177 * simplify-rtx.c (simplify_binary_operation_1): Use
17178 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17181 Alan Hayward <alan.hayward@arm.com>
17182 David Sherwood <david.sherwood@arm.com>
17184 * wide-int.h (int_traits<unsigned char>) New class.
17185 (int_traits<unsigned short>) Likewise.
17186 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17187 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17189 * combine.c (set_nonzero_bits_and_sign_copies): Use
17190 is_a <scalar_int_mode>.
17191 (find_split_point): Likewise.
17192 (combine_simplify_rtx): Likewise.
17193 (simplify_logical): Likewise.
17194 (expand_compound_operation): Likewise.
17195 (expand_field_assignment): Likewise.
17196 (make_compound_operation): Likewise.
17197 (extended_count): Likewise.
17198 (change_zero_ext): Likewise.
17199 (simplify_comparison): Likewise.
17200 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17201 (ucompare_loc_descriptor): Likewise.
17202 (minmax_loc_descriptor): Likewise.
17203 (mem_loc_descriptor): Likewise.
17204 (loc_descriptor): Likewise.
17205 * expmed.c (init_expmed_one_mode): Likewise.
17206 * lra-constraints.c (lra_constraint_offset): Likewise.
17207 * optabs.c (prepare_libcall_arg): Likewise.
17208 * postreload.c (move2add_note_store): Likewise.
17209 * reload.c (operands_match_p): Likewise.
17210 * rtl.h (load_extend_op): Likewise.
17211 * rtlhooks.c (gen_lowpart_general): Likewise.
17212 * simplify-rtx.c (simplify_truncation): Likewise.
17213 (simplify_unary_operation_1): Likewise.
17214 (simplify_binary_operation_1): Likewise.
17215 (simplify_const_binary_operation): Likewise.
17216 (simplify_const_relational_operation): Likewise.
17217 (simplify_subreg): Likewise.
17218 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17219 * var-tracking.c (adjust_mems): Likewise.
17220 (prepare_call_arguments): Likewise.
17222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17223 Alan Hayward <alan.hayward@arm.com>
17224 David Sherwood <david.sherwood@arm.com>
17226 * machmode.h (is_int_mode): New fuction.
17227 * combine.c (find_split_point): Use it.
17228 (combine_simplify_rtx): Likewise.
17229 (simplify_if_then_else): Likewise.
17230 (simplify_set): Likewise.
17231 (simplify_shift_const_1): Likewise.
17232 (simplify_comparison): Likewise.
17233 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17234 * cse.c (notreg_cost): Likewise.
17235 (cse_insn): Likewise.
17236 * cselib.c (cselib_lookup_1): Likewise.
17237 * dojump.c (do_jump_1): Likewise.
17238 (do_compare_rtx_and_jump): Likewise.
17239 * dse.c (get_call_args): Likewise.
17240 * dwarf2out.c (rtl_for_decl_init): Likewise.
17241 (native_encode_initializer): Likewise.
17242 * expmed.c (emit_store_flag_1): Likewise.
17243 (emit_store_flag): Likewise.
17244 * expr.c (convert_modes): Likewise.
17245 (store_field): Likewise.
17246 (expand_expr_real_1): Likewise.
17247 * fold-const.c (fold_read_from_constant_string): Likewise.
17248 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17249 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17250 * optabs.c (expand_binop): Likewise.
17251 (expand_unop): Likewise.
17252 (expand_abs_nojump): Likewise.
17253 (expand_one_cmpl_abs_nojump): Likewise.
17254 * simplify-rtx.c (mode_signbit_p): Likewise.
17255 (val_signbit_p): Likewise.
17256 (val_signbit_known_set_p): Likewise.
17257 (val_signbit_known_clear_p): Likewise.
17258 (simplify_relational_operation_1): Likewise.
17259 * tree.c (vector_type_mode): Likewise.
17261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17262 Alan Hayward <alan.hayward@arm.com>
17263 David Sherwood <david.sherwood@arm.com>
17265 * machmode.h (smallest_mode_for_size): Fix formatting.
17266 (smallest_int_mode_for_size): New function.
17267 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17268 instead of smallest_mode_for_size.
17269 * combine.c (make_extraction): Likewise.
17270 * config/arc/arc.c (arc_expand_movmem): Likewise.
17271 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17272 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17273 * config/s390/s390.c (s390_expand_insv): Likewise.
17274 * config/sparc/sparc.c (assign_int_registers): Likewise.
17275 * config/spu/spu.c (spu_function_value): Likewise.
17276 (spu_function_arg): Likewise.
17277 * coverage.c (get_gcov_type): Likewise.
17278 (get_gcov_unsigned_t): Likewise.
17279 * dse.c (find_shift_sequence): Likewise.
17280 * expmed.c (store_bit_field_1): Likewise.
17281 * expr.c (convert_move): Likewise.
17282 (store_field): Likewise.
17283 * internal-fn.c (expand_arith_overflow): Likewise.
17284 * optabs-query.c (get_best_extraction_insn): Likewise.
17285 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17286 * stor-layout.c (layout_type): Likewise.
17287 (initialize_sizetypes): Likewise.
17288 * targhooks.c (default_get_mask_mode): Likewise.
17289 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17292 Alan Hayward <alan.hayward@arm.com>
17293 David Sherwood <david.sherwood@arm.com>
17295 * machmode.h (opt_mode::else_blk): New function.
17296 (int_mode_for_mode): Declare.
17297 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17298 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17300 * cfgexpand.c (expand_debug_expr): Likewise.
17301 * combine.c (gen_lowpart_or_truncate): Likewise.
17302 (gen_lowpart_for_combine): Likewise.
17303 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17304 * config/avr/avr.c (avr_to_int_mode): Likewise.
17305 (avr_out_plus_1): Likewise.
17306 (avr_out_plus): Likewise.
17307 (avr_out_round): Likewise.
17308 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17309 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17310 (s390_expand_vcond): Likewise.
17311 * config/spu/spu.c (spu_split_immediate): Likewise.
17312 (spu_expand_mov): Likewise.
17313 * dse.c (get_stored_val): Likewise.
17314 * expmed.c (store_bit_field_1): Likewise.
17315 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17317 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17318 (extract_low_bits): Likewise.
17319 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17320 handling rather than repeating the check.
17321 (emit_group_store): Likewise.
17322 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17323 * optabs.c (expand_absneg_bit): Likewise.
17324 (expand_copysign_absneg): Likewise.
17325 (expand_copysign_bit): Likewise.
17326 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17327 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17328 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17329 * var-tracking.c (prepare_call_arguments): Likewise.
17330 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17331 int_mode_for_mode instead of mode_for_size.
17332 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17335 Alan Hayward <alan.hayward@arm.com>
17336 David Sherwood <david.sherwood@arm.com>
17338 * machmode.h (int_mode_for_size): New function.
17339 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17340 instead of mode_for_size.
17341 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17343 * combine.c (expand_field_assignment): Use int_mode_for_size
17344 instead of mode_for_size.
17345 (make_extraction): Likewise.
17346 (simplify_shift_const_1): Likewise.
17347 (simplify_comparison): Likewise.
17348 * dojump.c (do_jump): Likewise.
17349 * dwarf2out.c (mem_loc_descriptor): Likewise.
17350 * emit-rtl.c (init_derived_machine_modes): Likewise.
17351 * expmed.c (flip_storage_order): Likewise.
17352 (convert_extracted_bit_field): Likewise.
17353 * expr.c (copy_blkmode_from_reg): Likewise.
17354 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17355 * internal-fn.c (expand_mul_overflow): Likewise.
17356 * lower-subreg.c (simple_move): Likewise.
17357 * optabs-libfuncs.c (init_optabs): Likewise.
17358 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17359 * tree.c (vector_type_mode): Likewise.
17360 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17361 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17362 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17363 * tree-vect-stmts.c (vectorizable_load): Likewise.
17364 (vectorizable_store): Likewise.
17366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17367 Alan Hayward <alan.hayward@arm.com>
17368 David Sherwood <david.sherwood@arm.com>
17370 * coretypes.h (pod_mode): New type.
17371 (scalar_int_mode_pod): New typedef.
17372 * machmode.h (pod_mode): New class.
17373 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17374 * genmodes.c (emit_mode_int_n): Update accordingly.
17375 * lower-subreg.h (target_lower_subreg): Change type to
17376 scalar_int_mode_pod.
17377 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17378 scalar_int_mode_pod.
17380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17381 Alan Hayward <alan.hayward@arm.com>
17382 David Sherwood <david.sherwood@arm.com>
17384 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17385 machine_mode to scalar_int_mode.
17386 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17387 (rs6000_option_override_internal): Remove cast to int.
17388 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17389 machine_mode to scalar_int_mode.
17390 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17391 (rs6000_option_override_internal): Remove cast to int.
17392 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17393 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17395 * config/s390/s390.c (s390_expand_builtin): Likewise.
17396 * coretypes.h (scalar_int_mode): New type.
17397 (opt_scalar_int_mode): New typedef.
17398 * machmode.h (scalar_int_mode): New class.
17399 (scalar_int_mode::includes_p): New function.
17400 (byte_mode): Change type to scalar_int_mode.
17401 (word_mode): Likewise.
17402 (ptr_mode): Likewise.
17403 * emit-rtl.c (byte_mode): Likewise.
17404 (word_mode): Likewise.
17405 (ptr_mode): Likewise.
17406 (init_derived_machine_modes): Update accordingly.
17407 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17408 and MODE_PARTIAL_INT.
17409 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17410 opt_scalar_int_mode.
17412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17413 Alan Hayward <alan.hayward@arm.com>
17414 David Sherwood <david.sherwood@arm.com>
17416 * target.def (libgcc_floating_mode_supported_p): Take a
17418 * doc/tm.texi: Regenerate.
17419 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17421 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17422 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17426 Alan Hayward <alan.hayward@arm.com>
17427 David Sherwood <david.sherwood@arm.com>
17429 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17430 * doc/tm.texi: Regenerate.
17431 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17432 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17433 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17434 * targhooks.h (default_floatn_mode): Likewise.
17435 * targhooks.c (default_floatn_mode): Likewise.
17436 * tree.c (build_common_tree_nodes): Update accordingly.
17438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17439 Alan Hayward <alan.hayward@arm.com>
17440 David Sherwood <david.sherwood@arm.com>
17442 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17443 (mode_iterator::iterate_p): Likewise.
17444 (mode_iterator::get_wider): Likewise.
17445 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17448 Alan Hayward <alan.hayward@arm.com>
17449 David Sherwood <david.sherwood@arm.com>
17451 * coretypes.h (opt_scalar_float_mode): New typedef.
17452 * machmode.h (float_mode_for_size): New function.
17453 * emit-rtl.c (double_mode): Delete.
17454 (init_emit_once): Use float_mode_for_size.
17455 * stor-layout.c (layout_type): Likewise.
17456 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17459 Alan Hayward <alan.hayward@arm.com>
17460 David Sherwood <david.sherwood@arm.com>
17462 * output.h (assemble_real): Take a scalar_float_mode.
17463 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17464 * config/arm/arm.md (consttable_4): Likewise.
17465 (consttable_8): Likewise.
17466 (consttable_16): Likewise.
17467 * config/mips/mips.md (consttable_float): Likewise.
17468 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17469 * varasm.c (assemble_real): Take a scalar_float_mode.
17470 (output_constant_pool_2): Update accordingly.
17471 (output_constant): Likewise.
17473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17474 Alan Hayward <alan.hayward@arm.com>
17475 David Sherwood <david.sherwood@arm.com>
17477 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17478 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17479 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17480 (native_encode_real): Likewise.
17481 (native_interpret_real): Likewise.
17482 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17483 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17486 Alan Hayward <alan.hayward@arm.com>
17487 David Sherwood <david.sherwood@arm.com>
17489 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17490 <scalar_float_mode>. Simplify.
17491 (gen_extend_conv_libfunc): Likewise.
17493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17494 Alan Hayward <alan.hayward@arm.com>
17495 David Sherwood <david.sherwood@arm.com>
17497 * coretypes.h (scalar_float_mode): New type.
17498 * machmode.h (mode_traits::from_int): Use machine_mode if
17499 USE_ENUM_MODES is defined.
17500 (is_a): New function.
17502 (dyn_cast): Likewise.
17503 (scalar_float_mode): New class.
17504 (scalar_float_mode::includes_p): New function.
17505 (is_float_mode): Likewise.
17506 * gdbhooks.py (MachineModePrinter): New class.
17507 (build_pretty_printer): Use it for scalar_float_mode.
17508 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17509 (format_helper::format_helper): Turn into a template.
17510 * genmodes.c (get_mode_class): New function.
17511 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17512 or machine_mode if none.
17513 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17514 as_a <scalar_float_mode>.
17515 * dwarf2out.c (mem_loc_descriptor): Likewise.
17516 (insert_float): Likewise.
17517 (add_const_value_attribute): Likewise.
17518 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17519 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17520 (expand_unop): Update accordingly.
17521 (expand_abs_nojump): Likewise.
17522 (expand_copysign_absneg): Take a scalar_float_mode.
17523 (expand_copysign_bit): Likewise.
17524 (expand_copysign): Update accordingly.
17526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17527 Alan Hayward <alan.hayward@arm.com>
17528 David Sherwood <david.sherwood@arm.com>
17530 * coretypes.h (opt_mode): New class.
17531 * machmode.h (opt_mode): Likewise.
17532 (opt_mode::else_void): New function.
17533 (opt_mode::require): Likewise.
17534 (opt_mode::exists): Likewise.
17535 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17536 (GET_MODE_2XWIDER_MODE): Likewise.
17537 (mode_iterator::get_wider): Update accordingly.
17538 (mode_iterator::get_2xwider): Likewise.
17539 (mode_iterator::get_known_wider): Likewise, turning into a template.
17540 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17541 forcing a wider mode to exist.
17542 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17543 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17544 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17545 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17546 * lower-subreg.c (init_lower_subreg): Likewise.
17547 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17548 on the final iteration.
17549 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17550 a wider mode exists before asking for a move pattern.
17551 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17552 forcing a wider mode to exist.
17553 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17554 returning false if no such mode exists.
17555 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17556 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17557 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17558 Avoid checking for a MODE_INT if we already know the mode is not a
17560 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17561 forcing a wider mode to exist.
17562 (expmed_mult_highpart_optab): Likewise.
17563 (expmed_mult_highpart): Likewise.
17564 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17566 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17567 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17568 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17569 * internal-fn.c (expand_mul_overflow): Update use of
17570 GET_MODE_2XWIDER_MODE.
17571 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17572 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17573 GET_MODE_WIDER_MODE.
17574 (convert_plusminus_to_widen): Likewise.
17575 * tree-switch-conversion.c (array_value_type): Likewise.
17576 * var-tracking.c (emit_note_insn_var_location): Likewise.
17577 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17578 Return false inside rather than outside the loop if no wider mode
17580 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17581 and GET_MODE_2XWIDER_MODE
17582 (can_compare_p): Use else_void.
17583 * gdbhooks.py (OptMachineModePrinter): New class.
17584 (build_pretty_printer): Use it for opt_mode.
17586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17587 Alan Hayward <alan.hayward@arm.com>
17588 David Sherwood <david.sherwood@arm.com>
17590 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17591 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17594 Alan Hayward <alan.hayward@arm.com>
17595 David Sherwood <david.sherwood@arm.com>
17597 * machmode.h (mode_traits): New structure.
17598 (get_narrowest_mode): New function.
17599 (mode_iterator::start): Likewise.
17600 (mode_iterator::iterate_p): Likewise.
17601 (mode_iterator::get_wider): Likewise.
17602 (mode_iterator::get_known_wider): Likewise.
17603 (mode_iterator::get_2xwider): Likewise.
17604 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17605 (FOR_EACH_MODE): Likewise.
17606 (FOR_EACH_MODE_FROM): Likewise.
17607 (FOR_EACH_MODE_UNTIL): Likewise.
17608 (FOR_EACH_WIDER_MODE): Likewise.
17609 (FOR_EACH_2XWIDER_MODE): Likewise.
17610 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17611 * combine.c (simplify_comparison): Likewise
17612 * config/i386/i386.c (type_natural_mode): Likewise.
17613 * cse.c (cse_insn): Likewise.
17614 * dse.c (find_shift_sequence): Likewise.
17615 * emit-rtl.c (init_derived_machine_modes): Likewise.
17616 (init_emit_once): Likewise.
17617 * explow.c (hard_function_value): Likewise.
17618 * expmed.c (extract_fixed_bit_field_1): Likewise.
17619 (extract_bit_field_1): Likewise.
17620 (expand_divmod): Likewise.
17621 (emit_store_flag_1): Likewise.
17622 * expr.c (init_expr_target): Likewise.
17623 (convert_move): Likewise.
17624 (alignment_for_piecewise_move): Likewise.
17625 (widest_int_mode_for_size): Likewise.
17626 (emit_block_move_via_movmem): Likewise.
17627 (copy_blkmode_to_reg): Likewise.
17628 (set_storage_via_setmem): Likewise.
17629 (compress_float_constant): Likewise.
17630 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17631 * optabs-query.c (get_best_extraction_insn): Likewise.
17632 * optabs.c (expand_binop): Likewise.
17633 (expand_twoval_unop): Likewise.
17634 (expand_twoval_binop): Likewise.
17635 (widen_leading): Likewise.
17636 (widen_bswap): Likewise.
17637 (expand_parity): Likewise.
17638 (expand_unop): Likewise.
17639 (prepare_cmp_insn): Likewise.
17640 (prepare_float_lib_cmp): Likewise.
17641 (expand_float): Likewise.
17642 (expand_fix): Likewise.
17643 (expand_sfix_optab): Likewise.
17644 * postreload.c (move2add_use_add2_insn): Likewise.
17645 * reg-stack.c (reg_to_stack): Likewise.
17646 * reginfo.c (choose_hard_reg_mode): Likewise.
17647 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17648 * stor-layout.c (mode_for_size): Likewise.
17649 (smallest_mode_for_size): Likewise.
17650 (mode_for_vector): Likewise.
17651 (finish_bitfield_representative): Likewise.
17652 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17653 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17654 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17655 * var-tracking.c (prepare_call_arguments): Likewise.
17657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17658 Alan Hayward <alan.hayward@arm.com>
17659 David Sherwood <david.sherwood@arm.com>
17661 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17662 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17663 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17664 * machmode.h (mode_size): Move earlier in file.
17665 (mode_precision): Likewise.
17666 (mode_inner): Likewise.
17667 (mode_nunits): Likewise.
17668 (mode_unit_size): Likewise.
17669 (unit_unit_precision): Likewise.
17670 (mode_wider): Likewise.
17671 (mode_2xwider): Likewise.
17672 (machine_mode): New class.
17673 (mode_to_bytes): New function.
17674 (mode_to_bits): Likewise.
17675 (mode_to_precision): Likewise.
17676 (mode_to_inner): Likewise.
17677 (mode_to_unit_size): Likewise.
17678 (mode_to_unit_precision): Likewise.
17679 (mode_to_nunits): Likewise.
17680 (GET_MODE_SIZE): Use mode_to_bytes.
17681 (GET_MODE_BITSIZE): Use mode_to_bits.
17682 (GET_MODE_PRECISION): Use mode_to_precision.
17683 (GET_MODE_INNER): Use mode_to_inner.
17684 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17685 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17686 (GET_MODE_NUNITS): Use mode_to_nunits.
17687 * system.h (ALWAYS_INLINE): New macro.
17688 * config/powerpcspe/powerpcspe-c.c
17689 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17690 int for arg1_mode and arg2_mode.
17692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17693 Alan Hayward <alan.hayward@arm.com>
17694 David Sherwood <david.sherwood@arm.com>
17696 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17697 Prefix mode names with E_ in case statements.
17698 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17699 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
17700 (aarch64_split_simd_move): Likewise.
17701 (aarch64_gen_storewb_pair): Likewise.
17702 (aarch64_gen_loadwb_pair): Likewise.
17703 (aarch64_gen_store_pair): Likewise.
17704 (aarch64_gen_load_pair): Likewise.
17705 (aarch64_get_condition_code_1): Likewise.
17706 (aarch64_constant_pool_reload_icode): Likewise.
17707 (get_rsqrte_type): Likewise.
17708 (get_rsqrts_type): Likewise.
17709 (get_recpe_type): Likewise.
17710 (get_recps_type): Likewise.
17711 (aarch64_gimplify_va_arg_expr): Likewise.
17712 (aarch64_simd_container_mode): Likewise.
17713 (aarch64_emit_load_exclusive): Likewise.
17714 (aarch64_emit_store_exclusive): Likewise.
17715 (aarch64_expand_compare_and_swap): Likewise.
17716 (aarch64_gen_atomic_cas): Likewise.
17717 (aarch64_emit_bic): Likewise.
17718 (aarch64_emit_atomic_swap): Likewise.
17719 (aarch64_emit_atomic_load_op): Likewise.
17720 (aarch64_evpc_trn): Likewise.
17721 (aarch64_evpc_uzp): Likewise.
17722 (aarch64_evpc_zip): Likewise.
17723 (aarch64_evpc_ext): Likewise.
17724 (aarch64_evpc_rev): Likewise.
17725 (aarch64_evpc_dup): Likewise.
17726 (aarch64_gen_ccmp_first): Likewise.
17727 (aarch64_gen_ccmp_next): Likewise.
17728 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17729 (alpha_emit_xfloating_libcall): Likewise.
17730 (emit_insxl): Likewise.
17731 (alpha_arg_type): Likewise.
17732 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17733 (arc_preferred_simd_mode): Likewise.
17734 (arc_secondary_reload): Likewise.
17735 (get_arc_condition_code): Likewise.
17736 (arc_print_operand): Likewise.
17737 (arc_legitimate_constant_p): Likewise.
17738 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17739 * config/arc/arc.md (casesi_load): Likewise.
17740 (casesi_compact_jump): Likewise.
17741 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17742 (cc_use_register): Likewise.
17743 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17744 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17745 (arm_init_iwmmxt_builtins): Likewise.
17746 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17747 (neon_expand_vector_init): Likewise.
17748 (arm_attr_length_move_neon): Likewise.
17749 (maybe_get_arm_condition_code): Likewise.
17750 (arm_emit_vector_const): Likewise.
17751 (arm_preferred_simd_mode): Likewise.
17752 (arm_output_iwmmxt_tinsr): Likewise.
17753 (thumb1_output_casesi): Likewise.
17754 (thumb2_output_casesi): Likewise.
17755 (arm_emit_load_exclusive): Likewise.
17756 (arm_emit_store_exclusive): Likewise.
17757 (arm_expand_compare_and_swap): Likewise.
17758 (arm_evpc_neon_vuzp): Likewise.
17759 (arm_evpc_neon_vzip): Likewise.
17760 (arm_evpc_neon_vrev): Likewise.
17761 (arm_evpc_neon_vtrn): Likewise.
17762 (arm_evpc_neon_vext): Likewise.
17763 (arm_validize_comparison): Likewise.
17764 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17765 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17766 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17767 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17768 (c6x_preferred_simd_mode): Likewise.
17769 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17770 (epiphany_rtx_costs): Likewise.
17771 * config/epiphany/predicates.md (proper_comparison_operator):
17773 * config/frv/frv.c (condexec_memory_operand): Likewise.
17774 (frv_emit_move): Likewise.
17775 (output_move_single): Likewise.
17776 (output_condmove_single): Likewise.
17777 (frv_hard_regno_mode_ok): Likewise.
17778 (frv_matching_accg_mode): Likewise.
17779 * config/h8300/h8300.c (split_adds_subs): Likewise.
17780 (h8300_rtx_costs): Likewise.
17781 (h8300_print_operand): Likewise.
17782 (compute_mov_length): Likewise.
17783 (output_logical_op): Likewise.
17784 (compute_logical_op_length): Likewise.
17785 (compute_logical_op_cc): Likewise.
17786 (h8300_shift_needs_scratch_p): Likewise.
17787 (output_a_shift): Likewise.
17788 (compute_a_shift_length): Likewise.
17789 (compute_a_shift_cc): Likewise.
17790 (expand_a_rotate): Likewise.
17791 (output_a_rotate): Likewise.
17792 * config/i386/i386.c (classify_argument): Likewise.
17793 (function_arg_advance_32): Likewise.
17794 (function_arg_32): Likewise.
17795 (function_arg_64): Likewise.
17796 (function_value_64): Likewise.
17797 (ix86_gimplify_va_arg): Likewise.
17798 (ix86_legitimate_constant_p): Likewise.
17799 (put_condition_code): Likewise.
17800 (split_double_mode): Likewise.
17801 (ix86_avx256_split_vector_move_misalign): Likewise.
17802 (ix86_expand_vector_logical_operator): Likewise.
17803 (ix86_split_idivmod): Likewise.
17804 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17805 (ix86_build_const_vector): Likewise.
17806 (ix86_build_signbit_mask): Likewise.
17807 (ix86_match_ccmode): Likewise.
17808 (ix86_cc_modes_compatible): Likewise.
17809 (ix86_expand_branch): Likewise.
17810 (ix86_expand_sse_cmp): Likewise.
17811 (ix86_expand_sse_movcc): Likewise.
17812 (ix86_expand_int_sse_cmp): Likewise.
17813 (ix86_expand_vec_perm_vpermi2): Likewise.
17814 (ix86_expand_vec_perm): Likewise.
17815 (ix86_expand_sse_unpack): Likewise.
17816 (ix86_expand_int_addcc): Likewise.
17817 (ix86_split_to_parts): Likewise.
17818 (ix86_vectorize_builtin_gather): Likewise.
17819 (ix86_vectorize_builtin_scatter): Likewise.
17820 (avx_vpermilp_parallel): Likewise.
17821 (inline_memory_move_cost): Likewise.
17822 (ix86_tieable_integer_mode_p): Likewise.
17823 (x86_maybe_negate_const_int): Likewise.
17824 (ix86_expand_vector_init_duplicate): Likewise.
17825 (ix86_expand_vector_init_one_nonzero): Likewise.
17826 (ix86_expand_vector_init_one_var): Likewise.
17827 (ix86_expand_vector_init_concat): Likewise.
17828 (ix86_expand_vector_init_interleave): Likewise.
17829 (ix86_expand_vector_init_general): Likewise.
17830 (ix86_expand_vector_set): Likewise.
17831 (ix86_expand_vector_extract): Likewise.
17832 (emit_reduc_half): Likewise.
17833 (ix86_emit_i387_round): Likewise.
17834 (ix86_mangle_type): Likewise.
17835 (ix86_expand_round_sse4): Likewise.
17836 (expand_vec_perm_blend): Likewise.
17837 (canonicalize_vector_int_perm): Likewise.
17838 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17839 (expand_vec_perm_1): Likewise.
17840 (expand_vec_perm_interleave3): Likewise.
17841 (expand_vec_perm_even_odd_pack): Likewise.
17842 (expand_vec_perm_even_odd_1): Likewise.
17843 (expand_vec_perm_broadcast_1): Likewise.
17844 (ix86_vectorize_vec_perm_const_ok): Likewise.
17845 (ix86_expand_vecop_qihi): Likewise.
17846 (ix86_expand_mul_widen_hilo): Likewise.
17847 (ix86_expand_sse2_abs): Likewise.
17848 (ix86_expand_pextr): Likewise.
17849 (ix86_expand_pinsr): Likewise.
17850 (ix86_preferred_simd_mode): Likewise.
17851 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17852 * config/i386/sse.md (*andnot<mode>3): Likewise.
17853 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17854 (*<code><mode>3): Likewise.
17855 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17856 (ia64_expand_atomic_op): Likewise.
17857 (ia64_arg_type): Likewise.
17858 (ia64_mode_to_int): Likewise.
17859 (ia64_scalar_mode_supported_p): Likewise.
17860 (ia64_vector_mode_supported_p): Likewise.
17861 (expand_vec_perm_broadcast): Likewise.
17862 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17863 (iq2000_function_arg_advance): Likewise.
17864 (iq2000_function_arg): Likewise.
17865 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17866 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17867 (m68k_libcall_value): Likewise.
17868 (m68k_function_value): Likewise.
17869 (sched_attr_op_type): Likewise.
17870 * config/mcore/mcore.c (mcore_output_move): Likewise.
17871 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17873 (microblaze_function_arg): Likewise.
17874 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17875 (mips_print_operand): Likewise.
17876 (mips_mode_ok_for_mov_fmt_p): Likewise.
17877 (mips_vector_mode_supported_p): Likewise.
17878 (mips_preferred_simd_mode): Likewise.
17879 (mips_expand_vpc_loongson_even_odd): Likewise.
17880 (mips_expand_vec_unpack): Likewise.
17881 (mips_expand_vi_broadcast): Likewise.
17882 (mips_expand_vector_init): Likewise.
17883 (mips_expand_vec_reduc): Likewise.
17884 (mips_expand_msa_cmp): Likewise.
17885 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17886 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17887 (cc_flags_for_mode): Likewise.
17888 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17889 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17890 (nds32_output_casesi_pc_relative): Likewise.
17891 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17892 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17893 (nvptx_gen_unpack): Likewise.
17894 (nvptx_gen_pack): Likewise.
17895 (nvptx_gen_shuffle): Likewise.
17896 (nvptx_gen_wcast): Likewise.
17897 (nvptx_preferred_simd_mode): Likewise.
17898 * config/pa/pa.c (pa_secondary_reload): Likewise.
17899 * config/pa/predicates.md (base14_operand): Likewise.
17900 * config/powerpcspe/powerpcspe-c.c
17901 (altivec_resolve_overloaded_builtin): Likewise.
17902 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17904 (rs6000_preferred_simd_mode): Likewise.
17905 (output_vec_const_move): Likewise.
17906 (rs6000_expand_vector_extract): Likewise.
17907 (rs6000_split_vec_extract_var): Likewise.
17908 (reg_offset_addressing_ok_p): Likewise.
17909 (rs6000_legitimate_offset_address_p): Likewise.
17910 (rs6000_legitimize_address): Likewise.
17911 (rs6000_emit_set_const): Likewise.
17912 (rs6000_const_vec): Likewise.
17913 (rs6000_emit_move): Likewise.
17914 (spe_build_register_parallel): Likewise.
17915 (rs6000_darwin64_record_arg_recurse): Likewise.
17916 (swap_selector_for_mode): Likewise.
17917 (spe_init_builtins): Likewise.
17918 (paired_init_builtins): Likewise.
17919 (altivec_init_builtins): Likewise.
17920 (do_load_for_compare): Likewise.
17921 (rs6000_generate_compare): Likewise.
17922 (rs6000_expand_float128_convert): Likewise.
17923 (emit_load_locked): Likewise.
17924 (emit_store_conditional): Likewise.
17925 (rs6000_output_function_epilogue): Likewise.
17926 (rs6000_handle_altivec_attribute): Likewise.
17927 (rs6000_function_value): Likewise.
17928 (emit_fusion_gpr_load): Likewise.
17929 (emit_fusion_p9_load): Likewise.
17930 (emit_fusion_p9_store): Likewise.
17931 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17932 (fusion_gpr_mem_load): Likewise.
17933 (fusion_addis_mem_combo_load): Likewise.
17934 (fusion_addis_mem_combo_store): Likewise.
17935 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17936 (fusion_gpr_mem_load): Likewise.
17937 (fusion_addis_mem_combo_load): Likewise.
17938 (fusion_addis_mem_combo_store): Likewise.
17939 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17941 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17942 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17943 (rs6000_preferred_simd_mode): Likewise.
17944 (output_vec_const_move): Likewise.
17945 (rs6000_expand_vector_extract): Likewise.
17946 (rs6000_split_vec_extract_var): Likewise.
17947 (reg_offset_addressing_ok_p): Likewise.
17948 (rs6000_legitimate_offset_address_p): Likewise.
17949 (rs6000_legitimize_address): Likewise.
17950 (rs6000_emit_set_const): Likewise.
17951 (rs6000_const_vec): Likewise.
17952 (rs6000_emit_move): Likewise.
17953 (rs6000_darwin64_record_arg_recurse): Likewise.
17954 (swap_selector_for_mode): Likewise.
17955 (paired_init_builtins): Likewise.
17956 (altivec_init_builtins): Likewise.
17957 (rs6000_expand_float128_convert): Likewise.
17958 (emit_load_locked): Likewise.
17959 (emit_store_conditional): Likewise.
17960 (rs6000_output_function_epilogue): Likewise.
17961 (rs6000_handle_altivec_attribute): Likewise.
17962 (rs6000_function_value): Likewise.
17963 (emit_fusion_gpr_load): Likewise.
17964 (emit_fusion_p9_load): Likewise.
17965 (emit_fusion_p9_store): Likewise.
17966 * config/rx/rx.c (rx_gen_move_template): Likewise.
17967 (flags_from_mode): Likewise.
17968 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17969 (s390_slb_comparison): Likewise.
17970 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17971 (s390_vector_mode_supported_p): Likewise.
17972 (s390_cc_modes_compatible): Likewise.
17973 (s390_match_ccmode_set): Likewise.
17974 (s390_canonicalize_comparison): Likewise.
17975 (s390_emit_compare_and_swap): Likewise.
17976 (s390_branch_condition_mask): Likewise.
17977 (s390_rtx_costs): Likewise.
17978 (s390_secondary_reload): Likewise.
17979 (__SECONDARY_RELOAD_CASE): Likewise.
17980 (s390_expand_cs): Likewise.
17981 (s390_preferred_simd_mode): Likewise.
17982 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17983 * config/sh/sh.c (sh_print_operand): Likewise.
17984 (dump_table): Likewise.
17985 (sh_secondary_reload): Likewise.
17986 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17987 * config/sh/sh.md (casesi_worker_1): Likewise.
17988 (casesi_worker_2): Likewise.
17989 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17990 (fcc_comparison_operator): Likewise.
17991 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17992 (emit_soft_tfmode_cvt): Likewise.
17993 (sparc_preferred_simd_mode): Likewise.
17994 (output_cbranch): Likewise.
17995 (sparc_print_operand): Likewise.
17996 (sparc_expand_vec_perm_bmask): Likewise.
17997 (vector_init_bshuffle): Likewise.
17998 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17999 (spu_vector_mode_supported_p): Likewise.
18000 (spu_expand_insv): Likewise.
18001 (spu_emit_branch_or_set): Likewise.
18002 (spu_handle_vector_attribute): Likewise.
18003 (spu_builtin_splats): Likewise.
18004 (spu_builtin_extract): Likewise.
18005 (spu_builtin_promote): Likewise.
18006 (spu_expand_sign_extend): Likewise.
18007 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18008 (tilegx_simd_int): Likewise.
18009 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18010 (tilepro_simd_int): Likewise.
18011 * config/v850/v850.c (const_double_split): Likewise.
18012 (v850_print_operand): Likewise.
18013 (ep_memory_offset): Likewise.
18014 * config/vax/vax.c (vax_rtx_costs): Likewise.
18015 (vax_output_int_move): Likewise.
18016 (vax_output_int_add): Likewise.
18017 (vax_output_int_subtract): Likewise.
18018 * config/visium/predicates.md (visium_branch_operator): Likewise.
18019 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18020 (visium_print_operand_address): Likewise.
18021 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18022 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18023 (xtensa_expand_conditional_branch): Likewise.
18024 (xtensa_copy_incoming_a7): Likewise.
18025 (xtensa_output_literal): Likewise.
18026 * dfp.c (decimal_real_maxval): Likewise.
18027 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18030 Alan Hayward <alan.hayward@arm.com>
18031 David Sherwood <david.sherwood@arm.com>
18033 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18034 (mode_nunits_inline): Likewise.
18035 (mode_inner_inline): Likewise.
18036 (mode_unit_size_inline): Likewise.
18037 (mode_unit_precision_inline): Likewise.
18038 (emit_insn_modes_h): Likewise. Also emit a #define of the
18040 (emit_mode_wider): Add an E_ prefix to mode names.
18041 (emit_mode_complex): Likewise.
18042 (emit_mode_inner): Likewise.
18043 (emit_mode_adjustments): Likewise.
18044 (emit_mode_int_n): Likewise.
18045 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18046 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18047 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18048 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18049 (CRC32_BUILTIN, ENTRY): Likewise.
18050 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18051 (aarch64_pop_regs): Likewise.
18052 (aarch64_process_components): Likewise.
18053 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18054 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18055 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18056 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18057 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18058 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18060 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18061 (pcmpistr): Likewise.
18062 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18063 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18064 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18066 * config/rl78/rl78.c (mduc_regs): Likewise.
18067 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18068 (htm_expand_builtin): Likewise.
18069 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18070 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18071 * config/xtensa/xtensa.c (print_operand): Likewise.
18072 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18073 (NUM_MODE_VECTOR_INT): Likewise.
18074 * genoutput.c (null_operand): Likewise.
18075 (output_operand_data): Likewise.
18076 * genrecog.c (print_parameter_value): Likewise.
18077 * lra.c (debug_operand_data): Likewise.
18079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18081 * dwarf2out.c (scompare_loc_descriptor_wide)
18082 (scompare_loc_descriptor_narrow): New functions, split out from...
18083 (scompare_loc_descriptor): ...here.
18084 * expmed.c (emit_store_flag_int): New function, split out from...
18085 (emit_store_flag): ...here.
18087 2017-08-30 Richard Biener <rguenther@suse.de>
18089 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18090 (dwarf2out_early_finish): Move setting of AT_pubnames from
18091 early debug output to early finish.
18093 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18095 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18096 and -mdata-region to the assembler.
18098 2017-08-30 Richard Biener <rguenther@suse.de>
18100 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18102 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18103 (dwarf2out_early_global_decl): Only generate a DIE for the
18104 abstract origin if it doesn't already exist or is a declaration DIE.
18105 (resolve_addr): Do not add the linkage name twice when
18106 generating a stub DIE for the DW_TAG_GNU_call_site target.
18108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18110 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18111 Use machine_mode rather than int for arg1_mode.
18113 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18116 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18117 that the second argument of the built-in functions to unpack
18118 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18119 switch statement instead a lot of if statements.
18120 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18121 Allow 64-bit values to be in Altivec registers as well as
18122 traditional floating point registers.
18123 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18125 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18127 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18128 MAX_REGS_PER_ADDRESS == 1.
18130 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18132 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18133 * config/i386/i386.c (ix86_option_override_internal): Simplify
18134 setting of opts->x_flag_entry.
18136 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18137 Jakub Jelinek <jakub@redhat.com>
18138 Richard Biener <rguenther@suse.de>
18140 PR tree-optimization/81503
18141 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18142 folded constant fits in the target type; reorder tests for clarity.
18144 2017-08-29 Martin Liska <mliska@suse.cz>
18146 * passes.def: Include pass_lower_switch.
18147 * stmt.c (dump_case_nodes): Remove and move to
18148 tree-switch-conversion.
18149 (case_values_threshold): Likewise.
18150 (expand_switch_as_decision_tree_p): Likewise.
18151 (emit_case_decision_tree): Likewise.
18152 (expand_case): Likewise.
18153 (balance_case_nodes): Likewise.
18154 (node_has_low_bound): Likewise.
18155 (node_has_high_bound): Likewise.
18156 (node_is_bounded): Likewise.
18157 (emit_case_nodes): Likewise.
18158 (struct simple_case_node): New struct.
18159 (add_case_node): Remove.
18160 (emit_case_dispatch_table): Use vector instead of case_list.
18161 (reset_out_edges_aux): Remove.
18162 (compute_cases_per_edge): Likewise.
18163 (expand_case): Build list of simple_case_node.
18164 (expand_sjlj_dispatch_table): Use it.
18165 * tree-switch-conversion.c (struct case_node): Moved from
18166 stmt.c and adjusted.
18167 (emit_case_nodes): Likewise.
18168 (node_has_low_bound): Likewise.
18169 (node_has_high_bound): Likewise.
18170 (node_is_bounded): Likewise.
18171 (case_values_threshold): Likewise.
18172 (reset_out_edges_aux): Likewise.
18173 (compute_cases_per_edge): Likewise.
18174 (add_case_node): Likewise.
18175 (dump_case_nodes): Likewise.
18176 (balance_case_nodes): Likewise.
18177 (expand_switch_as_decision_tree_p): Likewise.
18178 (emit_jump): Likewise.
18179 (emit_case_decision_tree): Likewise.
18180 (try_switch_expansion): Likewise.
18181 (do_jump_if_equal): Likewise.
18182 (emit_cmp_and_jump_insns): Likewise.
18183 (fix_phi_operands_for_edge): New function.
18184 (record_phi_operand_mapping): Likewise.
18185 (class pass_lower_switch): New pass.
18186 (pass_lower_switch::execute): New function.
18187 (make_pass_lower_switch): Likewise.
18188 (conditional_probability):
18189 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18190 * tree-pass.h: Add make_pass_lower_switch.
18192 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18195 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18198 2017-08-29 Richard Biener <rguenther@suse.de>
18200 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18201 we do not add a DW_AT_inline attribute twice.
18202 (gen_subprogram_die): Remove code setting DW_AT_inline on
18203 DECL_ABSTRACT_P nodes.
18205 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18207 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18208 calls to internal functions.
18209 (gimplify_modify_expr): Likewise.
18210 * tree-call-cdce.c (use_internal_fn): Likewise.
18211 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18212 (convert_to_divmod): Set the nothrow flag.
18213 * tree-if-conv.c (predicate_mem_writes): Likewise.
18214 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18215 (vectorizable_call): Likewise.
18216 (vectorizable_store): Likewise.
18217 (vectorizable_load): Likewise.
18218 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18219 (vect_recog_mask_conversion_pattern): Likewise.
18221 2017-08-29 Martin Liska <mliska@suse.cz>
18224 * gcc.c (driver_handle_option): Add new argument.
18225 * opts-common.c (handle_option): Pass
18226 target_option_override_hook.
18227 * opts-global.c (lang_handle_option): Add new option.
18228 (set_default_handlers): Add new argument.
18229 (decode_options): Likewise.
18230 * opts.c (target_handle_option): Likewise.
18231 (common_handle_option): Call target_option_override_hook.
18232 * opts.h (struct cl_option_handler_func): Add hook for
18233 target option override.
18234 (struct cl_option_handlers): Likewise.
18235 (set_default_handlers): Add new argument.
18236 (decode_options): Likewise.
18237 (common_handle_option): Likewise.
18238 (target_handle_option): Likewise.
18239 * toplev.c (toplev::main): Pass targetm.target_option.override
18242 2017-08-29 Richard Biener <rguenther@suse.de>
18243 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18245 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18246 life to the active subtree.
18248 2017-08-28 Jeff Law <law@redhat.com>
18250 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18251 derive_equivalences.
18252 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18253 Code moved into....
18254 (edge_info::derive_equivalences): New private member function
18256 * tree-ssa-dom.c (class edge_info): Changed from a struct
18257 to a class. Add ctor/dtor, methods and data members.
18258 (edge_info::edge_info): Renamed from allocate_edge_info.
18259 Initialize additional members.
18260 (edge_info::~edge_info): New.
18261 (free_dom_edge_info): Delete the edge info.
18262 (record_edge_info): Use new class & associated member functions.
18263 Tighten forms for testing for edge equivalences.
18264 (record_temporary_equivalences): Iterate over the simple
18265 equivalences rather than assuming there's only one per edge.
18266 (cprop_into_successor_phis): Iterate over the simple
18267 equivalences rather than assuming there's only one per edge.
18268 (optimize_stmt): Use operand_equal_p rather than pointer
18269 equality for mini-DSE code.
18271 2017-08-28 Nathan Sidwell <nathan@acm.org>
18273 * gcc.c (execute): Fold SIGPIPE handling into switch
18274 statement. Adjust internal error message.
18276 2017-08-28 Richard Biener <rguenther@suse.de>
18279 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18280 Do nothing for removed DIEs.
18282 2017-08-28 Richard Biener <rguenther@suse.de>
18284 PR tree-optimization/81977
18285 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18288 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18291 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18292 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18293 using targetm.gen_mem_thread_fence.
18295 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18298 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18299 predicate to register_operand. Reorder operands.
18300 (*btr<mode>): Ditto.
18301 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18302 (*btr<mode>_mask): Ditto.
18304 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18306 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18307 * config/rs6000/xmmintrin.h: New file.
18308 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18310 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18313 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18314 parameter and_insn and return it.
18315 (recombine_lvx_pattern): Insert a copy to ensure availability of
18316 the base register of the copied masking operation at the point of
18317 the instruction replacement.
18318 (recombine_stvx_pattern): Likewise.
18320 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18322 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18323 undocumented switches.
18324 (-mpower9-dform-vector): Likewise.
18325 (-mpower9-dform): Likewise.
18326 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18327 comments to delete references to -mpower9-dform* switches.
18328 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18329 Delete reference to -mpower9-dform* switches, test for
18330 -mpower9-vector instead.
18331 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18332 (OTHER_P9_VECTOR_MASKS): Likewise.
18333 (POWERPC_MASKS): Likewise.
18334 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18335 tests against -mpower9-dform* to -mpower9-vector. Delete code
18336 that checked for -mpower9-dform* consistancy with other options.
18337 Add test for -mpower9-misc to enable other power9 switches.
18338 (rs6000_init_hard_regno_mode_ok): Likewise.
18339 (rs6000_option_override_internal): Likewise.
18340 (rs6000_emit_prologue): Likewise.
18341 (rs6000_emit_epilogue): Likewise.
18342 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18343 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18344 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18346 (emit_fusion_p9_store): Likewise.
18347 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18348 resetting these macros if the assembler does not support ISA 3.0
18350 (TARGET_P9_DFORM_VECTOR): Likewise.
18351 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18352 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18354 2017-08-25 Alan Modra <amodra@gmail.com>
18356 PR rtl-optimization/81747
18357 * cse.c (cse_extended_basic_block): Don't attempt to record
18358 equivalences for degenerate conditional branches that branch
18359 to their fall-through.
18361 2017-08-24 Martin Sebor <msebor@redhat.com>
18363 PR middle-end/81908
18364 * gimple-fold.c (size_must_be_zero_p): New function.
18365 (gimple_fold_builtin_memory_op): Call it.
18367 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18369 * config/rs6000/mm_malloc.h: New file.
18371 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18373 PR tree-optimization/81913
18374 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18375 analysis when either IVs in condition can wrap.
18377 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18379 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18380 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18382 2017-08-24 Richard Biener <rguenther@suse.de>
18385 * targhooks.c (default_target_can_inline_p): Properly
18386 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18387 is present and always compare.
18388 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18389 infer -mfpmath=sse from TARGET_SSE_P.
18390 (ix86_can_inline_p): Properly use target_option_default_node when
18391 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18393 2017-08-24 Richard Biener <rguenther@suse.de>
18396 * dwarf2out.c (output_die): Handle flag_generate_offload like
18398 (output_comp_unit): Likewise.
18399 (gen_array_type_die): Likewise.
18400 (dwarf2out_early_finish): Likewise.
18401 (note_variable_value_in_expr): Likewise.
18402 (dwarf2out_finish): Likewise. Adjust assert.
18403 * cgraphunit.c (symbol_table::compile): Move setting of
18404 flag_generate_offload earlier ...
18405 (symbol_table::finalize_compilation_unit): ... here, before
18406 early debug finalization.
18408 2017-08-24 Richard Biener <rguenther@suse.de>
18410 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18411 and ipa-fnsummary.h.
18412 (ix86_can_inline_p): When ix86_fpmath flags do not match
18413 check whether the callee uses FP math at all.
18415 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18417 PR middle-end/81931
18418 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18419 instead of TYPE_PRECISION.
18421 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18422 Alan Hayward <alan.hayward@arm.com>
18423 David Sherwood <david.sherwood@arm.com>
18425 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18427 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18429 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18431 (ix86_frame::stack_realign_allocate): New field.
18432 (struct machine_frame_state): Modify comments.
18433 (machine_frame_state::sp_realigned_fp_end): New field.
18434 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18435 layout calculation.
18436 (sp_valid_at): Add assertion to assure no attempt to access invalid
18437 offset of a realigned stack.
18438 (fp_valid_at): Likewise.
18439 (choose_baseaddr): Modify comments.
18440 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18441 ix86_expand_prologue.
18442 (ix86_expand_prologue): Modify stack realignment and allocation.
18443 (ix86_expand_epilogue): Modify comments.
18444 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18445 avx2_runtime, avx512f, and avx512f_runtime.
18447 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18449 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18450 (mstackrealign): Do not init to -1.
18451 * config/i386/i386.c (ix86_option_override_internal):
18452 Check opts_set, not opts when setting default value of
18453 opts->x_ix86_force_align_arg_pointer.
18455 2017-08-23 Richard Biener <rguenther@suse.de>
18457 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18458 lang_hooks.decl_printable_name.
18459 * print-rtl-function.c (print_rtx_function): Likewise.
18460 * tree-pretty-print.c (dump_function_header): Likewise.
18462 2017-08-23 Richard Biener <rguenther@suse.de>
18465 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18466 -g0 at compile-time.
18468 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18470 PR middle-end/19706
18471 * doc/sourcebuild.texi (Other hardware attributes):
18474 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18476 PR middle-end/19706
18477 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18478 Fix single-use check.
18480 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18482 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18484 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18486 * doc/install.texi: Modify to add more details on running selected
18489 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18491 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18492 is combined with -mabi=ms.
18493 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18496 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18498 PR tree-optimization/81488
18499 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18500 and cached_basis fields.
18501 (MAX_SPREAD): New constant.
18502 (alloc_cand_and_find_basis): Initialize new fields.
18503 (clear_visited): New function.
18504 (create_phi_basis_1): Rename from create_phi_basis, set visited
18505 and cached_basis fields.
18506 (create_phi_basis): New wrapper function.
18507 (phi_add_costs_1): Rename from phi_add_costs, add spread
18508 parameter, set visited field, short-circuit when limits reached.
18509 (phi_add_costs): New wrapper function.
18510 (record_phi_increments_1): Rename from record_phi_increments, set
18512 (record_phi_increments): New wrapper function.
18513 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18514 (phi_incr_cost): New wrapper function.
18515 (all_phi_incrs_profitable_1): Rename from
18516 all_phi_incrs_profitable, set visited field.
18517 (all_phi_incrs_profitable): New wrapper function.
18519 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18520 Alan Hayward <alan.hayward@arm.com>
18521 David Sherwood <david.sherwood@arm.com>
18523 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18524 that takes the outer and inner modes.
18525 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18526 comparison as the canonical test for a paradoxical subreg.
18527 * combine.c (simplify_set): Use paradoxical_subreg_p.
18528 (make_extraction): Likewise.
18529 (force_to_mode): Likewise.
18530 (rtx_equal_for_field_assignment_p): Likewise.
18531 (gen_lowpart_for_combine): Likewise.
18532 (simplify_comparison): Likewise.
18533 * cse.c (equiv_constant): Likewise.
18534 * expmed.c (store_bit_field_1): Likewise.
18535 * final.c (alter_subreg): Likewise.
18536 * fwprop.c (propagate_rtx): Likewise.
18537 (forward_propagate_subreg): Likewise.
18538 * ira-conflicts.c (ira_build_conflicts): Likewise.
18539 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18540 * lra-constraints.c (curr_insn_transform): Likewise.
18541 (split_reg): Likewise.
18542 * lra-eliminations.c (move_plus_up): Likewise.
18543 (lra_eliminate_regs_1): Likewise.
18544 * recog.c (general_operand): Likewise.
18545 * ree.c (combine_reaching_defs): Likewise.
18546 * reload.c (push_reload): Likewise.
18547 (find_reloads): Likewise.
18548 * reload1.c (elimination_effects): Likewise.
18549 (compute_reload_subreg_offset): Likewise.
18550 (choose_reload_regs): Likewise.
18551 * rtlanal.c (subreg_lsb_1): Likewise.
18552 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18553 (simplify_subreg): Likewise.
18554 * var-tracking.c (track_loc_p): Likewise.
18555 * emit-rtl.c (byte_lowpart_offset): Likewise.
18556 (paradoxical_subreg_p): Delete out-of-line definition.
18558 2017-08-22 Jeff Law <law@redhat.com>
18560 PR tree-optimization/81741
18561 PR tree-optimization/71947
18562 * tree-ssa-dom.c: Include tree-inline.h.
18563 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18564 equivalences if one is more expensive to compute than the other.
18565 * tree-ssa-scopedtables.h (class const_or_copies): Make
18566 record_const_or_copy_raw method private.
18567 (class avail_exprs_stack): New method simplify_binary_operation.
18568 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18569 avail_exprs_stack::simplify_binary_operation as needed.
18570 (avail_exprs_stack::simplify_binary_operation): New function.
18572 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18574 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18575 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18576 (DOT_SYMBOLS): Likewise.
18577 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18578 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18579 (RS6000_ABI_NAME): Likewise.
18580 (TARGET_CMODEL): Likewise.
18581 (TOC_SECTION_ASM_OP): Likewise.
18582 (SET_CMODEL): New macro.
18583 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18585 2017-08-22 Richard Biener <rguenther@suse.de>
18587 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18588 to allow for free-lang-data replacements similar to verify_type_variant.
18590 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18591 Alan Hayward <alan.hayward@arm.com>
18592 David Sherwood <david.sherwood@arm.com>
18594 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18595 VOIDmode for the LABEL_REF.
18597 2017-08-22 Richard Biener <rguenther@suse.de>
18599 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18601 2017-08-22 Richard Biener <rguenther@suse.de>
18603 * common.opt (feliminate-dwarf2-dups): Ignore.
18604 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18605 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18606 same_die_p_wrap, compute_section_prefix,
18607 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18608 (comdat_symbol_id, comdat_symbol_number): Likewise.
18609 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18611 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18612 (output_die): Mark unreachable path unreachable.
18613 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18614 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18615 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18616 (dwarf2out_early_finish): Likewise.
18618 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18620 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18622 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18625 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18626 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18627 (avr_attribute_table) <io, io_low, address>: Initialize
18628 .decl_required with true.
18630 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18632 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18633 undocumented debugging options.
18634 (-mvsx-scalar-double): Likewise.
18635 (-mallow-df-permute): Likewise.
18636 (-mvectorize-builtins): Likewise.
18637 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18638 (rs6000_builtin_vectorized_function): Likewise.
18639 (rs6000_builtin_md_vectorized_function): Likewise.
18640 (rs6000_opt_vars): Likewise.
18642 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18645 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18646 (*btrq_imm): Rename from *btrq.
18647 (*btcq_imm): Rename from *btcq.
18648 (btsc): New code attribute.
18649 (*<btsc><mode>): New insn pattern.
18650 (*btr<mode>): Ditto.
18651 (*<btsc><mode>_mask): New insn_and_split pattern.
18652 (*btr<mode>_mask): Ditto.
18654 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18655 Alan Hayward <alan.hayward@arm.com>
18656 David Sherwood <david.sherwood@arm.com>
18658 * function.c (pad_below): Simplify padding calculation.
18660 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18661 Alan Hayward <alan.hayward@arm.com>
18662 David Sherwood <david.sherwood@arm.com>
18664 * target.def (function_prologue): Remove frame size argument.
18665 (function_epilogue): Likewise.
18666 * doc/tm.texi: Regenerate.
18667 * final.c (final_start_function): Update call to function_prologue.
18668 (final_end_function): Update call to function_epilogue.
18669 (default_function_pro_epilogue): Remove frame size argument.
18670 * output.h (default_function_pro_epilogue): Likewise.
18671 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18672 (arm_output_function_prologue): Likewise.
18673 * config/frv/frv.c (frv_function_prologue): Likewise.
18674 (frv_function_epilogue): Likewise.
18675 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18676 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18677 (ia64_output_function_epilogue): Likewise.
18678 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18679 (m32r_output_function_epilogue): Likewise.
18680 * config/microblaze/microblaze.c (microblaze_function_prologue)
18681 (microblaze_function_epilogue): Likewise.
18682 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18683 (mips_output_function_epilogue): Likewise.
18684 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18685 (mmix_target_asm_function_epilogue): Likewise.
18686 * config/msp430/msp430.c (msp430_start_function): Likewise.
18687 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18688 (nds32_asm_function_epilogue): Likewise.
18689 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18690 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18691 (pa_output_function_epilogue): Likewise.
18692 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18693 (rs6000_output_function_epilogue): Likewise.
18694 * config/rl78/rl78.c (rl78_start_function): Likewise.
18695 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
18696 (rs6000_output_function_epilogue): Likewise.
18697 * config/rx/rx.c (rx_output_function_prologue): Likewise.
18698 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
18699 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
18700 (sparc_asm_function_epilogue): Likewise.
18702 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18704 * tree.h (type_has_mode_precision_p): New function.
18705 * convert.c (convert_to_integer_1): Use it.
18706 * expr.c (expand_expr_real_2): Likewise.
18707 (expand_expr_real_1): Likewise.
18708 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
18709 * match.pd: Likewise.
18710 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
18711 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
18712 * tree-tailcall.c (process_assignment): Likewise.
18713 * tree-vect-loop.c (vectorizable_reduction): Likewise.
18714 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
18715 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
18716 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18717 (vectorizable_assignment): Likewise.
18718 (vectorizable_shift): Likewise.
18719 (vectorizable_operation): Likewise.
18720 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18722 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18724 * match.pd: Add pow (C, x) simplification.
18726 2017-08-21 Richard Biener <rguenther@suse.de>
18728 PR tree-optimization/81900
18729 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18730 for blocks with abnormal predecessors.
18731 (compute_antic): Do not set visited flag prematurely.
18733 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18736 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18738 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18740 * stor-layout.h (vector_type_mode): Move to...
18741 * tree.h (vector_type_mode): ...here.
18742 * stor-layout.c (vector_type_mode): Move to...
18743 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18745 2017-08-21 Richard Biener <rguenther@suse.de>
18747 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18748 register_external_die hooks.
18749 (debug_false_tree_charstarstar_uhwistar): Declare.
18750 (debug_nothing_tree_charstar_uhwi): Likewise.
18751 * debug.c (do_nothing_debug_hooks): Adjust.
18752 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18753 (debug_nothing_tree_charstar_uhwi): Likewise.
18754 * dbxout.c (dbx_debug_hooks): Adjust.
18755 (xcoff_debug_hooks): Likewise.
18756 * sdbout.c (sdb_debug_hooks): Likewise.
18757 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18758 * dwarf2out.c (macinfo_label_base): New global.
18759 (dwarf2out_register_external_die): New function for the
18760 register_external_die hook.
18761 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18762 (dwarf2_debug_hooks): Use them.
18763 (dwarf2_lineno_debug_hooks): Adjust.
18764 (struct die_struct): Add with_offset flag.
18765 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18766 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18767 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18768 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18769 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18770 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18771 defining section names for the early LTO debug variants.
18772 (reset_indirect_string): New helper.
18773 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18774 (print_dw_val): Add support for offsetted symbol references.
18775 (get_ultimate_context): Split out from is_cxx.
18776 (is_cxx): Use get_ultimate_context.
18777 (is_fortran): Add decl overload.
18778 (compute_comp_unit_symbol): Split out worker from
18779 compute_section_prefix.
18780 (compute_section_prefix): Call compute_comp_unit_symbol and
18781 set comdat_type_p here.
18782 (output_die): Skip DIE symbol output for the LTO added one.
18783 Handle DIE symbol references with offset.
18784 (output_comp_unit): Guard section name mangling properly.
18785 For LTO debug sections emit a symbol at the section beginning
18786 which we use to refer to its DIEs.
18787 (add_abstract_origin_attribute): For DIEs registered via
18788 dwarf2out_register_external_die directly refer to the early
18789 DIE rather than indirectly through the shadow one we created.
18790 Remove obsolete call to dwarf2out_abstract_function for
18791 non-function/block origins.
18792 (gen_array_type_die): When generating early LTO debug do
18793 not emit DW_AT_string_length.
18794 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18795 late when in LTO. As suggested place a gcc_unreachable for
18796 the DECL_ABSTRACT_P case.
18797 (gen_subprogram_die): Avoid another specification DIE
18798 for early built declarations/definitions for the late LTO case.
18799 (gen_variable_die): Add type references for late duplicated VLA dies
18801 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18802 we have the abstract instance already.
18803 (process_scope_var): Adjust decl DIE contexts in LTO which
18804 first puts them in limbo.
18805 (gen_decl_die): Do not generate type DIEs late apart from
18806 types for VLAs or for decls we do not yet have a DIE. Do not
18807 call dwarf2out_abstract_function late.
18808 (dwarf2out_early_global_decl): Make sure to create DIEs
18809 for abstract instances of a decl first.
18810 (dwarf2out_late_global_decl): Adjust comment.
18811 (output_macinfo_op): With multiple macro sections use
18812 macinfo_label_base to distinguish labels.
18813 (output_macinfo): Likewise. Update macinfo_label_base.
18814 Pass in the line info label.
18815 (note_variable_value_in_expr): When generating LTO resolve
18816 all variable values here by generating DIEs as needed.
18817 (init_sections_and_labels): Add early LTO debug flag parameter
18818 and generate different sections and names if set. Add generation
18819 counter for the labels so we can have multiple of them.
18820 (reset_dies): Helper to allow DIEs to be output multiple times.
18821 (dwarf2out_finish): When outputting DIEs to the fat part of an
18822 LTO object first reset DIEs.
18823 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18824 (modified_type_die): Check for decl_ultimate_origin being self
18826 (gen_type_die_with_usage): Likewise.
18827 (gen_typedef_die): Allow decl_ultimate_origin being self.
18828 (set_decl_abstract_flags): Remove.
18829 (set_block_abstract_flags): Likewise.
18830 (dwarf2out_abstract_function): Treat the early generated DIEs
18831 as the abstract copy and only add DW_AT_inline and
18832 DW_AT_artificial here and call set_decl_origin_self.
18833 If the DIE has an abstract origin don't do anything.
18834 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18835 if we have none yet (Go fails to build one, PR78628).
18836 (variably_modified_type_p): Prevent endless recursion for Ada
18837 cyclic pointer types.
18838 * lto-streamer-in.c: Include debug.h.
18839 (dref_queue): New global.
18840 (lto_read_tree_1): Stream in DIE references.
18841 (lto_input_tree): Register DIE references.
18842 (input_function): Stream DECL_DEBUG_ARGS.
18843 * lto-streamer-out.c: Include debug.h.
18844 (lto_write_tree_1): Output DIE references.
18845 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18846 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18847 (output_function): Stream DECL_DEBUG_ARGS.
18848 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18849 Stream DECL_ABSTRACT_ORIGIN.
18850 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18851 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18852 DECL_CONTEXT for file-scope decls.
18853 * lto-streamer.h (struct dref_entry): Declare.
18854 (dref_queue): Likewise.
18855 * cfgexpand.c (pass_expand::execute): Do not call the
18856 outlining_inline_function hook here.
18857 * lto-wrapper.c (debug_obj): New global.
18858 (tool_cleanup): Unlink it if required.
18859 (debug_objcopy): New function.
18860 (run_gcc): Handle early debug sections in the IL files by
18861 extracting them to separate files, partially linkin them and
18862 feeding the result back as result to the linker.
18863 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18864 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18865 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18866 sections into a separate segment.
18867 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18869 (darwin_asm_dwarf_section): Likewise.
18870 (darwin_asm_output_dwarf_offset): Likewise.
18871 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18873 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18874 Alan Hayward <alan.hayward@arm.com>
18875 David Sherwood <david.sherwood@arm.com>
18877 * read-md.h (md_reader::record_potential_iterator_use): Replace
18878 pointer argument with an rtx and an index.
18879 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18880 (apply_mode_iterator): Likewise.
18881 (apply_code_iterator): Likewise.
18882 (apply_int_iterator): Likewise.
18883 (apply_subst_iterator): Likewise.
18884 (record_iterator_use): Likewise.
18885 (record_attribute_use): Likewise.
18886 (md_reader::record_potential_iterator_use): Likewise. Update calls
18887 to record_iterator_use and apply_iterator.
18888 (iterator_use): Replace ptr with x and index.
18889 (attribute_use): Likewise.
18890 (apply_attribute_uses): Update calls to apply_iterator.
18891 (apply_iterators): Likewise. Update initialization of iterator_use.
18892 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18893 and record_potential_iterator_use.
18894 (rtx_reader::read_rtx_operand): Likewise.
18896 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18897 Alan Hayward <alan.hayward@arm.com>
18898 David Sherwood <david.sherwood@arm.com>
18900 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18903 2017-08-21 Richard Biener <rguenther@suse.de>
18905 PR middle-end/81884
18906 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18907 at struct end conservatively when comparing common bases.
18909 2017-08-21 Richard Biener <rguenther@suse.de>
18911 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18912 (mem_ref_in_stmt): Remove.
18913 (determine_max_movement): Use ref index to get at the reference.
18914 (invariantness_dom_walker::before_dom_children): Deal with
18915 lim data already initialized.
18916 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18918 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18920 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18921 (TARGET_ISA_ROUND): Ditto.
18922 (TARGET_ROUND): Ditto.
18923 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18924 * config/i386/i386.md: Ditto.
18925 * config/i386/sse.md: Ditto.
18926 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18927 with OPTION_MASK_ISA_SSE4_1.
18929 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18932 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18933 __builtin_ia32_lzcnt_u16.
18935 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18938 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18939 (rs6000_set_current_function): Rewrite function to use it.
18941 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18944 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18945 and TYPE_WARN_IF_NOT_ALIGN.
18946 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18947 (handle_warn_if_not_align): New.
18948 (place_union_field): Call handle_warn_if_not_align.
18949 (place_field): Call handle_warn_if_not_align.
18950 Copy TYPE_WARN_IF_NOT_ALIGN.
18951 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18952 (layout_type): Likewise.
18953 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18955 (tree_decl_common): Add warn_if_not_align.
18956 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18957 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18958 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18959 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18960 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18961 * doc/extend.texi: Document warn_if_not_aligned attribute.
18962 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18964 2017-08-17 Martin Liska <mliska@suse.cz>
18967 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18968 (get_data_dependence): Use it as pointer type.
18969 (distribute_loop): Likewise.
18971 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18973 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18974 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18975 (p8_vmrgow_<mode>_direct): New define_insn.
18976 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18977 handle endianness for vmrgew and vmrgow permute patterns.
18979 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18981 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18982 * config/rs6000/rs6000-cpus.def: Remove comment.
18983 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18984 (POWERPC_MASKS): Likewise.
18985 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18986 use of TARGET_VSX_TIMODE.
18987 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18988 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18989 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18990 (rs6000_option_override_internal): Remove dead code.
18991 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18992 (rs6000_legitimize_reload_address): Likewise.
18993 (rs6000_legitimate_address_p): Likewise.
18994 (rs6000_opt_masks): Delete "vsx-timode".
18995 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18996 from function comment.
18997 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18998 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18999 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19001 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19002 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19003 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19004 (VSX_TI): Likewise.
19006 (define_peephole2): Likewise.
19008 2017-08-17 Martin Sebor <msebor@redhat.com>
19011 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19012 past the end of an array.
19013 (test_pp_format): Add test cases.
19015 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19017 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19018 missing ECF_NOTHROW flags.
19020 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19023 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19024 operands residing in integer registers.
19025 (*vsx_le_perm_load_<mode>): Likewise.
19026 (*vsx_le_perm_store_<mode>): Likewise.
19027 (define_peephole2): Add peepholes to optimize the above.
19029 2017-08-17 Marek Polacek <polacek@redhat.com>
19031 PR middle-end/81814
19032 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19033 to mimic what shorten_compare did. Change the return type to bool.
19034 (fold_cond_expr_with_comparison): Update call to
19035 operand_equal_for_comparison_p.
19036 (fold_ternary_loc): Likewise.
19038 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19040 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19042 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19044 2017-08-17 Richard Biener <rguenther@suse.de>
19046 * tree-ssa-structalias.c (solve_graph): When propagating
19047 to successors update the graphs succ edges and avoid duplicate work.
19049 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19052 * config/i386/i386.c (ix86_option_override_internal): Save target
19053 specific options after ix86_stack_protector_guard_reg was changed.
19055 2017-08-17 Richard Biener <rguenther@suse.de>
19057 PR tree-optimization/81827
19058 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19059 (new_var_info): Initialize it conservatively.
19060 (get_call_vi): Mark register vars.
19061 (new_scalar_tmp_constraint_exp): Likewise.
19062 (handle_rhs_call): Likewise.
19063 (handle_const_call): Likewise.
19064 (create_function_info_for): Likewise.
19065 (solve_constraints): Sort varinfos to separate register from
19066 non-register vars to pack points-to solution bitmaps during
19069 2017-08-17 Marek Polacek <polacek@redhat.com>
19071 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19073 2017-08-17 Richard Biener <rguenther@suse.de>
19075 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19076 to true when overflow is undefined and we saturated the result.
19078 2017-08-17 Alan Modra <amodra@gmail.com>
19081 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19082 Don't use store multiple if only one reg needs saving.
19083 (interesting_frame_related_regno): New function.
19084 (rs6000_frame_related): Don't emit frame info for regs that
19086 (rs6000_emit_epilogue): Likewise.
19088 2017-08-16 Nathan Sidwell <nathan@acm.org>
19090 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19091 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19092 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19093 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19094 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19095 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19096 (verify_type): Adjust for TYPE_BINFO move.
19097 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19098 process TYPE_BINFO directly.
19099 (hash_tree): Likewise.
19100 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19102 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19105 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19107 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19109 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19112 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19113 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19114 Add ix86_binary_operator_ok to insn constraint.
19115 (*iordi_1_bts): Ditto.
19116 (*xordi_1_btc): Ditto.
19117 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19118 Update corresponding peephole2 pattern.
19122 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19124 PR tree-optimization/81832
19125 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19126 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19128 2017-08-16 Marek Polacek <polacek@redhat.com>
19131 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19132 regarding size_zero_node.
19134 2017-08-16 Martin Liska <mliska@suse.cz>
19137 * config.gcc: Respect previously set extra_objs in case
19140 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19142 PR tree-optimization/81835
19143 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19144 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19145 not depend on the phi.
19147 2017-08-16 Alan Modra <amodra@gmail.com>
19149 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19152 2017-08-16 Alan Modra <amodra@gmail.com>
19154 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19155 (save_reg_p): ..into this. Update all callers.
19156 (first_reg_to_save): Simplify.
19158 2017-08-16 Alan Modra <amodra@gmail.com>
19160 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19163 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19167 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19168 constant count if that count is less than 32.
19170 2017-08-15 Nathan Sidwell <nathan@acm.org>
19172 * gcc.c (execute): Emit friendlier message if inferior is killed
19173 by an external cause.
19175 2017-08-15 Richard Biener <rguenther@suse.de>
19177 PR tree-optimization/81790
19178 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19179 CONSTRUCTORs from simplifying and VN.
19181 2017-08-14 Martin Sebor <msebor@redhat.com>
19183 * builtin-attrs.def: Add comments.
19185 2017-08-14 Martin Sebor <msebor@redhat.com>
19188 * doc/extend.texi (attribute nonstring): Document new attribute.
19190 2017-08-14 Martin Sebor <msebor@redhat.com>
19193 * tree-diagnostic.c (default_tree_printer): Handle %G.
19194 * gimple-pretty-print.h (percent_G_format): Declare new function.
19195 * gimple-pretty-print.c (percent_G_format): Define.
19196 * tree-pretty-print.c (percent_K_format): Add argument.
19198 2017-08-14 Martin Sebor <msebor@redhat.com>
19200 PR translation/79998
19201 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19202 Remove a stray space.
19204 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19207 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19208 (*iordi_1_bts): Ditto.
19209 (*xordi_1_btc): Ditto.
19211 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19214 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19215 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19217 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19218 quoted strings, and make more translator-friendly.
19219 (darwin_rs6000_override_options): Likewise.
19220 (rs6000_option_override_internal): Likewise.
19221 (rs6000_return_in_memory): Fix overlong line.
19222 (init_cmulative_args): Use quoted strings, and make more
19223 translator-friendly.
19224 (rs6000_pass_by_reference): Fix overlong line.
19225 (def_builtin): Use quoted strings.
19226 (altivec_expand_predicate_builtin): Use quoted strings, and make
19227 more translator-friendly.
19228 (htm_expand_builtin): Use quoted strings.
19229 (cpu_expand_builtin): Use quoted strings, and make more
19230 translator-friendly.
19231 (altivec_expand_builtin): Likewise.
19232 (paired_expand_predicate_builtin): Likewise.
19233 (rs6000_invalid_builtin): Likewise.
19234 (builtin_function_type): Use quoted strings.
19235 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19236 more translator-friendly.
19237 (rs6000_trampoline_init): Likewise.
19238 (rs6000_handle_altivec_attribute): Likewise.
19239 (rs6000_inner_target_options): Use quoted strings.
19240 (rs6000_disable_incompatible_switches): Likewise.
19241 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19242 strings, and make more translator-friendly.
19243 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19245 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19247 PR tree-optimization/81799
19248 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19249 cond_expr to simple gimple operand.
19251 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19253 PR middle-end/46932
19254 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19256 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19260 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19261 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19262 TARGET_GASISR_PROLOGUES.
19263 * config/avr/avr.c (avr_option_override): Same.
19264 (avr_pass_pre_proep::execute): Same.
19266 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19269 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19270 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19272 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19274 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19275 define_insn_and_split. Split to a memory load from 0 in
19276 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19277 using PTR mode iterator.
19278 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19279 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19280 (*add_tp_<mode>): Redefine as define_insn_and_split.
19281 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19282 address space. Merge with *add_tp_x32 using PTR mode iterator.
19283 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19284 Split to an add with a memory load from 0 in
19285 DEFAULT_TLS_SEG_REG address space.
19287 2017-08-12 Andrew Pinski <apinski@cavium.com>
19289 * config/aarch64/aarch64-option-extensions.def (rdma):
19290 Fix feature string to what Linux prints out in /proc/cpuinfo.
19292 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19295 * dwarf2out.c (modified_type_die): For C typedef types that have
19296 an ultimate origin, process the ultimate origin instead of the
19298 (gen_typedef_die): Assert that input DECLs have no ultimate
19300 (gen_type_die_with_usage): For typedef variants that have an
19301 ultimate origin, just call gen_decl_die on the original DECL.
19302 (process_scope_var): Avoid creating DIEs for local typedefs and
19303 concrete static variables.
19305 2017-08-12 Alan Modra <amodra@gmail.com>
19309 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19310 match gnu-user.h startfile.
19311 (ENDFILE_LINUX_SPEC): Similarly.
19313 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19316 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19318 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19320 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19321 * config/aarch64/aarch64.md (mov<mode>): Change.
19322 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19323 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19324 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19326 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19328 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19329 for storage order barriers.
19331 2017-08-11 Martin Liska <mliska@suse.cz>
19334 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19335 variables of void type.
19337 2017-08-11 Martin Liska <mliska@suse.cz>
19339 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19340 TARGET_SUPPORTS_ALIASES.
19341 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19342 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19343 (optimize_weakref): Likewise.
19344 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19345 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19346 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19348 2017-08-11 Martin Liska <mliska@suse.cz>
19351 * config/i386/i386.c (make_resolver_func): Do complete
19352 refactoring of the function.
19354 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19357 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19358 * config/i386/i386.c (ix86_stack_protect_guard): Use
19359 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19360 * doc/invoke.texi (x86 Options): Document
19361 -mstack-protector-guard-symbol= option.
19363 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19365 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19366 * config/i386/i386.c (ix86_split_stack_guard): New function.
19367 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19368 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19369 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19370 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19371 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19372 (split_stack_space_check): Call ix86_split_stack_guard.
19374 2017-08-10 Martin Sebor <msebor@redhat.com>
19376 * print-tree.c (print_node): Print location using the established
19378 Replace spaces with colons.
19379 (debug_raw, debug): Ditto.
19381 2017-08-10 Martin Sebor <msebor@redhat.com>
19384 * pretty-print.c (pp_format): Correct the handling of %s precision.
19386 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19389 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19391 (ix86_finalize_stack_frame_flags): This. Also clear
19392 frame_pointer_needed if -fno-omit-frame-pointer is used without
19394 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19395 with ix86_finalize_stack_frame_flags.
19396 (ix86_expand_epilogue): Likewise.
19397 (ix86_expand_split_stack_prologue): Likewise.
19398 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19400 2017-08-10 Martin Liska <mliska@suse.cz>
19403 * c-attribs.c (handle_target_attribute):
19404 Report warning for an empty string argument of target attribute.
19406 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19409 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19411 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19412 or DECL_NONLOCAL labels.
19413 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19414 or DECL_NONLOCAL labels here.
19416 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19418 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19419 to indicate when early gimple folding has been disabled.
19420 (rs6000_gimple_fold_builtin): Add debug content.
19421 (rs6000_invalid_builtin): Fix whitespace.
19422 (rs6000_expand_builtin): Fix whitespace.
19423 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19425 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19428 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19429 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19431 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19433 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19434 (qdf24xx): Likewise.
19435 * config/aarch64/aarch64.md: Include falkor.md.
19436 * config/aarch64/falkor.md: New.
19438 2017-08-09 Marek Polacek <polacek@redhat.com>
19441 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19442 * diagnostic.c (emit_diagnostic): Add a comment.
19443 (emit_diagnostic_valist): New function.
19445 2017-08-09 Marek Polacek <polacek@redhat.com>
19448 * input.c (make_location): New overload.
19449 * input.h (make_location): Declare.
19451 2017-08-08 Alan Modra <amodra@gmail.com>
19452 H.J. Lu <hongjiu.lu@intel.com>
19455 * gcc.c (NO_PIE_SPEC): Delete.
19456 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19458 (LINK_PIE_SPEC): ..to here.
19459 (LINK_COMMAND_SPEC): Support -no-pie.
19460 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19461 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19463 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19464 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19465 (ENDFILE_CRTEND_SPEC): Similarly.
19467 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19470 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19471 (mstack-protector-guard-offset=): Ditto.
19472 * config/i386/i386.c (ix86_option_override): Handle
19473 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19475 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19476 ix86_stack_protect_guard_offset variables.
19477 (TARGET_STACK_PROTECT_GUARD): Always define.
19478 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19479 and -mstack-protector-guard-offset= options.
19481 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19483 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19484 boundary case for the last candidate.
19486 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19488 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19489 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19491 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19493 PR middle-end/19706
19494 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19495 * config/aarch64/aarch64-builtins.c
19496 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19497 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19498 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19500 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19501 Andrew Pinski <pinskia@gmail.com>
19503 PR middle-end/19706
19504 * internal-fn.def (XORSIGN): New.
19505 * optabs.def (xorsign_optab): New.
19506 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19507 (convert_expand_mult_copysign): New.
19508 (pass_optimize_widening_mul::execute): Call
19509 convert_expand_mult_copysign.
19511 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19513 PR tree-optimization/81354
19514 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19515 Insert on edges rather than explicitly creating landing pads.
19516 (analyze_candidates_and_replace): Commit edge inserts.
19518 2017-08-08 Richard Biener <rguenther@suse.de>
19520 PR middle-end/81719
19521 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19522 (expand_simple_operations): Also look through ADDR_EXPRs with
19523 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19525 2017-08-08 Richard Biener <rguenther@suse.de>
19527 PR tree-optimization/81723
19528 * tree-vect-slp.c (struct bst_traits): New hash traits.
19529 (bst_fail): New global.
19530 (vect_build_slp_tree_2): New worker, split out from ...
19531 (vect_build_slp_tree): ... this now wrapping it with using
19532 bst_fail set to cache SLP tree build fails. Properly handle
19534 (vect_analyze_slp_instance): Allocate and free bst_fail.
19536 2017-08-08 Martin Liska <mliska@suse.cz>
19539 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19540 LABEL_DECLs that can be from a different function.
19542 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19544 PR tree-optimization/81744
19545 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19546 loop's number of iterations.
19548 2017-08-08 Martin Liska <mliska@suse.cz>
19550 * asan.c: Include header files.
19551 * attribs.c (build_decl_attribute_variant): New function moved
19553 (build_type_attribute_qual_variant): Likewise.
19554 (cmp_attrib_identifiers): Likewise.
19555 (simple_cst_list_equal): Likewise.
19556 (omp_declare_simd_clauses_equal): Likewise.
19557 (attribute_value_equal): Likewise.
19558 (comp_type_attributes): Likewise.
19559 (build_type_attribute_variant): Likewise.
19560 (lookup_ident_attribute): Likewise.
19561 (remove_attribute): Likewise.
19562 (merge_attributes): Likewise.
19563 (merge_type_attributes): Likewise.
19564 (merge_decl_attributes): Likewise.
19565 (merge_dllimport_decl_attributes): Likewise.
19566 (handle_dll_attribute): Likewise.
19567 (attribute_list_equal): Likewise.
19568 (attribute_list_contained): Likewise.
19569 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19570 (lookup_attribute_by_prefix): Likewise.
19571 * bb-reorder.c: Include header files.
19572 * builtins.c: Likewise.
19573 * calls.c: Likewise.
19574 * cfgexpand.c: Likewise.
19575 * cgraph.c: Likewise.
19576 * cgraphunit.c: Likewise.
19577 * convert.c: Likewise.
19578 * dwarf2out.c: Likewise.
19579 * final.c: Likewise.
19580 * fold-const.c: Likewise.
19581 * function.c: Likewise.
19582 * gimple-expr.c: Likewise.
19583 * gimple-fold.c: Likewise.
19584 * gimple-pretty-print.c: Likewise.
19585 * gimple.c: Likewise.
19586 * gimplify.c: Likewise.
19587 * hsa-common.c: Likewise.
19588 * hsa-gen.c: Likewise.
19589 * internal-fn.c: Likewise.
19590 * ipa-chkp.c: Likewise.
19591 * ipa-cp.c: Likewise.
19592 * ipa-devirt.c: Likewise.
19593 * ipa-fnsummary.c: Likewise.
19594 * ipa-inline.c: Likewise.
19595 * ipa-visibility.c: Likewise.
19597 * lto-cgraph.c: Likewise.
19598 * omp-expand.c: Likewise.
19599 * omp-general.c: Likewise.
19600 * omp-low.c: Likewise.
19601 * omp-offload.c: Likewise.
19602 * omp-simd-clone.c: Likewise.
19603 * opts-global.c: Likewise.
19604 * passes.c: Likewise.
19605 * predict.c: Likewise.
19606 * sancov.c: Likewise.
19607 * sanopt.c: Likewise.
19608 * symtab.c: Likewise.
19609 * toplev.c: Likewise.
19610 * trans-mem.c: Likewise.
19611 * tree-chkp.c: Likewise.
19612 * tree-eh.c: Likewise.
19613 * tree-into-ssa.c: Likewise.
19614 * tree-object-size.c: Likewise.
19615 * tree-parloops.c: Likewise.
19616 * tree-profile.c: Likewise.
19617 * tree-ssa-ccp.c: Likewise.
19618 * tree-ssa-live.c: Likewise.
19619 * tree-ssa-loop.c: Likewise.
19620 * tree-ssa-sccvn.c: Likewise.
19621 * tree-ssa-structalias.c: Likewise.
19622 * tree-ssa.c: Likewise.
19623 * tree-streamer-in.c: Likewise.
19624 * tree-vectorizer.c: Likewise.
19625 * tree-vrp.c: Likewise.
19626 * tsan.c: Likewise.
19627 * ubsan.c: Likewise.
19628 * varasm.c: Likewise.
19629 * varpool.c: Likewise.
19630 * tree.c: Remove functions moved to attribs.[ch].
19631 * tree.h: Likewise.
19632 * config/aarch64/aarch64.c: Add attrs.h header file.
19633 * config/alpha/alpha.c: Likewise.
19634 * config/arc/arc.c: Likewise.
19635 * config/arm/arm.c: Likewise.
19636 * config/avr/avr.c: Likewise.
19637 * config/bfin/bfin.c: Likewise.
19638 * config/c6x/c6x.c: Likewise.
19639 * config/cr16/cr16.c: Likewise.
19640 * config/cris/cris.c: Likewise.
19641 * config/darwin.c: Likewise.
19642 * config/epiphany/epiphany.c: Likewise.
19643 * config/fr30/fr30.c: Likewise.
19644 * config/frv/frv.c: Likewise.
19645 * config/ft32/ft32.c: Likewise.
19646 * config/h8300/h8300.c: Likewise.
19647 * config/i386/winnt.c: Likewise.
19648 * config/ia64/ia64.c: Likewise.
19649 * config/iq2000/iq2000.c: Likewise.
19650 * config/lm32/lm32.c: Likewise.
19651 * config/m32c/m32c.c: Likewise.
19652 * config/m32r/m32r.c: Likewise.
19653 * config/m68k/m68k.c: Likewise.
19654 * config/mcore/mcore.c: Likewise.
19655 * config/microblaze/microblaze.c: Likewise.
19656 * config/mips/mips.c: Likewise.
19657 * config/mmix/mmix.c: Likewise.
19658 * config/mn10300/mn10300.c: Likewise.
19659 * config/moxie/moxie.c: Likewise.
19660 * config/msp430/msp430.c: Likewise.
19661 * config/nds32/nds32-isr.c: Likewise.
19662 * config/nds32/nds32.c: Likewise.
19663 * config/nios2/nios2.c: Likewise.
19664 * config/nvptx/nvptx.c: Likewise.
19665 * config/pa/pa.c: Likewise.
19666 * config/pdp11/pdp11.c: Likewise.
19667 * config/powerpcspe/powerpcspe.c: Likewise.
19668 * config/riscv/riscv.c: Likewise.
19669 * config/rl78/rl78.c: Likewise.
19670 * config/rx/rx.c: Likewise.
19671 * config/s390/s390.c: Likewise.
19672 * config/sh/sh.c: Likewise.
19673 * config/sol2.c: Likewise.
19674 * config/sparc/sparc.c: Likewise.
19675 * config/spu/spu.c: Likewise.
19676 * config/stormy16/stormy16.c: Likewise.
19677 * config/tilegx/tilegx.c: Likewise.
19678 * config/tilepro/tilepro.c: Likewise.
19679 * config/v850/v850.c: Likewise.
19680 * config/vax/vax.c: Likewise.
19681 * config/visium/visium.c: Likewise.
19682 * config/xtensa/xtensa.c: Likewise.
19684 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19687 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19688 constraints since the -mupper-regs-* switches have been
19690 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19691 into a vector from a double word element that was extracted from
19692 another vector, and eliminate extra XXPERMDI instructions.
19693 (vsx_concat_<mode>_2): Likewise.
19694 (vsx_concat_<mode>_3): Likewise.
19695 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
19696 concat to allow optimizing inserts from previous extracts.
19698 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
19700 * config/i386/i386.c (ix86_stack_protect_guard): Generate
19701 memory reference to a SSP offset in TLS address space.
19702 (ix86_print_operand) <case '@'>: Remove.
19703 (ix86_print_operand_punct_valid_p): Remove '@' code.
19704 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
19705 UNSPEC_SP_TLS_TEST.
19706 (stack_tls_protect_set_<mode>): Remove.
19707 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
19708 (stack_tls_protect_test_<mode>): Remove.
19709 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
19711 2017-08-07 Olivier Hainque <hainque@adacore.com>
19714 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
19716 2017-08-07 Douglas Rupp <rupp@adacore.com>
19718 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19719 variable was referenced as multidir in command.
19721 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19724 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19727 2017-08-07 Martin Liska <mliska@suse.cz>
19729 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19730 * config/rl78/rl78.c: Add include of attribs.h.
19731 * config/sh/sh.c: Likewise.
19732 * config/v850/v850.c: Likewise.
19734 2017-08-07 Tom de Vries <tom@codesourcery.com>
19736 PR middle-end/78266
19737 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19739 2017-08-07 Martin Liska <mliska@suse.cz>
19741 * config/mips/mips.c: Include attribs.h.
19743 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19746 * doc/invoke.texi: Document change in behvaior for -Ofast for
19749 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19751 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19753 (aarch64_pop_regs): Likewise.
19754 (aarch64_gen_load_pair): Likewise.
19755 (aarch64_save_callee_saves): Likewise.
19756 (aarch64_restore_callee_saves): Likewise.
19758 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19760 * config/i386/i386.c: Revert the last change.
19762 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19765 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19767 (ix86_finalize_stack_frame_flags): This. Also clear
19768 frame_pointer_needed if -fno-omit-frame-pointer is used without
19770 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19771 with ix86_finalize_stack_frame_flags.
19772 (ix86_expand_epilogue): Likewise.
19773 (ix86_expand_split_stack_prologue): Likewise.
19775 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19778 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19779 to P_AES for Westmere.
19781 2017-08-07 Jonathan Yong <10walls@gmail.com>
19783 * config/i386/mingw.opt (fset-stack-executable): Removed.
19784 * config/i386/cygming.opt (fset-stack-executable): Moved
19786 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19788 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19790 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19792 2017-08-07 Marek Polacek <polacek@redhat.com>
19794 PR middle-end/81737
19795 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19797 2017-08-07 Martin Liska <mliska@suse.cz>
19799 * attribs.h (canonicalize_attr_name): New function.
19800 (cmp_attribs): Move from c-format.c and adjusted.
19801 (is_attribute_p): Moved from tree.h.
19802 * tree-inline.c: Add new includes.
19803 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19804 (private_is_attribute_p): Remove.
19805 (private_lookup_attribute): Likewise.
19806 (private_lookup_attribute_by_prefix): Simplify.
19807 (remove_attribute): Use is_attribute_p.
19808 * tree.h: Remove removed declarations.
19810 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19812 PR middle-end/81698
19813 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19814 instead of computing it in the function. Formatting fix.
19815 (expand_case): Don't rely on default_edge being the first edge,
19816 clear it if removing it, pass default_edge to
19817 emit_case_dispatch_table.
19818 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19821 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19823 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19824 insn in the function, emit NOP after the insn.
19826 2017-08-06 Tom de Vries <tom@codesourcery.com>
19828 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19831 2017-08-06 Tom de Vries <tom@codesourcery.com>
19833 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19836 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19838 PR tree-optimization/57371
19839 * match.pd: New pattern.
19841 2017-08-04 Marek Polacek <polacek@redhat.com>
19843 PR middle-end/81695
19844 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19845 perform the computation in offset_int.
19847 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19849 PR tree-optimization/81136
19850 * tree-vectorizer.h: Include tree-hash-traits.h.
19851 (vec_base_alignments): New typedef.
19852 (vec_info): Add a base_alignments field.
19853 (vect_record_base_alignments): Declare.
19854 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19856 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19857 (create_data_ref): Add an is_conditional_in_stmt argument.
19858 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19859 the is_conditional_in_stmt field.
19860 (data_ref_loc): Add an is_conditional_in_stmt field.
19861 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19862 (find_data_references_in_stmt): Update call to create_data_ref.
19863 (graphite_find_data_references_in_stmt): Likewise.
19864 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19865 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19866 (vect_record_base_alignment): New function.
19867 (vect_record_base_alignments): Likewise.
19868 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19869 for nested statements even if we fail to compute a misalignment.
19870 Use pooled base alignments for unconditional references.
19871 (vect_find_same_alignment_drs): Compare base addresses instead
19873 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19874 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19876 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19878 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19879 Add an explicit name for the enum. Use auto_vec for slp_instances
19880 and grouped_stores.
19881 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19883 (_bb_vec_info): Add a constructor and destructor.
19884 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19885 (destroy_loop_vec_info): Delete.
19886 (vect_destroy_datarefs): Likewise.
19887 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19888 (vec_info::vec_info): New function.
19889 (vec_info::~vec_info): Likewise.
19890 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19891 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19892 destroy_loop_vec_info.
19893 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19894 (_loop_vec_info::_loop_vec_info): ...this.
19895 (destroy_loop_vec_info): Replace with...
19896 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19897 the stmt_vec_infos. Leave handling of vec_info information to its
19898 destructor. Remove explicit vector releases.
19899 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19900 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19901 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19902 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19903 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19904 (destroy_bb_vec_info): Replace with...
19905 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19906 information to its destructor.
19907 (vect_slp_analyze_bb_1): Use new and delete instead of
19908 new_bb_vec_info and destroy_bb_vec_info.
19909 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19912 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19914 * tree-data-ref.h (subscript): Add access_fn field.
19915 (data_dependence_relation): Add could_be_independent_p.
19916 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19917 (same_access_functions): Move to tree-data-ref.c.
19918 * tree-data-ref.c (ref_contains_union_access_p): New function.
19919 (access_fn_component_p): Likewise.
19920 (access_fn_components_comparable_p): Likewise.
19921 (dr_analyze_indices): Add a reference to access_fn_component_p.
19922 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19924 (constant_access_functions): Likewise.
19925 (add_other_self_distances): Likewise.
19926 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19927 (initialize_data_dependence_relation): Use XCNEW and remove
19928 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19929 of access functions that have the same type. Allow the
19930 subsequence to end with different bases in some circumstances.
19931 Record the chosen access functions in SUB_ACCESS_FN.
19932 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19933 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19934 (subscript_dependence_tester_1): Likewise dra and drb.
19935 (build_classic_dist_vector): Update calls accordingly.
19936 (subscript_dependence_tester): Likewise.
19937 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19938 DDR_COULD_BE_INDEPENDENT_P.
19939 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19940 comp_alias_ddrs instead of may_alias_ddrs.
19941 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19943 (vect_analyze_data_ref_dependence): Use it if
19944 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19945 distance vectors if that fails.
19946 (dependence_distance_ge_vf): New function.
19947 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19948 LOOP_VINFO_MAY_ALIAS_DDRS.
19950 2017-08-04 Richard Biener <rguenther@suse.de>
19952 PR middle-end/81705
19953 * fold-const.c (fold_binary_loc): Properly restrict
19954 minus_var0 && minus_var1 case when associating undefined overflow
19957 2017-08-04 Tom de Vries <tom@codesourcery.com>
19959 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19961 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19963 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19964 Don't start diagnostic messages with a capital letter.
19965 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19967 (rs6000_invalid_builtin): Likewise.
19968 (rs6000_trampoline_init): Likewise.
19970 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19973 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19974 after setting changeable df flags.
19976 2017-08-03 Richard Biener <rguenther@suse.de>
19978 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19979 up if the use is in USE - X.
19981 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19983 * toplev.c (dumpfile.h): New include.
19984 (internal_error_reentered): New static function. Use it...
19985 (internal_error_function): ...here to handle reentered internal_error.
19987 2017-08-03 Richard Biener <rguenther@suse.de>
19989 PR middle-end/81148
19990 * fold-const.c (split_tree): Add minus_var and minus_con
19991 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19992 here but always use minus_*.
19993 (associate_trees): Assert we never associate with MINUS_EXPR
19994 and NULL first operand. Do not recurse for PLUS_EXPR operands
19995 when associating as MINUS_EXPR either.
19996 (fold_binary_loc): Track minus_var and minus_con.
19998 2017-08-03 Tom de Vries <tom@codesourcery.com>
20001 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20002 ACCEL_COMPILER, apply finish_options on
20003 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20005 2017-08-03 Tom de Vries <tom@codesourcery.com>
20008 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20009 function_entry_patch_area_size > 0.
20011 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20014 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20015 instead of 10??LU, perform unit multiplication in wide_int,
20016 don't change alloc_object_size_limit if the limit is larger
20019 PR tree-optimization/81655
20020 PR tree-optimization/81588
20021 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20022 the case when ranges[i].low and high are 1 for unsigned type with
20025 PR middle-end/81052
20026 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20027 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20029 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20031 * tree-vrp.h: Add include guard.
20033 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20036 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20037 (ud2): New insn pattern.
20038 * config/i386/i386.c (ix86_expand_epilogue):
20039 For naked functions, generate ud2 instead of trap insn.
20041 2017-08-02 Marek Polacek <polacek@redhat.com>
20044 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20046 2017-08-02 Tom de Vries <tom@codesourcery.com>
20047 Cesar Philippidis <cesar@codesourcery.com>
20049 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20050 Add missing edge probabilities.
20052 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20054 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20055 Correct endianness.
20057 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20059 PR middle-end/79499
20060 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20061 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20062 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20063 prologue_seq sequences - if any.
20065 2017-08-02 Richard Biener <rguenther@suse.de>
20067 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20068 via vectors if supported, integer extracts via punning if supported
20069 or otherwise vector extracts.
20071 2017-08-02 Richard Biener <rguenther@suse.de>
20073 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20075 (bitmap_insert_into_set): ... this.
20077 2017-08-02 Richard Biener <rguenther@suse.de>
20079 PR tree-optimization/81633
20081 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20083 PR tree-optimization/71752
20084 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20086 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20088 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20089 (machine_function::call_ms2sysv_pad_out): Remove field.
20090 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20091 (ix86_compute_frame_layout): Likewise.
20093 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20096 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20097 attribute with interrupt attribute.
20099 2017-08-01 Andrew Pinski <apinski@cavium.com>
20101 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20102 BIT_INSERT_EXPR's operand 1
20103 to see if the types precision matches.
20105 2017-08-01 Martin Liska <mliska@suse.cz>
20107 PR middle-end/70140
20108 * builtins.c (expand_builtin_memcpy_args): Remove.
20109 (expand_builtin_memcpy): Call newly added function
20110 expand_builtin_memory_copy_args.
20111 (expand_builtin_memcpy_with_bounds): Likewise.
20112 (expand_builtin_mempcpy): Remove last argument.
20113 (expand_builtin_mempcpy_with_bounds): Likewise.
20114 (expand_builtin_memory_copy_args): New function created from
20115 expand_builtin_mempcpy_args with small modifications.
20116 (expand_builtin_mempcpy_args): Remove.
20117 (expand_builtin_stpcpy): Remove unused argument.
20118 (expand_builtin): Likewise.
20119 (expand_builtin_with_bounds): Likewise.
20121 2017-08-01 Martin Liska <mliska@suse.cz>
20124 Make mempcpy more optimal (PR middle-end/70140).
20126 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20129 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20130 __builtin_vec_cmpne verify both arguments are compatible vectors
20131 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20132 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20133 move computation of aligned to after checking the argument types.
20137 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20138 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20141 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20142 Doug Rupp <rupp@adacore.com>
20143 Olivier Hainque <hainque@adacore.com>
20145 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20146 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20148 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20149 for TARGET_OS_CPP_BUILTIN.
20150 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20151 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20153 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20154 passing required abi options to the assembler for EABI configurations.
20155 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20156 of .text.hot and .text.unlikely sections for kernel modules when
20157 using ARM style exceptions.
20158 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20159 options. Add EXTRA_CC1_SPEC.
20160 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20162 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20164 (ARM_TARGET2_DWARF_FORMAT): Define.
20165 * config/arm/t-vxworks: Adjust multilib control to removal of the
20166 Diab command line options.
20168 2017-08-01 Martin Liska <mliska@suse.cz>
20170 PR gcov-profile/81561
20171 * gcov.c (unblock): Make unblocking safe as we need to preserve
20172 index correspondence of blocks and block_lists.
20174 2017-08-01 Richard Biener <rguenther@suse.de>
20176 PR tree-optimization/81181
20177 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20178 (compute_antic): ... end of iteration here.
20180 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20182 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20183 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20184 (ftree-slp-vectorize): Likewise.
20185 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20186 can no longer be set independent of flag_tree_loop_vectorize.
20187 * omp-general.c (emp_max_vf): Likewise.
20188 * opts.c (enable_fdo_optimizations): Remove references to
20189 flag_tree_vectorize, these are now implicit.
20190 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20191 and leave it for the options machinery.
20193 2017-08-01 Martin Liska <mliska@suse.cz>
20195 PR middle-end/70140
20196 * builtins.c (expand_builtin_memcpy_args): Remove.
20197 (expand_builtin_memcpy): Call newly added function
20198 expand_builtin_memory_copy_args.
20199 (expand_builtin_memcpy_with_bounds): Likewise.
20200 (expand_builtin_mempcpy): Remove last argument.
20201 (expand_builtin_mempcpy_with_bounds): Likewise.
20202 (expand_builtin_memory_copy_args): New function created from
20203 expand_builtin_mempcpy_args with small modifications.
20204 (expand_builtin_mempcpy_args): Remove.
20205 (expand_builtin_stpcpy): Remove unused argument.
20206 (expand_builtin): Likewise.
20207 (expand_builtin_with_bounds): Likewise.
20209 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20212 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20213 print "ds:" only for immediates in generic address space.
20215 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20218 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20219 (ix86_function_ok_for_sibcall): Return false for naked functions.
20221 2017-08-01 Richard Biener <rguenther@suse.de>
20223 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20224 (compute_antic): Seed worklist with exit block predecessors.
20225 * cfganal.c (dfs_find_deadend): For a cycle return the source
20226 of the edge closing it.
20228 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20230 * config/aarch64/aarch64.c
20231 (aarch64_can_const_movi_rtx_p): Move 0 check.
20233 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20235 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20236 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20238 * match.pd: Ditto in address comparison pattern.
20240 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20242 PR tree-optimization/81627
20243 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20244 closed ssa form for store-store chain.
20246 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20248 PR tree-optimization/81620
20249 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20250 for store-store chain.
20252 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20254 PR tree-optimization/81588
20255 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20256 ranges[i].in_p, invert comparison code ccode. For >/>=,
20257 swap rhs1 and rhs2 and comparison code unconditionally,
20258 for </<= don't do that. Don't swap rhs1/rhs2 again if
20259 ranges[i].in_p, instead invert comparison code ccode if
20260 opcode or oe->rank is BIT_IOR_EXPR.
20263 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20264 a direct optab to conversion optab.
20265 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20266 with GET_MODE_INNER as last argument instead of optab_handler.
20267 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20268 vector extraction if possible and optab is available.
20269 * expr.c (store_constructor): Use convert_optab_handler instead
20270 of optab_handler. Use vector initialization from smaller
20271 vectors if possible and optab is available.
20272 * tree-vect-stmts.c (vectorizable_load): Likewise.
20273 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20274 now have two modes.
20275 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20276 of vec_init from half-sized vectors with the same element mode.
20277 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20278 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20279 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20280 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20281 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20282 after mode in gen_vec_extract* calls.
20283 (vec_extract<mode>): Renamed to ...
20284 (vec_extract<mode><ssescalarmodelower>): ... this.
20285 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20286 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20287 element mode after mode in gen_vec_init* calls.
20288 (VEC_INIT_HALF_MODE): New mode iterator.
20289 (vec_init<mode>): Renamed to ...
20290 (vec_init<mode><ssescalarmodelower>): ... this.
20291 (vec_init<mode><ssehalfvecmodelower>): New expander.
20292 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20293 (vec_extractv2sfsf): ... this.
20294 (vec_initv2sf): Renamed to ...
20295 (vec_initv2sfsf): ... this.
20296 (vec_extractv2si): Renamed to ...
20297 (vec_extractv2sisi): ... this.
20298 (vec_initv2si): Renamed to ...
20299 (vec_initv2sisi): ... this.
20300 (vec_extractv4hi): Renamed to ...
20301 (vec_extractv4hihi): ... this.
20302 (vec_initv4hi): Renamed to ...
20303 (vec_initv4hihi): ... this.
20304 (vec_extractv8qi): Renamed to ...
20305 (vec_extractv8qiqi): ... this.
20306 (vec_initv8qi): Renamed to ...
20307 (vec_initv8qiqi): ... this.
20308 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20309 (vec_init<mode>): Renamed to ...
20310 (vec_init<mode><VEC_base_l>): ... this.
20311 (vec_extract<mode>): Renamed to ...
20312 (vec_extract<mode><VEC_base_l>): ... this.
20313 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20314 (vec_initv2sfsf): ... this.
20315 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20316 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20317 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20318 element mode after mode in gen_vec_init* calls.
20319 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20320 (vec_init<mode><Vel>): ... this.
20321 (vec_extract<mode>): Renamed to ...
20322 (vec_extract<mode><Vel>): ... this.
20323 * config/aarch64/iterators.md (Vel): New mode attribute.
20324 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20325 Add element mode after mode in gen_vec_extract* calls.
20326 * config/s390/vector.md (non_vec_l): New mode attribute.
20327 (vec_extract<mode>): Renamed to ...
20328 (vec_extract<mode><non_vec_l>): ... this.
20329 (vec_init<mode>): Renamed to ...
20330 (vec_init<mode><non_vec_l>): ... this.
20331 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20332 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20334 * config/arm/iterators.md (V_elem_l): New mode attribute.
20335 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20336 (vec_extract<mode><V_elem_l>): ... this.
20337 (vec_extractv2di): Renamed to ...
20338 (vec_extractv2didi): ... this.
20339 (vec_init<mode>): Renamed to ...
20340 (vec_init<mode><V_elem_l>): ... this.
20341 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20342 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20343 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20344 Add element mode after gen_vec_extract* calls.
20345 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20346 (vec_init<mode><unitmode>): ... this.
20347 (vec_extract<mode>): Renamed to ...
20348 (vec_extract<mode><unitmode>): ... this.
20349 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20350 (vec_init<mode><unitmode>): ... this.
20351 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20352 (vec_initv2sfsf): ... this.
20353 (vec_extractv2sf): Renamed to ...
20354 (vec_extractv2sfsf): ... this.
20355 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20356 Add element mode after gen_vec_extract* calls.
20357 * config/mips/mips.md (unitmode): New mode iterator.
20358 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20359 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20360 * config/spu/spu.md (inner_l): New mode attribute.
20361 (vec_init<mode>): Renamed to ...
20362 (vec_init<mode><inner_l>): ... this.
20363 (vec_extract<mode>): Renamed to ...
20364 (vec_extract<mode><inner_l>): ... this.
20365 * config/sparc/sparc.md (veltmode): New mode iterator.
20366 (vec_init<VMALL:mode>): Renamed to ...
20367 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20368 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20369 (vec_initv2sisi): ... this.
20370 (vec_initv2sf): Renamed to ...
20371 (vec_initv2sfsf): ... this.
20372 (vec_extractv2sf): Renamed to ...
20373 (vec_extractv2sfsf): ... this.
20374 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20375 (vec_init<mode>): Renamed to ...
20376 (vec_init<mode><VEC_base_l>): ... this.
20377 (vec_extract<mode>): Renamed to ...
20378 (vec_extract<mode><VEC_base_l>): ... this.
20379 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20380 (vec_initv2sfsf): ... this.
20381 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20382 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20383 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20384 gen_vec_init* calls.
20386 2017-08-01 Richard Biener <rguenther@suse.de>
20388 PR tree-optimization/81297
20389 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20390 TREE_OVERFLOW from INTEGER_CSTs.
20392 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20394 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20396 2017-07-31 Carl Love <cel@us.ibm.com>
20398 * config/rs6000/rs6000-c: Add support for built-in functions
20399 vector signed char vec_xl_be (signed long long, signed char *);
20400 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20401 vector signed int vec_xl_be (signed long long, signed int *);
20402 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20403 vector signed long long vec_xl_be (signed long long, signed long long *);
20404 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20405 vector signed short vec_xl_be (signed long long, signed short *);
20406 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20407 vector double vec_xl_be (signed long long, double *);
20408 vector float vec_xl_be (signed long long, float *);
20409 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20410 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20411 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20413 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20414 (altivec_expand_builtin): Add switch statement to call
20415 altivec_expand_xl_be for each builtin.
20416 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20417 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20418 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20419 __builtin_vsx_le_be_v16qi.
20420 * doc/extend.texi: Update the built-in documentation file for the
20421 new built-in functions.
20423 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20426 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20428 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20430 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20432 * config.gcc: Add z14.
20433 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20434 CPU model numbers for z13s and z14.
20435 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20437 * config/s390/s390-opts.h (enum processor_type): Rename
20438 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20439 * config/s390/s390.c (processor_table): Add field for CPU name to
20440 be passed to Binutils.
20441 (s390_asm_output_machine_for_arch): Use the new field in
20442 processor_table for Binutils.
20443 (s390_expand_builtin): Replace arch12 with z14.
20444 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20445 (s390_get_sched_attrmask): Likewise.
20446 (s390_get_unit_mask): Likewise.
20447 * config/s390/s390.opt: Add z14 to processor_type enum.
20449 2017-07-31 Martin Jambor <mjambor@suse.cz>
20452 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20453 regardless of optimization level.
20455 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20456 Martin Liska <mliska@suse.cz>
20458 * predict.def: Remove old comment and adjust probability.
20459 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20460 PREDICT statements.
20462 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20465 * config/i386/i386.c (ix86_function_naked): New function.
20466 (ix86_can_use_return_insn_p): Return false for naked functions.
20467 (ix86_expand_prologue): Skip prologue for naked functions.
20468 (ix86_expand_epilogue): Skip epilogue for naked functions
20469 and emit trap instruction.
20470 (ix86_warn_func_return): New function.
20471 (ix86_attribute_table): Add "naked" attribute specification.
20472 (TARGET_WARN_FUNC_RETURN): Define.
20473 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20475 2017-07-31 Martin Liska <mliska@suse.cz>
20477 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20478 (dump_gimple_bb_header): Always dump BB info.
20479 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20481 2017-07-31 Martin Liska <mliska@suse.cz>
20484 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20485 also with current_function_decl non-null equality.
20487 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20490 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20491 change type to the element type, instead add eltype variable and
20492 use it where we are interested in the element type.
20494 PR tree-optimization/81603
20495 * ipa-polymorphic-call.c
20496 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20497 offset arithmetic in offset_int, bail out if the resulting bit offset
20498 doesn't fit into shwi.
20500 2017-07-31 Martin Liska <mliska@suse.cz>
20502 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20503 (gimplify_save_expr): Fix comment.
20505 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20508 * config/i386/i386.c (ix86_function_arg): Update arguments for
20510 (ix86_compute_frame_layout): Set the initial stack offset to
20511 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20512 INCOMING_FRAME_SP_OFFSET.
20513 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20514 stack before exception handler returns.
20515 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20516 the 'ERROR_CODE' for exception handler.
20518 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20520 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20521 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20522 (ASM_OUTPUT_REG_POP): Ditto.
20523 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20524 instead of asm_fprintf to output pure string.
20526 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20528 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20529 to imported_module_or_decl hook.
20530 (debug_nothing_tree_tree_tree_bool): Remove.
20531 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20532 * debug.c (do_nothing_debug_hooks): Use
20533 debug_nothing_tree_tree_tree_bool_bool instead of
20534 debug_nothing_tree_tree_tree_bool.
20535 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20536 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20537 * sdbout.c (sdb_debug_hooks): Likewise.
20538 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20539 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20541 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20542 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20543 attribute, don't add anything.
20545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20547 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20548 (fold_build2_stat_loc): Likewise.
20549 (fold_build3_stat_loc): Likewise.
20550 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20551 (fold_build1_loc): Remove macro.
20552 (fold_build2_loc): Likewise.
20553 (fold_build3_loc): Likewise.
20555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20557 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20558 (gimple_build_debug_bind_source_stat): Likewise.
20559 * gimple.h (gimple_build_debug_bind): Remove macro.
20560 (gimple_build_debug_bind_source): Likewise.
20562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20564 * bitmap.c (bitmap_alloc): Adjust.
20565 (bitmap_gc_alloc): Likewise.
20566 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20568 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20570 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20571 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20572 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20573 (bitmap_gc_alloc_stat): Likewise.
20574 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20576 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20578 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20579 * rtl.h (shallow_copy_rtx): Remove macro.
20581 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20583 * emit-rtl.c (gen_raw_REG): Adjust.
20584 * gengenrtl.c (gendef): Likewise.
20585 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20586 * rtl.h (rtx_alloc): Remove macro.
20588 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20590 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20591 (build_tree_list_stat): Likewise.
20592 * tree.h (build_tree_list): Remove macro.
20593 (build_tree_list_vec): Likewise.
20595 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20597 * tree.c (make_vector_stat): Remove _stat from name.
20598 (build_vector_stat): Likewise.
20599 * tree.h (make_vector_stat): Remove macro.
20600 (build_vector_stat): Likewise.
20602 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20604 * tree.h (build_var_debug_value): Remove prototype.
20606 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20608 * tree.c (tree_cons_stat): Remove _stat from name.
20609 * tree.h (tree_cons): Remove macro.
20611 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20613 * tree.c (build_vl_exp_stat): Remove _stat from name.
20614 * tree.h (build_vl_exp): Remove macro.
20616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20618 * tree.c (build_decl_stat): Remove _stat from name.
20619 * tree.h (build_decl): Remove macro.
20621 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20623 * gimple.c (gimple_build_with_ops_stat): Adjust.
20624 (gimple_alloc_stat): Remove _stat from name.
20625 * gimple.h (gimple_alloc): Remove macro.
20627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20629 * tree.c (make_tree_vec_stat): Remove _stat from name.
20630 (grow_tree_vec_stat): Likewise.
20631 * tree.h (make_tree_vec_stat): Adjust prototype.
20632 (grow_tree_vec_stat): Likewise.
20633 (make_tree_vec): Remove macro.
20634 (grow_tree_vec): Likewise.
20636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20638 * fold-const.c (fold_build1_stat_loc): Adjust.
20639 (fold_build2_stat_loc): Likewise.
20640 (fold_build3_stat_loc): Likewise.
20641 * tree.c (build0_stat): Remove _stat from name.
20642 (build1_stat): Likewise.
20643 (build2_stat): Likewise.
20644 (build3_stat): Likewise.
20645 (build4_stat): Likewise.
20646 (build5_stat): Likewise.
20647 * tree.h (build1_loc): Remove macro, and rename _stat function
20649 (build2_loc): Likewise.
20650 (build3_loc): Likewise.
20651 (build4_loc): Likewise.
20652 (build5_loc): Likewise.
20654 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20656 * tree.c (make_int_cst_stat): Remove _stat from name.
20657 * tree.h (make_int_cst_stat): Adjust prototype.
20658 (make_int_cst): Remove macro.
20660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20662 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20663 * tree.h (make_tree_binfo_stat): Adjust prototype.
20664 (make_tree_binfo): Remove.
20666 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20668 * tree.c (copy_node_stat): Rename to copy_node.
20669 (build_distinct_type_copy): Adjust.
20670 * tree.h (copy_node_stat): Adjust prototype.
20671 (copy_node): Remove macro.
20673 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20675 * tree.c (make_node_stat): rename to make_node.
20676 (build_tree_list_stat): Adjust.
20677 (build0_stat): Likewise.
20678 (build2_stat): Likewise.
20679 (build3_stat): Likewise.
20680 (build4_stat): Likewise.
20681 (build5_stat): Likewise.
20682 (build_decl_stat): Likewise.
20683 * tree.h (make_node_stat): Adjust prototype.
20684 (make_node): remove macro.
20686 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20688 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20689 (PPC_FEATURE2_SCV): Likewise.
20690 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20692 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20694 * config/aarch64/aarch64.c
20695 (aarch64_internal_mov_immediate): Add new special pattern.
20696 * config/aarch64/aarch64.md (*movdi_aarch64):
20697 Add reg/32bit const mov case.
20699 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20700 Richard Sandiford <richard.sandiford@linaro.org>
20702 * config/aarch64/aarch64.md (mov<mode>): Generalize.
20703 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20704 Add integer and movi cases.
20705 (movi-split-hf-df-sf split, fp16): New.
20706 (enabled): Added TARGET_FP_F16INST.
20707 * config/aarch64/iterators.md (GPF_HF): New.
20708 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
20710 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20712 * config/aarch64/aarch64.c
20713 (aarch64_simd_container_mode): Add prototype.
20714 (aarch64_expand_mov_immediate): Add HI support.
20715 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
20716 (aarch64_can_const_movi_rtx_p): New.
20717 (aarch64_preferred_reload_class):
20718 Remove restrictions of using FP registers for certain SIMD operations.
20719 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20720 (aarch64_valid_floating_const): Add integer move validation.
20721 (aarch64_simd_imm_scalar_p): Remove.
20722 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20723 (aarch64_legitimate_constant_p): Expand list of supported cases.
20724 * config/aarch64/aarch64-protos.h
20725 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20726 (aarch64_reinterpret_float_as_int): New.
20727 (aarch64_simd_imm_scalar_p): Remove.
20728 * config/aarch64/constraints.md (Uvi): New.
20729 (Dd): Split into Ds and new Dd.
20730 * config/aarch64/aarch64.md (*movsi_aarch64):
20732 (*movdi_aarch64): Add SIMD mov case.
20734 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20736 * tree-predcom.c: (struct chain): Handle store-store chain in which
20737 stores for elimination only store loop invariant values.
20738 (execute_pred_commoning_chain): Ditto.
20739 (prepare_initializers_chain_store_elim): Ditto.
20740 (prepare_finalizers): Ditto.
20741 (is_inv_store_elimination_chain): New function.
20742 (initialize_root_vars_store_elim_1): New function.
20744 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20746 * tree-predcom.c: Revise general description of the pass.
20747 (enum chain_type): New enum type for store elimination.
20748 (struct chain): New field supporting store elimination.
20749 (struct component): Ditto.
20750 (dump_chain): Dump store-stores chain.
20751 (release_chain): Release resources.
20752 (split_data_refs_to_components): Compute and create component
20753 contains only stores for elimination.
20754 (get_chain_last_ref_at): New function.
20755 (make_invariant_chain): Initialization.
20756 (make_rooted_chain): Specify chain type in parameter and record it.
20757 (add_looparound_copies): Skip for store-stores chain.
20758 (determine_roots_comp): Compute type of chain and pass it to
20760 (initialize_root_vars_store_elim_2): New function.
20761 (finalize_eliminated_stores): New function.
20762 (remove_stmt): Handle store for elimination.
20763 (execute_pred_commoning_chain): Execute predictive commoning on
20764 store-store chains.
20765 (determine_unroll_factor): Skip unroll for store-stores chain.
20766 (prepare_initializers_chain_store_elim): New function.
20767 (prepare_initializers_chain): Hanlde store-store chain.
20768 (prepare_finalizers_chain, prepare_finalizers): New function.
20769 (tree_predictive_commoning_loop): Return integer value indicating
20770 if loop is unrolled or lcssa form is corrupted.
20771 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20773 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20775 * tree-predcom.c (initialize_root): Delete.
20776 (execute_pred_commoning_chain): Initialize root vars and replace
20777 reference of non-combined chain directly, rather than call above
20780 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20782 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20783 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20785 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20787 * tree-predcom.c (struct chain): New field init_seq.
20788 (release_chain): Release init_seq.
20789 (prepare_initializers_chain): Record intialization stmts in above
20791 (insert_init_seqs): New function.
20792 (tree_predictive_commoning_loop): Call insert_init_seqs.
20794 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20796 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20798 2017-07-28 Richard Biener <rguenther@suse.de>
20800 * match.pd: Remove superfluous :c.
20801 * genmatch.c (simplify::id): Add member.
20802 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20804 (current_id): New global.
20805 (dt_node::parent): Move from ...
20806 (dt_operand::parent): ... here. Add for_id member.
20807 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20808 (decision_tree::find_node): Relax order requirement when
20809 merging DT_TRUE nodes to ones inbetween the current simplify
20810 and the one we try to merge with. Add diagnostic whenever
20811 we need to enforce pattern order by not merging.
20812 (decision_tree::insert): Set current_id.
20813 (decision_tree::print_node): Dump parent node and for_id.
20814 (parser::last_id): Add member.
20815 (parser::push_simplify): Assign unique id.
20816 (parser::parser): Initialize last_id.
20818 2017-07-28 Martin Liska <mliska@suse.cz>
20821 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20822 gimple_build_debug_bind.
20824 2017-07-28 Richard Biener <rguenther@suse.de>
20826 PR tree-optimization/81502
20827 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20829 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20831 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20832 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20833 for BIT_FIELD_REF args.
20834 * fold-const.c (make_bit_field_ref): Likewise.
20835 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20837 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20840 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20841 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20842 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20843 Or it into SANITIZER_UNDEFINED.
20844 * ubsan.c: Include gimple-fold.h and varasm.h.
20845 (ubsan_expand_ptr_ifn): New function.
20846 (instrument_pointer_overflow): New function.
20847 (maybe_instrument_pointer_overflow): New function.
20848 (instrument_object_size): Formatting fix.
20849 (pass_ubsan::execute): Call instrument_pointer_overflow
20850 and maybe_instrument_pointer_overflow.
20851 * internal-fn.c (expand_UBSAN_PTR): New function.
20852 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20853 * sanitizer.def (__ubsan_handle_pointer_overflow,
20854 __ubsan_handle_pointer_overflow_abort): New builtins.
20855 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20856 * internal-fn.def (UBSAN_PTR): New internal function.
20857 * opts.c (sanitizer_opts): Add pointer-overflow.
20858 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20859 * fold-const.c (build_range_check): Compute pointer range check in
20860 integral type if pointer arithmetics would be needed. Formatting
20863 2017-07-28 Martin Liska <mliska@suse.cz>
20866 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20867 parameters that are of a variable-length.
20869 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20871 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20873 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20874 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20875 (CRT_CALL_STATIC_FUNCTION): Likewise.
20876 (ASM_DEFAULT_SPEC): New define.
20877 (ASM_SPEC32): Likewise.
20878 (ASM_SPEC64): Likewise.
20879 (ASM_SPEC_COMMON): Likewise.
20880 (ASM_SPEC): Likewise.
20881 (INVALID_64BIT): Likewise.
20882 (LINK_OS_DEFAULT_SPEC): Likewise.
20883 (LINK_OS_SPEC32): Likewise.
20884 (LINK_OS_SPEC64): Likewise.
20885 (POWERPC_LINUX): Likewise.
20886 (PTRDIFF_TYPE): Likewise.
20887 (RESTORE_FP_PREFIX): Likewise.
20888 (RESTORE_FP_SUFFIX): Likewise.
20889 (SAVE_FP_PREFIX): Likewise.
20890 (SAVE_FP_SUFFIX): Likewise.
20891 (SIZE_TYPE): Likewise.
20892 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20893 (TARGET_64BIT): Likewise.
20894 (TARGET_64BIT): Likewise.
20895 (TARGET_AIX): Likewise.
20896 (WCHAR_TYPE_SIZE): Likewise.
20897 (WCHAR_TYPE): Undefine.
20898 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20899 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20900 (CPP_OS_RTEMS_SPEC): Delete.
20901 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20902 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20904 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20906 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20908 PR tree-optimization/81578
20909 * tree-parloops.c (build_new_reduction): Bail out if
20910 reduction_code isn't one of the standard OpenMP reductions.
20911 Move the details printing after that decision.
20913 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20915 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20916 related to reload_in_progress.
20917 (splat_input_operand): Likewise.
20918 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20920 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20922 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20923 (TARGET_INSTANTIATE_DECLS): Likewise.
20924 (legitimate_indexed_address_p): Delete reload_in_progress code.
20925 (rs6000_debug_legitimate_address_p): Likewise.
20926 (rs6000_eliminate_indexed_memrefs): Likewise.
20927 (rs6000_emit_le_vsx_store): Likewise.
20928 (rs6000_emit_move_si_sf_subreg): Likewise.
20929 (rs6000_emit_move): Likewise.
20930 (register_to_reg_type): Likewise.
20931 (rs6000_pre_atomic_barrier): Likewise.
20932 (rs6000_machopic_legitimize_pic_address): Likewise.
20933 (rs6000_allocate_stack_temp): Likewise.
20934 (rs6000_address_for_fpconvert): Likewise.
20935 (rs6000_address_for_altivec): Likewise.
20936 (rs6000_secondary_memory_needed_rtx): Delete function.
20937 (rs6000_check_sdmode): Likewise.
20938 (rs6000_alloc_sdmode_stack_slot): Likewise.
20939 (rs6000_instantiate_decls): Likewise.
20940 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20941 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20942 Delete reload_in_progress.
20943 (*vec_reload_and_plus_<mptrsize>): Likewise.
20944 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20945 (vsx_div_v2di): Likewise.
20946 (vsx_udiv_v2di): Likewise.
20948 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20950 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20951 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20952 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20953 (rs6000_debug_reg_global): Delete print of LRA status.
20954 (rs6000_option_override_internal): Delete dead LRA related code.
20955 (rs6000_lra_p): Delete function.
20956 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20958 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20960 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20961 * config/riscv/rtems.h: New file.
20963 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20964 Sudakshina Das <sudi.das@arm.com>
20966 * config/aarch64/aarch64.md
20967 (define_split for and<mode>3nr_compare): Move
20968 non aarch64_logical_operand to a register.
20969 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20970 register immediate operand to a register.
20971 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20973 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20975 PR middle-end/81564
20976 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20978 2017-07-27 Richard Biener <rguenther@suse.de>
20980 PR tree-optimization/81573
20981 PR tree-optimization/81494
20982 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20983 multi defuse cycle case.
20985 2017-07-27 Richard Biener <rguenther@suse.de>
20987 PR tree-optimization/81571
20988 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20991 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20993 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20994 earlier and only if MASK_FPU is set. Adjust formatting.
20996 2017-07-27 Martin Liska <mliska@suse.cz>
20998 * opt-functions.awk: Add validation of value of Init.
20999 * optc-gen.awk: Pass new argument.
21001 2017-07-27 Martin Liska <mliska@suse.cz>
21003 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21004 Fix wrong condition.
21006 2017-07-27 Martin Liska <mliska@suse.cz>
21008 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21009 BBs and edges seen by autoFDO.
21011 2017-07-27 Richard Biener <rguenther@suse.de>
21013 PR tree-optimization/81502
21014 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21015 with incompatible but same sized type.
21016 (execute_update_addresses_taken): Likewise.
21018 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21020 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21021 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21023 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21026 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21027 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21028 Change s_operand to memory_operand.
21030 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21032 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21033 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21034 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21035 Emit instructions rather than returning an expression. Handle TFmode
21036 and KFmode by casting to TImode.
21037 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21038 (rs6000_emit_le_vsx_store): Likewise.
21039 * config/rs6000/vsx.md (VSX_TI): New iterator.
21040 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21041 (*vsx_le_undo_permute_<mode>): Likewise.
21042 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21043 emit the split sequence.
21044 (*vsx_le_perm_store_<mode>): Likewise.
21046 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21048 PR tree-optimization/81555
21049 PR tree-optimization/81556
21050 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21051 if true, force CHANGED for the recursive invocation.
21052 (reassociate_bb): Remember original length of ops array, pass
21053 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21055 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21056 attributes for noipa attribute. For naked attribute use
21057 lookup_attribute first before lookup_attribute_spec.
21058 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21060 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21061 for functions with noipa attribute.
21062 (cgraph_externally_visible_p): Return true for functions with noipa
21064 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21065 for functions with noipa attribute.
21066 * doc/extend.texi: Document noipa function attribute.
21067 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21068 also for functions with noipa attribute.
21069 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21071 2017-07-26 Andrew Pinski <apinski@cavium.com>
21073 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21074 vec_unalign_load_cost and vec_unalign_store_cost.
21076 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21078 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21079 -mvsx-small-integer option.
21080 (ISA_3_0_MASKS_IEEE): Likewise.
21081 (OTHER_VSX_VECTOR_MASKS): Likewise.
21082 (POWERPC_MASKS): Likewise.
21083 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21084 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21085 code, only testing for DImode being allowed in non-VSX floating
21087 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21088 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21090 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21091 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21092 TARGET_P8_VECTOR test.
21093 (rs6000_secondary_reload_simple_move): Likewise.
21094 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21095 since TARGET_P9_VECTOR was already tested.
21096 (rs6000_opt_masks): Remove -mvsx-small-integer.
21097 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21098 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21100 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21101 test for TARGET_VEXTRACTUB was used, and that uses
21103 (p9 extract splitter): Likewise.
21104 (vsx_extract_<mode>_di_p9): Likewise.
21105 (vsx_extract_<mode>_store_p9): Likewise.
21106 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21107 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21108 the elimination of TARGET_VSX_SMALL_INTEGER.
21109 (vsx_extract_<mode>_p8): Likewise.
21110 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21111 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21112 (vsx_set_<mode>_p9): Likewise.
21113 (vsx_set_v4sf_p9): Likewise.
21114 (vsx_set_v4sf_p9_zero): Likewise.
21115 (vsx_insert_extract_v4sf_p9): Likewise.
21116 (vsx_insert_extract_v4sf_p9_2): Likewise.
21117 * config/rs6000/rs6000.md (sign extend splitter): Change
21118 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21119 (floatsi<mode>2_lfiwax_mem): Likewise.
21120 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21121 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21122 since a test for TARGET_P9_VECTOR was used.
21123 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21124 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21125 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21126 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21127 TARGET_P8_VECTOR test.
21128 (fix_trunc<mode>si2_stfiwx): Likewise.
21129 (fix_trunc<mode>si2_internal): Likewise.
21130 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21131 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21133 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21134 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21135 TARGET_P8_VECTOR test.
21136 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21137 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21138 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21140 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21141 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21142 since a test for TARGET_P9_VECTOR was used.
21143 (splitter for loading small constants): Likewise.
21145 2017-07-26 Andrew Pinski <apinski@cavium.com>
21147 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21150 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21153 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21154 (fp_valid_at): Likewise.
21156 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21158 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21159 (qdf24xx_addrcost_table): Likewise.
21160 (cortexa57_tunings): Update to use generic_branch_cost.
21161 (cortexa72_tunings): Likewise.
21162 (cortexa73_tunings): Likewise.
21163 (qdf24xx_tunings): Likewise.
21165 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21167 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21168 (thunderx2t99_branch_cost): Likewise.
21169 (cortexa35_tunings): Update to use generic_branch_cost.
21170 (cortexa53_tunings): Likewise.
21171 (cortexa57_tunings): Likewise.
21172 (cortexa72_tunings): Likewise.
21173 (cortexa73_tunings): Likewise.
21174 (thunderx2t99_tunings): Likewise.
21176 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21178 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21179 (sparc_option_override): Honour MASK_FSMULD.
21180 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21181 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21182 * config/sparc/sparc.opt (mfsmuld): New option.
21183 * doc/invoke.texi (mfsmuld): Document option.
21185 2017-07-26 Marek Polacek <polacek@redhat.com>
21187 PR middle-end/70992
21188 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21190 2017-07-26 Richard Biener <rguenther@suse.de>
21192 * gimple-match-head.c (do_valueize): Return OP if valueize
21194 (get_def): New helper to get at the def stmt of a SSA name
21195 if valueize allows.
21196 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21197 do_valueize to get at the def stmt.
21198 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21200 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21202 PR middle-end/46932
21203 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21205 2017-07-26 Martin Liska <mliska@suse.cz>
21208 * function.c (expand_function_start): Make expansion of
21209 nonlocal_goto_save_area after parm_birth_insn.
21211 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21213 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21214 from all CPU target flags enable members.
21216 2017-07-26 Richard Biener <rguenther@suse.de>
21218 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21219 (decision_tree::gen): Make 'type' const.
21220 (write_predicate): Likewise.
21222 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21224 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21225 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21226 (rs6000_option_override_internal): Likewise.
21227 (rs6000_expand_vector_set): Likewise.
21228 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21229 (TARGET_UPPER_REGS_SF): Likewise.
21230 (TARGET_UPPER_REGS_DI): Likewise.
21231 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21232 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21233 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21234 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21235 (Splitters for DI constants in Altivec registers): Likewise.
21236 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21237 (vsx_set_v4sf_p9): Likewise.
21238 (vsx_set_v4sf_p9_zero): Likewise.
21239 (vsx_insert_extract_v4sf_p9): Likewise.
21240 (vsx_insert_extract_v4sf_p9_2): Likewise.
21242 2017-07-25 Carl Love <cel@us.ibm.com>
21244 * doc/extend.texi: Update the built-in documentation file for the
21245 existing built-in functions
21246 vector signed char vec_cnttz (vector signed char);
21247 vector unsigned char vec_cnttz (vector unsigned char);
21248 vector signed short vec_cnttz (vector signed short);
21249 vector unsigned short vec_cnttz (vector unsigned short);
21250 vector signed int vec_cnttz (vector signed int);
21251 vector unsigned int vec_cnttz (vector unsigned int);
21252 vector signed long long vec_cnttz (vector signed long long);
21253 vector unsigned long long vec_cnttz (vector unsigned long long);
21255 2017-07-25 Andrew Pinski <apinski@cavium.com>
21257 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21258 accesses where the use is for the first operand of a BIT_INSERT.
21260 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21263 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21264 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21266 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21268 * config/i386/gstabs.h: Delete.
21269 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21271 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21273 * config/i386/i386.c (ix86_decompose_address): Do not check for
21274 register RTX when looking at index_reg or base_reg.
21275 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21277 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21279 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21280 to update EH info here.
21282 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21284 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21286 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21288 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21290 2017-07-25 Torsten Duwe <duwe@suse.de>
21292 * common.opt: Introduce -fpatchable-function-entry
21293 command line option, and its variables function_entry_patch_area_size
21294 and function_entry_patch_area_start.
21295 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21296 including a two-value parser.
21297 * target.def (print_patchable_function_entry): New target hook.
21298 * targhooks.h (default_print_patchable_function_entry): New function.
21299 * targhooks.c (default_print_patchable_function_entry): Likewise.
21300 * toplev.c (process_options): Switch off IPA-RA if
21301 patchable function entries are being generated.
21302 * varasm.c (assemble_start_function): Look at the
21303 patchable-function-entry command line switch and current
21304 function attributes and maybe generate NOP instructions by
21305 calling the print_patchable_function_entry hook.
21306 * doc/extend.texi: Document patchable_function_entry attribute.
21307 * doc/invoke.texi: Document -fpatchable_function_entry
21308 command line option.
21309 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21311 * doc/tm.texi: Re-generate.
21313 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21316 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21317 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21319 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21321 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21323 2017-07-25 Richard Biener <rguenther@suse.de>
21325 PR tree-optimization/81455
21326 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21327 not walk in cycles when looking for guards.
21329 2017-07-25 Richard Biener <rguenther@suse.de>
21331 PR tree-optimization/81529
21332 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21333 when optimizing backedge uses.
21335 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21337 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21339 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21340 to dl_section_ref. On AIX, append an expression to subtract
21341 the size of the section length to dl_section_ref.
21343 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21345 * configure.ac: If any of the config.* scripts fail, exit 1.
21346 * configure: Regenerate.
21348 2017-07-25 Richard Biener <rguenther@suse.de>
21350 PR middle-end/81546
21351 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21352 of immediate uses, be more verbose on errors.
21354 2017-07-25 Richard Biener <rguenther@suse.de>
21356 PR tree-optimization/81510
21357 * tree-vect-loop.c (vect_is_simple_reduction): When the
21358 reduction stmt is not inside the loop bail out.
21360 2017-07-25 Richard Biener <rguenther@suse.de>
21362 PR tree-optimization/81303
21363 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21364 profitability check against LOOP_VINFO_NITERSM1.
21366 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21368 * domwalk.c (cmp_bb_postorder): Simplify.
21369 (sort_bbs_postorder): New function. Use it...
21370 (dom_walker::walk): ...here to optimize common cases.
21372 2017-07-25 Martin Liska <mliska@suse.cz>
21375 * ipa-visibility.c (function_and_variable_visibility): Make the
21376 redirection just on target that supports aliasing.
21377 Fix GNU coding style.
21379 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21382 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21384 * config/arm/rtems.h: Likewise.
21385 * config/bfin/rtems.h: Likewise.
21386 * config/i386/rtemself.h: Likewise.
21387 * config/lm32/rtems.h: Likewise.
21388 * config/m32c/rtems.h: Likewise.
21389 * config/m68k/rtemself.h: Likewise.
21390 * config/microblaze/rtems.h: Likewise.
21391 * config/mips/rtems.h: Likewise.
21392 * config/moxie/rtems.h: Likewise.
21393 * config/nios2/rtems.h: Likewise.
21394 * config/powerpcspe/rtems.h: Likewise.
21395 * config/rs6000/rtems.h: Likewise.
21396 * config/rtems.h: Likewise.
21397 * config/sh/rtems.h: Likewise.
21398 * config/sh/rtemself.h: Likewise.
21399 * config/sparc/rtemself.h: Likewise.
21401 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21404 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21405 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21406 * tree-ssa-structalias.c (alias_get_name): Same.
21408 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21411 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21412 instructions if no du chain is found.
21414 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21416 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21418 2017-07-25 Richard Biener <rguenther@suse.de>
21420 PR middle-end/81505
21421 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21424 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21426 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21427 upper-regs options.
21428 (ISA_2_7_MASKS_SERVER): Likewise.
21429 (ISA_3_0_MASKS_IEEE): Likewise.
21430 (OTHER_P8_VECTOR_MASKS): Likewise.
21431 (OTHER_VSX_VECTOR_MASKS): Likewise.
21432 (POWERPC_MASKS): Likewise.
21433 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21434 duplicate list of options.
21435 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21436 explicit -mupper-regs options.
21437 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21438 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21439 alias for -mupper-regs-df.
21440 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21441 (rs6000_init_hard_regno_mode_ok): Likewise.
21442 (rs6000_option_override_internal): Likewise.
21443 (rs6000_opt_masks): Likewise.
21444 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21445 options in terms of whether -mvsx or -mpower8-vector was used.
21446 (TARGET_UPPER_REGS_DI): Likewise.
21447 (TARGET_UPPER_REGS_SF): Likewise.
21448 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21449 -mupper-regs-* options.
21451 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21453 * passes.c (emergency_dump_function): Print some empty lines and a
21454 header before the RTL dump.
21456 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21458 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21460 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21463 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21464 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21466 2017-07-24 Carl Love <cel@us.ibm.com>
21468 * config/rs6000/rs6000-c.c: Add support for built-in functions
21469 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21470 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21471 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21472 vec_extract_fp_from_shortl): Add defines for the two builtins.
21473 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21474 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21476 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21477 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21478 * doc/extend.texi: Update the built-in documentation file for the
21479 new built-in function.
21481 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21484 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21486 * doc/generic.texi: Likewise.
21487 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21488 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21490 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21492 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21493 (aarch64_mls_elt_merge<mode>): Likewise.
21495 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21497 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21498 having __cxa_atexit.
21500 2017-07-23 Michael Collison <michael.collison@arm.com>
21502 * config/arm/arm.c (arm_option_override): Deprecate
21503 use of -mstructure-size-boundary.
21504 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21505 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21507 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21510 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21511 Reduce cost estimate for direct moves.
21513 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21516 * config/i386/i386.c (ix86_option_override_internal): Disable
21517 BMI, BMI2 and TBM instructions for -m16.
21519 2017-07-21 Carl Love <cel@us.ibm.com>
21521 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21522 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21523 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21524 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21525 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21526 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21527 VMULOSW): New enum "unspec" values.
21528 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21529 altivec_vmulosw): New patterns.
21530 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21531 VMULOSW): Add definitions.
21533 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21535 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21536 (qdf24xx): Likewise.
21537 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21538 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21539 (AARCH64_FL_V8_1): Renumber.
21540 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21541 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21542 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21543 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21544 rdma to feature modifiers list.
21546 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21548 PR middle-end/56727
21549 * ipa-visibility (function_and_variable_visibility): Convert
21550 recursive PLT call to direct call if appropriate.
21552 2017-07-21 Andrew Pinski <apinski@cavium.com>
21554 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21555 operand 1 to see if the types precision matches.
21556 * fold-const.c (operand_equal_p): Likewise.
21558 2017-07-21 Richard Biener <rguenther@suse.de>
21560 PR tree-optimization/81303
21561 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21562 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21563 (vect_peeling_hash_get_lowest_cost): Adjust.
21564 (vect_enhance_data_refs_alignment): Likewise. Use
21565 vect_get_peeling_costs_all_drs to compute the penalty for no
21566 peeling to match up costs.
21568 2017-07-21 Richard Biener <rguenther@suse.de>
21570 PR tree-optimization/81500
21571 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21572 we didn't identify a reduction path.
21574 2017-07-21 Tom de Vries <tom@codesourcery.com>
21575 Cesar Philippidis <cesar@codesourcery.com>
21577 PR gcov-profile/81442
21578 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21581 2017-07-21 Tom de Vries <tom@codesourcery.com>
21584 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21586 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21587 nvptx_override_options_after_change.
21589 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21591 * dwarf2out.c (output_file_names): Avoid double testing for
21592 dwarf_version >= 5.
21594 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21596 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21598 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21600 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21602 (try_crossjump_to_edge): Do not punt on partitioned functions.
21604 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21606 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21607 Put all BBs reachable only via paths crossing cold region to cold
21609 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21611 2016-07-21 Richard Biener <rguenther@suse.de>
21613 PR tree-optimization/81303
21614 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21615 into account prologue and epilogue iterations when raising
21616 min_profitable_iters to sth at least covering one vector iteration.
21618 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21620 * config/arm/arm.c (arm_test_cpu_arch_dat):
21623 2017-07-20 Nathan Sidwell <nathan@acm.org>
21625 Remove TYPE_METHODS.
21626 * tree.h (TYPE_METHODS): Delete.
21627 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21628 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21629 (dbxout_type_methods): Scan TYPE_FIELDS.
21630 (dbxout_type): Don't check TYPE_METHODS here.
21631 * function.c (use_register_for_decl): Always ignore register for
21632 class types when not optimizing.
21633 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21634 * tree.c (free_lang_data_in_type): Stitch out member functions and
21635 templates from TYPE_FIELDS.
21636 (build_distinct_type_copy, verify_type_variant,
21637 verify_type): Member fns are on TYPE_FIELDS.
21638 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21639 * tree-pretty-print.c (dump_generic_node): Likewise.
21641 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21644 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21645 V2TImode and V4TImode.
21646 (ix86_expand_vector_extract): Likewise.
21647 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21648 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21649 (ssescalarmode): Handle V4TImode and V2TImode.
21650 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21651 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21652 (VEXTRACTI128_MODE): New mode iterator.
21653 (splitter for *vec_extractv?ti first element): New.
21654 (VEC_INIT_MODE): New mode iterator.
21655 (vec_init<mode>): Consolidate 3 expanders into one using
21656 VEC_INIT_MODE mode iterator.
21658 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21660 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21661 non_spilled_static_chain_regno_p.
21663 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21665 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21667 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21669 * bb-reorder.c (connect_traces): Allow copying of blocks within
21672 2017-07-20 Richard Biener <rguenther@suse.de>
21674 * gimple.h (gimple_phi_result): Add gphi * overload.
21675 (gimple_phi_result_ptr): Likewise.
21676 (gimple_phi_arg): Likewise. Adjust index assert to only
21677 allow actual argument accesses rather than all slots available
21679 (gimple_phi_arg_def): Add gphi * overload.
21680 * tree-phinodes.c (make_phi_node): Initialize only actual
21682 (resize_phi_node): Clear memory not covered by old node,
21683 do not initialize excess argument slots.
21684 (reserve_phi_args_for_new_edge): Initialize new argument slot
21687 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21689 PR tree-optimization/81388
21691 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21693 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21694 by removing computation of may_be_zero.
21696 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21697 Tom de Vries <tom@codesourcery.com>
21699 PR middle-end/81030
21700 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
21701 when gimple level profile disagrees with what RTL expander did.
21703 2017-07-20 Richard Biener <rguenther@suse.de>
21705 PR tree-optimization/61171
21706 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
21707 (vect_analyze_stmt): Add slp instance parameter.
21708 (vectorizable_reduction): Likewise.
21709 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
21710 (vect_is_simple_reduction): Deal with chains not detected
21711 as SLP reduction chain, specifically not properly associated
21712 chains containing a mix of plus/minus.
21713 (get_reduction_op): Remove.
21714 (get_initial_defs_for_reduction): Simplify, pass in whether
21715 this is a reduction chain, pass in the SLP node for the PHIs.
21716 (vect_create_epilog_for_reduction): Get the SLP instance as
21718 (vectorizable_reduction): Get the SLP instance as arg.
21719 During analysis remember the SLP node with the PHIs in the
21720 instance. Simplify getting at the vectorized reduction PHIs.
21721 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21722 through SLP instance.
21723 (vect_slp_analyze_operations): Likewise.
21724 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21725 (vect_transform_stmt): Likewise.
21727 2017-07-20 Tom de Vries <tom@codesourcery.com>
21729 PR tree-optimization/81489
21730 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21731 read of phi arg location to before loop that modifies phi.
21733 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21735 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21738 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21740 PR middle-end/81331
21741 * except.c (execute): Fix ordering issue.
21743 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21745 PR rtl-optimization/81423
21746 * combine.c (make_compound_operation_int): Don't try to optimize
21747 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21749 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21751 PR rtl-optimization/81423
21752 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21753 with a constant that is -1 in the truncated to mode.
21755 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21757 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21758 (determine_unlikely_bbs): ... here.
21759 * predict.h (propagate_unlikely_bbs_forward): Declare.
21760 * cfgexpand.c (pass_expand::execute): Use it.
21761 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21763 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21764 propagate_unlikely_bbs_forward.
21766 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21768 PR middle-end/81331
21769 * except.c (maybe_add_nop_after_section_switch): New function.
21772 2017-07-19 Tom de Vries <tom@codesourcery.com>
21774 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21776 2017-07-19 Tom de Vries <tom@codesourcery.com>
21778 * gimple.h (gimple_phi_arg): Make assert more strict.
21780 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21782 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21783 * config/rs6000/mmintrin.h: New file.
21784 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21786 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21788 PR tree-optimization/81346
21789 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21791 2017-07-19 Tom de Vries <tom@codesourcery.com>
21793 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21795 2017-07-19 Tom de Vries <tom@codesourcery.com>
21797 * config/nvptx/nvptx-modes.def: Add V2DImode.
21798 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21799 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21800 (nvptx_output_mov_insn): Handle lack of mov.b128.
21801 (nvptx_print_operand): Handle 'H' and 'L' codes.
21802 (nvptx_vector_mode_supported): Allow V2DImode.
21803 (nvptx_preferred_simd_mode): New function.
21804 (nvptx_data_alignment): New function.
21805 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21806 nvptx_preferred_simd_mode.
21807 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21809 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21811 2017-07-19 Tom de Vries <tom@codesourcery.com>
21813 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21814 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21815 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21816 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21817 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21818 (mov<VECIM>_insn): New define_insn.
21819 (define_expand "mov<VECIM>): New define_expand.
21821 2017-07-19 Tom de Vries <tom@codesourcery.com>
21823 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21825 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21827 PR tree-optimization/81346
21828 * fold-const.h (fold_div_compare, range_check_type): Declare.
21829 * fold-const.c (range_check_type): New function.
21830 (build_range_check): Use range_check_type.
21831 (fold_div_compare): No longer static, rewritten into
21832 a match.pd helper function.
21833 (fold_comparison): Don't call fold_div_compare here.
21834 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21835 as helper function.
21837 2017-07-19 Nathan Sidwell <nathan@acm.org>
21839 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21840 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21841 * tree.c (find_decls_types_r, verify_type): Use
21842 TYPE_{MIN,MAX}_VALUE_RAW.
21843 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21844 (hash_tree): Likewise.
21845 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21847 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21850 2017-07-18 Tom de Vries <tom@codesourcery.com>
21852 PR middle-end/81464
21853 * omp-expand.c (expand_omp_for_static_chunk): Handle
21854 equal-argument loop exit phi.
21856 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21859 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21860 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21861 operand 2 predicate.
21862 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21863 operand 2 predicate.
21864 (ror,rol -> rorx splitters): Use const_int_operand as
21865 operand 2 predicate.
21867 2017-06-18 Richard Biener <rguenther@suse.de>
21869 PR tree-optimization/81410
21870 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21871 the gap in the ! slp_perm SLP case after each group.
21873 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21875 PR middle-end/81463
21876 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21879 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21881 PR middle-end/81462
21882 * predict.c (set_even_probabilities): Cleanup; do not affect
21883 probabilities that are already known.
21884 (combine_predictions_for_bb): Call even when count is set.
21886 2017-07-18 Nathan Sidwell <nathan@acm.org>
21888 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21891 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21894 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21895 optimization for loop niter analysis.
21897 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21900 * config/avr/avr.c (avr_optimize_casesi): Don't use
21901 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21903 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21905 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21906 body_cost_vec from _vect_peel_extended_info.
21907 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21908 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21911 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21913 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21915 2017-07-18 Richard Biener <rguenther@suse.de>
21917 PR tree-optimization/80620
21918 PR tree-optimization/81403
21919 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21920 info when re-using a VN table entry.
21922 2017-07-18 Richard Biener <rguenther@suse.de>
21924 PR tree-optimization/81418
21925 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21926 vectype_in. Verify that with lane-reducing reduction operations
21927 we have a single def-use cycle.
21929 2017-07-17 Carl Love <cel@us.ibm.com>
21931 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21933 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21934 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21935 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21936 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21937 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21938 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21939 VMULOSW): New enum "unspec" values.
21940 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21941 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21942 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21943 altivec_vmulosw): New patterns.
21944 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21945 VMULOSW): Add definitions.
21947 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21949 * config/alpha/alpha.c: Include predict.h.
21951 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21953 * tree-vrp.c (compare_assert_loc): Fix comparison function
21954 to return predictable results.
21956 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21958 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21960 (subdi3): Likewise.
21961 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21962 * doc/invoke.texi (mexpand-adddi): Update text.
21964 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21966 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21967 that also clobbers the CC register. The old expand code is moved
21969 (*arc_clzsi2): ... here.
21970 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21971 the CC register. The old expand code is moved to ...
21972 (arc_ctzsi2): ... here.
21974 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21976 * config/arc/arc.opt (mindexed-loads): Use initial value
21977 TARGET_INDEXED_LOADS_DEFAULT.
21978 (mauto-modify-reg): Use initial value
21979 TARGET_AUTO_MODIFY_REG_DEFAULT.
21980 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21981 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21982 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21983 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21985 2017-07-17 Martin Liska <mliska@suse.cz>
21988 * opts.c (finish_options): Do not allow -fgnu-tm
21989 w/ -fsanitize={kernel-,}address. Say sorry.
21991 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21994 * tree-loop-distribution.c (classify_partition): Only assert on
21995 numer of iterations.
21996 (merge_dep_scc_partitions): Delete prameter. Update function call.
21997 (distribute_loop): Remove code handling loop with unknown niters.
21998 (pass_loop_distribution::execute): Skip loop with unknown niters.
22000 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22003 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22004 function sort_partitions_by_post_order.
22006 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22008 PR tree-optimization/81374
22009 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22010 the max index of basic blocks, rather than number of basic blocks.
22012 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22014 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22016 (arc_legitimate_pic_operand_p): Likewise.
22017 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22019 (arc_needs_pcl_p): Likewise.
22020 (arc_legitimate_pc_offset_p): Likewise.
22021 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22022 function is also used in constrains.md.
22023 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22024 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22025 PLUS. Only return true/false in known cases, otherwise assert.
22026 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22027 is already called in arc_legitimate_constant_p.
22028 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22030 (LEGITIMATE_PIC_OPERAND_P): Use
22031 arc_raw_symbolic_reference_mentioned_p function.
22032 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22037 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22038 Andrew Burgess <andrew.burgess@embecosm.com>
22040 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22041 (arc_return_address_register): New function.
22042 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22043 (arc_handle_fndecl_attribute): Add naked attribute.
22044 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22045 (TARGET_WARN_FUNC_RETURN): Likewise.
22046 (arc_allocate_stack_slots_for_args): New function.
22047 (arc_warn_func_return): Likewise.
22048 (machine_function): Change type fn_type.
22049 (arc_compute_function_type): Consider new naked function type,
22050 change function return type.
22051 (arc_must_save_register): Adapt to handle new
22052 arc_compute_function_type's return type.
22053 (arc_expand_prologue): Likewise.
22054 (arc_expand_epilogue): Likewise.
22055 (arc_return_address_regs): Delete.
22056 (arc_return_address_register): New function.
22057 (arc_epilogue_uses): Use above function.
22058 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22059 (arc_function_type): Change encoding, add naked type.
22060 (ARC_INTERRUPT_P): Change to handle the new encoding.
22061 (ARC_FAST_INTERRUPT_P): Likewise.
22062 (ARC_NORMAL_P): Define.
22063 (ARC_NAKED_P): Likewise.
22064 (arc_compute_function_type): Delete prototype.
22065 * config/arc/arc.md (in_ret_delay_slot): Use
22066 arc_return_address_register function.
22067 (simple_return): Likewise.
22068 (p_return_i): Likewise.
22070 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22072 PR tree-optimization/81428
22073 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22074 can't be built for those types.
22076 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22078 Remove stuff dead since r239246.
22080 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22081 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22082 (avr_inform_devices): Remove dead stuff.
22084 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22086 * config/arm/arm_neon.h: Fix softp typo.
22088 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22090 PR tree-optimization/81365
22091 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22092 aggregate moves onto bb predecessor edges, make sure there are no
22093 loads that could alias the lhs in between the start of bb and the
22096 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22099 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22100 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22101 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22103 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22105 PR tree-optimization/81396
22106 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22107 (init_symbolic_number): Initialize it to 1.
22108 (perform_symbolic_merge): Add n_ops from both operands into the new
22110 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22111 without base_addr as useless if they need more than one operation.
22112 (bswap_replace): Handle !bswap case for NULL base_addr.
22114 2017-07-17 Tom de Vries <tom@codesourcery.com>
22117 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22120 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22122 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22123 conditional builtin define __FIX_LEON3FT_B2BST.
22125 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22127 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22128 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22131 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22133 PR rtl-optimization/81424
22134 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22135 to remove potential trapping from operands if -fnon-call-exceptions.
22137 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22139 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22140 profile_proability for scalling.
22141 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22143 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22145 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22147 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22149 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22150 fixpoint arithmetics.
22152 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22154 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22155 fixpoint arithmetics.
22157 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22159 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22160 fixpoint arithmetics.
22162 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22164 * profile-count.h (profile_probability::from_reg_br_prob_note,
22165 profile_probability::to_reg_br_prob_note): New functions.
22166 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22167 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22168 * predict.c (probability_reliable_p): Update.
22169 (edge_probability_reliable_p): Update.
22170 (br_prob_note_reliable_p): Update.
22171 (invert_br_probabilities): Update.
22172 (add_reg_br_prob_note): New function.
22173 (combine_predictions_for_insn): Update.
22174 * asan.c (asan_clear_shadow): Update.
22175 * cfgbuild.c (compute_outgoing_frequencies): Update.
22176 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22177 (update_br_prob_note): Update.
22178 (rtl_verify_edges): Update.
22179 (purge_dead_edges): Update.
22180 (fixup_reorder_chain): Update.
22181 * emit-rtl.c (try_split): Update.
22182 * ifcvt.c (cond_exec_process_insns): Update.
22183 (cond_exec_process_if_block): Update.
22184 (dead_or_predicable): Update.
22185 * internal-fn.c (expand_addsub_overflow): Update.
22186 (expand_neg_overflow): Update.
22187 (expand_mul_overflow): Update.
22188 * loop-doloop.c (doloop_modify): Update.
22189 * loop-unroll.c (compare_and_jump_seq): Update.
22190 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22191 * predict.h: Update.
22192 * reorg.c (mostly_true_jump): Update.
22194 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22195 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22196 * config/arc/arc.c: (emit_unlikely_jump): Update.
22197 * config/arm/arm.c: (emit_unlikely_jump): Update.
22198 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22199 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22200 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22201 (ix86_print_operand): Update.
22202 (ix86_split_fp_branch): Update.
22203 (predict_jump): Update.
22204 * config/ia64/ia64.c (ia64_print_operand): Update.
22205 * config/mmix/mmix.c (mmix_print_operand): Update.
22206 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22207 (rs6000_expand_split_stack_prologue): Update.
22208 * config/rs6000/rs6000.c: Update.
22209 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22210 (s390_expand_vec_movstr): Update.
22211 (s390_expand_cs_tdsi): Update.
22212 (s390_expand_split_stack_prologue): Update.
22213 * config/sh/sh.c (sh_print_operand): Update.
22214 (expand_cbranchsi4): Update.
22215 (expand_cbranchdi4): Update.
22216 * config/sparc/sparc.c (output_v9branch): Update.
22217 * config/spu/spu.c (get_branch_target): Update.
22218 (ea_load_store_inline): Update.
22219 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22220 * config/tilepro/tilepro.c: Update.
22222 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22224 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22225 (gimplify_save_expr): Likewise.
22227 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22229 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22231 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22233 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22235 * timevar.def (TV_IPA_FNSUMMARY): Define.
22237 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22239 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22240 to back store errata sensitive sequence from being generated.
22241 (sqrtdf2_fix): Likewise.
22243 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22245 * tree-ssa-threadupdate.c (compute_path_counts,
22246 update_joiner_offpath_counts): Use profile_probability.
22248 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22251 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22253 * config/arm/arm-c.c (arm_cpu_builtins): Define
22254 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22256 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22258 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22259 array entries to represent __ieee128 versions of the
22260 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22261 scalar_extract_sig, and scalar_insert_exp built-in functions.
22262 (altivec_resolve_overloaded_builtin): Add special case handling
22263 for the __builtin_scalar_insert_exp function, as represented by
22264 the P9V_BUILTIN_VEC_VSIEDP constant.
22265 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22266 exponent support for __ieee128 argument.
22267 (VSESQP): Add scalar extract signature support for __ieee128
22269 (VSTDCNQP): Add scalar test negative support for __ieee128
22271 (VSIEQP): Add scalar insert exponent support for __int128 argument
22272 with __ieee128 result.
22273 (VSIEQPF): Add scalar insert exponent support for __ieee128
22274 argument with __ieee128 result.
22275 (VSTDCQP): Add scalar test data class support for __ieee128
22277 (VSTDCNQP): Add overload support for scalar test negative with
22278 __ieee128 argument.
22279 (VSTDCQP): Add overload support for scalar test data class
22280 __ieee128 argument.
22281 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22282 UNSPEC_VSX_SXSIGDP.
22283 (UNSPEC_VSX_SIEXPQP): New constant.
22284 (xsxexpqp): New insn for VSX scalar extract exponent quad
22286 (xsxsigqp): New insn for VSX scalar extract significand quad
22288 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22289 precision with floating point argument.
22290 (xststdcqp): New expand for VSX scalar test data class quad
22292 (xststdcnegqp): New expand for VSX scalar test negative quad
22294 (xststdcqp): New insn to match expansions for VSX scalar test data
22295 class quad precision and VSX scalar test negative quad precision.
22296 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22297 special case operand checking to enforce that second operand of
22298 VSX scalar test data class with quad precision argument is a 7-bit
22300 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22301 prototypes and descriptions of __ieee128 versions of
22302 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22303 scalar_test_data_class, and scalar_test_neg built-in functions.
22305 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22307 PR tree-optimization/81162
22308 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22309 replace a negate with an add.
22311 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22313 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22315 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22317 * config/arm/arm-c.c (arm_cpu_builtins): Define
22318 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22320 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22322 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22323 (armv8-r): Set ARM Cortex-R52 as default CPU.
22324 * config/arm/arm-tables.opt: Regenerate.
22325 * config/arm/arm-tune.md: Regenerate.
22326 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22328 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22329 extension for -mcpu=cortex-r52.
22331 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22333 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22334 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22335 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22336 (fp-armv8): Define it as FP_ARMv8 only.
22337 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22338 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22340 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22341 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22342 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22343 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22344 than TARGET_FPU_ARMV8.
22345 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22346 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22347 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22349 * config/arm/neon.md (neon_vrint): Likewise.
22350 (neon_vcvt): Likewise.
22351 (neon_<fmaxmin_op><mode>): Likewise.
22352 (<fmaxmin><mode>3): Likewise.
22353 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22354 * config/arm/predicates.md (arm_cond_move_operator): Check against
22355 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22357 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22359 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22360 to top of function.
22362 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22364 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22365 loop in comment with memset.
22367 2017-07-14 Martin Liska <mliska@suse.cz>
22369 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22370 * dwarf2out.c (is_java): Remove the function.
22371 (output_pubname): Remove usage of the function.
22372 (lower_bound_default): Remove usage of DW_LANG_Java.
22373 (gen_compile_unit_die): Likewise.
22374 * gcc.c: Remove compiler defaults for .java and .zip files.
22375 * gimple-expr.c (remove_suffix): Change as there's no longer
22376 extension than 4-letter one.
22377 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22378 (gimplify_save_expr): Likewise.
22379 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22380 as it's possible even for other languages than Java.
22381 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22382 * lto-opts.c (lto_write_options): Remove reference to Java.
22383 * opts.c (strip_off_ending): Update file extension handling.
22384 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22385 * tree-eh.c (lower_resx): Likewise.
22386 * tree.c (free_lang_data_in_type): Remove dead code.
22387 (find_decls_types_r): Likewise.
22388 (build_common_builtin_nodes): Remove Java from a comment.
22389 (verify_type): Remove dead code.
22390 * varasm.c (assemble_external): Remove Java from a comment.
22392 2017-07-14 Martin Liska <mliska@suse.cz>
22394 * opts.c (finish_options): Add quotes.
22395 (common_handle_option): Likewise.
22397 2017-07-14 Martin Liska <mliska@suse.cz>
22399 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22400 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22401 Remove N_SO_PASCAL.
22402 * dwarf2out.c (lower_bound_default): Do not handle
22404 (gen_compile_unit_die): Likewise.
22405 * gcc.c: Remove default extension binding for GNU Pascal.
22406 * stmt.c: Remove Pascal language from a comment.
22407 * xcoffout.c: Likewise.
22409 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22412 * diagnostic-show-locus.c (fixit_cmp): New function.
22413 (layout::layout): Sort m_fixit_hints.
22414 (column_range::column_range): Assert that the values are valid.
22415 (struct char_span): New struct.
22416 (correction::overwrite): New method.
22417 (struct source_line): New struct.
22418 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22419 calls in terms of classes source_line and char_span, and
22420 correction::overwrite.
22421 (selftest::test_overlapped_fixit_printing_2): New function.
22422 (selftest::diagnostic_show_locus_c_tests): Call it.
22424 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22426 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22427 early if there is no lhs.
22429 2017-07-13 Martin Liska <mliska@suse.cz>
22431 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22432 (gen_reference_type_die): Likewise.
22433 * stor-layout.c: Remove Pascal-related comment.
22435 2017-07-13 Martin Liska <mliska@suse.cz>
22437 * opts.c (finish_options): Add quotes to error messages.
22438 (parse_sanitizer_options): Likewise.
22440 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22442 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22444 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22446 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22448 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22450 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22452 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22454 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22457 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22458 provides the hardware capability bits, define the macro
22459 __BUILTIN_CPU_SUPPORTS__.
22460 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22461 if GLIBC does not provide the hardware capability bits. Add a
22462 gcc_unreachable call if the built-in cpu function is neither
22463 __builtin_cpu_is nor __builtin_cpu_supports.
22464 (rs6000_get_function_versions_dispatcher): Change the warning
22465 that an old GLIBC is used which does not export the capability
22466 bits to be an error.
22467 * doc/extend.texi (target_clones attribute): Document the
22468 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22469 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22470 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22471 the macros defined by GCC if the newer GLIBC is available.
22473 2017-07-12 Jeff Law <law@redhat.com>
22475 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22476 remaining includes slightly.
22477 * config/riscv/riscv-builtins.c: Include profile-count.h.
22479 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22482 * config/avr/avr.c (avr_set_current_function): In diagnostic
22483 messages: Quote keywords and (parts of) identifiers.
22484 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22487 2017-07-12 Carl Love <cel@us.ibm.com>
22489 * config/rs6000/rs6000-c.c: Add support for built-in functions
22490 vector bool char vec_revb (vector bool char);
22491 vector bool short vec_revb (vector short char);
22492 vector bool int vec_revb (vector bool int);
22493 vector bool long long vec_revb (vector bool long long);
22494 * doc/extend.texi: Update the built-in documentation file for the
22495 new built-in functions.
22497 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22499 * config/s390/s390.md: Remove movcc splitter.
22501 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22503 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22504 load/store on condition.
22506 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22509 * config/avr/avr.c (avr_encode_section_info)
22510 [progmem && !TREE_READONLY]: Error if progmem object needs
22513 2017-07-11 Michael Collison <michael.collison@arm.com>
22515 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22518 2017-07-11 Carl Love <cel@us.ibm.com>
22520 * config/rs6000/rs6000-c.c: Add support for builtins
22521 vector unsigned int vec_parity_lsbb (vector signed int);
22522 vector unsigned int vec_parity_lsbb (vector unsigned int);
22523 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22524 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22525 vector unsigned long long vec_parity_lsbb (vector signed long long);
22526 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22527 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22528 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22529 * doc/extend.texi: Update the built-in documentation file for the
22530 new built-in functions.
22532 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22534 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22535 (layout::m_primary_loc): New field.
22536 (layout::layout): Initialize new field. Move location filtering
22537 logic from here to...
22538 (layout::maybe_add_location_range): ...this new method. Add
22539 support for filtering to just the lines already specified by other
22541 (layout::will_show_line_p): New method.
22542 (gcc_rich_location::add_location_if_nearby): New method.
22543 (selftest::test_add_location_if_nearby): New test function.
22544 (selftest::diagnostic_show_locus_c_tests): Call it.
22545 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22548 2017-07-11 Tom de Vries <tom@codesourcery.com>
22550 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22551 (bb_first_real_insn): New function.
22552 (nvptx_single): Add extra initialization of broadcasted condition
22555 2017-07-11 Nathan Sidwell <nathan@acm.org>
22557 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22559 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22561 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22562 Binutils doc as TEXI will mess them up.
22563 * doc/invoke.texi (AVR Options): Same here.
22565 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22567 * config/sparc/sparc.opt (mfix-ut700): New option.
22568 (mfix-gr712rc): Likewise.
22569 (sparc_fix_b2bst): New variable.
22570 * doc/invoke.texi (SPARC options): Document them.
22571 (ARM options): Fix warnings.
22572 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22573 instructions to prevent sequences that can trigger the store-store
22574 errata for certain LEON3FT processors.
22575 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22576 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22577 * config/sparc/sparc.md (fix_b2bst): New attribute.
22578 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22580 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22583 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22585 (*rsqrtsf2_sse): Ditto.
22587 (div<mode>3): Macroize insn from divdf3 and divsf3
22588 using MODEF mode iterator.
22590 2017-07-10 Martin Sebor <msebor@redhat.com>
22592 PR tree-optimization/80397
22593 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22594 instead of testing for equality to INTEGER_TYPE.
22596 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22598 * config.gcc: Remove uclibc from arc target spec.
22600 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22602 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22604 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22607 * lto-wrapper.c (remove_option): New function.
22608 (merge_and_complain): Merge PIC/PIE options more realistically.
22610 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22612 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22616 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22617 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22618 * config.in: Regenerate.
22619 * configure: Regenerate.
22620 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22621 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22622 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22623 (TARGET_GASISR_PROLOGUES): ...target mask.
22624 * common/config/avr/avr-common.c
22625 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22626 Set -mgas-isr-prologues.
22627 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22628 INSERT_PASS_BEFORE for it.
22629 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22630 * config/avr/avr.c (avr_option_override)
22631 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22632 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22633 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22634 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22635 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22637 (make_avr_pass_pre_proep): New function.
22638 (emit_push_sfr) <treg>: Add argument to function and use it
22639 instead of TMP_REG.
22640 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22641 and set machine->gasisr.yes.
22642 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22643 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22644 __gcc_isr.n_pushed to .L__stack_usage.
22645 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22646 (avr_asm_final_postscan_insn): ...this new static function.
22647 * config/avr/avr.h (machine_function)
22648 <is_no_gccisr, use_L__stack_usage>: New fields.
22649 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22650 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22651 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22652 (gasisr, *gasisr): New expander and insn.
22653 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22654 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22655 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22657 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22659 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22662 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22664 Move jump-tables out of .text again.
22667 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22668 (ASM_OUTPUT_ADDR_VEC): New function.
22669 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22670 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22671 INSN_ADDRESSes as asm comment.
22672 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22673 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22674 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22675 * config/avr/avr.md (*tablejump): Adjust comment.
22676 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22677 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22679 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22680 (avr_output_addr_vec): New proto.
22681 (avr_log_t) <insn_addresses>: New field.
22683 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22686 * config/i386/i386.c (ix86_function_arg_advance): Set
22687 outgoing_args_on_stack to true if there are outgoing arguments
22689 (ix86_function_arg): Likewise.
22690 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22691 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22692 * config/i386/i386.h (machine_function): Add
22693 outgoing_args_on_stack.
22695 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
22697 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22698 supporting pthreds.
22699 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
22701 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
22703 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
22704 (REAL_H): Remove $(MACHMODE_H).
22705 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
22707 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
22708 $(MACHMODE_H) and double-int.h.
22709 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
22711 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
22714 2017-07-07 Andrew Pinski <apinski@cavium.com>
22716 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
22717 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22719 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22721 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22722 Add warning if GCC was not configured to link against a GLIBC that
22723 exports the hardware capability bits.
22724 (make_resolver_func): Make resolver function private and not a
22725 COMDAT function. Create the name with clone_function_name instead
22726 of make_unique_name.
22729 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22730 correct operand in doing the split.
22732 2017-07-07 Carl Love <cel@us.ibm.com>
22734 * config/rs6000/rs6000-c: Add support for built-in function
22735 vector unsigned short vec_pack_to_short_fp32 (vector float,
22737 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22738 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22739 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22740 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22741 (convert_4f32_8i16): Add define_expand.
22742 * doc/extend.texi: Update the built-in documentation file for the
22743 new built-in function.
22745 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22747 * config/sparc/m8.md: New file.
22748 * config/sparc/sparc.md: Include m8.md.
22750 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22752 * config/sparc/sparc.opt: New option -mvis4b.
22753 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22754 (sparc_option_override): Handle VIS4B.
22755 (enum sparc_builtins): Define
22756 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22757 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22758 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22759 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22760 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22761 (check_constant_argument): New function.
22762 (sparc_vis_init_builtins): Define builtins
22763 __builtin_vis_dictunpack{8,16,32},
22764 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22765 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22766 __builtin_vis_fpcmpde{8,16,32}shl and
22767 __builtin_vis_fpcmpur{8,16,32}shl.
22768 (sparc_expand_builtin): Check that the constant operands to
22769 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22770 constant and in range.
22771 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22773 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22774 (SPARC_IMM5_P): Likewise.
22775 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22776 (enabled): Handle vis4b.
22777 (UNSPEC_DICTUNPACK): New unspec.
22778 (UNSPEC_FPCMPSHL): Likewise.
22779 (UNSPEC_FPUCMPSHL): Likewise.
22780 (UNSPEC_FPCMPDESHL): Likewise.
22781 (UNSPEC_FPCMPURSHL): Likewise.
22782 (cpu_feature): New CPU feature `vis4b'.
22783 (dictunpack{8,16,32}): New insns.
22784 (FPCSMODE): New mode iterator.
22785 (fpcscond): New code iterator.
22786 (fpcsucond): Likewise.
22787 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22788 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22789 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22790 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22791 * config/sparc/constraints.md: Define constraints `q' for unsigned
22792 2-bit integer constants and `t' for unsigned 5-bit integer
22794 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22796 (imm5_operand_dictunpack16): Likewise.
22797 (imm5_operand_dictunpack32): Likewise.
22798 (imm2_operand): Likewise.
22799 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22800 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22801 ditunpack* and fpcmp*shl builtins.
22803 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22805 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22806 * config.in: Add HAVE_AS_SPARC6 define.
22807 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22809 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22811 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22812 (CPP_CPU_SPEC): Handle m8.
22813 (ASM_CPU_SPEC): Likewise.
22814 * config/sparc/sparc-opts.h (enum processor_type): Add
22816 * config/sparc/sparc.c (m8_costs): New struct.
22817 (sparc_option_override): Handle TARGET_CPU_m8.
22818 (sparc32_initialize_trampoline): Likewise.
22819 (sparc64_initialize_trampoline): Likewise.
22820 (sparc_issue_rate): Likewise.
22821 (sparc_register_move_cost): Likewise.
22822 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22823 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22824 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22825 (CPP_CPU_SPEC): Handle M8.
22826 (ASM_CPU_SPEC): Likewise.
22827 (AS_M8_FLAG): Define.
22828 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22829 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22830 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22832 * configure: Regenerate.
22833 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22836 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22838 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22840 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22841 ("*movdi_insn_sp32"): Do not set v3pipe.
22842 ("*movsi_insn"): Likewise.
22843 ("*movdi_insn_sp64"): Likewise.
22844 ("*movsf_insn"): Likewise.
22845 ("*movdf_insn_sp32"): Likewise.
22846 ("*movdf_insn_sp64"): Likewise.
22847 ("*zero_extendsidi2_insn_sp64"): Likewise.
22848 ("*sign_extendsidi2_insn"): Likewise.
22849 ("*mov<VM32:mode>_insn"): Likewise.
22850 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22851 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22852 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22853 ("<vlop:code><VL:mode>3"): Likewise.
22854 ("*not_<vlop:code><VL:mode>3"): Likewise.
22855 ("*nand<VL:mode>_vis"): Likewise.
22856 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22857 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22858 ("one_cmpl<VL:mode>2"): Likewise.
22859 ("faligndata<VM64:mode>_vis"): Likewise.
22860 ("alignaddrsi_vis"): Likewise.
22861 ("alignaddrdi_vis"): Likweise.
22862 ("alignaddrlsi_vis"): Likewise.
22863 ("alignaddrldi_vis"): Likewise.
22864 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22865 ("bmaskdi_vis"): Likewise.
22866 ("bmasksi_vis"): Likewise.
22867 ("bshuffle<VM64:mode>_vis"): Likewise.
22868 ("cmask8<P:mode>_vis"): Likewise.
22869 ("cmask16<P:mode>_vis"): Likewise.
22870 ("cmask32<P:mode>_vis"): Likewise.
22871 ("pdistn<P:mode>_vis"): Likewise.
22872 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22874 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22876 * config/sparc/sparc.md ("subtype"): New insn attribute.
22877 ("*wrgsr_sp64"): Set insn subtype.
22878 ("*rdgsr_sp64"): Likewise.
22879 ("alignaddrsi_vis"): Likewise.
22880 ("alignaddrdi_vis"): Likewise.
22881 ("alignaddrlsi_vis"): Likewise.
22882 ("alignaddrldi_vis"): Likewise.
22883 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22884 ("fexpand_vis"): Likewise.
22885 ("fpmerge_vis"): Likewise.
22886 ("faligndata<VM64:mode>_vis"): Likewise.
22887 ("bshuffle<VM64:mode>_vis"): Likewise.
22888 ("cmask8<P:mode>_vis"): Likewise.
22889 ("cmask16<P:mode>_vis"): Likewise.
22890 ("cmask32<P:mode>_vis"): Likewise.
22891 ("fchksm16_vis"): Likewise.
22892 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22893 ("fmean16_vis"): Likewise.
22894 ("fp<plusminus_insn>64_vis"): Likewise.
22895 ("<plusminus_insn>v8qi3"): Likewise.
22896 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22897 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22898 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22899 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22900 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22901 ("*movqi_insn"): Likewise.
22902 ("*movhi_insn"): Likewise.
22903 ("*movsi_insn"): Likewise.
22904 ("movsi_pic_gotdata_op"): Likewise.
22905 ("*movdi_insn_sp32"): Likewise.
22906 ("*movdi_insn_sp64"): Likewise.
22907 ("movdi_pic_gotdata_op"): Likewise.
22908 ("*movsf_insn"): Likewise.
22909 ("*movdf_insn_sp32"): Likewise.
22910 ("*movdf_insn_sp64"): Likewise.
22911 ("*zero_extendhisi2_insn"): Likewise.
22912 ("*zero_extendqihi2_insn"): Likewise.
22913 ("*zero_extendqisi2_insn"): Likewise.
22914 ("*zero_extendqidi2_insn"): Likewise.
22915 ("*zero_extendhidi2_insn"): Likewise.
22916 ("*zero_extendsidi2_insn_sp64"): Likewise.
22917 ("ldfsr"): Likewise.
22918 ("prefetch_64"): Likewise.
22919 ("prefetch_32"): Likewise.
22920 ("tie_ld32"): Likewise.
22921 ("tie_ld64"): Likewise.
22922 ("*tldo_ldub_sp32"): Likewise.
22923 ("*tldo_ldub1_sp32"): Likewise.
22924 ("*tldo_ldub2_sp32"): Likewise.
22925 ("*tldo_ldub_sp64"): Likewise.
22926 ("*tldo_ldub1_sp64"): Likewise.
22927 ("*tldo_ldub2_sp64"): Likewise.
22928 ("*tldo_ldub3_sp64"): Likewise.
22929 ("*tldo_lduh_sp32"): Likewise.
22930 ("*tldo_lduh1_sp32"): Likewise.
22931 ("*tldo_lduh_sp64"): Likewise.
22932 ("*tldo_lduh1_sp64"): Likewise.
22933 ("*tldo_lduh2_sp64"): Likewise.
22934 ("*tldo_lduw_sp32"): Likewise.
22935 ("*tldo_lduw_sp64"): Likewise.
22936 ("*tldo_lduw1_sp64"): Likewise.
22937 ("*tldo_ldx_sp64"): Likewise.
22938 ("*mov<VM32:mode>_insn"): Likewise.
22939 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22940 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22942 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22944 * config/sparc/sparc.md ("type"): New insn type viscmp.
22945 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22947 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22948 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22949 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22950 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22952 ("n7_vis_logical_11cycle"): Likewise.
22953 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22954 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22955 * config/sparc/niagara.md ("niag_vis"): Likewise.
22956 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22957 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22959 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22961 * config/sparc/sparc.md: New instruction type `bmask'.
22962 (bmaskdi_vis): Use the `bmask' type.
22963 (bmasksi_vis): Likewise.
22964 * config/sparc/ultra3.md (us3_array): Likewise.
22965 * config/sparc/niagara7.md (n7_array): Likewise.
22966 * config/sparc/niagara4.md (n4_array): Likewise.
22967 * config/sparc/niagara2.md (niag2_vis): Likewise.
22968 (niag3_vis): Likewise.
22969 * config/sparc/niagara.md (niag_vis): Likewise.
22971 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22973 * ipa-comdats.c: Remove optimize check from gate.
22974 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22975 for functions not optimized.
22976 (ipa_fn_summary_read): Skip optimize check.
22977 (ipa_fn_summary_write): Likewise.
22978 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22980 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22982 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22983 (check_callers): Give up on caller which is not optimized.
22984 (inline_small_functions): Likewise.
22985 (ipa_inline): Do not give up when not optimizing.
22986 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22987 away unoptimizes cdtors.
22988 (whole_program_function_and_variable_visibility): Do
22989 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22990 * ipa.c (process_references): Do not check optimize.
22991 (symbol_table::remove_unreachable_nodes): Update optimize check.
22992 (set_writeonly_bit): Update optimize check.
22993 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22994 (pass_ipa_single_use::gate): Remove.
22996 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22998 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22999 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23000 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23001 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23002 permute_load, permute_store, adjust_extract, adjust_splat,
23003 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23004 replace_swap_with_copy, dump_swap_insn_table,
23005 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23006 recombine_lvx_pattern, recombine_stvx_pattern,
23007 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23008 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23009 to file rs6000-p8swap.c.
23010 * config/rs6000/rs6000-p8swap.c: New file.
23011 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23012 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23013 and rs6000*-*-* targets.
23015 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23017 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23019 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23021 * lto-wrapper.c (merge_and_complain): Do not merge
23022 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23023 fsigned_zeros, ftrapping_math, fwrapv.
23024 (append_compiler_options): Do not track these options.
23025 (append_linker_options): Likewie
23027 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23029 * cgraphunit.c (cgraph_node::finalize_function): When
23030 !flag_toplevel_reorde set no_reorder flag.
23031 (varpool_node::finalize_decl): Likewise.
23032 (symbol_table::compile): Drop no toplevel reorder path.
23034 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23036 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23037 edges; zero probability is not better than uninitialized.
23039 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23041 * asan.h (asan_sanitize_allocas_p): Declare.
23042 * asan.c (asan_sanitize_allocas_p): New function.
23043 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23044 (handle_builtin_alloca): Likewise.
23045 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23046 if !asan_sanitize_allocas_p.
23047 * params.def (asan-instrument-allocas): Add new option.
23048 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23049 * opts.c (common_handle_option): Disable allocas sanitization for
23052 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23054 * asan.c: Include gimple-fold.h.
23055 (get_last_alloca_addr): New function.
23056 (handle_builtin_stackrestore): Likewise.
23057 (handle_builtin_alloca): Likewise.
23058 (asan_emit_allocas_unpoison): Likewise.
23059 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23060 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23061 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23062 (instrument_builtin_call): Pass gimple iterator to
23063 get_mem_refs_of_builtin_call.
23064 (last_alloca_addr): New global.
23065 * asan.h (asan_emit_allocas_unpoison): Declare.
23066 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23067 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23068 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23069 if function calls alloca.
23070 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23071 * gimple-fold.h (replace_call_with_value): Declare.
23072 * internal-fn.c: Include asan.h.
23073 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23074 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23076 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23078 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23079 (C_SELFTEST_FLAGS): New.
23080 (CPP_SELFTEST_FLAGS): New.
23081 (SELFTEST_DEPS): New, from deps of s-selftest.
23082 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23083 (CPP_SELFTEST_DEPS): New.
23084 (selftest): Add dependency on s-selftest-c++.
23085 (s-selftest): Rename to...
23086 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23087 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23088 than SELFTEST_FLAGS.
23089 (selftest-gdb): Rename to...
23090 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23092 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23093 (selftest-valgrind): Rename to...
23094 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23096 (selftest-valgrind): Reintroduce as an alias for
23097 selftest-c-valgrind.
23098 (s-selftest-c++): New.
23099 (selftest-c++-gdb): New.
23100 (selftest-c++-valgrind): New.
23102 2017-07-06 Olivier Hainque <hainque@adacore.com>
23104 * gcc.c (process_command): When deciding if undefined variables
23105 should be ignored when processing specs, accept "gcc -v" as well.
23107 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23109 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23110 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23112 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23114 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23115 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23116 * config/arm/arm-tables.opt: Regenerate.
23117 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23119 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23121 2017-07-06 Carl Love <cel@us.ibm.com>
23123 * ChangeLog: Clean up from mid air collision
23125 2017-07-06 Carl Love <cel@us.ibm.com>
23127 * config/rs6000/rs6000-c.c: Add support for built-in functions
23128 vector signed int vec_subc (vector signed int, vector signed int);
23129 vector signed __int128 vec_subc (vector signed __int128,
23130 vector signed __int128);
23131 vector unsigned __int128 vec_subc (vector unsigned __int128,
23132 vector unsigned __int128);
23133 vector signed int vec_sube (vector signed int, vector signed int,
23134 vector signed int);
23135 vector unsigned int vec_sube (vector unsigned int,
23136 vector unsigned int,
23137 vector unsigned int);
23138 vector signed __int128 vec_sube (vector signed __int128,
23139 vector signed __int128,
23140 vector signed__int128);
23141 vector unsigned __int128 vec_sube (vector unsigned __int128,
23142 vector unsigned __int128,
23143 vector unsigned __int128);
23144 vector signed int vec_subec (vector signed int, vector signed int,
23145 vector signed int);
23146 vector unsigned int vec_subec (vector unsigned int,
23147 vector unsigned int,
23148 vector unsigned int);
23149 vector signed __int128 vec_subec (vector signed __int128,
23150 vector signed __int128,
23151 vector signed__int128);
23152 vector unsigned __int128 vec_subec (vector unsigned __int128,
23153 vector unsigned __int128,
23154 vector unsigned __int128);
23155 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23156 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23157 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23158 BU_ALTIVEC_OVERLOAD_X definitions.
23159 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23160 * doc/extend.texi: Update the built-in documentation file for the new
23161 built-in functions.
23163 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23166 * diagnostic-show-locus.c (layout::layout): Use start and finish
23167 spelling location for the start and finish of each range.
23168 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23169 Add unused aspect param.
23170 * input.c (expand_location_1): Add "aspect" param, and use it
23171 to access the correct part of the location.
23172 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23174 (expand_location_to_spelling_point): Likewise.
23175 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23176 param, and pass it to expand_location_1.
23178 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23180 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23181 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23182 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23183 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23184 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23185 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23186 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23187 _mm_maskz_getmant_ss): New intrinsics.
23188 (__builtin_ia32_getexpss128_mask): Changed to ...
23189 __builtin_ia32_getexpss128_round ... this.
23190 (__builtin_ia32_getexpsd128_mask): Changed to ...
23191 __builtin_ia32_getexpsd128_round ... this.
23192 * config/i386/i386-builtin-types.def
23193 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23194 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23195 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23196 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23197 __builtin_ia32_getmantss_mask_round): New builtins.
23198 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23199 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23200 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23201 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23202 * config/i386/sse.md
23203 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23204 avx512f_sgetexp<mode><mask_scalar_name>
23205 <round_saeonly_scalar_name> ... this.
23206 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23207 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23208 vgetexp<ssescalarmodesuffix>
23209 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23210 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23211 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23212 avx512f_vgetmant<mode><mask_scalar_name>
23213 <round_saeonly_scalar_name> ... this.
23214 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23215 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23216 vgetmant<ssescalarmodesuffix>
23217 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23218 %0<mask_scalar_operand4>, %1, %2
23219 <round_saeonly_scalar_mask_op4>, %3} ... this.
23220 * config/i386/subst.md (mask_scalar_operand4,
23221 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23222 round_saeonly_scalar_nimm_predicate): New subst attributes.
23224 2017-07-06 Julia Koval <julia.koval@intel.com>
23226 * config/i386/i386.c (ix86_erase_embedded_rounding):
23227 Remove code for old rounding pattern.
23229 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23231 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23233 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23235 * doc/sourcebuild.texi (Test Directives, Variants of
23236 dg-require-support): Add documentation for dg-require-stack-check.
23238 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23240 * config/i386/subst.md (mask_scalar, round_scalar,
23241 round_saeonly_scalar): New meta-templates.
23242 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23243 round_scalar_mask_operand3, round_scalar_mask_op3,
23244 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23245 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23246 round_saeonly_scalar_constraint,
23247 round_saeonly_scalar_prefix): New subst attribute.
23248 * config/i386/sse.md
23249 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23250 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23251 <round_scalar_name> ... this.
23252 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23253 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23254 <round_scalar_name> ... this.
23255 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23256 <sse>_vm<code><mode>3<mask_scalar_name>
23257 <round_saeonly_scalar_name> ... this.
23258 (v<plusminus_mnemonic><ssescalarmodesuffix>
23259 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23260 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23261 v<plusminus_mnemonic><ssescalarmodesuffix>
23262 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23263 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23264 (v<multdiv_mnemonic><ssescalarmodesuffix>
23265 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23266 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23267 v<multdiv_mnemonic><ssescalarmodesuffix>
23268 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23269 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23270 (v<maxmin_float><ssescalarmodesuffix>
23271 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23272 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23273 v<maxmin_float><ssescalarmodesuffix>
23274 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23275 %0<mask_scalar_operand3>, %1, %<iptr>2
23276 <round_saeonly_scalar_mask_op3>} ... this.
23278 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23280 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23281 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23283 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23284 Alan Hayward <alan.hayward@arm.com>
23285 David Sherwood <david.sherwood@arm.com>
23287 * combine.c (simplify_if_then_else): Remove "enum" before
23289 * compare-elim.c (can_eliminate_compare): Likewise.
23290 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23292 (aarch64_lookup_simd_builtin_type): Likewise.
23293 (aarch64_simd_builtin_type): Likewise.
23294 (aarch64_init_simd_builtin_types): Likewise.
23295 (aarch64_simd_expand_args): Likewise.
23296 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23298 (aarch64_reverse_mask): Likewise.
23299 (aarch64_simd_emit_reg_reg_move): Likewise.
23300 (aarch64_gen_adjusted_ldpstp): Likewise.
23301 (aarch64_ccmp_mode_to_code): Likewise.
23302 (aarch64_operands_ok_for_ldpstp): Likewise.
23303 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23304 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23306 (aarch64_min_divisions_for_recip_mul): Likewise.
23307 (aarch64_reassociation_width): Likewise.
23308 (aarch64_get_condition_code_1): Likewise.
23309 (aarch64_simd_emit_reg_reg_move): Likewise.
23310 (aarch64_simd_attr_length_rglist): Likewise.
23311 (aarch64_reverse_mask): Likewise.
23312 (aarch64_operands_ok_for_ldpstp): Likewise.
23313 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23314 (aarch64_gen_adjusted_ldpstp): Likewise.
23315 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23317 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23318 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23319 (arm_lookup_simd_builtin_type): Likewise.
23320 (arm_simd_builtin_type): Likewise.
23321 (arm_init_simd_builtin_types): Likewise.
23322 (arm_expand_builtin_args): Likewise.
23323 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23324 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23325 (ft32_setup_incoming_varargs): Likewise.
23326 (ft32_function_arg): Likewise.
23327 (ft32_function_arg_advance): Likewise.
23328 (ft32_pass_by_reference): Likewise.
23329 (ft32_arg_partial_bytes): Likewise.
23330 (ft32_valid_pointer_mode): Likewise.
23331 (ft32_addr_space_pointer_mode): Likewise.
23332 (ft32_addr_space_legitimate_address_p): Likewise.
23333 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23335 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23336 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23337 (iamcu_alignment): Likewise.
23338 (canonicalize_vector_int_perm): Likewise.
23339 (ix86_noce_conversion_profitable_p): Likewise.
23340 (ix86_mpx_bound_mode): Likewise.
23341 (ix86_operands_ok_for_move_multiple): Likewise.
23342 * config/microblaze/microblaze-protos.h
23343 (microblaze_expand_conditional_branch_reg): Likewise.
23344 * config/microblaze/microblaze.c
23345 (microblaze_expand_conditional_branch_reg): Likewise.
23346 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23348 (rs6000_reassociation_width): Likewise.
23349 (rs6000_invalid_binary_op): Likewise.
23350 (fusion_p9_p): Likewise.
23351 (emit_fusion_p9_load): Likewise.
23352 (emit_fusion_p9_store): Likewise.
23353 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23355 (riscv_hard_regno_mode_ok_p): Likewise.
23356 (riscv_address_insns): Likewise.
23357 (riscv_split_symbol): Likewise.
23358 (riscv_legitimize_move): Likewise.
23359 (riscv_function_value): Likewise.
23360 (riscv_hard_regno_nregs): Likewise.
23361 (riscv_expand_builtin): Likewise.
23362 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23363 (riscv_build_integer): Likewise.
23364 (riscv_split_integer): Likewise.
23365 (riscv_legitimate_constant_p): Likewise.
23366 (riscv_cannot_force_const_mem): Likewise.
23367 (riscv_regno_mode_ok_for_base_p): Likewise.
23368 (riscv_valid_base_register_p): Likewise.
23369 (riscv_valid_offset_p): Likewise.
23370 (riscv_valid_lo_sum_p): Likewise.
23371 (riscv_classify_address): Likewise.
23372 (riscv_legitimate_address_p): Likewise.
23373 (riscv_address_insns): Likewise.
23374 (riscv_load_store_insns): Likewise.
23375 (riscv_force_binary): Likewise.
23376 (riscv_split_symbol): Likewise.
23377 (riscv_force_address): Likewise.
23378 (riscv_legitimize_address): Likewise.
23379 (riscv_move_integer): Likewise.
23380 (riscv_legitimize_const_move): Likewise.
23381 (riscv_legitimize_move): Likewise.
23382 (riscv_address_cost): Likewise.
23383 (riscv_subword): Likewise.
23384 (riscv_output_move): Likewise.
23385 (riscv_canonicalize_int_order_test): Likewise.
23386 (riscv_emit_int_order_test): Likewise.
23387 (riscv_function_arg_boundary): Likewise.
23388 (riscv_pass_mode_in_fpr_p): Likewise.
23389 (riscv_pass_fpr_single): Likewise.
23390 (riscv_pass_fpr_pair): Likewise.
23391 (riscv_get_arg_info): Likewise.
23392 (riscv_function_arg): Likewise.
23393 (riscv_function_arg_advance): Likewise.
23394 (riscv_arg_partial_bytes): Likewise.
23395 (riscv_function_value): Likewise.
23396 (riscv_pass_by_reference): Likewise.
23397 (riscv_setup_incoming_varargs): Likewise.
23398 (riscv_print_operand): Likewise.
23399 (riscv_elf_select_rtx_section): Likewise.
23400 (riscv_save_restore_reg): Likewise.
23401 (riscv_for_each_saved_reg): Likewise.
23402 (riscv_register_move_cost): Likewise.
23403 (riscv_hard_regno_mode_ok_p): Likewise.
23404 (riscv_hard_regno_nregs): Likewise.
23405 (riscv_class_max_nregs): Likewise.
23406 (riscv_memory_move_cost): Likewise.
23407 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23408 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23409 (rl78_addr_space_address_mode): Likewise.
23410 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23412 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23413 (rs6000_reassociation_width): Likewise.
23414 (rs6000_invalid_binary_op): Likewise.
23415 (fusion_p9_p): Likewise.
23416 (emit_fusion_p9_load): Likewise.
23417 (emit_fusion_p9_store): Likewise.
23418 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23419 (ok_for_simple_move_operands): Likewise.
23420 (ok_for_simple_move_strict_operands): Likewise.
23421 (ok_for_simple_arith_logic_operands): Likewise.
23422 (visium_legitimize_reload_address): Likewise.
23423 (visium_select_cc_mode): Likewise.
23424 (output_cbranch): Likewise.
23425 (visium_split_double_move): Likewise.
23426 (visium_expand_copysign): Likewise.
23427 (visium_expand_int_cstore): Likewise.
23428 (visium_expand_fp_cstore): Likewise.
23429 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23430 (visium_function_arg): Likewise.
23431 (visium_function_arg_advance): Likewise.
23432 (visium_libcall_value): Likewise.
23433 (visium_setup_incoming_varargs): Likewise.
23434 (visium_legitimate_constant_p): Likewise.
23435 (visium_legitimate_address_p): Likewise.
23436 (visium_legitimize_address): Likewise.
23437 (visium_secondary_reload): Likewise.
23438 (visium_register_move_cost): Likewise.
23439 (visium_memory_move_cost): Likewise.
23440 (prepare_move_operands): Likewise.
23441 (ok_for_simple_move_operands): Likewise.
23442 (ok_for_simple_move_strict_operands): Likewise.
23443 (ok_for_simple_arith_logic_operands): Likewise.
23444 (visium_function_value_1): Likewise.
23445 (rtx_ok_for_offset_p): Likewise.
23446 (visium_legitimize_reload_address): Likewise.
23447 (visium_split_double_move): Likewise.
23448 (visium_expand_copysign): Likewise.
23449 (visium_expand_int_cstore): Likewise.
23450 (visium_expand_fp_cstore): Likewise.
23451 (visium_split_cstore): Likewise.
23452 (visium_select_cc_mode): Likewise.
23453 (visium_split_cbranch): Likewise.
23454 (output_cbranch): Likewise.
23455 (visium_print_operand_address): Likewise.
23456 * expmed.c (flip_storage_order): Likewise.
23457 * expmed.h (emit_cstore): Likewise.
23458 (flip_storage_order): Likewise.
23459 * genrecog.c (validate_pattern): Likewise.
23460 * hsa-gen.c (gen_hsa_addr): Likewise.
23461 * internal-fn.c (expand_arith_overflow): Likewise.
23462 * ira-color.c (allocno_copy_cost_saving): Likewise.
23463 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23464 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23465 (process_invariant_for_inheritance): Likewise.
23466 * lra-eliminations.c (move_plus_up): Likewise.
23467 * omp-low.c (lower_oacc_reductions): Likewise.
23468 * simplify-rtx.c (simplify_subreg): Likewise.
23469 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23470 (TARGET_CHKP_BOUND_MODE): Likewise..
23471 * targhooks.c (default_chkp_bound_mode): Likewise.
23472 (default_setup_incoming_vararg_bounds): Likewise.
23473 * targhooks.h (default_chkp_bound_mode): Likewise.
23474 (default_setup_incoming_vararg_bounds): Likewise.
23475 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23476 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23477 (have_whole_vector_shift): Likewise.
23478 * tree-vect-stmts.c (vectorizable_load): Likewise.
23479 * doc/tm.texi: Regenerate.
23481 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23483 Graceful degrade if Binutils PR21472 is not available.
23486 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23487 .rodata in flash test fails.
23488 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23489 * confgure: Regenerate.
23490 * config.in: Regenerate.
23491 * config/avr/avr.c (avr_asm_named_section)
23492 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23493 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23494 (avr_asm_init_sections): Same.
23496 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23498 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23499 (fma<VH:mode>4_intrinsic): Likewise.
23500 (*fmsub<VCVTF:mode>4): Likewise.
23501 (*fmsub<VH:mode>4_intrinsic): Likewise.
23503 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23506 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23507 Don't depend on "optimize > 0".
23508 (out_movhi_r_mr, out_movqi_mr_r): Same.
23509 (out_movhi_mr_r, out_movqi_r_mr): Same.
23510 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23511 io_address_operand on "optimize > 0".
23513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23515 * tree-loop-distribution.c: Add general explanantion on the pass.
23516 (generate_loops_for_partition): Mark distributed loop.
23517 (pg_add_dependence_edges): New parameter. Handle alias data
23518 dependence specially and record it in the parameter if asked.
23519 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23520 (init_partition_graph_vertices, add_partition_graph_edge): New.
23521 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23522 (free_partition_graph_vdata, build_partition_graph): New.
23523 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23524 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23525 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23526 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23527 (version_for_distribution_p, finalize_partitions): New.
23528 (distribute_loop): Handle alias data dependence specially. Factor
23529 out loop fusion code as functions and call these functions.
23531 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23533 * tree-loop-distribution.c (classify_partition): New parameter and
23534 better handle reduction statement.
23535 (rdg_build_partitions): Revise comment.
23536 (distribute_loop): Compute statements in all partitions and pass it
23537 to classify_partition.
23539 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23541 * tree-loop-distribution.c (enum partition_type): New.
23542 (struct partition): New field type.
23543 (partition_merge_into): Add parameter. Update partition type.
23544 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23545 (build_rdg_partition_for_vertex): Compute partition type.
23546 (rdg_build_partitions): Dump partition type.
23547 (distribute_loop): Update calls to partition_merge_into.
23549 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23551 * tree-loop-distribution.c (struct ddr_hasher): New.
23552 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23554 (classify_partition): Call get_data_dependence.
23555 (pg_add_dependence_edges): Ditto.
23556 (distribute_loop): Release data dependence hash table.
23558 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23560 * tree-loop-distribution.c (ref_base_address): Delete.
23561 (similar_memory_accesses): Rename ...
23562 (share_memory_accesses): ... to this. Check if partitions access
23563 the same memory reference.
23564 (distribute_loop): Call share_memory_accesses.
23566 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23568 * tree-loop-distribution.c (struct partition): New field recording
23569 its data reference.
23570 (partition_alloc, partition_free): Init and release data refs.
23571 (partition_merge_into): Merge data refs.
23572 (build_rdg_partition_for_vertex): Collect data refs for partition.
23573 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23575 (distribute_loop): Remve data refs from vertice data of partition
23578 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23580 * tree-loop-distribution.c (params.h): Include header file.
23581 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23582 (datarefs_vec): New global var.
23583 (create_rdg_vertices): Use datarefs_vec directly.
23584 (free_rdg): Don't free data references.
23585 (build_rdg): Update use. Don't free data references.
23586 (distribute_loop): Compute global variable for data references.
23587 Bail out if there are too many data references.
23589 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23591 * tree-loop-distribution.c (loop_nest): New global var.
23592 (build_rdg): Use loop directly, rather than loop nest.
23593 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23595 (distribute_loop): Compute global variable loop nest. Update use.
23597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23599 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23600 (partition_merge_into): New parameter. Dump reason for fusion.
23601 (distribute_loop): Update use of partition_merge_into.
23603 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23605 * tree-loop-distribution.c (bb_top_order_index): New.
23606 (bb_top_order_index_size, bb_top_order_cmp): New.
23607 (stmts_from_loop): Use topological order.
23608 (pass_loop_distribution::execute): Compute and release topological
23609 order for basic blocks.
23611 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23613 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23616 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23618 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23619 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23620 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23621 * internal-fn.def (LOOP_DIST_ALIAS): New.
23622 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23623 (fold_loop_internal_call): ... this.
23624 (vect_loop_dist_alias_call): New function.
23625 (set_uid_loop_bbs): Call fold_loop_internal_call.
23626 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23629 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23632 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23633 Require dead FLAGS_REG at the beginning of a peephole.
23635 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23638 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23639 arguments in the call to __builtin_ia32_sbb_u32.
23640 (_subborrow_u64): Swap _X and _Y arguments in the call to
23641 __builtin_ia32_sbb_u64.
23643 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23646 * tree-vrp.c (compare_assert_loc): Turn into a function template
23647 with stable template parameter. Only test if a->e is NULL,
23648 !a->e == !b->e has been verified already. Use e == NULL or
23649 e != NULL instead of e or ! e tests. If stable is true, don't use
23650 iterative_hash_expr, on the other side allow a or b or both NULL
23651 and sort the NULLs last.
23652 (process_assert_insertions): Sort using compare_assert_loc<false>
23653 instead of compare_assert_loc, later sort using
23654 compare_assert_loc<true> before calling process_assert_insertions_for
23655 in a loop. Use break instead of continue once seen NULL pointer.
23657 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23659 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23660 Cortex-R7 and Cortex-R8 processors.
23662 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23664 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23665 uninitialized while src is not.
23667 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23669 * common/config/arm/arm-common.c: Adjust include path for
23671 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23672 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23673 (arm-cpu-data.h): Likewise.
23674 (arm-cpu-cdata.h): Likewise.
23675 * config/arm/arm-cpu.h: Delete.
23676 * config/arm/arm-cpu-cdata.h: Delete.
23677 * config/arm/arm-cpu-data.h: Delete.
23679 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23681 * config/arm/arm-cpus.in (cortex-a55): New.
23682 (cortex-a75): Likewise.
23683 (cortex-a75.cortex-a55): Likewise.
23684 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23686 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23687 * config/arm/arm-cpu-cdata.h: Regenerate.
23688 * config/arm/arm-cpu-data.h: Regenerate.
23689 * config/arm/arm-cpu.h: Regenerate.
23690 * config/arm/arm-tables.opt: Regenerate.
23691 * config/arm/arm-tune.md: Regenerate.
23693 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23695 * haifa-sched.c (sched_create_recovery_edges): Update profile.
23697 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23699 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
23702 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
23704 PR tree-optimization/81292
23705 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
23706 full_string_p, also call adjust_related_strinfos if the adjustment
23707 is simple, otherwise invalidate related strinfos.
23709 2017-07-04 Martin Liska <mliska@suse.cz>
23712 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
23713 newly created variable as DECL_IGNORED_P.
23715 2017-07-04 Martin Liska <mliska@suse.cz>
23718 * ipa-inline.c (inline_small_functions):
23719 Use xstrdup_for_dump.
23721 2017-07-04 Tom de Vries <tom@codesourcery.com>
23723 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23725 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23728 * config/darwin.c (darwin_function_switched_text_sections):
23731 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23733 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23735 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23737 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23739 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23741 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23742 min_profitable_iters, and th as inclusive lower bounds.
23743 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23744 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23745 for min_profitable_iters and min_profitable_estimate.
23746 (vect_transform_loop): Treat th as an inclusive lower bound.
23747 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23749 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23752 * config/darwin.c (darwin_function_switched_text_sections):
23753 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23754 in two pieces, and suppress the use of buf.
23756 2017-07-03 Nathan Sidwell <nathan@acm.org>
23758 * hash-table.h (hash_table_mod1): Fix indentation.
23760 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23762 PR middle-end/81290
23763 * predict.c (force_edge_cold): Be more careful about propagation
23765 * profile-count.h (profile_probability::guessed,
23766 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23768 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23770 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23772 * doc/invoke.texi (rcpc architecture extension): Document it.
23774 2017-07-03 Richard Biener <rguenther@suse.de>
23776 PR tree-optimization/60510
23777 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23778 the scalar reduction PHI and use it.
23779 (vectorizable_reduction): Properly guard the single_defuse_cycle
23780 path for non-SLP reduction chains where we cannot use it.
23781 Rework reduc_def/index and vector type deduction. Rework
23782 vector operand gathering during reduction op code-gen.
23783 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23784 chains dissolve the chain and leave it to non-SLP reduction
23787 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23789 * tree-data-ref.h (dr_alignment): Declare.
23790 * tree-data-ref.c (dr_alignment): New function.
23791 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23792 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23794 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23796 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23798 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23799 and base_misalignment fields.
23800 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23801 * tree-data-ref.c: Include builtins.h.
23802 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23803 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23804 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23805 * tree-vect-data-refs.c: Include tree-cfg.h.
23806 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23807 fields instead of calculating an alignment here.
23808 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23809 innermost_loop_behavior fields.
23811 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23813 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23815 (DR_STEP_ALIGNMENT): New macro.
23816 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23817 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23818 (create_data_ref): Print it.
23819 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23820 to tell whether the step preserves vector (mis)alignment.
23821 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23822 Move the check for an integer step and generalise to all INTEGER_CST.
23823 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23824 Print the outer step alignment.
23826 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23828 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23829 with offset_alignment.
23830 (DR_ALIGNED_TO): Delete.
23831 (DR_OFFSET_ALIGNMENT): New macro.
23832 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23833 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23834 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23835 (create_data_ref): Likewise.
23836 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23837 (vect_analyze_data_refs): Likewise.
23838 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23839 creating dummy innermost behavior.
23841 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23843 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23844 with a "innermost_loop_behavior *" and refeence tree.
23845 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23846 (create_data_ref): Update call accordingly.
23847 * tree-predcom.c (find_looparound_phi): Likewise.
23849 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23851 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23852 fields with dr_wrt_vec_loop.
23853 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23854 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23855 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23856 (vect_dr_behavior): New function.
23857 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23858 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23859 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23860 track whether the step preserves the misalignment.
23861 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23862 Use vect_dr_behavior.
23863 (vect_setup_realignment): Update call accordingly.
23864 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23865 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23866 call to vect_create_addr_base_for_vector_ref.
23867 (vect_create_cond_for_align_checks): Likewise.
23868 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23869 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23870 (vect_recog_mask_conversion_pattern): Likewise.
23871 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23872 (new_stmt_vec_info): Remove redundant zeroing.
23874 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23876 * common/config/arm/arm-common.c (arm_be8_option): New function.
23877 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23878 (ISA_ARMv6): Add isa_bit_be8.
23879 * config/arm/arm.h (arm_be8_option): Add prototype.
23880 (BE8_SPEC_FUNCTION): New define.
23881 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23882 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23883 (mlittle-endian): Similarly.
23884 (mbe8, mbe32): New options.
23885 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23886 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23888 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23890 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23892 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23894 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23895 (cleanup_tree_cfg_bb): Use it.
23896 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23898 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23900 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23903 * loop-doloop.c (add_test): Update profile.
23905 2017-07-03 Martin Liska <mliska@suse.cz>
23908 * sanopt.c (rewrite_usage_of_param): New function.
23909 (sanitize_rewrite_addressable_params): Likewise.
23910 (pass_sanopt::execute): Call rewrite_usage_of_param.
23912 2017-07-03 Richard Biener <rguenther@suse.de>
23914 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23915 back to using VIEW_CONVERT_EXPR.
23917 2017-07-03 Martin Liska <mliska@suse.cz>
23920 * doc/extend.texi: Document when a resolver function is
23921 generated for target_clones.
23923 2017-07-03 Martin Liska <mliska@suse.cz>
23925 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23926 and shadow memory of auto variables which are subject of
23927 use-after-scope sanitization.
23928 (asan_expand_mark_ifn): Add do set only when is_poison.
23930 2016-07-03 Richard Biener <rguenther@suse.de>
23932 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23934 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23935 (vectorizable_reduction): Perform reduction PHI creation when
23936 visiting a reduction PHI and adjust and simplify code generation
23937 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23938 (vect_transform_loop): Visit reduction PHIs.
23939 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23940 defs into the SLP tree.
23941 (vect_build_slp_tree): Reduction defs terminate the recursion.
23942 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23944 (vect_get_vec_defs_for_stmt_copy): Export.
23945 (vect_get_vec_defs): Likewise.
23946 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23948 (vect_get_vec_defs_for_stmt_copy): Declare.
23949 (vect_get_vec_defs): Likewise.
23951 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23953 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23954 parameter with a "loop" parameter and use it instead of the
23955 loop containing DR_STMT. Don't check simple_iv when doing
23956 BB analysis. Describe the two analysis modes in the comment.
23958 2017-07-03 Tom de Vries <tom@codesourcery.com>
23960 PR tree-optimization/69468
23961 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23962 (find_same_succ_bb): Handle ignore_edge_flags.
23964 2017-07-03 Tom de Vries <tom@codesourcery.com>
23966 PR tree-optimization/81192
23967 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23969 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23971 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23973 2017-07-03 Tom de Vries <tom@codesourcery.com>
23975 PR tree-optimization/81192
23976 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23977 BB_SAME_SUCC (bb) == NULL.
23979 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23981 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23984 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23986 * dumpfile.c: Include profile-count.h
23987 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23989 (insert_cond_bb): Update profile.
23990 * tree-cfg.h (insert_cond_bb): Update prototype.
23991 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23992 * tree-dump.c: Do not include tree-cfg.
23994 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23996 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23998 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24000 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24003 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24005 * tree-complex.c (expand_complex_div_wide): update profile.
24007 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24008 Alan Hayward <alan.hayward@arm.com>
24009 David Sherwood <david.sherwood@arm.com>
24011 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24012 (CORETYPES_H): New define.
24013 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24014 (insn-modes-inline.h, s-modes-inline-h): New rules.
24015 (generated_files): Add insn-modes-inline.h.
24016 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24017 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24018 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24019 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24020 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24021 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24022 (build/gencodes.o, build/genconditions.o): Likewise.
24023 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24024 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24025 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24026 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24027 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24028 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24029 * coretypes.h: Include everything up to real.h for generators.
24030 Include insn-modes.h first. Include wide-int-print.h after
24031 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24032 * machmode.h: Don't include insn-modes.h here.
24033 * function-tests.c: Remove includes of signop.h, machmode.h,
24034 double-int.h and wide-int.h.
24036 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24038 * optc-save-gen.awk: Likewise.
24039 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24040 * godump.c: Remove include of wide-int-print.h.
24041 * pretty-print.h: Likewise.
24042 * wide-int-print.cc: Likewise.
24043 * wide-int.cc: Likewise.
24044 * hash-map-tests.c: Remove include of signop.h.
24045 * hash-set-tests.c: Likewise.
24046 * rtl-tests.c: Likewise.
24047 * mkconfig.sh: Remove include of machmode.h.
24048 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24050 (emit_insn_modes_inline_h): ...this new function. Emit the code
24051 into an insn-modes-inline.h header file, adding appropriate
24052 include guards and end comments.
24053 (emit_insn_modes_c_header): Remove include of machmode.h.
24054 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24056 (main): Handle -i flag and call emit_insn_modes_inline_h when
24059 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24061 * tree-ssa-strlen.c (strinfo): Rename the length field to
24062 nonzero_chars. Add a full_string_p field.
24063 (compare_nonzero_chars, zero_length_string_p): New functions.
24064 (get_addr_stridx): Add an offset_out parameter.
24065 Use compare_nonzero_chars.
24066 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24067 (new_strinfo): Update after above changes to strinfo.
24068 (set_endptr_and_length): Set full_string_p.
24069 (get_string_length): Update after above changes to strinfo.
24070 (unshare_strinfo): Update call to new_strinfo.
24071 (maybe_invalidate): Likewise.
24072 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24073 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24074 as a uhwi instead of an shwi. Update after above changes to
24075 strinfo and new_strinfo.
24076 (zero_length_string): Assert that chainsi contains full strings.
24077 Use zero_length_string_p. Update call to new_strinfo.
24078 (adjust_related_strinfos): Update after above changes to strinfo.
24079 Copy full_string_p from origsi.
24080 (adjust_last_stmt): Use zero_length_string_p.
24081 (handle_builtin_strlen): Update after above changes to strinfo and
24082 new_strinfo. Install the lhs as the string length if the previous
24083 entry didn't describe a full string.
24084 (handle_builtin_strchr): Update after above changes to strinfo
24086 (handle_builtin_strcpy): Likewise.
24087 (handle_builtin_strcat): Likewise.
24088 (handle_builtin_malloc): Likewise.
24089 (handle_pointer_plus): Likewise.
24090 (handle_builtin_memcpy): Likewise. Track nonzero characters
24091 that aren't necessarily followed by a nul terminator.
24092 (handle_char_store): Likewise.
24094 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24096 PR tree-optimization/80769
24097 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24098 for malloc and calloc. Document the new invariant that all related
24099 strinfos have delayed lengths or none do.
24100 (verify_related_strinfos): Move earlier in file.
24101 (set_endptr_and_length): New function, split out from...
24102 (get_string_length): ...here. Also set the lengths of related
24104 (zero_length_string): Assert that chainsi has known (rather than
24106 (adjust_related_strinfos): Likewise.
24108 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24110 PR tree-optimization/81136
24111 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24112 assert that two references with the same misalignment have the same
24113 compile-time misalignment if those compile-time misalignments
24116 2017-07-01 Andi Kleen <ak@linux.intel.com>
24118 * print-tree.c (print_node): Print all attributes.
24120 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24122 * cfg.c (scale_bbs_frequencies): New function.
24123 * cfg.h (scale_bbs_frequencies): Declare it.
24124 * cfgloopanal.c (single_likely_exit): Cleanup.
24125 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24127 (scale_loop_profile): Likewise.
24128 (loop_version): Likewise.
24129 (create_empty_loop_on_edge): Update.
24130 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24131 scale_loop_frequencies, scale_loop_profile, loopify,
24132 loop_version): Update prototypes.
24133 * modulo-sched.c (sms_schedule): Update.
24134 * predict.c (unlikely_executed_edge_p): Also check probability.
24135 (probably_never_executed_edge_p): Fix typo.
24136 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24137 * tree-parloops.c (gen_parallel_loop): Update.
24138 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24139 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24140 * tree-ssa-loop-split.c (split_loop): Update.
24141 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24142 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24143 (vect_loop_versioning): Update.
24144 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24146 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24148 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24150 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24152 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24153 to keep profile consistent.
24155 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24157 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24158 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24159 * profile-count.h (max_safe_multiplier): Make unsigned.
24160 (profile_count::guessed_zero): New.
24162 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24164 * bb-reorder.c (fix_up_crossing_landing_pad,
24165 fix_crossing_conditional_branches): Use make_single_succ_edge
24166 to keep profile consistent.
24168 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24170 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24173 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24176 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24177 the right scopes, make sure cond_jump isn't preserved between multiple
24178 iterations. Search for fallthru edge whenever there are 3+ edges and
24179 use find_fallthru_edge for it.
24181 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24183 Patch by Alexander Monakov <amonakov@ispras.ru>
24184 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24185 probabilities consistently.
24187 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24189 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24190 emit_cmp_and_jump_insns.
24192 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24195 * tree-inline.c (expand_call_inline): Combine profile statuses.
24197 2017-06-30 Andrew Pinski <apinski@cavium.com>
24199 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24200 fold_stmt returned true.
24202 2017-06-30 Nathan Sidwell <nathan@acm.org>
24204 * ggc.h (empty_string): Delete.
24205 * cfgexpand.c (expand_asm_stmt): Use plain "".
24206 * optabs.c (expand_asm_memory_barrier): Likewise.
24207 * stringpool.c (empty_string): Delete.
24208 (digit_vector, digit_string): Delete.
24209 (ggc_alloc_string): Use plain "", don't optimize single digit
24210 strings. Use ggc_alloc_atomic.
24212 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24214 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24215 comparison set and one other set, use the cost of the non-comparison
24218 2017-06-30 Nathan Sidwell <nathan@acm.org>
24220 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24223 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24225 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24226 loops. Remove now unneeded calls to gimple_switch_set_label() that
24227 just set removed labels to NULL_TREE.
24229 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24231 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24232 (set_range_info): ...here. Only call set_range_info_raw if domain
24234 (set_nonzero_bits): Call set_range_info_raw.
24235 * tree-ssanames.h (set_range_info_raw): New.
24237 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24240 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24241 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24242 of nonimmediate_operand and <store_mask_constraint> instead of m
24243 for the input operand. For V8FI iterator, always split if input
24244 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24245 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24246 <store_mask_predicate> instead of register_operand and
24247 <store_mask_constraint> instead of v for the input operand. Make
24248 sure both operands aren't MEMs for if not <mask_applied>.
24250 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24252 * lto-wrapper.c (copy_file) Close both file descriptors before
24255 2017-06-30 Martin Liska <mliska@suse.cz>
24258 * multiple_target.c (create_dispatcher_calls): Make ifunc
24259 also for function that don't have calls or are not referenced.
24261 2017-06-30 Richard Biener <rguenther@suse.de>
24263 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24264 analyze the first scalar stmt. Move vector type computation
24265 for the BB case here from ...
24266 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24267 live operation processing in the SLP case properly.
24269 2017-06-30 Richard Biener <rguenther@suse.de>
24271 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24273 2017-06-30 Martin Liska <mliska@suse.cz>
24276 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24277 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24279 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24281 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24284 2017-06-30 Richard Biener <rguenther@suse.de>
24286 PR tree-optimization/81249
24287 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24288 condition reduction result to original scalar type.
24290 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24292 * profile-count.h (enum profile_quality): Fix typos and whitespace
24295 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24297 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24298 type for branch probabilities.
24300 2017-06-29 Julian Brown <julian@codesourcery.com>
24301 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24303 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24304 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24305 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24306 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24308 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24310 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24311 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24312 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24313 CC usage from generic code to here.
24314 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24315 CC usage into the target macros.
24317 2017-06-29 Maya Rashish <coypu@sdf.org>
24319 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24322 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24324 * arm/arm-builtins.c: Include profile-count.h
24325 * except.c (sjlj_emit_function_enter): Use
24326 profile_probability::unlikely.
24328 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24330 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24331 and tocrel_offset be pointer args rather than implicitly using
24333 (legitimate_constant_pool_address_p, rs6000_emit_move,
24334 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24335 tocrel_offset and use in toc_relative_expr_p call.
24336 (print_operand, print_operand_address): Use static tocrel_base_oac
24337 and tocrel_offset_oac.
24338 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24341 2017-06-29 Maya Rashish <coypu@sdf.org>
24343 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24345 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24347 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24348 objects, take into account only the alignment of 'op0' and 'mode1' if
24351 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24353 * ccmp.c (ccmp_tree_comparison_p): New function.
24354 (ccmp_candidate_p): Update to use above function.
24355 (get_compare_parts): New function.
24356 (expand_ccmp_next): Update to use new functions.
24357 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24359 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24360 take mode as argument.
24361 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24362 * expr.c (expand_expr_real_1): Pass mode as argument.
24364 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24366 * combine.c (combine_instructions): Print insns to dump_file, together
24369 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24371 * asan.c (asan_emit_stack_protection): Update.
24372 (create_cond_insert_point): Update.
24373 * auto-profile.c (afdo_propagate_circuit): Update.
24374 * basic-block.h (struct edge_def): Turn probability to
24375 profile_probability.
24376 (EDGE_FREQUENCY): Update.
24377 * bb-reorder.c (find_traces_1_round): Update.
24378 (better_edge_p): Update.
24379 (sanitize_hot_paths): Update.
24380 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24381 (make_single_succ_edge): Update.
24382 (check_bb_profile): Update.
24383 (dump_edge_info): Update.
24384 (update_bb_profile_for_threading): Update.
24385 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24386 probabilitycount to 0.
24387 * cfgbuild.c (compute_outgoing_frequencies): Update.
24388 * cfgcleanup.c (try_forward_edges): Update.
24389 (outgoing_edges_match): Update.
24390 (try_crossjump_to_edge): Update.
24391 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24392 (expand_gimple_tailcall): Update.
24393 (construct_init_block): Use make_single_succ_edge.
24394 (construct_exit_block): Use make_single_succ_edge.
24395 * cfghooks.c (verify_flow_info): Update.
24396 (redirect_edge_succ_nodup): Update.
24397 (split_edge): Update.
24398 (account_profile_record): Update.
24399 * cfgloopanal.c (single_likely_exit): Update.
24400 * cfgloopmanip.c (scale_loop_profile): Update.
24401 (set_zero_probability): Remove.
24402 (duplicate_loop_to_header_edge): Update.
24403 * cfgloopmanip.h (loop_version): Update prototype.
24404 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24405 (force_nonfallthru_and_redirect): Update.
24406 (update_br_prob_note): Update.
24407 (rtl_verify_edges): Update.
24408 (purge_dead_edges): Update.
24409 (rtl_lv_add_condition_to_bb): Update.
24410 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24411 * cgraphunit.c (init_lowered_empty_function): Update.
24412 (cgraph_node::expand_thunk): Update.
24413 * cilk-common.c: Include profile-count.h
24414 * dojump.c (inv): Remove.
24415 (jumpifnot): Update.
24416 (jumpifnot_1): Update.
24417 (do_jump_1): Update.
24419 (do_jump_by_parts_greater_rtx): Update.
24420 (do_compare_rtx_and_jump): Update.
24421 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24422 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24423 * dwarf2cfi.c: Include profile-count.h
24424 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24425 (sjlj_emit_dispatch_table): Likewise.
24426 * explow.c: Include profile-count.h
24427 * expmed.c (emit_store_flag_force): Update.
24428 (do_cmp_and_jump): Update.
24429 * expr.c (compare_by_pieces_d::generate): Update.
24430 (compare_by_pieces_d::finish_mode): Update.
24431 (emit_block_move_via_loop): Update.
24432 (store_expr_with_bounds): Update.
24433 (store_constructor): Update.
24434 (expand_expr_real_2): Update.
24435 (expand_expr_real_1): Update.
24436 * expr.h (try_casesi, try_tablejump): Update prototypes.
24437 * gimple-pretty-print.c (dump_probability): Update.
24438 (dump_profile): New.
24439 (dump_gimple_label): Update.
24440 (dump_gimple_bb_header): Update.
24441 * graph.c (draw_cfg_node_succ_edges): Update.
24442 * hsa-gen.c (convert_switch_statements): Update.
24443 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24444 (find_if_case_1): Update.
24445 (find_if_case_2): Update.
24446 * internal-fn.c (expand_arith_overflow_result_store): Update.
24447 (expand_addsub_overflow): Update.
24448 (expand_neg_overflow): Update.
24449 (expand_mul_overflow): Update.
24450 (expand_vector_ubsan_overflow): Update.
24451 * ipa-cp.c (good_cloning_opportunity_p): Update.
24452 * ipa-split.c (split_function): Use make_single_succ_edge.
24453 * ipa-utils.c (ipa_merge_profiles): Update.
24454 * loop-doloop.c (add_test): Update.
24455 (doloop_modify): Update.
24456 * loop-unroll.c (compare_and_jump_seq): Update.
24457 (unroll_loop_runtime_iterations): Update.
24458 * lra-constraints.c (lra_inheritance): Update.
24459 * lto-streamer-in.c (input_cfg): Update.
24460 * lto-streamer-out.c (output_cfg): Update.
24461 * mcf.c (adjust_cfg_counts): Update.
24462 * modulo-sched.c (sms_schedule): Update.
24463 * omp-expand.c (expand_omp_for_init_counts): Update.
24464 (extract_omp_for_update_vars): Update.
24465 (expand_omp_ordered_sink): Update.
24466 (expand_omp_for_ordered_loops): Update.
24467 (expand_omp_for_generic): Update.
24468 (expand_omp_for_static_nochunk): Update.
24469 (expand_omp_for_static_chunk): Update.
24470 (expand_cilk_for): Update.
24471 (expand_omp_simd): Update.
24472 (expand_omp_taskloop_for_outer): Update.
24473 (expand_omp_taskloop_for_inner): Update.
24474 * omp-simd-clone.c (simd_clone_adjust): Update.
24475 * optabs.c (expand_doubleword_shift): Update.
24476 (expand_abs): Update.
24477 (emit_cmp_and_jump_insn_1): Update.
24478 (expand_compare_and_swap_loop): Update.
24479 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24480 * predict.c (predictable_edge_p): Update.
24481 (edge_probability_reliable_p): Update.
24482 (set_even_probabilities): Update.
24483 (combine_predictions_for_insn): Update.
24484 (combine_predictions_for_bb): Update.
24485 (propagate_freq): Update.
24486 (estimate_bb_frequencies): Update.
24487 (force_edge_cold): Update.
24488 * profile-count.c (profile_count::dump): Add missing space into dump.
24489 (profile_count::debug): Add newline.
24490 (profile_count::differs_from_p): Explicitly convert to unsigned.
24491 (profile_count::stream_in): Update.
24492 (profile_probability::dump): New member function.
24493 (profile_probability::debug): New member function.
24494 (profile_probability::differs_from_p): New member function.
24495 (profile_probability::differs_lot_from_p): New member function.
24496 (profile_probability::stream_in): New member function.
24497 (profile_probability::stream_out): New member function.
24498 * profile-count.h (profile_count_quality): Rename to ...
24499 (profile_quality): ... this one.
24500 (profile_probability): New.
24501 (profile_count): Update.
24502 * profile.c (compute_branch_probabilities): Update.
24503 * recog.c (peep2_attempt): Update.
24504 * sched-ebb.c (schedule_ebbs): Update.
24505 * sched-rgn.c (find_single_block_region): Update.
24506 (compute_dom_prob_ps): Update.
24507 (schedule_region): Update.
24508 * sel-sched-ir.c (compute_succs_info): Update.
24509 * stmt.c (struct case_node): Update.
24510 (do_jump_if_equal): Update.
24511 (get_outgoing_edge_probs): Update.
24512 (conditional_probability): Update.
24513 (emit_case_dispatch_table): Update.
24514 (expand_case): Update.
24515 (expand_sjlj_dispatch_table): Update.
24516 (emit_case_nodes): Update.
24517 * targhooks.c: Update.
24518 * tracer.c (better_p): Update.
24519 (find_best_successor): Update.
24520 * trans-mem.c (expand_transaction): Update.
24521 * tree-call-cdce.c: Update.
24522 * tree-cfg.c (gimple_split_edge): Upate.
24523 (move_sese_region_to_fn): Upate.
24524 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24525 * tree-eh.c (lower_resx): Upate.
24526 (cleanup_empty_eh_move_lp): Upate.
24527 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24528 * tree-inline.c (copy_edges_for_bb): Update.
24529 (copy_cfg_body): Update.
24530 * tree-parloops.c (gen_parallel_loop): Update.
24531 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24532 (gimple_gen_time_profiler): Update.
24533 * tree-ssa-dce.c (remove_dead_stmt): Update.
24534 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24535 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24536 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24537 (unloop_loops): Update.
24538 (try_peel_loop): Update.
24539 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24540 * tree-ssa-loop-split.c (connect_loops): Update.
24541 (split_loop): Update.
24542 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24543 (hoist_guard): Update.
24544 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24545 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24546 (value_replacement): Update.
24547 * tree-ssa-reassoc.c (branch_fixup): Update.
24548 * tree-ssa-tail-merge.c (replace_block_by): Update.
24549 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24550 (create_edge_and_update_destination_phis): Update.
24551 (compute_path_counts): Update.
24552 (recompute_probabilities): Update.
24553 (update_joiner_offpath_counts): Update.
24554 (freqs_to_counts_path): Update.
24555 (duplicate_thread_path): Update.
24556 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24557 (struct switch_conv_info): Update.
24558 (gen_inbound_check): Update.
24559 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24560 (vect_do_peeling): Update.
24561 (vect_loop_versioning): Update.
24562 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24563 (optimize_mask_stores): Update.
24564 * ubsan.c (ubsan_expand_null_ifn): Update.
24565 * value-prof.c (gimple_divmod_fixed_value): Update.
24566 (gimple_divmod_fixed_value_transform): Update.
24567 (gimple_mod_pow2): Update.
24568 (gimple_mod_pow2_value_transform): Update.
24569 (gimple_mod_subtract): Update.
24570 (gimple_mod_subtract_transform): Update.
24571 (gimple_ic): Update.
24572 (gimple_stringop_fixed_value): Update.
24573 (gimple_stringops_transform): Update.
24574 * value-prof.h: Update.
24576 2017-06-29 Carl Love <cel@us.ibm.com>
24578 * config/rs6000/rs6000-c.c: Add support for built-in functions
24579 vector signed int vec_signed (vector float);
24580 vector signed long long vec_signed (vector double);
24581 vector signed int vec_signed2 (vector double, vector double);
24582 vector signed int vec_signede (vector double);
24583 vector signed int vec_signedo (vector double);
24584 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24585 instruction generator.
24586 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24587 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24588 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24590 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24591 vunsignede_v2df): Add define_expands.
24592 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24593 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24594 VEC_UNSIGNEDO): Add definitions.
24595 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24596 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24597 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24598 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24599 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24600 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24601 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24602 vec_unsignede, vec_unsignedo): Add builtin defines.
24603 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24605 * doc/extend.texi: Update the built-in documentation file for the
24606 new built-in functions.
24608 2017-06-29 Richard Biener <rguenther@suse.de>
24610 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24611 reduction chains to LOOP_VINFO_REDUCTIONS.
24612 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24613 SLP reductions after processing reduction chains.
24615 2017-06-29 Nathan Sidwell <nathan@acm.org>
24617 * builtins.c (fold_builtin_FUNCTION): Use
24618 lang_hooks.decl_printable_name.
24620 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24622 PR middle-end/81194
24623 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24624 with only one label.
24625 * stmt.c (expand_case): Assert NCASES is greater than one.
24627 2017-06-29 Richard Biener <rguenther@suse.de>
24629 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24631 (group_case_labels): Likewise.
24632 (find_taken_edge): Push sanity checking on val to workers...
24633 (find_taken_edge_cond_expr): ... here
24634 (find_taken_edge_switch_expr): ... and here, handle cases
24635 with just a default label.
24636 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24637 (group_case_labels): Likewise.
24638 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24639 group_case_labels does anything cleanup the CFG again.
24641 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24643 PR tree-optimization/81196
24644 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24645 exit condition comparing two IVs.
24647 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24649 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24650 profile to the dummy entry at the end of the list of architectures.
24651 * config/arm/arm-cpu-cdata.h: Regenerated.
24653 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24654 Michael Collison <michael.collison@arm.com>
24657 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24659 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24660 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24661 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24662 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24663 of shift when the shift amount is masked with constant equal to
24664 the size of the mode.
24665 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24668 2017-06-29 Martin Liska <mliska@suse.cz>
24670 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24672 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24675 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24678 (edge_growth_cache_entry::edge_growth_cache_entry): New
24680 (reset_edge_growth_cache): Update to use constructor.
24682 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24684 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24685 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24686 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24688 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24690 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24691 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24693 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24695 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
24696 (*-linux-uclibc*): Add t-uclibc tmake_file.
24697 * config/t-musl: New.
24698 * config/t-uclibc: New.
24700 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
24702 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
24704 (gen_comm_data): Emit architectural setting of arch_prof.
24705 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
24707 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
24708 (armv8-m.base, armv8-m.main): Likewise.
24709 * arm-protos.h (arm_build_target): Add profile field.
24710 (arch_option): Likewise.
24711 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
24713 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
24714 arm_active_target.profile.
24716 2017-06-28 Richard Biener <rguenther@suse.de>
24718 PR middle-end/81227
24719 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24720 TYPE_OVERFLOW_WRAPS.
24721 * match.pd (negate_expr_p): Likewise.
24722 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24723 fold_build2, not fold_binary.
24725 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24727 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24728 Convert memory address to Pmode.
24729 (aarch64_print_operand): Assert MEM operands are always Pmode.
24731 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24734 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24735 Remove redundant if.
24736 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24737 * config/arm/aarch-common-protos.h
24738 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24739 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24741 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24744 * multiple_target.c (create_dispatcher_calls): Set the default
24745 clone to be static, not public.
24747 2017-06-28 Richard Biener <rguenther@suse.de>
24749 * tree-vect-loop.c (vectorizable_reduction): Move special
24750 cond reduction IV var creation ...
24751 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24752 parameter. Use STMT_VINFO_VECTYPE.
24753 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24756 2017-06-28 Martin Liska <mliska@suse.cz>
24759 * ipa-visibility.c (non_local_p): Handle visibility.
24761 2017-06-28 Martin Liska <mliska@suse.cz>
24764 * common.opt: Add IntegerRange to various options.
24765 * opt-functions.awk (integer_range_info): New function.
24766 * optc-gen.awk: Add integer_range_info to cl_options struct.
24767 * opts-common.c (decode_cmdline_option): Handle
24768 CL_ERR_INT_RANGE_ARG.
24769 (cmdline_handle_error): Likewise.
24770 * opts.c (print_filtered_help): Show valid interval in
24771 when --help is provided.
24772 * opts.h (struct cl_option): Add range_min and range_max fields.
24773 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24775 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24777 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24778 (x * C EQ/NE y * C): New transformation.
24780 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24782 * genmultilib (combination_space): Accept '+' in option names.
24784 2017-06-28 Martin Liska <mliska@suse.cz>
24787 * asan.c (instrument_derefs): Bail out inner references
24788 that are hard register variables.
24790 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24793 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24794 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24796 2017-06-28 Richard Biener <rguenther@suse.de>
24798 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24799 (vect_get_slp_defs): Adjust.
24800 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24802 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24804 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24805 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24806 (vectorizable_reduction): Adjust.
24807 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24809 (vect_get_slp_defs): Likewise.
24810 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24811 (vectorizable_bswap): Adjust.
24812 (vectorizable_call): Likewise.
24813 (vectorizable_conversion): Likewise.
24814 (vectorizable_assignment): Likewise.
24815 (vectorizable_shift): Likewise.
24816 (vectorizable_operation): Likewise.
24817 (vectorizable_store): Likewise.
24818 (vectorizable_condition): Likewise.
24819 (vectorizable_comparison): Likewise.
24821 2017-06-28 Michael Collison <michael.collison@arm.com>
24824 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24826 (arm_gen_movmemqi): Removed unused variable 'i'.
24827 Convert 'for' loop into 'while' loop.
24828 (arm_expand_prologue): Remove last unnecessary set of insn.
24829 (thumb_pop): Remove unused variable 'pushed_words'.
24830 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24832 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24834 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24835 * config/s390/s390-protos.h: Add prototype of
24836 s390_rel_address_ok_p.
24837 * config/s390/s390.c (s390_got_symbol): New function.
24838 (s390_rel_address_ok_p): New function.
24839 (legitimize_pic_address): Use s390_rel_address_ok_p.
24840 (s390_load_got): Use s390_got_symbol.
24841 (s390_option_override): Issue error if
24842 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24843 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24845 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24847 2017-06-27 Andrew Pinski <apinski@cavium.com>
24849 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24850 (X * copysign (1.0, X)): New pattern.
24851 (X * copysign (1.0, -X)): New pattern.
24852 (copysign (-1.0, CST)): New pattern.
24854 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24856 * genmultilib (combination_space): Remove variable.
24857 Validate reuse rules against regular expression for any sequence
24858 of multilib options in any order.
24860 2017-06-27 Michael Collison <michael.collison@arm.com>
24862 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24863 call aarch64_split_simd_combine.
24864 * (aarch64_combine_internal<mode>): Delete pattern.
24865 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24866 Allow register and subreg operands.
24868 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24870 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24871 specific need, just fallback on defaults.
24872 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24874 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24875 Olivier Hainque <hainque@adacore.com>
24877 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24879 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24880 targets. Pick a default if no particular attempt applied.
24881 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24884 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24886 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24887 (x86_64-wrs-vxworks7): Likewise.
24889 2017-06-27 Marek Polacek <polacek@redhat.com>
24892 * ubsan.c (instrument_null): Check get_base_address's result for null.
24894 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24896 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24898 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24900 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24901 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24902 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24903 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24904 New function types.
24905 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24906 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24907 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24908 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24909 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24910 BUILT_IN_FEUPDATEENV): New builtins.
24911 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24912 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24913 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24914 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24916 (builtin_structptr_types): Adjust size.
24917 * tree.c (builtin_structptr_types): Add four entries.
24919 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24920 Olivier Hainque <hainque@adacore.com>
24922 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24923 (TLS_SYM): New local macro, forcing reference to __tls__ on
24924 link command lines for VxWorks 7 RTPs, triggering initialization
24926 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24927 OS features TLS support, true for RTPs on VxWorks 7.
24928 * config/vxworks.c (vxworks_override_options): Setup emutls
24931 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24933 * predict.c (test_prediction_value_range): Use -1U instead of -1
24934 to avoid narrowing conversion warning.
24935 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24936 to avoid narrowing conversion warning.
24937 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24939 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24941 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24943 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24944 64bit configurations.
24945 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24946 (SIZE_TYPE): Likewise.
24947 * config/vxworks.c (vxworks_emutls_var_fields): Use
24948 long_unsigned_type_node instead of unsigned_type_node as the offset
24949 field type, which is "pointer" mode in emutls.c.
24951 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24954 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24956 PR middle-end/81207
24957 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24958 gimple_vuse copying separately from gimple_vdef copying.
24960 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24962 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24965 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24966 Olivier Hainque <hainque@adacore.com>
24968 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24969 for all vxworks7 targets.
24970 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24971 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24972 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24973 variations for VX6/VX7 and 32/64bits later on in ...
24974 (VXWORKS_LIB_SPEC): Leverage new macros.
24975 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24976 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24978 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24980 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24981 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24983 2017-06-26 Carl Love <cel@us.ibm.com>
24985 * config/rs6000/rs6000-c.c: Add support for built-in functions
24986 vector bool char vec_reve (vector bool char);
24987 vector signed char vec_reve (vector signed char);
24988 vector unsigned char vec_reve (vector unsigned char);
24989 vector bool int vec_reve (vector bool int);
24990 vector signed int vec_reve (vector signed int);
24991 vector unsigned int vec_reve (vector unsigned int);
24992 vector bool long long vec_reve (vector bool long long);
24993 vector signed long long vec_reve (vector signed long long);
24994 vector unsigned long long vec_reve (vector unsigned long long);
24995 vector bool short vec_reve (vector bool short);
24996 vector signed short vec_reve (vector signed short);
24997 vector double vec_reve (vector double);
24998 vector float vec_reve (vector float);
24999 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25000 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25001 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25002 (altivec_vreve): New pattern.
25003 * config/rs6000/altivec.h (vec_reve): New define.
25004 * doc/extend.texi (vec_rev): Update the built-in documentation file
25005 for the new built-in functions.
25007 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25009 PR tree-optimization/71815
25010 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25012 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25014 (slsr_process_phi): Likewise.
25015 (replace_uncond_cands_and_profitable_phis): Don't replace a
25016 multiply candidate with a stride of 1 (copy or cast).
25017 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25019 (lowest_cost_path): Likewise.
25020 (total_savings): Likewise.
25022 2017-06-26 Richard Biener <rguenther@suse.de>
25025 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25026 Use def_builtin_pure for all gather builtins.
25028 2017-06-26 Richard Biener <rguenther@suse.de>
25030 PR tree-optimization/81203
25031 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25032 non-dominating BBs.
25034 2017-06-26 Marek Polacek <polacek@redhat.com>
25037 * doc/invoke.texi: Document -Wmultistatement-macros.
25039 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25041 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25042 arm_neon_ok_no_float_abi effective target.
25044 2017-06-26 Richard Biener <rguenther@suse.de>
25046 PR tree-optimization/80928
25047 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25048 (copy_bbs): Set BB_DUPLICATED flag early.
25049 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25051 (execute_on_shrinking_pred): Likewise.
25052 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25053 BB_DUPLICATED blocks.
25054 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25055 iterate over all PHIs considering removal of *gsi.
25057 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25059 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25062 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25064 * config/rs6000/rs6000-string.c: (expand_block_clear,
25065 do_load_for_compare, select_block_compare_mode,
25066 compute_current_alignment, expand_block_compare,
25067 expand_strncmp_align_check, expand_strn_compare,
25068 expand_block_move, rs6000_output_load_multiple)
25069 Move functions related to string/block move/compare
25070 to a separate file.
25071 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25072 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25073 for this function which is now used in two files.
25074 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25075 * config.gcc: Add rs6000-string.o to extra_objs for
25076 targets powerpc*-*-* and rs6000*-*-*.
25078 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25081 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25082 32-bit, since indexed is not valid for DImode.
25083 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25084 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25085 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25086 (define_peephole2 for Altivec d-form store): Likewise.
25089 * multiple_target.c (create_dispatcher_calls): Only create the
25090 dispatcher call if the function is the default clone of a
25091 versioned function.
25093 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25095 PR middle-end/80902
25096 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25097 a call, force the call to not be a tail call.
25099 2017-06-23 Jeff Law <law@redhat.com>
25101 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25104 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25106 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25107 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25108 handling for early expansion of vector shifts (sl,sr,sra,rl).
25109 (builtin_function_type): Add vector shift right instructions
25110 to the unsigned argument list.
25112 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25114 rtl-optimizatoin/79286
25115 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25116 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25117 trap. PIC register plus a const unspec without offset can never trap.
25119 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25121 * tree.h (builtin_structptr_type): New type.
25122 (builtin_structptr_types): Declare new array.
25123 * tree.c (builtin_structptr_types): New array.
25124 (free_lang_data, build_common_tree_nodes): Use it.
25126 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25129 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25132 2017-06-22 Matt Turner <mattst88@gmail.com>
25134 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25135 Lake models to skylake case. Assume skylake for unknown
25136 models with clflushopt.
25138 2017-06-22 Jeff Law <law@redhat.com>
25140 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25141 frame sizes that do not satisfy aarch64_uimm12_shift.
25143 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25145 * profile-count.h (apply_probability,
25146 apply_scale, probability_in): Fix checks for zero.
25148 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25150 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25151 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25153 2016-06-22 Richard Biener <rguenther@suse.de>
25155 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25156 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25157 REDUC_MAX_EXPR support.
25158 (vectorizable_reduction): Likewise.
25159 (vect_create_epilog_for_reduction): Likewise.
25161 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25163 * match.pd (A / (1 << B) -> A >> B): New.
25164 * generic-match-head.c: Include optabs-tree.h.
25165 * gimple-match-head.c: Likewise.
25166 * optabs-tree.h (target_supports_op_p): New.
25167 * optabs-tree.c (target_supports_op_p): New.
25169 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25171 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25172 $gcc_cv_ld --help output.
25173 (gcc_cv_ld_demangle): Likewise.
25174 (gcc_cv_ld_eh_frame_hdr): Likewise.
25175 (gcc_cv_ld_pie): Likewise.
25176 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25177 (gcc_cv_ld_buildid): Likewise.
25178 (gcc_cv_ld_sysroot): Likewise.
25179 (ld_bndplt_support): Likewise.
25180 (ld_pushpopstate_support): Likewise.
25181 * configure: Regenerate.
25182 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25184 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25187 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25188 operands indexes to avoid gap between operands and match_dups.
25190 2017-06-21 Andrew Pinski <apinski@cavium.com>
25192 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25193 Increment Arith_shift and Arith_shift_reg by 1.
25194 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25196 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25197 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25198 (aarch64_strip_extend): Add new argument and test for it.
25199 (aarch64_cheap_mult_shift_p): New function.
25200 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25201 add a cost if it is true.
25202 Update calls to aarch64_strip_extend.
25203 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25205 2017-06-21 Andrew Pinski <apinski@cavium.com>
25207 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25209 (thunderxt88): Likewise.
25210 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25211 (thunderx_prefetch_tune): New variable.
25212 (thunderx2t99_prefetch_tune): Update for the correct values.
25213 (thunderxt88_tunings): New variable.
25214 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25215 generic_prefetch_tune.
25216 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25218 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25220 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25221 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25222 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25223 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25224 (aarch64_atomic_cas<mode>, GPI): Likewise.
25226 2017-06-21 Martin Liska <mliska@suse.cz>
25228 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25229 statements on cold and hot labels.
25230 * predict.c (tree_estimate_probability_bb): Remove the
25231 prediction from this place.
25233 2017-06-21 Martin Liska <mliska@suse.cz>
25235 PR tree-optimization/79489
25236 * gimplify.c (maybe_add_early_return_predict_stmt): New
25238 (gimplify_return_expr): Call the function.
25239 * predict.c (tree_estimate_probability_bb): Remove handling
25241 * predict.def: Update comment about early return predictor.
25242 * gimple-predict.h (is_gimple_predict): New function.
25243 * predict.def: Change default value of early return to 66.
25244 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25246 * passes.def: Put pass_strip_predict_hints to the beginning of
25249 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25251 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25252 FUNCTION_DECL declarations.
25253 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25255 (dwaf2out_decl): Likewise.
25256 * godump.c (go_early_global_decl): Skip call to the real debug hook
25257 for FUNCTION_DECL declarations.
25258 * passes.c (rest_of_decl_compilation): Skip call to the
25259 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25260 -fdump-go-spec is passed.
25262 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25264 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25265 Reorder for compactness.
25266 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25267 (def_builtin_pure, def_builtin_pure2): New functions.
25268 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25270 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25272 * match.pd (nop_convert): New predicate.
25273 ((A +- CST1) +- CST2): Allow some NOP conversions.
25275 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25278 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25279 with ctors/dtors if GOVD_SHARED is set.
25281 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25283 * config/aarch64/aarch64.md (movti_aarch64):
25284 Emit mov rather than orr.
25285 (movtf_aarch64): Likewise.
25286 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25287 Emit mov rather than orr.
25289 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25291 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25292 Swap alternatives, make integer dup more expensive.
25294 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25296 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25297 Return true for non-tls symbols.
25299 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25301 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25302 (cortex-a75): Likewise.
25303 (cortex-a75.cortex-a55): Likewise.
25304 * config/aarch64/aarch64-tune.md: Regenerate.
25305 * doc/invoke.texi (-mtune): Document new values for -mtune.
25307 2017-06-21 Tom de Vries <tom@codesourcery.com>
25309 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25310 stack_size feature.
25311 (Effective-Target Keywords, Other attributes): Suggest using
25312 dg-add-options stack_size feature to get stack limit in stack_size
25313 effective target documentation.
25315 2017-06-21 Julian Brown <julian@codesourcery.com>
25316 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25318 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25319 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25320 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25322 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25323 attribute type list for neon_multiply.
25324 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25325 attribute type list for neon_multiply.
25326 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25327 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25328 attribute type list for neon_multiply.
25329 * config/arm/types.md (crypto_pmull): Add.
25330 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25331 attribute type list.
25333 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25335 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25338 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25340 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25341 to make sure not to dereference a NULL cost_classes_ptr pointer.
25343 2017-06-20 Carl Love <cel@us.ibm.com>
25345 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25346 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25347 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25348 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25349 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25350 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25351 VMULOSW): New enum "unspec" values.
25352 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25353 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25354 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25355 altivec_vmulosw): New patterns.
25356 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25357 VMULOSW): Add definitions.
25359 2017-06-20 Julia Koval <julia.koval@intel.com>
25361 * config/i386/i386.c: Fix rounding expand for new pattern.
25362 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25364 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25366 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25367 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25369 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25371 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25374 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25376 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25377 architecture, then by implementer ID.
25378 * config/aarch64/aarch64-tune.md: Regenerate.
25380 2017-06-20 Richard Biener <rguenther@suse.de>
25382 PR middle-end/81097
25383 * fold-const.c (split_tree): Fold to type before negating.
25385 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25387 * diagnostic-show-locus.c
25388 (selftest::test_fixit_deletion_affecting_newline): New function.
25389 (selftest::diagnostic_show_locus_c_tests): Call it.
25391 2017-06-20 Andreas Schwab <schwab@suse.de>
25394 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25397 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25399 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25400 __ARM_FEATURE_COPROC according to support.
25402 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25404 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25405 Rewritten to avoid overflow for > 32-bit pointers.
25408 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25409 by removing enum keyword.
25410 (ubsan_type_descriptor): Likewise. Formatting fix.
25413 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25414 splitter): Require TARGET_SSE2 in the condition.
25416 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25419 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25420 for doing vector set of SFmode on ISA 3.0.
25421 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25422 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25424 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25425 SFmode value into a V4SF variable that was extracted from another
25426 V4SF variable without converting the element to double precision
25427 and back to single precision vector format.
25428 (vsx_insert_extract_v4sf_p9_2): Likewise.
25430 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25432 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25433 in UWHI to avoid undefined overflow.
25436 * ubsan.h (enum ubsan_encode_value_phase): New.
25437 (ubsan_encode_value): Change second argument to
25438 enum ubsan_encode_value_phase with default value of
25439 UBSAN_ENCODE_VALUE_GENERIC.
25440 * ubsan.c (ubsan_encode_value): Change second argument to
25441 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25442 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25443 create_tmp_var_raw instead of create_tmp_var and use a
25445 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25446 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25447 ubsan_encode_value callers.
25450 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25451 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25452 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25454 2017-06-19 Richard Biener <rguenther@suse.de>
25456 PR middle-end/81118
25457 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25458 estimates if we changed anything.
25460 2017-06-19 Richard Biener <rguenther@suse.de>
25462 PR tree-optimization/80887
25463 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25464 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25465 simplified lookups, then reset mprts_hook.
25466 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25468 (try_to_simplify): Likewise.
25470 2017-06-19 Martin Liska <mliska@suse.cz>
25473 * gimplify.c (gimplify_switch_expr):
25474 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25476 2017-06-19 Martin Liska <mliska@suse.cz>
25478 * doc/install.texi: Document that PGO runs in 4 stages.
25480 2017-06-19 Martin Liska <mliska@suse.cz>
25483 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25484 to dispatcher function name.
25485 * multiple_target.c (replace_function_decl): New function.
25486 (create_dispatcher_calls): Redirect both edges and references.
25488 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25490 * profile-count.c (profile_count::dump): Dump quality.
25491 (profile_count::differs_from_p): Update for unsigned val.
25492 * profile-count.h (profile_count_quality): New enum.
25493 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25495 2017-06-19 Richard Biener <rguenther@suse.de>
25497 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25498 struct function as arg.
25499 (estimate_numbers_of_iterations): Export overload with loop arg.
25500 (free_numbers_of_iterations_estimates_loop): Use an overload of
25501 free_numbers_of_iterations_estimates instead.
25502 * tree-cfg.c (remove_bb): Adjust.
25503 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25504 * tree-parloops.c (gen_parallel_loop): Likewise.
25505 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25507 (tree_unroll_loops_completely): Likewise.
25508 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25509 Use an overload instead and export.
25510 (estimated_loop_iterations): Adjust.
25511 (max_loop_iterations): Likewise.
25512 (likely_max_loop_iterations): Likewise.
25513 (estimate_numbers_of_iterations): Take struct function as arg
25515 (loop_exits_before_overflow): Adjust.
25516 (free_numbers_of_iterations_estimates_loop): Use an overload.
25517 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25518 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25520 2017-06-19 Richard Biener <rguenther@suse.de>
25523 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25524 RANGE_EXPR conservatively.
25526 2017-06-16 Carl Love <cel@us.ibm.com>
25528 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25529 definitions for vec_float, vec_float2, vec_floato,
25530 vec_floate built-ins.
25531 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25532 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25534 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25535 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25536 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25537 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25538 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25539 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25540 vec_floato): Add builtin defines.
25541 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25542 Update the built-in documentation file for the new built-in
25545 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25547 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25548 (mthumb): Mark as the negative of -marm.
25550 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25552 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25554 (ARM Options, -mtune): Document that this accepts the same
25555 extension options as -mcpu.
25556 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25560 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25561 permitted extensions.
25563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25565 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25566 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25567 (armv8-m.main): Add option +nodsp.
25568 * config/arm/arm-cpu-cdata.h: Regenerated.
25570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25572 * config/arm/t-fuchsia: New file.
25573 * config.gcc (arm*-*-fuchsia*): Use it.
25575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25577 * config/arm/t-symbian: Rewrite for new option infrastructure.
25579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25581 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25582 (MULTILIB_REQUIRED): Likewise.
25584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25586 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25587 (MULTILIB_RESUE): Likewise.
25588 (MULTILIB_MATCHES): Likewise.
25589 (MULTLIB_REQUIRED): Likewise.
25591 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25593 * config/arm/t-rtems: Rewrite for new option framework.
25595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25597 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25598 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25599 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25600 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25601 * config/arm/t-multilib: ... here.
25602 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25603 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25604 armv7-a and armv8*-a when A-profile libraries have not been built.
25605 * config/arm/t-rmprofile: Rewrite.
25607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25609 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25610 with a backslash. Remove the backslash after substituting unescaped
25612 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25614 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25616 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25617 the driver to pass through the default setting of -mfloat-abi.
25618 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25620 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25621 (all_feat_combs): New rule.
25622 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25624 * config/arm/t-aprofile: Rewrite.
25626 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25628 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25629 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25630 fpu is not specified by the user/command-line.
25631 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25632 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25633 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25634 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25635 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25636 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25640 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25641 * config/arm/t-arm-elf: Rewritten.
25643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25645 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25646 have some floating-point instructions.
25647 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25648 (TARGET_MAYBE_HARD_FLOAT): New macro.
25649 * config/arm/arm-builtins.c (arm_init_builtins): Use
25650 TARGET_MAYBE_HARD_FLOAT.
25651 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25653 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25655 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25656 (configargs.h): Include it.
25657 (arm_print_hint_for_fpu_option): New function.
25658 (arm_parse_fpu_option): New function.
25659 (candidate_extension): New class.
25660 (arm_canon_for_multilib): New function.
25661 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25662 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25663 (ARCH_CANONICAL_SPECS): New macro.
25664 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25668 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25669 are set after handling multilib fragments. Set target_cpu_default2
25672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25674 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25676 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25677 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25679 (check_arch): Likewise.
25680 * config/arm/arm.c (arm_configure_build_target): Handle
25681 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25682 options in the default.
25684 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25686 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25687 when an option is an alias of another.
25688 * config/arm/parsecpu.awk (optalias): New parser token.
25689 (gen_comm_data): Mark non-alias options as such. Emit entries
25690 for extension aliases.
25691 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25692 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25693 (armv6kz, armv6zk, armv6t2): Likewise.
25694 (armv7): Make vfpv3-d16 an alias.
25695 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
25697 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
25698 Sort in canonical order.
25699 (armv8-a): Sort in canonical order.
25700 (armv8.1-a, armv8.2-a): Likewise.
25701 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
25703 (cortex-a9): Sort in canonical order.
25704 * config/arm/arm.c (selftests.h): Include it.
25705 (arm_test_cpu_arch_data): New function.
25706 (arm_run_self_tests): New function.
25707 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
25708 (targetm): Move declaration to the end of the file.
25709 * arm-cpu-cdata.h: Regenerated.
25711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25713 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
25714 call to target_mode_check describing the type of option passed.
25715 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
25716 (arm_target_thumb_only): Use arm_parse_arch_option_name or
25717 arm_parse_cpu_option_name to match parameters against list of
25719 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25720 arm_arch_core_flags data structure.
25721 * config/arm/arm-cpu_cdata.h: Regenerated.
25723 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25725 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25727 (arm_print_hint_for_cpu_option): Likewise.
25728 (arm_print_hint_for_arch_option): Likewise.
25729 (arm_parse_cpu_option_name): Likewise.
25730 (arm_parse_arch_option_name): Likewise.
25731 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25732 of entries in the all_fpus list.
25733 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25734 (arm_parse_cpu_option_name): Declare.
25735 (arm_parse_arch_option_name): Declare.
25736 (arm_parse_option_features): Declare.
25737 (arm_intialize_isa): Declare.
25738 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25740 (gen_comm_data): ... here. Make definitions non-static.
25741 * config/arm/arm-cpu-data.h: Regenerated.
25742 * config/arm/arm-cpu-cdata.h: Regenerated.
25744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25746 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25747 (cpu_arch_extension): New structure.
25748 (cpu_arch_option, arch_option, cpu_option): New structures.
25749 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25750 architecture types.
25751 (gen_data): Generate new format data tables.
25752 * config/arm/arm.c (cpu_tune): New structure.
25753 (cpu_option, processors): Delete.
25754 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25755 (arm_print_hint_for_cpu_option): ... this and ...
25756 (arm_print_hint_for_arch_option): ... this.
25757 (arm_parse_arch_cpu_name): Delete. Replace with ...
25758 (arm_parse_cpu_option_name): ... this and ...
25759 (arm_parse_arch_option_name): ... this.
25760 (arm_unrecognized_feature): Change type of target parameter to
25762 (arm_parse_arch_cpu_features): Delete. Replace with ...
25763 (arm_parse_option_features): ... this.
25764 (arm_configure_build_target): Rework to use new configuration data
25766 (arm_print_tune_info): Rework for new configuration data tables.
25767 * config/arm/arm-cpu-data.h: Regenerated.
25768 * config/arm/arm-cpu.h: Regenerated.
25770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25772 * Makefile.in (OBJS): Move sbitmap.o from here ...
25773 (OBJS-libcommon): ... to here.
25775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25777 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25778 (ISA_ALL_CRYPTO): New macro.
25779 (ISA_ALL_SIMD): New macro
25780 (ISA_ALL_FP): New macro.
25781 * config/arm/arm.c (fpu_bitlist): Update initializer.
25782 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25784 (arm9e): Add fpu. Add option for nofp
25785 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25786 (arm926ej-s, arm1026ej-s): Likewise.
25787 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25788 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25789 neon-fp16, neon-vfpv4, nofp and nosimd.
25790 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25791 (cortex-a8): Add fpu. Add option for nofp.
25792 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25793 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25794 (cortex-r4f): Add fpu.
25795 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25796 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25798 (cortex-r8): Likewise.
25799 (cortex-m4): Add fpu. Add option for nofp.
25800 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25801 (cortex-a17.cortex-a7): Likewise.
25802 (cortex-a32): Add fpu. Add options for crypto and nofp.
25803 (cortex-a35, cortex-a53): Likewise.
25804 (cortex-a57): Add fpu. Add option for crypto.
25805 (cortex-a72, cortex-a73): Likewise.
25806 (exynos-m1): Likewise.
25807 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25808 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25809 (cortex-m33): Add fpu. Add option for nofp.
25810 * config/arm/arm-cpu-cdata.h: Regenerated
25811 * config/arm/arm-cpu-data.h: Regenerated.
25813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25815 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25816 (armv5te, armv5tej): Likewise.
25817 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25818 (armv7): Add options fp and vfpv3-d16.
25819 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25820 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25822 (armv7ve): Likewise.
25823 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25824 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25825 (armv8-a): Add nocrypto option.
25826 (armv8.1-a, armv8.2-a): Likewise.
25827 (armv8-m.main): add options fp, fp.dp and nofp.
25829 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25831 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25833 (armv8-a+crc): Delete.
25834 (armv8.1-a): Add options simd, crypto and nofp.
25835 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25836 (armv8.2-a+fp16): Delete.
25837 (armv8-m.main): Add option dsp.
25838 (armv8-m.main+dsp): Delete.
25839 (cortex-a8): Add fpu. Add nofp option.
25840 (cortex-a9): Add fpu. Add nofp and nosimd options.
25841 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25842 link to main cpu and architecture data structures.
25843 (gen_comm_data): Only put isa attributes from the main architecture
25845 (option): New statement for architecture and CPU entries.
25846 * arm.c (struct cpu_option): New structure.
25847 (struct processors): Add entry for options.
25848 (arm_unrecognized_feature): New function.
25849 (arm_parse_arch_cpu_name): Ignore any characters after the first
25851 (arm_parse_arch_cpu_feature): New function.
25852 (arm_configure_build_target): Separate out any CPU and architecture
25853 features and parse separately. Don't error out if -mfpu=auto is
25854 used with only an architecture string.
25855 (arm_print_asm_arch_directives): New function.
25856 (arm_file_start): Call it.
25857 * config/arm/arm-cpu-cdata.h: Regenerated.
25858 * config/arm/arm-cpu-data.h: Likewise.
25859 * config/arm/arm-tables.opt: Likewise.
25861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25863 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25864 assembler when it is not -mfpu=auto.
25866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25868 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25869 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25870 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25871 (ASM_CPU_SPEC): Rewrite.
25872 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25873 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25874 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25875 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25876 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25877 copied string is NUL-terminated. Also strip any characters prefixed
25879 (arm_rewrite_selected_arch): New function.
25880 (arm_rewrite_march): New function.
25882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25884 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25885 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25886 (march, mcpu, mtune): Convert to string-based options.
25887 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25888 (arm_parse_arch_cpu_name): New function.
25889 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25890 identify selected architecture or CPU.
25891 (arm_option_save): New function.
25892 (TARGET_OPTION_SAVE): Redefine.
25893 (arm_option_restore): Restore string options.
25894 (arm_option_print): Print string options.
25896 2017-06-16 Martin Sebor <msebor@redhat.com>
25898 PR tree-optimization/80933
25899 PR tree-optimization/80934
25900 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25901 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25902 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25903 (gimple_fold_builtin): Call them.
25905 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25907 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25908 as unlikely; update profile.
25910 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25912 * predict.c (force_edge_cold): Handle declaring edges impossible
25915 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25917 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25919 (try_unroll_loop_completely): Fix reporting.
25921 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25923 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25925 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25928 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25929 if given a non-constant argument for an intrinsic which requires a
25932 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25934 * profile.c (compare_freqs): New function.
25935 (branch_prob): Sort edge list.
25936 (find_spanning_tree): Assume that the list is priority sorted.
25938 2017-06-16 Richard Biener <rguenther@suse.de>
25940 PR tree-optimization/81090
25941 * passes.def (pass_record_bounds): Remove.
25942 * tree-pass.h (make_pass_record_bounds): Likewise.
25943 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25944 make_pass_record_bounds): Likewise.
25945 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25946 not free niter estimates at the beginning but at the end.
25947 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25949 2017-06-16 Richard Biener <rguenther@suse.de>
25951 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25952 initializer to workaround ICE in host GCC 4.8.
25954 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25956 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25958 (clone_inlined_nodes): Update.
25960 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25962 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25963 prefetch settings, and enable prefetching by default at -O3.
25965 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25967 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25968 Set flag_prefetch_loop_arrays according to tuning data.
25970 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25972 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25973 New tune structure.
25974 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25975 [Unrelated to main purpose of the patch] Place the pointer field last
25976 to enable type checking errors when tune structure are wrongly merged.
25977 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25978 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25979 (thunderx2t99_prefetch_tune): New tune constants.
25980 (tune_params *_tunings): Update all tunings (no functional change).
25981 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25982 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25983 from tunings structures.
25985 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25988 * ubsan.c (instrument_null): Add T argument, use it instead
25989 of computing it based on IS_LHS.
25990 (instrument_object_size): Likewise.
25991 (pass_ubsan::execute): Adjust instrument_null and
25992 instrument_object_size callers to pass gimple_get_lhs or
25993 gimple_assign_rhs1 result to it. Use instrument_null instead of
25994 calling get_base_address and instrument_mem_ref. Handle
25995 aggregate call arguments for object-size sanitization.
25997 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25999 PR tree-optimization/81089
26000 * tree-vrp.c (is_masked_range_test): Validate operands of
26003 2017-06-15 Martin Sebor <msebor@redhat.com>
26006 * dumpfile.c (dump_register): Avoid calling memset to initialize
26007 a class with a default ctor.
26008 * gcc.c (struct compiler): Remove const qualification.
26009 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26010 * hash-table.h: Ditto.
26011 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26013 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26014 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26016 * params.h (struct param_info): Make struct members non-const.
26017 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26018 with default initialization.
26019 * vec.h (vec_copy_construct, vec_default_construct): New helper
26021 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26022 with vec_copy_construct.
26023 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26024 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26025 * doc/invoke.texi (-Wclass-memaccess): Document.
26027 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26029 * emit-rtl.h (is_leaf): Update comment about local
26030 register allocator.
26032 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26035 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26036 for a variable to have a section before checking if the section has a
26038 Set section to.persistent if persistent attribute is set.
26039 Warn if .persistent attribute is used on an automatic variable.
26041 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26043 PR rtl-optimization/80474
26044 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26046 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26048 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26051 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26053 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26054 (casesi): Emit bounds checking as RTL.
26055 (casesi_internal_mips16_<mode>): Remove bounds checking.
26057 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26059 * config/xtensa/xtensa.c (xtensa_option_override): Append
26060 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26061 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26062 xtensa_doloop_hooks): Define unconditionally.
26063 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26065 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26066 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26067 for it in xtensa_option_override.
26068 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26069 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26071 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26073 * doc/cppopts.texi: Document '-' special value to -MF.
26075 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26077 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26078 (cortex_a53_fconst): Likewise.
26079 (cortex_a53_fpmul): Likewise.
26080 (cortex_a53_f_load_64): Likewise.
26081 (cortex_a53_f_load_many): Likewise.
26082 (cortex_a53_advsimd_alu): Likewise.
26083 (cortex_a53_advsimd_alu_q): Likewise.
26084 (cortex_a53_advsimd_mul): Likewise.
26085 (cortex_a53_advsimd_mul_q): Likewise.
26086 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26087 Add missing fmul, r2f_cvt and fconst cases.
26089 2017-06-14 Richard Biener <rguenther@suse.de>
26091 PR middle-end/81088
26092 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26094 (fold_binary_loc): When associating do not treat pre-existing
26095 TREE_OVERFLOW on literal constants as a reason to allow
26096 TREE_OVERFLOW on associated literal constants.
26098 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26100 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26101 (MASK_FEATURES): New macro.
26102 * config/sparc/sparc.c (sparc_option_override): Remove the special
26103 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26105 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26107 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26108 a division of 0 if non-call exceptions are enabled.
26110 2017-06-14 Andrew Pinski <apinski@cavium.com>
26111 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26114 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26115 Improve vector initialization code gen for only variable case.
26117 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26119 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26121 2017-06-14 Richard Biener <rguenther@suse.de>
26123 PR tree-optimization/81083
26124 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26129 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26130 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26131 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26132 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26133 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26134 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26136 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26138 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26139 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26141 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26143 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26145 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26147 * config/rs6000/t-rtems: Don't handle SPE.
26149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26151 * config/rs6000/t-linux: Don't handle SPE.
26153 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26155 * config/rs6000/eabispe.h: Delete file.
26157 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26159 * config/rs6000/t-spe: Delete file.
26161 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26163 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26164 (rs6000_legitimate_offset_address_p): Return false for anything in
26165 V2SImode or V2SFmode.
26167 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26169 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26170 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26171 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26173 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26174 (rs6000_legitimate_offset_address_p): Ditto.
26175 (rs6000_emit_move): Ditto.
26176 (rs6000_init_builtins): Remove V4HI_type_node.
26178 2017-06-13 Martin Liska <mliska@suse.cz>
26181 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26182 (gate_asan): Likewise.
26183 * asan.h (asan_no_sanitize_address_p): Remove the function.
26184 (sanitize_flags_p): New function.
26185 * builtins.def: Fix coding style.
26186 * common.opt: Use renamed enum value.
26187 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26188 * doc/extend.texi: Document no_sanitize attribute.
26189 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26190 to SANITIZE_UNDEFINED_NONDEFAULT.
26191 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26192 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26193 Use sanitize_flags_p.
26194 * gimplify.c (gimplify_function_tree): Likewise.
26195 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26196 * opts.c (parse_no_sanitize_attribute): New function.
26197 (common_handle_option): Use renamed enum value.
26198 * opts.h (parse_no_sanitize_attribute): Declare.
26199 * tree.c (sanitize_flags_p): New function.
26200 * tree.h: Declared here.
26201 * tsan.c: Use sanitize_flags_p.
26202 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26203 (instrument_mem_ref): Likewise.
26204 (instrument_bool_enum_load): Likewise.
26205 (do_ubsan_in_current_function): Remove the function.
26206 (pass_ubsan::execute): Use sanitize_flags_p.
26207 * ubsan.h: Remove do_ubsan_in_current_function
26208 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26209 (dump_function_to_file): Use it here.
26211 2017-06-13 Martin Jambor <mjambor@suse.cz>
26213 PR tree-optimization/80803
26214 PR tree-optimization/81063
26215 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26216 (propagate_subaccesses_across_link): Enqueue subtree whenever
26217 necessary instead of relying on the caller.
26219 2017-06-13 Martin Jambor <mjambor@suse.cz>
26221 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26222 that have a first_link.
26223 (sort_and_splice_var_accesses): Do not check first_link before
26225 (subtree_mark_written_and_enqueue): Likewise.
26226 (propagate_all_subaccesses): Likewise and do not stop at first
26227 parent with a first_link.
26229 2017-06-13 Martin Jambor <mjambor@suse.cz>
26231 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26234 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26236 * match.pd: New pattern.
26238 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26240 * tree-vrp.c (is_masked_range_test): New function.
26241 (register_edge_assert_for): Determine ranges for
26244 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26246 PR tree-optimization/67328
26247 * fold-const.c (maskable_range_p): New function.
26248 (build_range_check): Generate bittests if possible.
26250 2017-06-13 Martin Liska <mliska@suse.cz>
26252 * gimple-pretty-print.c (dump_probability): Add new argument.
26253 (dump_edge_probability): Dump both probability and count.
26254 (dump_gimple_label): Likewise.
26255 (dump_gimple_bb_header): Likewise.
26257 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26260 * config/avr/avr-devices.c: Fix indentation.
26261 * config/avr/gen-avr-mmcu-specs.c: Dito.
26263 2017-06-13 Richard Biener <rguenther@suse.de>
26265 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26266 instead get vector type from stmt_info.
26267 (vectorizable_reduction): Adjust. Remove dead code.
26269 2017-06-13 Richard Biener <rguenther@suse.de>
26271 PR middle-end/81065
26272 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26273 case of C * (x * C2 + C3).
26274 (fold_addr_of_array_ref_difference): Properly fold index difference.
26276 2017-06-12 David S. Miller <davem@davemloft.net>
26279 * config/sparc/sparc.md (return expander): Emit frame blockage if
26280 function uses alloca.
26282 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26284 * combine.c (make_field_assignment): Check len rather than the mode
26285 precision when calling force_to_mode.
26287 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26289 Support multilibs and devices that see flash in RAM address range.
26292 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26293 (avr_mcu_t) <flash_pm_offset>: New field.
26294 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26295 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26296 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26297 (AVR_TINY_PM_OFFSET): Remove macro.
26298 * config/avr/avr.opt (-mshort-calls): New option.
26299 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26300 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26301 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26302 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26303 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26304 instead of avr_arch->have_jmp_call.
26305 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26306 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26307 avr_arch->flash_pm_offset to define.
26308 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26309 new field flash_pm_offset. Add entry for avrxmega3.
26310 (avr_texinfo): Add entry for avrxmega3.
26311 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26312 attiny212, attiny214,
26313 attiny412, attiny414, attiny416, attiny417,
26314 attiny814, attiny816, attiny817,
26315 attiny1614, attiny1616, attiny1617,
26316 attiny3214, attiny3216, attiny3217.
26317 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26318 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26319 (avr_print_operand_address) [AVR_TINY]: Same.
26320 (avr_asm_init_sections) <readonly_data_section>: Only patch
26321 callback if avr_arch->flash_pm_offset = 0.
26322 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26323 for rodata if avr_arch->flash_pm_offset != 0.
26324 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26325 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26326 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26327 (m_options): Append opt_rcall.
26328 (m_dirnames): Append dir_rcall.
26329 * config/avr/t-multilib: Regenerate.
26331 * configure.ac [target=avr]: Check whether avrxmega3 default
26332 linker description file works as needed.
26333 * configure: Regenerate.
26334 * doc/avr-mmcu.texi: Regenerate.
26335 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26336 <__AVR_ARCH__>: Document avrxmega3 and 103.
26337 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26338 <__AVR_SHORT_CALLS__>: Document it.
26339 <__AVR_PM_BASE_ADDRESS__>: Document it.
26340 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26341 (AVR Variable Attributes) <progmem>: Document this is
26342 not needed for avrxmega3.
26343 (AVR Named Address Spaces) <__flash>: Dito.
26345 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26347 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26349 2017-06-12 Doug Rupp <rupp@adacore.com>
26351 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26352 Append vxworks-stdint.h to the tm_file list.
26353 * config/vxworks-stdint.h: New file.
26355 2017-06-12 Martin Liska <mliska@suse.cz>
26357 PR tree-optimization/81041
26358 * tree-profile.c (gimple_gen_ic_func_profiler):
26359 Create an extra BB in profile-generate
26360 (gimple_gen_time_profiler): Likewise.
26362 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26364 PR tree-optimization/81003
26365 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26366 (update_range_test): Use it instead of force_gimple_operand_gsi.
26368 2017-06-12 Richard Biener <rguenther@suse.de>
26370 PR tree-optimization/81053
26371 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26372 with backedge value not defined in loop. Simplify def stmt
26375 2017-06-11 Tom de Vries <tom@codesourcery.com>
26378 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26380 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26381 nvptx_cannot_force_const_mem.
26383 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26385 * opts.c (finish_options): Move test for flag_split_stack after
26386 it has been initialized.
26388 2017-06-11 Jason Merrill <jason@redhat.com>
26390 * tree.h (id_equal): New.
26391 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26392 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26393 instead of strcmp of IDENTIFIER_POINTER.
26395 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26397 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26398 (mark_all_inlined_calls_cdtor): Fix formating.
26399 (inline_transform): Rescale profile before inlining.
26401 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26403 * cgraph.h (cgraph_edge::clone): Update prototype.
26404 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26405 (cgraph_node::create_clone): Update.
26406 (cgraph_node::create_version_clone): Update.
26407 * tree-inline.c (copy_bb): Update.
26408 (expand_call_inline): Update.
26410 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26412 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26413 factored out from ...
26414 (rs6000_emit_prologue): ... here.
26416 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26418 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26419 factored out from ...
26420 (rs6000_emit_prologue): ... here.
26422 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26424 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26426 (handle_missing_profiles): Fix computation of tp_first_run.
26427 (counts_to_freqs): Do not touch freqs when count is 0.
26429 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26431 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26434 2017-06-10 Tom de Vries <tom@codesourcery.com>
26436 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26437 attributes): Document signal effective target.
26439 2017-06-10 Tom de Vries <tom@codesourcery.com>
26441 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26442 Document effective target stack_size.
26444 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26446 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26447 to the edit_context if they can be auto-applied.
26449 2017-06-9 Ian Lance Taylor <iant@golang.org>
26451 * opts.c (finish_options): If -fsplit-stack, disable implicit
26452 -forder-blocks-and-partition.
26453 * doc/invoke.texi (Optimize Options): Document that when using
26454 -fsplit-stack -forder-blocks-and-partition is not implicitly
26457 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26459 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26460 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26461 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26462 * builtins.def (abort, trap, unreachable): Declare cold.
26463 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26464 * tree-core.h (ECF_COLD): New.
26465 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26466 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26468 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26470 * predict.c (unlikely_executed_stmt_p): Cleanup.
26472 2017-06-09 Richard Biener <rguenther@suse.de>
26474 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26475 model if the ref is always written to.
26477 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26479 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26481 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26483 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26486 2017-06-09 Tom de Vries <tom@codesourcery.com>
26489 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26490 "target cannot support label values" when encountering LABEL_REF.
26492 2017-06-09 Martin Liska <mliska@suse.cz>
26494 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26495 (gimple_gen_ic_func_profiler): Emit direct comparison
26496 of __gcov_indirect_call_callee with NULL.
26497 (gimple_gen_time_profiler): Change probability from
26498 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26500 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26502 * profile.c (edge_gcov_counts): Turn to pointer.
26503 (compute_branch_probabilities, compute_branch_probabilities): Update.
26504 (branch_prob): Do not clear edge_gcov_count.
26505 * profile.h (edge_gcov_counts): Turn to pointer.
26506 (edge_gcov_count): Update.
26508 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26510 * gimple.h (gimple_check_failed): Mark cold.
26512 2017-06-09 Richard Biener <rguenther@suse.de>
26514 PR tree-optimization/66623
26515 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26516 refactor check_reduction into two parts, properly computing
26517 whether we have to check reduction validity for outer loop
26520 2017-06-09 Richard Biener <rguenther@suse.de>
26522 PR tree-optimization/79483
26523 * graphite-scop-detection.c (order): New global.
26524 (get_order): Compute bb to order mapping that satisfies code
26525 generation constraints.
26526 (cmp_pbbs): New helper.
26527 (build_scops): Start domwalk at entry block, sort generated
26530 2017-06-09 Richard Biener <rguenther@suse.de>
26532 PR middle-end/81007
26533 * ipa-polymorphic-call.c
26534 (ipa_polymorphic_call_context::restrict_to_inner_class):
26535 Skip FIELD_DECLs with error_mark_node type.
26536 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26539 2017-06-09 Martin Liska <mliska@suse.cz>
26541 * predict.c (struct branch_predictor): New struct.
26542 (test_prediction_value_range): New test.
26543 (predict_c_tests): New function.
26544 * selftest-run-tests.c (selftest::run_tests): Run the function.
26545 * selftest.h: Declare new tests.
26547 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26550 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26551 gen_add3_insn did not fail.
26552 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26553 r0, construct that number in a temporary reg and add that reg to r0.
26554 If asked to put the result in r0 as well, fail.
26556 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26558 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26559 for early expansion of vec_eqv.
26561 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26563 PR middle-end/81005
26564 * ubsan.c (instrument_null): Avoid pointless code temporary.
26565 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26567 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26571 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26574 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26577 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26579 * predict.c (unlikely_executed_edge_p): Move ahead.
26580 (probably_never_executed_edge_p): Use it.
26582 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26584 PR middle-end/79988
26585 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26586 gimple_call_builtin_p call.
26588 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26590 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26591 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26592 rtl_check_failed_type2, rtl_check_failed_code1,
26593 rtl_check_failed_code2, rtl_check_failed_code_mode,
26594 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26595 rtvec_check_failed_bounds, rtl_check_failed_flag,
26596 _fatal_insn_not_found, _fatal_insn): Likewise.
26597 * tree.h (tree_contains_struct_check_failed,
26598 tree_check_failed, tree_not_check_failed,
26599 tree_class_check_failed, tree_range_check_failed,
26600 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26601 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26602 tree_operand_check_failed, omp_clause_check_failed,
26603 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26606 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26608 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26609 flag_branch_probabilities.
26610 * ipa-inline.c (edge_badness): Likewise.
26611 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26612 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26613 * predict.c (maybe_hot_frequency_p): Likewise.
26614 (probably_never_executed): Likewise.
26615 * sched-ebb.c (schedule_ebbs): Likewise.
26616 * sched-rgn.c (find_single_block_region): Likewise.
26617 * tracer.c (tail_duplicate): Likewise.
26619 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26621 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26622 longer requires x_flag_profile_use.
26624 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26626 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26627 instead of flag_reorder_blocks_and_partition.
26628 * dbxout.c (dbxout_function_end): Likewise.
26629 * dwarf2out.c (gen_subprogram_die): Likewise.
26630 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26631 * hw-doloop.c (reorg_loops): Likewise.
26632 * varasm.c (assemble_start_function,
26633 assemble_end_function): Likewise.
26634 (decide_function_section): Do not check for
26635 flag_reorder_blocks_and_partition.
26637 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26639 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26641 (chkp_get_hard_register_fake_addr_expr): Ditto.
26642 (chkp_build_addr_expr): Add check for hard reg case.
26643 (chkp_parse_array_and_component_ref): Ditto.
26644 (chkp_find_bounds_1): Ditto.
26645 (chkp_process_stmt): Don't generate bounds store for
26648 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26650 * predict.c (maybe_hot_bb_p): Do not check profile status.
26651 (maybe_hot_edge_p): Likewise.
26652 (probably_never_executed): Check for zero counts even if profile
26654 (unlikely_executed_edge_p): New function.
26655 (unlikely_executed_stmt_p): New function.
26656 (unlikely_executed_bb_p): New function.
26657 (set_even_probabilities): Use unlikely predicates.
26658 (combine_predictions_for_bb): Likewise.
26659 (predict_paths_for_bb): Likewise.
26660 (predict_paths_leading_to_edge): Likewise.
26661 (determine_unlikely_bbs): New function.
26662 (estimate_bb_frequencies): Use it.
26663 (compute_function_frequency): Use zero counts even if profile is
26665 * profile-count.h: Fix typo.
26667 2017-08-08 Julia Koval <julia.koval@intel.com>
26669 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26670 _mm512_mask_cvtsepi16_storeu_epi8,
26671 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26672 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26673 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26674 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26675 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26676 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26677 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26678 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26679 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26680 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26681 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26682 __builtin_ia32_pmovuswb256mem_mask,
26683 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26684 __builtin_ia32_pmovwb512mem_mask): New builtins.
26686 2017-08-08 Julia Koval <julia.koval@intel.com>
26688 PR target/73350,80862
26689 * config/i386/subst.md (round): Fix round pattern.
26690 * config/i386/i386.c (ix86_erase_embedded_rounding):
26691 Fix erasing rounding for the fixed pattern.
26693 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26695 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
26697 2017-06-08 Martin Liska <mliska@suse.cz>
26699 PR gcov-profile/80911
26700 * gcov.c (block_info::block_info): New constructor.
26702 2017-06-07 Carl Love <cel@us.ibm.com>
26704 * config/rs6000/rs6000-c: The return type of the following
26705 built-in functions was implemented as int not long long. Fix sign
26706 of return value for the unsigned version of vec_mulo and vec_mule.
26707 vector unsigned long long vec_bperm (vector unsigned long long,
26708 vector unsigned char)
26709 vector signed long long vec_mule (vector signed int,
26711 vector unsigned long long vec_mule (vector unsigned int,
26712 vector unsigned int)
26713 vector signed long long vec_mulo (vector signed int,
26715 vector unsigned long long vec_mulo (vector unsigned int,
26716 vector unsigned int)
26717 * doc/extend.texi: Fix the documentation for the built-in
26720 2017-06-07 Carl Love <cel@us.ibm.com>
26723 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26726 2017-06-07 Carl Love <cel@us.ibm.com>
26728 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26729 support, Generate doublehv for signed int/float for BE case only.
26731 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26733 * doc/invoke.texi (mcx16): Rewrite.
26735 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26737 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26738 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26739 *mov<mode>_softfloat, and an anonymous splitter): Use
26740 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26742 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26744 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26746 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26747 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26748 (rs6000_debug_reg_global): Adjust.
26749 (rs6000_init_hard_regno_mode_ok): Adjust.
26750 (rs6000_dbx_register_number): Adjust.
26751 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26752 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26753 Remove SPE_ACC and SPEFSCR.
26754 (REG_ALLOC_ORDER): Ditto.
26755 (FRAME_POINTER_REGNUM): Change to 111.
26756 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26757 (REG_CLASS_NAMES): Ditto.
26758 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26759 (REGISTER_NAMES): Ditto.
26760 (ADDITIONAL_REG_NAMES): Ditto.
26761 (rs6000_reg_names): Ditto.
26762 * config/rs6000/rs6000.md: Renumber some register number
26765 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26767 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26769 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26770 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26772 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26773 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26774 Delete the SPE high registers.
26775 (REG_ALLOC_ORDER): Ditto.
26776 (enum reg_class): Remove SPE_HIGH_REGS.
26777 (REG_CLASS_NAMES): Ditto.
26778 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26779 (REGISTER_NAMES): Ditto.
26780 (rs6000_reg_names): Ditto.
26781 * doc/tm.texi.in: Remove SPE as example.
26782 * doc/tm.texi: Regenerate.
26784 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26786 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26787 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26788 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26789 * config/rs6000/rs6000.md (type): Remove "brinc".
26791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26793 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26794 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26795 * config/rs6000/linuxspe.h: Delete file.
26796 * config/rs6000/rs6000.md: Don't include spe.md.
26797 * config/rs6000/spe.h: Delete file.
26798 * config/rs6000/spe.md: Delete file.
26799 * config/rs6000/t-rs6000: Remove spe.md.
26801 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26803 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26804 (reg_or_none500mem_operand): Delete.
26805 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26806 instead of reg_or_none500mem_operand.
26808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26811 handling of SPE flags.
26812 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26814 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26816 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26818 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26819 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26820 paired_divv2sf3): Similar.
26821 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26822 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26823 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26825 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26826 Rename the paired_* instruction patterns.
26827 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26829 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26830 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26831 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26832 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26833 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26834 PAIRED_VECTOR_MODE.
26835 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26836 (spe_func_has_64bit_regs_p): Delete.
26837 (spe_expand_predicate_builtin): Delete.
26838 (spe_expand_evsel_builtin): Delete.
26839 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26840 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26841 (invalid_e500_subreg): Delete.
26842 (rs6000_legitimize_address): Always force_reg op2 as well, for
26843 paired single memory accesses.
26844 (rs6000_member_type_forces_blk): Delete.
26845 (rs6000_spe_function_arg): Delete.
26846 (rs6000_expand_unop_builtin): Delete SPE handling.
26847 (rs6000_expand_binop_builtin): Ditto.
26848 (spe_expand_stv_builtin): Delete.
26849 (bdesc_2arg_spe): Delete.
26850 (spe_expand_builtin): Delete.
26851 (spe_expand_predicate_builtin): Delete.
26852 (spe_expand_evsel_builtin): Delete.
26853 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26854 (spe_init_builtins): Delete.
26855 (spe_func_has_64bit_regs_p): Delete.
26856 (savres_routine_name): Delete "info" parameter. Adjust callers.
26857 (rs6000_emit_stack_reset): Ditto.
26858 (rs6000_dwarf_register_span): Delete.
26859 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26860 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26861 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26863 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26865 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26866 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26867 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26868 mulv2sf3, divv2sf3): Delete expanders.
26870 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26872 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26876 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26877 * config/rs6000/rs6000.c: Ditto.
26879 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26881 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26882 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26883 comparison_operator.
26885 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26887 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26888 * config/rs6000/rs6000.opt: Ditto.
26889 * config/rs6000/t-rtems: Ditto.
26891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26893 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26894 TARGET_E500_SINGLE by 0, simplify.
26895 * config/rs6000/rs6000.c: Ditto.
26896 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26897 (spe_build_register_parallel): Delete.
26898 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26899 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26900 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26901 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26902 (E500_CONVERT): Delete.
26903 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26905 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26907 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26908 * config/rs6000/dfp.md: Ditto.
26909 (negdd2, *negdd2_fpr): Merge.
26910 (absdd2, *absdd2_fpr): Merge.
26911 (negtd2, *negtd2_fpr): Merge.
26912 (abstd2, *abstd2_fpr): Merge.
26913 * config/rs6000/e500.h: Delete file.
26914 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26915 TARGET_FPRS by 1 and simplify.
26916 * config/rs6000/rs6000-c.c: Ditto.
26917 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26918 TARGET_DF_SPE by 0.
26919 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26921 * config/rs6000/rs6000.md: Ditto.
26922 (floatdidf2, *floatdidf2_fpr): Merge.
26923 (move_from_CR_gt_bit): Delete.
26924 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26925 (E500_CR_IOR_COMPARE): Delete.
26926 (All patterns that require !TARGET_FPRS): Delete.
26927 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26929 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26931 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26933 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26935 * graphds.c (add_edge): Intitialize edge's attached data.
26936 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26937 pointer parameter. Call pointed function on each edge during
26938 graph traversing. Skip traversing the edge when the function
26940 (graphds_dfs, graphds_scc): Ditto.
26941 (for_each_edge): New parameter. Pass the new parameter to callback
26943 * graphds.h (skip_edge_callback): New function pointer type.
26944 (graphds_dfs, graphds_scc): New function pointer parameter.
26945 (graphds_edge_callback, for_each_edge): New parameter.
26947 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26949 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26950 out code checking if runtime alias check is possible to below ...
26951 Call the new function.
26952 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26953 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26955 2017-06-07 Marek Polacek <polacek@redhat.com>
26958 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26959 TYPE_OVERFLOW_WRAPS check.
26961 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26963 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26964 if versioning is required.
26965 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26966 peeling with the check for versioning.
26968 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26970 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26971 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26972 Set true to new parameter if new ssa variable is defined.
26973 (vect_gen_vector_loop_niters): Refactor. Set range information
26974 for the new vector loop bound variable.
26975 (vect_do_peeling): Ditto.
26977 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26979 * tree-affine.c (ssa.h): Include header file.
26980 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26981 has wrapping overflow behavior.
26983 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26985 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26987 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26989 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26990 (tree_to_aff_combination): ... here.
26992 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26994 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26995 reg_pressure model function.
26996 (ivopts_global_cost_for_size): Delete.
26997 (determine_set_costs, iv_ca_recount_cost): Call new model function
26998 ivopts_estimate_reg_pressure.
27000 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27002 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27003 expensive than udiv. Remove floating point cases from mod.
27005 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27007 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27008 Increase idiv cost.
27010 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27012 * config/aarch64/aarch64.md
27013 (copysignsf3): Fix mask generation.
27015 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27017 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27019 (class dump_manager): Add register_dumps method.
27020 * dumpfile.c: Include langhooks.h.
27021 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27022 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27023 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27024 (dump_manager::dump_register): Start with 512 entries instead of 32.
27025 (dump_manager::register_dumps): New method.
27026 * toplev.c (general_init): Instead of invoking register_dumps
27027 langhook, invoke register_dumps method on the dump manager.
27028 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27031 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27033 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27034 Say that # requires an associated define_split to exist, and that
27035 the define_split must be suitable for use after register allocation.
27037 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27039 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27040 (compute_outgoing_frequencies): Also initialize zero counts.
27041 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27042 around loops; preserve more of profile when nothing changes.
27044 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27046 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27048 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27049 * config/arm/arm-cpu-cdata.h: Regenerate.
27050 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27051 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27052 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27053 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27054 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27056 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27058 * config/arm/t-rmprofile: Likewise.
27059 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27061 2017-06-06 David S. Miller <davem@davemloft.net>
27064 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27065 blockage if function uses alloca.
27067 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27069 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27070 New "uid" fields to hold pretty-print IDs of group and ref.
27071 Memory references are now identified as <group_id>:<ref_id>
27072 instead of using [random] addresses.
27073 (dump_mem_details): Simplify, no functional change.
27074 (dump_mem_ref): Simplify and make output more concise.
27075 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27076 (find_or_create_group): Initialize group uid.
27077 (record_ref): Initialize ref uid. Improve debug output.
27078 (prune_group_by_reuse, should_issue_prefetch_p,)
27079 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27080 (mark_nontemporal_store, determine_loop_nest_reuse):
27081 Improve debug output.
27083 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27085 * dbgcnt.def (prefetch): New debug counter.
27086 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27087 (schedule_prefetches): Stop issueing prefetches if debug counter
27090 2017-06-06 Tom de Vries <tom@codesourcery.com>
27092 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27093 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27095 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27097 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27098 Use aarch64_reg_or_zero predicate for operand 4.
27099 (aarch64_compare_and_swap<mode> define_insn_and_split):
27100 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27101 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27103 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27105 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27106 (arm_compute_save_core_reg_mask): This.
27107 (thumb1_compute_save_reg_mask): Rename into ...
27108 (thumb1_compute_save_core_reg_mask): This.
27109 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27110 (arm_compute_frame_layout): Likewise.
27112 2017-06-06 Richard Biener <rguenther@suse.de>
27114 PR tree-optimization/80974
27115 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27116 keep or clear leaders SSA info.
27118 2017-06-06 Tom de Vries <tom@codesourcery.com>
27120 * config/nvptx/nvptx.c (split_mode_p): New function.
27121 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27123 2017-06-06 Tom de Vries <tom@codesourcery.com>
27125 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27127 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27130 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27133 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27135 * shrink-wrap.c (handle_simple_exit): Update profile.
27136 (try_shrink_wrapping): Upate profile.
27138 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27140 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27141 (tree_guess_outgoing_edge_probabilities): New.
27142 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27143 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27145 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27147 * ipa-split.c (split_function): Initialize return bb profile.
27149 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27151 * profile.c (compute_branch_probabilities): Also initialize
27152 EXIT_BLOCK profile.
27154 2017-06-06 Richard Biener <rguenther@suse.de>
27156 PR tree-optimization/80928
27157 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27158 (vect_analyze_loop_operations): Properly guard analysis for
27160 (vect_transform_loop): Likewise.
27161 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27162 (vect_model_induction_cost): Do not cost for pure SLP.
27163 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27164 of induction in inner loop vectorization.
27165 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27166 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27167 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27169 (vect_analyze_slp_cost_1): Cost induction.
27170 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27171 (vect_get_slp_vect_defs): Likewise.
27172 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27173 (vect_transform_stmt): Handle SLP reductions.
27174 * tree-vectorizer.h (vectorizable_induction): Adjust.
27176 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27178 * config/rs6000/rs6000.c (make_resolver_func): Update
27179 init_lowered_empty_function call.
27181 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27183 * doc/invoke.texi: Document the -fprofile-abs-path option.
27184 * common.opt (fprofile-abs-path): New option.
27185 * gcov-io.h (gcov_write_filename): Declare.
27186 * gcov-io.c (gcov_write_filename): New function.
27187 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27188 * profile.c (output_location): Likewise.
27190 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27192 * shring-wrap.c: Revert accidental commit.
27194 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27196 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27198 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27200 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27202 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27203 profile in callgraph edge.
27204 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27205 (apply_scale): Likewise.
27206 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27207 Also scale profile when inlining function with zero profile.
27208 (initialize_cfun): Update exit block profile even when it is zero.
27209 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27210 when profile is read.
27212 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27214 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27215 (CLONE_*): New constants to define the processors we can generate
27216 code for with the target_clone attribute.
27217 (rs6000_clone_map): New array to identify which clone processors
27218 the current program is running on.
27219 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27220 target_clone attribute.
27221 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27222 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27223 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27224 (cpu_expand_builtin): Add support for target_clone attribute.
27225 (rs6000_valid_attribute_p): Allow "default" attribute.
27226 (get_decl_name): New debug function to simplify printing the
27227 current function name in debugging statements.
27228 (rs6000_clone_priority): New functions to support the target_clone
27229 attribute, and be able to generate code to switch between ISA 2.05
27230 through ISA 3.0 (power6 through power9).
27231 (rs6000_compare_version_priority): Likewise.
27232 (rs6000_get_function_versions_dispatcher): Likewise.
27233 (make_resolver_func): Likewise.
27234 (add_condition_to_bb): Likewise.
27235 (dispatch_function_versions): Likewise.
27236 (rs6000_generate_version_dispatcher_body): Likewise.
27237 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27238 (fusion_gpr_load_p): Fix a spacing issue.
27239 * doc/extend.texi (Common Function Attributes): Document that the
27240 PowerPC supports the target_clone attribute.
27242 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27244 * config/arm/arm.h: explain F symbol found in description of ARM
27245 register allocation in its legend.
27247 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27249 * config/mips/frame-header-opt.c: Include profile-count.h.
27250 * config/riscv/riscv.c: Include profile-count.h
27252 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27254 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27256 (sm_set_flag_if_changed): Add bbs field.
27257 (execute_sm_if_changed_flag_set): Pass BBS.
27258 (execute_sm): Update.
27260 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27262 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27265 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27267 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27268 (peephole2): New peephole2 to emit the above.
27269 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27271 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27273 * config/aarch64/aarch64.c (define_peephole2 above
27274 *sub_<shift>_<mode>): New peephole.
27276 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27278 * config/i386/i386.c (make_resolver_func): Update.
27279 * Makefile.in: Add profile-count.h and profile-count.o
27280 * auto-profile.c (afdo_indirect_call): Update to new API.
27281 (afdo_set_bb_count): Update.
27282 (afdo_propagate_edge): Update.
27283 (afdo_propagate_circuit): Update.
27284 (afdo_calculate_branch_prob): Update.
27285 (afdo_annotate_cfg): Update.
27286 * basic-block.h: Include profile-count.h
27287 (struct edge_def): Turn count to profile_count.
27288 (struct basic_block_def): Likewie.
27289 (REG_BR_PROB_BASE): Move to profile-count.h
27290 (RDIV): Move to profile-count.h
27291 * bb-reorder.c (max_entry_count): Turn to profile_count.
27292 (find_traces): Update.
27293 (rotate_loop):Update.
27294 (connect_traces):Update.
27295 (sanitize_hot_paths):Update.
27296 * bt-load.c (migrate_btr_defs): Update.
27297 * cfg.c (RDIV): Remove.
27298 (init_flow): Use alloc_block.
27299 (alloc_block): Uninitialize count.
27300 (unchecked_make_edge): Uninitialize count.
27301 (check_bb_profile): Update.
27302 (dump_edge_info): Update.
27303 (dump_bb_info): Update.
27304 (update_bb_profile_for_threading): Update.
27305 (scale_bbs_frequencies_int): Update.
27306 (scale_bbs_frequencies_gcov_type): Update.
27307 (scale_bbs_frequencies_profile_count): New.
27308 * cfg.h (update_bb_profile_for_threading): Update.
27309 (scale_bbs_frequencies_profile_count): Declare.
27310 * cfgbuild.c (compute_outgoing_frequencies): Update.
27311 (find_many_sub_basic_blocks): Update.
27312 * cfgcleanup.c (try_forward_edges): Update.
27313 (try_crossjump_to_edge): Update.
27314 * cfgexpand.c (expand_gimple_tailcall): Update.
27315 (construct_exit_block): Update.
27316 * cfghooks.c (verify_flow_info): Update.
27317 (dump_bb_for_graph): Update.
27318 (split_edge): Update.
27319 (make_forwarder_block): Update.
27320 (duplicate_block): Update.
27321 (account_profile_record): Update.
27322 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27323 (get_estimated_loop_iterations): Update.
27324 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27325 (single_likely_exit): Update.
27326 * cfgloopmanip.c (scale_loop_profile): Update.
27328 (set_zero_probability): Update.
27329 (lv_adjust_loop_entry_edge): Update.
27330 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27331 (purge_dead_edges): Update.
27332 (rtl_account_profile_record): Update.
27333 * cgraph.c (cgraph_node::create): Uninitialize count.
27334 (symbol_table::create_edge): Uninitialize count.
27335 (cgraph_update_edges_for_call_stmt_node): Update.
27336 (cgraph_edge::dump_edge_flags): Update.
27337 (cgraph_node::dump): Update.
27338 (cgraph_edge::maybe_hot_p): Update.
27339 * cgraph.h: Include profile-count.h
27340 (create_clone), create_edge, create_indirect_edge): Update.
27341 (cgraph_node): Turn count to profile_count.
27342 (cgraph_edge0: Likewise.
27343 (make_speculative, clone): Update.
27344 (create_edge): Update.
27345 (init_lowered_empty_function): Update.
27346 * cgraphclones.c (cgraph_edge::clone): Update.
27347 (duplicate_thunk_for_node): Update.
27348 (cgraph_node::create_clone): Update.
27349 * cgraphunit.c (cgraph_node::analyze): Update.
27350 (cgraph_node::expand_thunk): Update.
27351 * final.c (dump_basic_block_info): Update.
27352 * gimple-streamer-in.c (input_bb): Update.
27353 * gimple-streamer-out.c (output_bb): Update.
27354 * graphite.c (print_global_statistics): Update.
27355 (print_graphite_scop_statistics): Update.
27356 * hsa-brig.c: Include basic-block.h.
27357 * hsa-dump.c: Include basic-block.h.
27358 * hsa-gen.c (T sum_slice): Update.
27359 (convert_switch_statements):Update.
27360 * hsa-regalloc.c: Include basic-block.h.
27361 * ipa-chkp.c (chkp_produce_thunks): Update.
27362 * ipa-cp.c (struct caller_statistics): Update.
27363 (init_caller_stats): Update.
27364 (gather_caller_stats): Update.
27365 (ipcp_cloning_candidate_p): Update.
27366 (good_cloning_opportunity_p): Update.
27367 (get_info_about_necessary_edges): Update.
27368 (dump_profile_updates): Update.
27369 (update_profiling_info): Update.
27370 (update_specialized_profile): Update.
27371 (perhaps_add_new_callers): Update.
27372 (decide_about_value): Update.
27373 (ipa_cp_c_finalize): Update.
27374 * ipa-devirt.c (struct odr_type_warn_count): Update.
27375 (struct decl_warn_count): Update.
27376 (struct final_warning_record): Update.
27377 (possible_polymorphic_call_targets): Update.
27378 (ipa_devirt): Update.
27379 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27380 * ipa-icf.c (sem_function::merge): Update.
27381 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27382 * ipa-inline.c (compute_uninlined_call_time): Update.
27383 (compute_inlined_call_time): Update.
27384 (want_inline_small_function_p): Update.
27385 (want_inline_self_recursive_call_p): Update.
27386 (edge_badness): Update.
27387 (lookup_recursive_calls): Update.
27388 (recursive_inlining): Update.
27389 (inline_small_functions): Update.
27390 (dump_overall_stats): Update.
27391 (dump_inline_stats): Update.
27392 * ipa-profile.c (ipa_profile_generate_summary): Update.
27393 (ipa_propagate_frequency): Update.
27394 (ipa_profile): Update.
27395 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27396 * ipa-utils.c (ipa_merge_profiles): Update.
27397 * loop-doloop.c (doloop_modify): Update.
27398 * loop-unroll.c (report_unroll): Update.
27399 (unroll_loop_runtime_iterations): Update.
27400 * lto-cgraph.c (lto_output_edge): Update.
27401 (lto_output_node): Update.
27402 (input_node): Update.
27403 (input_edge): Update.
27404 (merge_profile_summaries): Update.
27405 * lto-streamer-in.c (input_cfg): Update.
27406 * lto-streamer-out.c (output_cfg): Update.
27407 * mcf.c (create_fixup_graph): Update.
27408 (adjust_cfg_counts): Update.
27409 (sum_edge_counts): Update.
27410 * modulo-sched.c (sms_schedule): Update.
27411 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27412 * predict.c (maybe_hot_count_p): Update.
27413 (probably_never_executed): Update.
27414 (dump_prediction): Update.
27415 (combine_predictions_for_bb): Update.
27416 (propagate_freq): Update.
27417 (handle_missing_profiles): Update.
27418 (counts_to_freqs): Update.
27419 (rebuild_frequencies): Update.
27420 (force_edge_cold): Update.
27421 * predict.h: Include profile-count.h
27422 (maybe_hot_count_p, counts_to_freqs): UPdate.
27423 * print-rtl-function.c: Do not include cfg.h
27424 * print-rtl.c: Include basic-block.h
27425 * profile-count.c: New file.
27426 * profile-count.h: New file.
27427 * profile.c (is_edge_inconsistent): Update.
27428 (correct_negative_edge_counts): Update.
27429 (is_inconsistent): Update.
27430 (set_bb_counts): Update.
27431 (read_profile_edge_counts): Update.
27432 (compute_frequency_overlap): Update.
27433 (compute_branch_probabilities): Update; Initialize and deinitialize
27435 (branch_prob): Update.
27436 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27437 (edge_gcov_count): New.
27438 (bb_gcov_count): New.
27439 * shrink-wrap.c (try_shrink_wrapping): Update.
27440 * tracer.c (better_p): Update.
27441 * trans-mem.c (expand_transaction): Update.
27442 (ipa_tm_insert_irr_call): Update.
27443 (ipa_tm_insert_gettmclone_call): Update.
27444 * tree-call-cdce.c: Update.
27445 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27446 (gimple_duplicate_sese_tail): Update.
27447 (gimple_account_profile_record): Update.
27448 (execute_fixup_cfg): Update.
27449 * tree-inline.c (copy_bb): Update.
27450 (copy_edges_for_bb): Update.
27451 (initialize_cfun): Update.
27452 (freqs_to_counts): Update.
27453 (copy_cfg_body): Update.
27454 (expand_call_inline): Update.
27455 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27456 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27457 (try_unroll_loop_completely): Update.
27458 (try_peel_loop): Update.
27459 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27460 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27461 * tree-ssa-loop-split.c (connect_loops): Update.
27462 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27463 * tree-ssa-reassoc.c (branch_fixup): Update.
27464 * tree-ssa-tail-merge.c (replace_block_by): Update.
27465 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27466 (compute_path_counts): Update.
27467 (update_profile): Update.
27468 (recompute_probabilities): Update.
27469 (update_joiner_offpath_counts): Update.
27470 (estimated_freqs_path): Update.
27471 (freqs_to_counts_path): Update.
27472 (clear_counts_path): Update.
27473 (ssa_fix_duplicate_block_edges): Update.
27474 (duplicate_thread_path): Update.
27475 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27476 (struct switch_conv_info): Update.
27477 * tree-tailcall.c (decrease_profile): Update.
27478 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27479 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27480 * value-prof.c (check_counter): Update.
27481 (gimple_divmod_fixed_value): Update.
27482 (gimple_mod_pow2): Update.
27483 (gimple_mod_subtract): Update.
27484 (gimple_ic_transform): Update.
27485 (gimple_stringop_fixed_value): Update.
27486 * value-prof.h (gimple_ic): Update.
27488 2017-06-02 Carl Love <cel@us.ibm.com>
27490 * config/rs6000/rs6000-c: Add support for built-in functions
27491 vector double vec_doublee (vector signed int);
27492 vector double vec_doublee (vector unsigned int);
27493 vector double vec_doublee (vector float);
27494 vector double vec_doubleh (vector signed int);
27495 vector double vec_doubleh (vector unsigned int);
27496 vector double vec_doubleh (vector float);
27497 vector double vec_doublel (vector signed int);
27498 vector double vec_doublel (vector unsigned int);
27499 vector double vec_doublel (vector float);
27500 vector double vec_doubleo (vector signed int);
27501 vector double vec_doubleo (vector unsigned int);
27502 vector double vec_doubleo (vector float);.
27503 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27504 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27506 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27507 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27508 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27510 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27511 vec_doublel, vec_doubleh.
27512 * doc/extend.texi: Update the built-in documentation file for the
27513 new built-in functions.
27515 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27518 * ipa-inline-analysis.c (free_growth_caches): Set
27519 edge_removal_hook_holder to NULL after removing it.
27521 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27523 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27524 comparision with zero.
27526 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27527 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27528 for early expansion of vec_min and vec_max builtins.
27529 (builtin_function_type): Add min/max unsigned variants to those
27530 identified as having unsigned arguments.
27532 2017-06-02 Olivier Hainque <hainque@adacore.com>
27534 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27536 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27538 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27539 Use VALL_F16 iterator rather than VALL.
27541 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27543 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27544 Emit CBNZ inside loop when doing a strong exchange and comparing
27545 against zero. Generate the CC flags after the loop.
27547 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27549 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27550 (dl_section_ref): New.
27551 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27552 On AIX, append an expression to subtract the size of the
27553 section length to dl_section_ref.
27555 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27557 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27558 for early expansion of vector absolute builtins.
27560 2017-06-02 Richard Biener <rguenther@suse.de>
27562 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27563 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27565 2017-06-02 Richard Biener <rguenther@suse.de>
27567 PR tree-optimization/80948
27568 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27569 stmt order as well.
27571 2017-06-02 Richard Biener <rguenther@suse.de>
27573 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27575 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27576 for inductions as relevant.
27578 2017-06-02 Richard Biener <rguenther@suse.de>
27580 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27581 (vectorizable_induction): ... this. Remove dead code.
27583 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27585 * builtins. (expand_builtin_alloca): Remove second parameter and
27586 infer its value from the first parameter instead.
27587 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27589 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27591 PR rtl-optimization/80903
27592 * loop-doloop.c (add_test): Unshare sequence.
27594 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27596 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27598 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27600 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27602 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27603 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27604 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27605 (xlogue_layout::get_stub_rtx): Make static.
27606 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27607 (xlogue_layout::compute_stub_managed_regs): Rename to...
27608 (xlogue_layout::count_stub_managed_regs): ...this.
27609 (xlogue_layout::is_stub_managed_reg): New function.
27610 (xlogue_layout::m_stub_names): Rename to...
27611 (xlogue_layout::s_stub_names): ...this, make static.
27612 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27613 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27614 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27615 xlogue_layout::s_stub_names): Instantiate statics.
27616 (stub_managed_regs): Remove.
27617 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27618 (disable_call_ms2sysv_xlogues): Rename to...
27619 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27620 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27622 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27623 change after reload_completed.
27624 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27626 (ix86_expand_prologue): Likewise.
27627 (ix86_expand_epilogue): Likewise.
27628 (ix86_expand_split_stack_prologue): Likewise.
27629 (ix86_compute_frame_layout): Remove frame parameter ...
27630 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27631 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27633 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27634 (ix86_frame): Move from here ...
27635 * config/i386/i386.h (ix86_frame): ... to here.
27636 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27637 complete ix86_frame data structure instead. Remove some_ld_name.
27639 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27641 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27642 symbols that hold a DECL_VALUE_EXPR.
27644 2017-06-01 Martin Jambor <mjambor@suse.cz>
27646 PR tree-optimization/80898
27647 * tree-sra.c (process_subtree_disqualification): Removed.
27648 (disqualify_candidate): Do not acll
27649 process_subtree_disqualification.
27650 (subtree_mark_written_and_enqueue): New function.
27651 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27652 RHS has been disqualified and re-queue LHS if necessary. Apart
27653 from that, ignore disqualified RHS.
27655 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27657 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27658 address fetch for z10 or later.
27660 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27662 * config/arc/arc.md (tst_movb): Add guard when splitting.
27664 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27666 * config/arc/arc.c (arc_can_eliminate): Test against
27667 arc_frame_pointer_needed.
27669 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27671 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27672 to prevent store reordering.
27673 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27674 (type): Add block type.
27675 (stack_tie): Define special instruction to be used in
27678 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27680 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27681 constraint. It is not valid for the pattern.
27682 (noncommutative_binary_comparison): Likewise.
27684 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27686 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27689 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27691 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27692 be used by the reg-alloc.
27694 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27696 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
27697 reg-alloc when having mul64 or mul32x16 instructions.
27698 (mulsidi3): Likewise.
27699 (umulsidi3): Likewise.
27700 (mulsi32x16): New pattern.
27701 (mulsi64): Likewise.
27702 (mulsidi64): Likewise.
27703 (umulsidi64): Likewise.
27704 (MUL32x16_REG): Define.
27705 (mul64_600): Use MUL32x16_REG.
27706 (mac64_600): Likewise.
27707 (umul64_600): Likewise.
27708 (umac64_600): Likewise.
27710 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27712 * config/arc/arc.md (mulsi3_700): Make it commutative.
27714 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
27716 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27718 (*sign_extendsidi2_insn): Likewise for movstosw.
27720 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27722 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27723 the type of the input discriminant value. Convert the
27724 discriminant value of signedness vary.
27726 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27728 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27729 Add to -Wall section.
27731 2017-06-01 Richard Biener <rguenther@suse.de>
27733 PR middle-end/66313
27734 * fold-const.c (fold_plusminus_mult_expr): If the factored
27735 factor may be zero use a wrapping type for the inner operation.
27736 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27737 and handle moved defs.
27738 (process_assignment): Properly guard the unary op case. Return a
27739 tri-state indicating that moving the stmt before the call may allow
27740 to continue. Pass through to_move.
27741 (find_tail_calls): Handle moving unrelated defs before
27744 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27747 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27748 splitter result in the canonical way.
27750 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27752 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27753 also for 32bit target. Update insn attributes.
27754 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27756 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27758 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27759 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27760 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27761 (_mm_maskz_min_ss): New intrinsics.
27763 2017-05-31 Martin Liska <mliska@suse.cz>
27765 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27766 Change comment style to one we normally use.
27767 (vectorizable_reduction): Likewise.
27768 (vectorizable_induction): Likewise.
27769 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27770 (vectorizable_call): Likewise.
27771 (vectorizable_simd_clone_call): Likewise.
27772 (vectorizable_conversion): Likewise.
27773 (vectorizable_assignment): Likewise.
27774 (vectorizable_shift): Likewise.
27775 (vectorizable_operation): Likewise.
27776 (vectorizable_store): Likewise.
27777 (vectorizable_load): Likewise.
27778 * tree-vectorizer.h: Likewise.
27780 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27782 * passes.c (emergency_dump_function): New.
27783 * tree-pass.h (emergency_dump_function): Declare.
27784 * plugin.c (plugins_internal_error_function): Remove.
27785 * plugin.h (plugins_internal_error_function): Remove declaration.
27786 * toplev.c (internal_error_function): New static function. Use it...
27787 (general_init): ...here.
27789 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27791 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27792 (arc_rtx_costs): Add costs for new patterns.
27793 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27794 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27796 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27798 * tree-ssa-strlen.c (get_next_strinfo): New function.
27799 (get_stridx_plus_constant): Use it.
27800 (zero_length_string): Likewise.
27801 (adjust_related_strinfos): Likewise.
27802 (adjust_last_stmt): Likewise.
27804 2017-05-31 Richard Biener <rguenther@suse.de>
27807 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27808 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27810 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27812 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27813 loop_vinfo argument and use of dependence distance vectors.
27814 Check instead whether the two references differ only in their
27815 initial value and assume that they have the same alignment if the
27816 difference is a multiple of the vector alignment.
27817 (vect_analyze_data_refs_alignment): Update call accordingly.
27819 2017-05-31 Martin Liska <mliska@suse.cz>
27822 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27824 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27826 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27827 (create_intersect_range_checks): Move from ...
27828 * tree-data-ref.c (create_intersect_range_checks_index)
27829 (create_intersect_range_checks): ... to here.
27830 (create_runtime_alias_checks): New function factored from ...
27831 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27832 here. Call above function.
27833 * tree-data-ref.h (create_runtime_alias_checks): New function.
27835 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27837 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27838 segment length for dr_b and compute it in wide_int.
27840 2017-05-31 Richard Biener <rguenther@suse.de>
27842 PR tree-optimization/80906
27843 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27844 and pass through iv_map.
27845 (copy_bb_and_scalar_dependences): Adjust.
27846 (translate_pending_phi_nodes): Likewise.
27847 (copy_loop_close_phi_args): Handle code-generating IVs instead
27850 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27852 * diagnostic-color.c (color_dict): Add "type-diff".
27853 (parse_gcc_colors): Update comment.
27854 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27855 -fdiagnostics-show-template-tree and -fno-elide-type.
27856 (GCC_COLORS): Add type-diff to example.
27858 (-fdiagnostics-show-template-tree): New.
27859 (-fno-elide-type): New.
27860 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27861 the pp_format_decoder callback. Call any m_format_postprocessor's
27863 (pretty_printer::pretty_printer): Initialize
27864 m_format_postprocessor.
27865 (pretty_printer::~pretty_printer): Delete any
27866 m_format_postprocessor.
27867 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27868 (class format_postprocessor): New class.
27869 (struct pretty_printer::format_decoder): Document the new parameters.
27870 (struct pretty_printer::m_format_postprocessor): New field.
27871 * tree-diagnostic.c (default_tree_printer): Update for new
27872 bool and const char ** params.
27873 * tree-diagnostic.h (default_tree_printer): Likewise.
27875 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27877 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27878 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27879 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27880 rs6000_gen_cell_microcode code.
27881 (rs6000_final_prescan_insn): Delete.
27882 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27883 "warn-cell-microcode" entries.
27884 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27885 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27886 throughout. Change cc_reg_not_micro_cr0_operand to
27887 cc_reg_not_cr0_operand throughout.
27888 (*extendhi<mode>2_noload): Delete.
27889 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27890 (mwarn-cell-microcode): Delete.
27891 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27892 -mgen-cell-microcode and -mwarn-cell-microcode.
27894 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27897 * config/i386/constraints.md (Yd): New constraint.
27899 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27900 and (?Yd, r) alternatives. Update insn attributes.
27901 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27902 and (?*Yd, r) alternatives. Update insn attributes.
27903 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27905 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27907 * gimplify.c (gimplify_modify_expr): Don't create a
27908 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27911 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27913 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27915 2017-05-30 Richard Biener <rguenther@suse.de>
27917 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27918 and reduc_def fields.
27919 (STMT_VINFO_REDUC_TYPE): New define.
27920 (STMT_VINFO_REDUC_DEF): Likewise.
27921 (vect_force_simple_reduction): Adjust prototype.
27922 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27923 (vect_is_simple_reduction): Remove check_reduction argument.
27924 (vect_force_simple_reduction): Adjust and set
27925 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27926 (vectorizable_reduction): Do not re-do reduction analysis
27927 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27928 * tree-parloops.c (gather_scalar_reductions): Adjust.
27930 2017-05-30 Richard Biener <rguenther@suse.de>
27932 PR middle-end/80901
27933 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27936 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27938 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27939 Introduce unknown_misalignment parameter and remove vf.
27940 (vect_peeling_hash_get_lowest_cost):
27941 Pass unknown_misalignment parameter.
27942 (vect_enhance_data_refs_alignment):
27943 Fix unsupportable data ref treatment.
27945 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27947 * tree-vect-data-refs.c (vect_get_data_access_cost):
27948 Workaround for SLP handling.
27949 (vect_enhance_data_refs_alignment):
27950 Compute costs for doing no peeling at all, compare to the best
27951 peeling costs so far and avoid peeling if cheaper.
27953 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27955 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27956 Return peeling info and set costs to zero for unlimited cost
27958 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27959 with unknown misalignment. Compute and costs for unknown
27960 misalignment, compare them to the costs for known misalignment
27961 and choose the cheapest for peeling.
27963 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27965 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27966 (vect_get_peeling_costs_all_drs): Create function.
27967 (vect_peeling_hash_get_lowest_cost):
27968 Use vect_get_peeling_costs_all_drs.
27969 (vect_peeling_supportable): Create function.
27970 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27972 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27974 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27975 DR_HAS_NEGATIVE_STEP.
27976 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27977 (vect_enhance_data_refs_alignment): Use.
27978 (vect_duplicate_ssa_name_ptr_info): Use.
27979 * tree-vectorizer.h (dr_misalignment): Use.
27980 (known_alignment_for_access_p): Use.
27982 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27985 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27987 (has_section_name): New function.
27989 2017-05-30 Martin Liska <mliska@suse.cz>
27992 * auto-profile.c (get_function_decl_from_block): Fix
27995 2017-05-30 Richard Biener <rguenther@suse.de>
27997 PR middle-end/80876
27998 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28000 2017-05-30 Martin Liska <mliska@suse.cz>
28002 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28003 * dumpfile.h (struct dump_file_info): Remove ctors.
28005 2017-05-30 Martin Liska <mliska@suse.cz>
28007 * predict.def: Fix GNU coding style.
28009 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28011 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28012 Mark 'to' argument with ATTRIBUTE_UNUSED.
28014 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28016 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28017 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28018 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28021 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28023 * doc/install.texi (Options specification): Restore entry of
28024 --enable-sjlj-exceptions.
28026 2017-05-27 Michael Eager <eager@eagercon.com>
28029 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28031 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28033 * config/microblaze/microblaze.h
28034 (FIXED_REGISTERS): Update in macro.
28035 (CALL_USED_REGISTERS): Update in macro.
28037 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28039 * doc/install.texi: Add links to macOS binary distributions.
28041 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28045 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28047 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28049 2017-05-26 Martin Liska <mliska@suse.cz>
28051 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28053 2017-05-26 Martin Liska <mliska@suse.cz>
28055 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28056 always leading ';; '.
28057 (dump_bb_info): Likewise.
28058 (brief_dump_cfg): Likewise.
28059 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28060 * dumpfile.c: Remove usage of TDF_VERBOSE.
28061 * dumpfile.h (enum dump_kind): Likewise.
28062 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28063 * print-tree.c (debug_verbose): Remove.
28064 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28065 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28066 * tree-diagnostic.c (default_tree_printer): Replace
28067 TDF_DIAGNOSTIC with TDF_SLIM.
28069 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28071 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28072 in parameter loop, rather than loop_vinfo.
28073 (create_intersect_range_checks): Ditto.
28074 (vect_create_cond_for_alias_checks): Update call to above functions.
28076 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28078 PR tree-optimization/80815
28079 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28080 for merging runtime alias checks. Handle negative DR_STEPs.
28082 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28084 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28086 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28087 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28088 out code pruning runtime alias checks.
28089 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28090 factored out from above.
28091 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28093 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28095 (prune_runtime_alias_test_list): New decalaration.
28097 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28099 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28100 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28101 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28102 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28103 (operator==, comp_dr_with_seg_len_pair): Ditto.
28104 (vect_prune_runtime_alias_test_list): Ditto.
28106 2017-05-26 Martin Liska <mliska@suse.cz>
28109 * params.def: Bound partial-inlining-entry-probability param.
28111 2017-05-26 Marek Polacek <polacek@redhat.com>
28114 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28117 2017-05-26 Richard Biener <rguenther@suse.de>
28119 PR tree-optimization/80842
28120 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28123 2017-05-26 Richard Biener <rguenther@suse.de>
28125 PR tree-optimization/80844
28126 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28128 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28130 * doc/md.texi (Machine Constraints): Update x86 family
28131 machine constraints section to match 'config/i386/constraints.md'.
28133 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28135 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28137 2017-05-25 Nathan Sidwell <nathan@acm.org>
28139 * doc/invoke.texi (--enable-languages): Update documentation.
28141 2017-05-25 Martin Liska <mliska@suse.cz>
28143 * dumpfile.c: Add TDF_FOLDING.
28144 * dumpfile.h (enum dump_kind): Likewise.
28145 * genmatch.c (dt_simplify::gen_1): Use it.
28147 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28149 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28151 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28153 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28154 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28156 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28158 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28159 * match.pd (X == C): Rewrite it here.
28160 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28161 with_certain_nonzero_bits2): New predicates.
28162 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28164 2017-05-24 Nathan Sidwell <nathan@acm.org>
28166 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28169 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28172 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28174 * config/powerpcspe: New port. Files are copied from the rs6000
28175 port, with "rs6000" in filenames replaced by "powerpcspe".
28177 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28179 PR rtl-optimization/80754
28180 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28182 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28184 * config/sparc/sparc.md (length): Return the correct value for -mflat
28185 sibcalls to match output_sibcall.
28187 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28191 * config/rs6000/rs6000.c (struct machine_function): Add new field
28193 (rs6000_get_separate_components): Init that field, use it.
28194 (rs6000_components_for_bb): Use the field.
28196 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28198 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28200 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28202 PR middle-end/80823
28203 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28205 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28208 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28209 address against address_operand predicate.
28210 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28212 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28214 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28215 parameters passed indirectly.
28217 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28219 * config/i386/i386.md (*movdi_internal): Remove SSE4
28220 alternative 18 (?r, *v). Update insn attributes.
28221 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28222 Update insn attributes.
28223 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28224 Update insn attributes.
28225 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28226 alternative 1 (r, v). Remove isa attribute.
28227 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28228 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28229 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28231 2017-05-23 Tom de Vries <tom@codesourcery.com>
28233 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28236 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28238 * cgraphunit.c (symbol_table::process_new_functions): Update.
28239 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28240 (inline_generate_summary): Rename to ...
28241 (ipa_fn_summary_generate): ... this one.
28242 (inline_read_summary): Rename to ...
28243 (ipa_fn_summary_read): ... this one.
28244 (inline_write_summary): Rename to ...
28245 (ipa_fn_summary_write): ... this one.
28246 (inline_free_summary): Rename to ...
28247 (ipa_free_fn_summary): ... this one.
28248 (pass_data_local_fn_summary, pass_local_fn_summary,
28249 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28250 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28251 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28252 make_pass_ipa_fn_summary): New.
28253 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28254 inline_write_summary, inline_free_summary): Remove.
28255 (ipa_free_fn_summary) : New.
28256 * ipa-inline.c (ipa_inline): Update.
28257 (pass_ipa_inline): Do not generate summaries.
28258 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28260 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28261 and add pass_ipa_fn_summary.
28262 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28264 (make_pass_inline_parameters): Remove.
28266 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28268 * omp-low.c (struct omp_context): Remove "default_kind" member.
28271 * omp-offload.c (execute_oacc_device_lower): Remove the
28272 parallelism dimensions function attributes for unparallelized
28273 OpenACC kernels constructs.
28275 2017-05-23 Martin Liska <mliska@suse.cz>
28277 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28279 (cgraph_edge::make_speculative): Likewise.
28280 (cgraph_edge::resolve_speculation): Likewise.
28281 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28282 (cgraph_node::dump): Likewise.
28283 * cgraph.h: Likewise.
28284 * cgraphunit.c (analyze_functions): Likewise.
28285 (symbol_table::compile): Likewise.
28286 * ipa-cp.c (print_all_lattices): Likewise.
28287 (determine_versionability): Likewise.
28288 (initialize_node_lattices): Likewise.
28289 (ipcp_verify_propagated_values): Likewise.
28290 (estimate_local_effects): Likewise.
28291 (update_profiling_info): Likewise.
28292 (create_specialized_node): Likewise.
28293 (perhaps_add_new_callers): Likewise.
28294 (decide_about_value): Likewise.
28295 (decide_whether_version_node): Likewise.
28296 (identify_dead_nodes): Likewise.
28297 (ipcp_store_bits_results): Likewise.
28298 * ipa-devirt.c (dump_targets): Likewise.
28299 (ipa_devirt): Likewise.
28300 * ipa-icf.c (sem_item::dump): Likewise.
28301 (sem_function::equals): Likewise.
28302 (sem_variable::equals): Likewise.
28303 (sem_item_optimizer::read_section): Likewise.
28304 (sem_item_optimizer::execute): Likewise.
28305 (congruence_class::dump): Likewise.
28306 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28307 (dump_inline_summary): Likewise.
28308 (estimate_node_size_and_time): Likewise.
28309 (inline_analyze_function): Likewise.
28310 * ipa-inline-transform.c (inline_call): Likewise.
28311 * ipa-inline.c (report_inline_failed_reason): Likewise.
28312 (want_early_inline_function_p): Likewise.
28313 (edge_badness): Likewise.
28314 (update_edge_key): Likewise.
28315 (inline_small_functions): Likewise.
28316 * ipa-profile.c (ipa_profile): Likewise.
28317 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28318 (ipa_make_edge_direct_to_target): Likewise.
28319 (remove_described_reference): Likewise.
28320 (ipa_impossible_devirt_target): Likewise.
28321 (propagate_controlled_uses): Likewise.
28322 (ipa_print_node_params): Likewise.
28323 (ipcp_transform_function): Likewise.
28324 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28325 (propagate_pure_const): Likewise.
28326 * ipa-reference.c (generate_summary): Likewise.
28327 (read_write_all_from_decl): Likewise.
28328 (propagate): Likewise.
28329 (ipa_reference_read_optimization_summary): Likewise.
28330 * ipa-utils.c (ipa_merge_profiles): Likewise.
28331 * ipa.c (walk_polymorphic_call_targets): Likewise.
28332 (symbol_table::remove_unreachable_nodes): Likewise.
28333 (ipa_single_use): Likewise.
28334 * passes.c (execute_todo): Likewise.
28335 * predict.c (drop_profile): Likewise.
28336 * symtab.c (symtab_node::get_dump_name): New function.
28337 (symtab_node::dump_name): Likewise.
28338 (symtab_node::dump_asm_name): Likewise.
28339 (symtab_node::dump_references): Likewise.
28340 (symtab_node::dump_referring): Likewise.
28341 (symtab_node::dump_base): Likewise.
28342 (symtab_node::debug_symtab): Likewise.
28343 * tree-sra.c (convert_callers_for_node): Likewise.
28344 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28345 * value-prof.c (init_node_map): Likewise.
28347 2017-05-23 Martin Liska <mliska@suse.cz>
28349 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28350 and symtab_node::debug_symtab to symbol_table::debug.
28351 * cgraphunit.c (analyze_functions): Use the renamed function.
28352 (symbol_table::compile): Likewise.
28353 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28354 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28355 * passes.c (execute_todo): Likewise.
28356 * symtab.c (symbol_table::dump): New function.
28357 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28359 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28361 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28362 that nonconst implies exec.
28364 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28366 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28367 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28368 (inline_edge_summary_vec): Turn into ...
28369 (ipa_call_summaries): ... this one.
28370 (redirect_to_unreachable, edge_set_predicate,
28371 evaluate_properties_for_edge, inline_summary_alloc,
28372 reset_ipa_call_summary, reset_inline_summary,
28373 inline_summary_t::duplicate): Update.
28374 (inline_edge_duplication_hook): Turn to ...
28375 (ipa_call_summary_t::duplicate): ... this one.
28376 (inline_edge_removal_hook): Turn to ...
28377 (ipa_call_summary_t::remove): ... this one.
28378 (dump_inline_edge_summary): Turn to ...
28379 (dump_ipa_call_summary): ... this one.
28380 (estimate_function_body_sizes): Update.
28381 (inline_update_callee_summaries): Update.
28382 (remap_edge_change_prob): Update.
28383 (remap_edge_summaries): Update.
28384 (inline_merge_summary): Update.
28385 (do_estimate_edge_time): Update.
28386 (inline_generate_summary): Update.
28387 (inline_read_section): Update.
28388 (inline_read_summary): Update.
28389 (inline_free_summary): Update.
28390 * ipa-inline.c (can_inline_edge_p): Update.
28391 (compute_inlined_call_time): Update.
28392 (want_inline_small_function_p): Update.
28393 (edge_badness): Update.
28394 (early_inliner): Update.
28395 * ipa-inline.h (inline_edge_summary): Turn to ...
28396 (ipa_call_summary): ... this one.
28397 (ipa_call_summary_t): New class.
28398 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28399 (ipa_call_summaries): New.
28400 (inline_edge_summary): Remove.
28401 (estimate_edge_growth): Update.
28402 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28403 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28404 * ipa-split.c (execute_split_functions): Update.
28405 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28407 2017-05-23 Tom de Vries <tom@codesourcery.com>
28409 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28410 attributes): Document rdrand effective target.
28412 2017-05-23 Tom de Vries <tom@codesourcery.com>
28414 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28415 attributes): Sort alphabetically.
28417 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28419 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28421 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28424 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28425 V2DF/V2DI splat into two separate patterns, one that handles
28426 registers, and the other that only handles memory. Drop support
28427 for splatting from a GPR on ISA 2.07 and then splitting the
28428 splat into direct move and splat.
28429 (vsx_splat_<mode>_reg): Likewise.
28430 (vsx_splat_<mode>_mem): Likewise.
28432 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28434 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28436 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28438 PR middle-end/80809
28439 * omp-low.c (finish_taskreg_remap): New function.
28440 (finish_taskreg_scan): If unit size of ctx->record_type
28441 is non-constant, unshare the size expression and replace
28442 decls in it with possible outer var refs.
28444 PR middle-end/80809
28445 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28446 GOVD_SHARED rather than GOVD_PRIVATE with it.
28447 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28448 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28450 PR middle-end/80853
28451 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28452 as last argument to build_outer_var_ref for pointer bases of array
28453 section reductions.
28455 2017-05-19 Martin Sebor <msebor@redhat.com>
28457 * print-tree.c (print_node): Print DECL_READ_P flag.
28459 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28461 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28462 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28463 * cgraph.c: Likewise.
28464 * cgraphunit.c: Likewise.
28465 * gengtype.c: Likewise.
28466 * ipa-cp.c: Likewise.
28467 * ipa-devirt.c: Likewise.
28468 * ipa-icf.c: Likewise.
28469 * ipa-predicate.c: Likewise.
28470 * ipa-profile.c: Likewise.
28471 * ipa-prop.c: Likewise.
28472 * ipa-split.c: Likewise.
28474 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28475 edge_predicate_pool, dump_inline_hints,
28476 inline_summary::account_size_time, redirect_to_unreachable,
28477 edge_set_predicate, set_hint_predicate,
28478 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28479 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28480 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28481 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28482 ipa_call_summary_t::remove, initialize_growth_caches,
28483 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28484 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28485 mark_modified, unmodified_parm_1, unmodified_parm,
28486 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28487 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28488 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28489 will_be_nonconstant_predicate, record_modified_bb_info,
28490 get_minimal_bb, record_modified, param_change_prob,
28491 phi_result_unknown_predicate, predicate_for_phi_result,
28492 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28493 estimate_function_body_sizes, compute_inline_parameters,
28494 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28495 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28496 inline_update_callee_summaries, remap_edge_change_prob,
28497 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28498 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28499 inline_analyze_function, inline_summary_t::insert,
28500 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28501 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28502 inline_free_summary): Move to ipa-fnsummary.h
28503 (predicate_t): Remove.
28504 * ipa-fnsummary.c: New file.
28505 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28506 (enum inline_hints_vals, inline_hints, agg_position_info,
28507 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28508 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28509 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28510 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28511 inline_read_summary, inline_write_summary, inline_free_summary,
28512 inline_analyze_function, initialize_inline_failed,
28513 inline_merge_summary, inline_update_overall_summary,
28514 compute_inline_parameters): Move to ipa-fnsummary.h
28515 * ipa-fnsummary.h: New file.
28516 * ipa-inline-transform.h: Include ipa-inline.h.
28517 * ipa-inline.c: LIkewise.
28519 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28521 * ipa-inline.c (edge_badness): Use inlined_time instead of
28522 inline_summaries->get.
28524 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28526 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28528 2017-05-22 Nathan Sidwell <nathan@acm.org>
28530 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28531 (fdump-lang): Document 'raw' option.
28532 * dumpfile.h (TDI_tu): Delete.
28533 * dumpfile.c (dump_files): Remove translation-unit.
28534 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28536 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28538 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28539 command option from $(AWK) call.
28540 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28541 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28542 [FORMAT]: Remove handling of variable.
28543 * config/avr/t-multilib: Regenerate.
28545 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28547 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28549 (dump_inline_summary): Do not print self_time.
28550 (estimate_function_body_sizes): Do not set self_time.
28551 (compute_inline_parameters): Likewise.
28552 (inline_read_section, inline_write_summary): Do not stream self_time.
28553 * ipa-inline.h (inline_summary): Drop self_time.
28555 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28557 * ipa-inline-analysis.c (account_size_time): Rename to ...
28558 (inline_summary::account_size_time): ... this one.
28559 (reset_ipa_call_summary): Turn to ...
28560 (ipa_call_summary::reset): ... this one.
28561 (reset_inline_summary): Turn to ...
28562 (inline_summary::reset): ... this one.
28563 (inline_summary_t::remove): Update.
28564 (inline_summary_t::duplicate): Update.
28565 (ipa_call_summary_t::remove): Update.
28566 (dump_inline_summary): Update.
28567 (estimate_function_body_sizes): Update.
28568 (compute_inline_parameters): Update.
28569 (estimate_node_size_and_time): Update.
28570 (inline_merge_summary): Update.
28571 (inline_update_overall_summary): Update.
28572 (inline_read_section): Update.
28573 (inline_write_summary): Update.
28574 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28575 add account_size_time and reset member functions.
28576 (ipa_call_summary): Add reset function.
28577 * ipa-predicate.h (predicate::operator &): Constify.
28579 2017-05-22 Richard Biener <rguenther@suse.de>
28581 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28583 2017-05-19 Jason Merrill <jason@redhat.com>
28585 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28587 2017-05-19 Marek Polacek <polacek@redhat.com>
28590 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28591 TYPE_OVERFLOW_WRAPS checks.
28593 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28595 * tree-core.h (enum omp_clause_default_kind): Add
28596 "OMP_CLAUSE_DEFAULT_PRESENT".
28597 * tree-pretty-print.c (dump_omp_clause): Handle it.
28598 * gimplify.c (enum gimplify_omp_var_data): Add
28599 "GOVD_MAP_FORCE_PRESENT".
28600 (gimplify_adjust_omp_clauses_1): Map it to
28601 "GOMP_MAP_FORCE_PRESENT".
28602 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28604 * gimplify.c (oacc_default_clause): Clarify.
28606 2017-05-19 Nathan Sidwell <nathan@acm.org>
28608 LANG_HOOK_REGISTER_DUMPS
28609 * toplev.c (general_init): Call register dump lang hook.
28610 * doc/invoke.texi: Document -fdump-lang option family.
28611 * dumpfile.c (dump_files): Remove class dump here.
28612 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28613 * dumpfile.h (tree_dump_index): Remove TDI_class.
28614 * langhooks-def.h (lhd_register_dumps): Declare.
28615 (LANG_HOOKS_REGISTER_DUMPS): Define.
28616 (LANG_HOOKS_INITIALIZER): Add it.
28617 * langhooks.c (lhd_register_dumps): Define.
28618 * langhooks.h (struct lang_hooks): Add register_dumps.
28620 2017-05-19 Nathan Sidwell <nathan@acm.org>
28622 * context.h (context::set_passes): New.
28623 * context.c (context::context): Do not create pass manager.
28624 * toplev.c (general_init): Create pass manager here.
28626 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28628 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28629 use this splitter if two add or or instructions would also work for
28630 the constant we want to generate.
28632 2017-05-19 Richard Biener <rguenther@suse.de>
28635 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28636 predicate evaluation.
28638 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28640 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28642 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28645 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28647 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28648 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28649 (inline_edge_summary_vec): Turn into ...
28650 (ipa_call_summaries): ... this one.
28651 (redirect_to_unreachable, edge_set_predicate,
28652 evaluate_properties_for_edge, inline_summary_alloc,
28653 reset_ipa_call_summary, reset_inline_summary,
28654 inline_summary_t::duplicate): Update.
28655 (inline_edge_duplication_hook): Turn to ...
28656 (ipa_call_summary_t::duplicate): ... this one.
28657 (inline_edge_removal_hook): Turn to ...
28658 (ipa_call_summary_t::remove): ... this one.
28659 (dump_inline_edge_summary): Turn to ...
28660 (dump_ipa_call_summary): ... this one.
28661 (estimate_function_body_sizes): Update.
28662 (inline_update_callee_summaries): Update.
28663 (remap_edge_change_prob): Update.
28664 (remap_edge_summaries): Update.
28665 (inline_merge_summary): Update.
28666 (do_estimate_edge_time): Update.
28667 (inline_generate_summary): Update.
28668 (inline_read_section): Update.
28669 (inline_read_summary): Update.
28670 (inline_free_summary): Update.
28671 * ipa-inline.c (can_inline_edge_p): Update.
28672 (compute_inlined_call_time): Update.
28673 (want_inline_small_function_p): Update.
28674 (edge_badness): Update.
28675 (early_inliner): Update.
28676 * ipa-inline.h (inline_edge_summary): Turn to ...
28677 (ipa_call_summary): ... this one.
28678 (ipa_call_summary_t): New class.
28679 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28680 (ipa_call_summaries): New.
28681 (inline_edge_summary): Remove.
28682 (estimate_edge_growth): Update.
28683 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28684 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28685 * ipa-split.c (execute_split_functions): Update.
28686 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28688 2017-05-19 Richard Biener <rguenther@suse.de>
28690 PR middle-end/80764
28691 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28693 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28695 * config/rs6000/rs6000.c (struct machine_function): Add field
28696 fpr_is_wrapped_separately.
28697 (rs6000_get_separate_components): Use 64 components. Handle the
28698 new FPR components.
28699 (rs6000_components_for_bb): Handle the FPR components.
28700 (rs6000_emit_prologue_components): Handle the FPR components.
28701 (rs6000_emit_epilogue_components): Handle the FPR components.
28702 (rs6000_set_handled_components): Handle the FPR components.
28703 (rs6000_emit_prologue): Don't output prologue code for those FPRs
28704 that are already separately shrink-wrapped.
28705 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
28706 that are already separately shrink-wrapped.
28708 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28711 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
28714 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
28715 (define_peephole2 for Altivec d-form load): Add peepholes to catch
28716 cases where the register allocator uses a move and an offsettable
28717 memory operation to/from a FPR register on ISA 2.06/2.07.
28718 (define_peephole2 for Altivec d-form store): Likewise.
28720 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28723 * config/i386/mmx.md (*mov<mode>_internal): Enable
28724 alternatives 11, 12, 13 and 14 also for 32bit targets.
28725 Remove alternatives 15, 16, 17 and 18.
28726 * config/i386/sse.md (vec_concatv2di): Change
28727 alternative (!x, *y) to (x, ?!*Yn).
28729 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28731 * dumpfile.h (enum dump_kind): Remove stray comma.
28733 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28735 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28736 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28737 predicate::num_conditions
28738 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28739 (CHANGED): turn into predicate::changed.
28740 (agg_position_info): Move to ipa-predicate.h
28741 (add_condition, predicate::add_clause, predicate::operator &=,
28742 predicate::or_with, predicate::evaluate, predicate::probability,
28743 dump_condition, dump_clause, predicate::dump,
28744 predicate::remap_after_duplication, predicate::remap_after_inlining,
28745 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28746 (evaluate_conditions_for_known_args): Update.
28747 (set_cond_stmt_execution_predicate): Update.
28748 * ipa-inline.h: Include ipa-predicate.h
28749 (condition, inline_param_summary, conditions, agg_position_info,
28750 predicate): Move to ipa-predicate.h
28751 * ipa-predicate.c: New file.
28752 * ipa-predicate.h: New file.
28754 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28756 * final.c (leaf_function_p): Check we are not in a sequence.
28758 2017-05-18 Martin Liska <mliska@suse.cz>
28760 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28761 * dumpfile.c (dump_register): Use new enum dump_kind.
28762 (get_dump_file_name): Likewise.
28763 (dump_enable_all): Likewise.
28764 (dump_switch_p_1): Likewise.
28765 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28766 * dumpfile.h (enum dump_kind): New enum type.
28767 (struct dump_file_info): Create constructor and
28768 format fields and comments.
28769 * passes.c (pass_manager::register_one_dump_file):
28771 * statistics.c (statistics_early_init): Likewise.
28772 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28773 TDF_TREE with TDF_SLIM.
28774 (gather_memory_references_ref): Likewise.
28776 2017-05-18 Martin Liska <mliska@suse.cz>
28778 * vec.h (struct vnull): Use it.
28780 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28782 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28783 (true_predicate, false_predicate, true_predicate_p,
28784 false_predicate_p): Remove.
28785 (single_cond_predicate, not_inlined_predicate): Turn to member function
28787 (add_condition): Update.
28788 (add_clause): Turn to...
28789 (predicate::add_clause): ... this one; update; allow passing NULL
28791 (and_predicates): Turn to ...
28792 (predicate::operator &=): ... this one.
28793 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28794 (or_predicates): Turn to ...
28795 (predicate::or_with): ... this one.
28796 (evaluate_predicate): Turn to ...
28797 (predicate::evaluate): ... this one.
28798 (predicate_probability): Turn to ...
28799 (predicate::probability): ... this one.
28800 (dump_condition): Update.
28801 (dump_predicate): Turn to ...
28802 (predicate::dump): ... this one.
28803 (account_size_time): Update.
28804 (edge_set_predicate): Update.
28805 (set_hint_predicate): UPdate.
28806 (evaluate_conditions_for_known_args): Update.
28807 (evaluate_properties_for_edge): Update.
28808 (remap_predicate_after_duplication): Turn to...
28809 (predicate::remap_after_duplication): ... this one.
28810 (remap_hint_predicate_after_duplication): Update.
28811 (inline_summary_t::duplicate): UPdate.
28812 (dump_inline_edge_summary): Update.
28813 (dump_inline_summary): Update.
28814 (set_cond_stmt_execution_predicate): Update.
28815 (set_switch_stmt_execution_predicate): Update.
28816 (compute_bb_predicates): Update.
28817 (will_be_nonconstant_expr_predicate): Update.
28818 (will_be_nonconstant_predicate): Update.
28819 (phi_result_unknown_predicate): Update.
28820 (predicate_for_phi_result): Update.
28821 (array_index_predicate): Update.
28822 (estimate_function_body_sizes): Update.
28823 (estimate_node_size_and_time): Update.
28824 (estimate_ipcp_clone_size_and_time): Update.
28825 (remap_predicate): Rename to ...
28826 (predicate::remap_after_inlining): ... this one.
28827 (remap_hint_predicate): Update.
28828 (inline_merge_summary): Update.
28829 (inline_update_overall_summary): Update.
28830 (estimate_size_after_inlining): Update.
28831 (read_predicate): Rename to ...
28832 (predicate::stream_in): ... this one.
28833 (read_inline_edge_summary): Update.
28834 (write_predicate): Rename to ...
28835 (predicate::stream_out): ... this one.
28836 (write_inline_edge_summary): Update.
28837 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28838 (clause_t): Turn to uint32_t
28839 (predicate): Turn to class; implement constructor and operators
28841 (size_time_entry): Update.
28842 (inline_summary): Update.
28843 (inline_edge_summary): Update.
28845 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28847 * fold-const.c (fold_binary_loc): Move transformation...
28848 * match.pd (C - X CMP X): ... here.
28850 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28852 * config/sparc/sparc.c (sparc_option_override): Set function
28853 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28854 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28856 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28858 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28860 2017-05-18 Marek Polacek <polacek@redhat.com>
28863 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28864 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28865 gimple_assign_load_p.
28867 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28869 PR middle-end/80692
28870 * real.c (do_compare): Give decimal_do_compare preference over
28871 comparing just the signs.
28873 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28875 * doc/md.texi (Canonicalization of Instructions): Describe the
28876 canonical form of instructions that inherently set a condition
28879 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28881 PR middle-end/80775
28882 * tree-cfg.c: Move deletion of unreachable case statements to after
28883 the merging of consecutive case labels.
28885 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28887 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28888 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28889 restoring of callee-saved registers.
28891 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28893 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28894 * config/visium/visium.c (single_set_and_flags): Likewise.
28895 * config/visium/visium.md (Substitutions): Likewise.
28897 2017-05-17 Martin Liska <mliska@suse.cz>
28899 * cfg.c: Introduce dump_flags_t type and
28900 use it instead of int type.
28902 * cfghooks.c: Likewise.
28903 * cfghooks.h (struct cfg_hooks): Likewise.
28904 * cfgrtl.c: Likewise.
28905 * cfgrtl.h: Likewise.
28906 * cgraph.c (cgraph_node::get_body): Likewise.
28907 * coretypes.h: Likewise.
28908 * domwalk.c: Likewise.
28909 * domwalk.h: Likewise.
28910 * dumpfile.c (struct dump_option_value_info): Likewise.
28911 (dump_enable_all): Likewise.
28912 (dump_switch_p_1): Likewise.
28913 (opt_info_switch_p): Likewise.
28914 * dumpfile.h (enum tree_dump_index): Likewise.
28915 (struct dump_file_info): Likewise.
28916 * genemit.c: Likewise.
28917 * generic-match-head.c: Likewise.
28918 * gengtype.c (open_base_files): Likewise.
28919 * gimple-pretty-print.c: Likewise.
28920 * gimple-pretty-print.h: Likewise.
28921 * graph.c (print_graph_cfg): Likewise.
28922 * graphite-scop-detection.c (dot_all_sese): Likewise.
28923 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28924 * loop-unroll.c (report_unroll): Likewise.
28925 * passes.c (pass_manager::register_one_dump_file): Likewise.
28926 * print-tree.c: Likewise.
28927 * statistics.c: Likewise.
28928 * tree-cfg.c: Likewise.
28929 * tree-cfg.h: Likewise.
28930 * tree-dfa.c: Likewise.
28931 * tree-dfa.h: Likewise.
28932 * tree-dump.c (dump_function): Likewise.
28933 * tree-dump.h (struct dump_info): Likewise.
28934 * tree-pretty-print.c: Likewise.
28935 * tree-pretty-print.h: Likewise.
28936 * tree-ssa-live.c: Likewise.
28937 * tree-ssa-live.h: Likewise.
28938 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28939 * tree-vect-loop.c: Likewise.
28940 * tree-vect-slp.c: Likewise.
28942 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28943 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28945 PR tree-optimization/80457
28946 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28947 of all arguments to a statement as scalar_to_vec operations.
28948 (vectorizable_call): Adjust call to vect_model_simple_cost for
28950 (vectorizable_conversion): Likewise.
28951 (vectorizable_assignment): Likewise.
28952 (vectorizable_shift): Likewise.
28953 (vectorizable_operation): Likewise.
28954 (vectorizable_comparison): Likewise.
28955 (vect_is_simple_cond): Record the def types for operands.
28956 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28957 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28958 for statement argument count.
28960 2017-05-16 Carl Love <cel@us.ibm.com>
28962 * config/rs6000/rs6000-c: Add support for built-in functions
28963 vector unsigned long long vec_bperm (vector unsigned long long,
28964 vector unsigned char)
28965 vector signed long long vec_mule (vector signed int,
28967 vector unsigned long long vec_mule (vector unsigned int,
28968 vector unsigned int)
28969 vector signed long long vec_mulo (vector signed int,
28971 vector unsigned long long vec_mulo (vector unsigned int,
28972 vector unsigned int)
28973 vector signed char vec_sldw (vector signed char,
28974 vector signed char,
28976 vector unsigned char vec_sldw (vector unsigned char,
28977 vector unsigned char,
28979 vector signed short vec_sldw (vector signed short,
28980 vector signed short,
28982 vector unsigned short vec_sldw (vector unsigned short,
28983 vector unsigned short,
28985 vector signed int vec_sldw (vector signed int,
28988 vector unsigned int vec_sldw (vector unsigned int,
28989 vector unsigned int,
28991 vector signed long long vec_sldw (vector signed long long,
28992 vector signed long long,
28994 vector unsigned long long vec_sldw (vector unsigned long long,
28995 vector unsigned long long,
28997 * config/rs6000/rs6000-c: Add support for built-in functions
28998 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28999 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29000 * doc/extend.texi: Update the built-in documentation for the
29001 new built-in functions.
29003 2017-05-16 Marek Polacek <polacek@redhat.com>
29007 * tree.c (save_expr): Don't fold the expression.
29009 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29011 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29012 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29013 and (?*y,m). Update insn attributes.
29015 2017-05-16 Martin Liska <mliska@suse.cz>
29017 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29018 flags argument of print_gimple_stmt, print_gimple_expr,
29019 print_generic_stmt and print_generic_expr.
29020 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29021 * coretypes.h: Likewise.
29022 * except.c (dump_eh_tree): Likewise.
29023 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29024 * gimple-pretty-print.h: Likewise.
29025 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29026 (backprop::push_to_worklist): Likewise.
29027 (backprop::pop_from_worklist): Likewise.
29028 (backprop::process_use): Likewise.
29029 (backprop::intersect_uses): Likewise.
29030 (note_replacement): Likewise.
29031 * gimple-ssa-store-merging.c
29032 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29033 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29034 (pass_store_merging::execute): Likewise.
29035 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29036 (ssa_base_cand_dump_callback): Likewise.
29037 (dump_incr_vec): Likewise.
29038 (replace_refs): Likewise.
29039 (replace_mult_candidate): Likewise.
29040 (create_add_on_incoming_edge): Likewise.
29041 (create_phi_basis): Likewise.
29042 (insert_initializers): Likewise.
29043 (all_phi_incrs_profitable): Likewise.
29044 (introduce_cast_before_cand): Likewise.
29045 (replace_one_candidate): Likewise.
29046 * gimplify.c (gimplify_expr): Likewise.
29047 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29048 (set_rename): Likewise.
29049 (rename_uses): Likewise.
29050 (copy_loop_phi_nodes): Likewise.
29051 (add_close_phis_to_merge_points): Likewise.
29052 (copy_loop_close_phi_args): Likewise.
29053 (copy_cond_phi_args): Likewise.
29054 (graphite_copy_stmts_from_block): Likewise.
29055 (translate_pending_phi_nodes): Likewise.
29056 * graphite-poly.c (print_pdr): Likewise.
29057 (dump_gbb_cases): Likewise.
29058 (dump_gbb_conditions): Likewise.
29059 (print_scop_params): Likewise.
29060 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29061 (build_cross_bb_scalars_use): Likewise.
29062 (gather_bbs::before_dom_children): Likewise.
29063 * hsa-dump.c (dump_hsa_immed): Likewise.
29064 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29065 (get_replacement_map): Likewise.
29066 * ipa-inline-analysis.c (dump_condition): Likewise.
29067 (estimate_function_body_sizes): Likewise.
29068 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29069 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29070 * ipa-prop.c (ipa_dump_param): Likewise.
29071 (ipa_print_node_jump_functions_for_edge): Likewise.
29072 (ipa_modify_call_arguments): Likewise.
29073 (ipa_modify_expr): Likewise.
29074 (ipa_dump_param_adjustments): Likewise.
29075 (ipa_dump_agg_replacement_values): Likewise.
29076 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29077 * ipa-pure-const.c (check_stmt): Likewise.
29078 (pass_nothrow::execute): Likewise.
29079 * ipa-split.c (execute_split_functions): Likewise.
29080 * omp-offload.c (dump_oacc_loop_part): Likewise.
29081 (dump_oacc_loop): Likewise.
29082 * trans-mem.c (tm_log_emit): Likewise.
29083 (tm_memopt_accumulate_memops): Likewise.
29084 (dump_tm_memopt_set): Likewise.
29085 (dump_tm_memopt_transform): Likewise.
29086 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29087 (print_loop): Likewise.
29088 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29089 (chkp_gather_checks_info): Likewise.
29090 (chkp_get_check_result): Likewise.
29091 (chkp_remove_check_if_pass): Likewise.
29092 (chkp_use_outer_bounds_if_possible): Likewise.
29093 (chkp_reduce_bounds_lifetime): Likewise.
29094 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29095 (chkp_mark_completed_bounds): Likewise.
29096 (chkp_register_incomplete_bounds): Likewise.
29097 (chkp_mark_invalid_bounds): Likewise.
29098 (chkp_maybe_copy_and_register_bounds): Likewise.
29099 (chkp_build_returned_bound): Likewise.
29100 (chkp_get_bound_for_parm): Likewise.
29101 (chkp_build_bndldx): Likewise.
29102 (chkp_get_bounds_by_definition): Likewise.
29103 (chkp_generate_extern_var_bounds): Likewise.
29104 (chkp_get_bounds_for_decl_addr): Likewise.
29105 * tree-chrec.c (chrec_apply): Likewise.
29106 * tree-data-ref.c (dump_data_reference): Likewise.
29107 (dump_subscript): Likewise.
29108 (dump_data_dependence_relation): Likewise.
29109 (analyze_overlapping_iterations): Likewise.
29110 * tree-inline.c (expand_call_inline): Likewise.
29111 (tree_function_versioning): Likewise.
29112 * tree-into-ssa.c (dump_defs_stack): Likewise.
29113 (dump_currdefs): Likewise.
29114 (dump_names_replaced_by): Likewise.
29115 (dump_update_ssa): Likewise.
29116 (update_ssa): Likewise.
29117 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29118 * tree-parloops.c (build_new_reduction): Likewise.
29119 (try_create_reduction_list): Likewise.
29120 (ref_conflicts_with_region): Likewise.
29121 (oacc_entry_exit_ok_1): Likewise.
29122 (oacc_entry_exit_single_gang): Likewise.
29123 * tree-pretty-print.h: Likewise.
29124 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29125 (get_scalar_evolution): Likewise.
29126 (add_to_evolution): Likewise.
29127 (get_loop_exit_condition): Likewise.
29128 (analyze_evolution_in_loop): Likewise.
29129 (analyze_initial_condition): Likewise.
29130 (analyze_scalar_evolution): Likewise.
29131 (instantiate_scev): Likewise.
29132 (number_of_latch_executions): Likewise.
29133 (gather_chrec_stats): Likewise.
29134 (final_value_replacement_loop): Likewise.
29135 (scev_const_prop): Likewise.
29136 * tree-sra.c (dump_access): Likewise.
29137 (disqualify_candidate): Likewise.
29138 (create_access): Likewise.
29139 (reject): Likewise.
29140 (maybe_add_sra_candidate): Likewise.
29141 (create_access_replacement): Likewise.
29142 (analyze_access_subtree): Likewise.
29143 (analyze_all_variable_accesses): Likewise.
29144 (sra_modify_assign): Likewise.
29145 (initialize_constant_pool_replacements): Likewise.
29146 (find_param_candidates): Likewise.
29147 (decide_one_param_reduction): Likewise.
29148 (replace_removed_params_ssa_names): Likewise.
29149 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29150 * tree-ssa-copy.c (dump_copy_of): Likewise.
29151 (copy_prop_visit_cond_stmt): Likewise.
29152 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29153 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29154 (record_equivalences_from_stmt): Likewise.
29155 * tree-ssa-dse.c (compute_trims): Likewise.
29156 (delete_dead_call): Likewise.
29157 (delete_dead_assignment): Likewise.
29158 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29159 (forward_propagate_into_cond): Likewise.
29160 (pass_forwprop::execute): Likewise.
29161 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29162 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29164 (move_computations_worker): Likewise.
29165 (execute_sm): Likewise.
29166 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29167 (remove_exits_and_undefined_stmts): Likewise.
29168 (remove_redundant_iv_tests): Likewise.
29169 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29170 (adjust_iv_update_pos): Likewise.
29171 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29172 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29173 (value_replacement): Likewise.
29174 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29175 * tree-ssa-pre.c (print_pre_expr): Likewise.
29176 (get_representative_for): Likewise.
29177 (create_expression_by_pieces): Likewise.
29178 (insert_into_preds_of_block): Likewise.
29179 (eliminate_insert): Likewise.
29180 (eliminate_dom_walker::before_dom_children): Likewise.
29181 (eliminate): Likewise.
29182 (remove_dead_inserted_code): Likewise.
29183 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29184 * tree-ssa-reassoc.c (get_rank): Likewise.
29185 (eliminate_duplicate_pair): Likewise.
29186 (eliminate_plus_minus_pair): Likewise.
29187 (eliminate_not_pairs): Likewise.
29188 (undistribute_ops_list): Likewise.
29189 (eliminate_redundant_comparison): Likewise.
29190 (update_range_test): Likewise.
29191 (optimize_range_tests_var_bound): Likewise.
29192 (optimize_vec_cond_expr): Likewise.
29193 (rewrite_expr_tree): Likewise.
29194 (rewrite_expr_tree_parallel): Likewise.
29195 (linearize_expr): Likewise.
29196 (break_up_subtract): Likewise.
29197 (linearize_expr_tree): Likewise.
29198 (attempt_builtin_powi): Likewise.
29199 (attempt_builtin_copysign): Likewise.
29200 (transform_stmt_to_copy): Likewise.
29201 (transform_stmt_to_multiply): Likewise.
29202 (dump_ops_vector): Likewise.
29203 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29204 (print_scc): Likewise.
29205 (set_ssa_val_to): Likewise.
29206 (visit_reference_op_store): Likewise.
29207 (visit_use): Likewise.
29208 (sccvn_dom_walker::before_dom_children): Likewise.
29209 (run_scc_vn): Likewise.
29210 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29212 (expr_hash_elt::print): Likewise.
29213 (const_and_copies::pop_to_marker): Likewise.
29214 (const_and_copies::record_const_or_copy_raw): Likewise.
29215 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29216 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29217 (dump_predicates): Likewise.
29218 (find_uninit_use): Likewise.
29219 (warn_uninitialized_phi): Likewise.
29220 (pass_late_warn_uninitialized::execute): Likewise.
29221 * tree-ssa.c (verify_vssa): Likewise.
29222 (verify_ssa): Likewise.
29223 (maybe_optimize_var): Likewise.
29224 * tree-vrp.c (dump_value_range): Likewise.
29225 (dump_all_value_ranges): Likewise.
29226 (dump_asserts_for): Likewise.
29227 (register_edge_assert_for_2): Likewise.
29228 (vrp_visit_cond_stmt): Likewise.
29229 (vrp_visit_switch_stmt): Likewise.
29230 (vrp_visit_stmt): Likewise.
29231 (vrp_visit_phi_node): Likewise.
29232 (simplify_cond_using_ranges_1): Likewise.
29233 (fold_predicate_in): Likewise.
29234 (evrp_dom_walker::before_dom_children): Likewise.
29235 (evrp_dom_walker::push_value_range): Likewise.
29236 (evrp_dom_walker::pop_value_range): Likewise.
29237 (execute_early_vrp): Likewise.
29239 2017-05-16 Richard Biener <rguenther@suse.de>
29241 * dwarf2out.c (loc_list_from_tree_1): Do not create
29242 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29244 2017-05-16 Richard Biener <rguenther@suse.de>
29246 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29248 (note_variable_value_in_expr): If we resolved the decl ref
29249 do not push to the stack.
29251 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29253 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29254 operations in fast-math mode.
29255 (vaddq_f16): Likewise.
29256 (vmul_f16): Likewise.
29257 (vmulq_f16): Likewise.
29258 (vsub_f16): Likewise.
29259 (vsubq_f16): Likewise.
29260 * config/arm/neon.md (add<mode>3): New.
29262 (fma:<VH:mode>3): New. Also remove outdated comment.
29265 2017-05-16 Martin Liska <mliska@suse.cz>
29269 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29270 (odr_types_equivalent_p): Likewise.
29272 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29274 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29276 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29279 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29280 non-interunit SSE move alternatives with '?'.
29281 (zero-extendsidi peephole2): New peephole to skip intermediate
29282 general register in SSE zero-extend sequence.
29284 2017-05-15 Jeff Law <law@redhat.com>
29286 * reorg.c (relax_delay_slots): Create a new variable to hold
29287 the temporary target rather than clobbering TARGET_LABEL.
29289 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29290 missing argument to extract_bit_field call.
29291 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29293 2017-05-15 Martin Liska <mliska@suse.cz>
29296 * gcc.c (process_command): Do not allow empty argument of -o option.
29298 2017-05-15 Renlin Li <renlin.li@arm.com>
29300 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29301 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29302 * config/aarch64/constraints.md (Usf): Add long call check.
29303 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29304 (call_value): Likewise.
29305 (sibcall): Likewise.
29306 (sibcall_value): Likewise.
29308 (call_value_insn): New.
29309 (sibcall_insn): Update rtx pattern.
29310 (sibcall_value_insn): Likewise.
29311 (call_internal): Remove.
29312 (call_value_internal): Likewise.
29313 (sibcall_internal): Likewise.
29314 (sibcall_value_internal): Likewise.
29315 (call_reg): Likewise.
29316 (call_symbol): Likewise.
29317 (call_value_reg): Likewise.
29318 (call_value_symbol): Likewise.
29320 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29323 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29325 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29327 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29328 compatible with CCGOCmode and with CCZmode.
29330 2017-05-14 Martin Sebor <msebor@redhat.com>
29332 PR middle-end/77671
29333 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29334 (gimple_fold_builtin_snprintf): Same.
29335 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29336 (gimple_fold_builtin_snprintf): Same.
29337 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29338 of character types.
29339 (is_call_safe): New function.
29340 (try_substitute_return_value): Call it.
29341 (try_simplify_call): New function.
29342 (pass_sprintf_length::handle_gimple_call): Call it.
29344 2017-05-14 Martin Sebor <msebor@redhat.com>
29346 PR middle-end/80669
29347 * builtins.c (expand_builtin_stpncpy): Simplify.
29349 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29351 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29352 * config/i386/i386.h
29353 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29354 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29355 (struct machine_function): Add new members call_ms2sysv,
29356 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29357 (struct machine_frame_state): New fields sp_realigned and
29358 sp_realigned_offset.
29359 * config/i386/i386.c
29360 (enum xlogue_stub): New enum.
29361 (enum xlogue_stub_sets): New enum.
29362 (class xlogue_layout): New class.
29363 (struct ix86_frame): New fields stack_realign_allocate_offset,
29364 stack_realign_offset and outlined_save_offset. Modify comments to
29365 detail stack layout when using out-of-line stubs.
29366 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29367 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29368 -mcall-ms2sysv-xlogues.
29369 (stub_managed_regs): New static variable.
29370 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29371 registers managed by out-of-line stub.
29372 (disable_call_ms2sysv_xlogues): New function.
29373 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29374 m->call_ms2sysv when appropriate and compute frame layout for
29376 (sp_valid_at, fp_valid_at): New inline functions.
29377 (choose_basereg): New function.
29378 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29380 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29381 Use align parameter of choose_baseaddr to generated aligned SSE movs
29383 (pro_epilogue_adjust_stack): Modify to track
29384 machine_frame_state::sp_realigned.
29385 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29386 (ix86_nsaved_sseregs): Likewise.
29387 (ix86_emit_save_regs): Likewise.
29388 (ix86_emit_save_regs_using_mov): Likewise.
29389 (ix86_emit_save_sse_regs_using_mov): Likewise.
29390 (get_scratch_register_on_entry): Likewise.
29391 (gen_frame_set): New function.
29392 (gen_frame_load): Likewise.
29393 (gen_frame_store): Likewise.
29394 (emit_outlined_ms2sysv_save): Likewise.
29395 (emit_outlined_ms2sysv_restore): Likewise.
29396 (ix86_expand_prologue): Modify stack re-alignment code and call
29397 emit_outlined_ms2sysv_save when appropriate.
29398 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29399 parameter rtx_insn *insn, which allows the function to be used to only
29400 generate the notes.
29401 (ix86_expand_epilogue): Modify validity checks of frame and stack
29402 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29403 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29404 * config/i386/predicates.md
29405 (save_multiple): New predicate.
29406 (restore_multiple): Likewise.
29407 * config/i386/sse.md
29408 (save_multiple<mode>): New pattern.
29409 (save_multiple_realign<mode>): Likewise.
29410 (restore_multiple<mode>): Likewise.
29411 (restore_multiple_and_return<mode>): Likewise.
29412 (restore_multiple_leave_return<mode>): Likewise.
29413 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29415 2017-05-14 Julia Koval <julia.koval@intel.com>
29417 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29418 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29419 (__builtin_ia32_xsetbv): New builtins.
29420 * config/i386/i386.c (ix86_expand_special_args_builtin):
29422 (ix86_expand_builtin): Special expand for new intrinsics.
29423 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29424 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29425 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29427 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29429 * cfganal.c (inverted_post_order_compute): Change argument type
29431 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29432 * df-core.c (rest_of_handle_df_initialize): Adjust.
29433 (rest_of_handle_df_finish): Likewise.
29434 (df_analyze_1): Likewise.
29435 (df_analyze): Likewise.
29436 (loop_inverted_post_order_compute): Change argument to be a vec *.
29437 (df_analyze_loop): Adjust.
29438 (df_get_n_blocks): Likewise.
29439 (df_get_postorder): Likewise.
29440 * df.h (struct df_d): Change field to be a vec.
29441 * lcm.c (compute_laterin): Adjust.
29442 (compute_available): Likewise.
29443 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29444 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29445 * tree-ssa-pre.c (compute_antic): Likewise.
29447 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29449 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29450 (depth_first_search::depth_first_search): Change structure init
29451 function to this constructor.
29452 (depth_first_search::add_bb): Rename function to this member.
29453 (depth_first_search::execute): Likewise.
29454 (flow_dfs_compute_reverse_finish): Adjust.
29456 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29458 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29459 (longest_simple_path): Likewise.
29460 * shrink-wrap.c (spread_components): Likewise.
29461 (disqualify_problematic_components): Likewise.
29462 (emit_common_heads_for_components): Likewise.
29463 (emit_common_tails_for_components): Likewise.
29464 (insert_prologue_epilogue_for_components): Likewise.
29466 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29468 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29473 * df-core.c (df_set_blocks): Start using auto_bitmap.
29474 (df_compact_blocks): Likewise.
29475 * df-problems.c (df_rd_confluence_n): Likewise.
29476 * df-scan.c (df_insn_rescan_all): Likewise.
29477 (df_process_deferred_rescans): Likewise.
29478 (df_update_entry_block_defs): Likewise.
29479 (df_update_exit_block_uses): Likewise.
29480 (df_entry_block_bitmap_verify): Likewise.
29481 (df_exit_block_bitmap_verify): Likewise.
29482 (df_scan_verify): Likewise.
29483 * lra-constraints.c (lra_constraints): Likewise.
29484 (undo_optional_reloads): Likewise.
29485 (lra_undo_inheritance): Likewise.
29486 * lra-remat.c (calculate_gen_cands): Likewise.
29487 (do_remat): Likewise.
29488 * lra-spills.c (assign_spill_hard_regs): Likewise.
29489 (spill_pseudos): Likewise.
29490 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29491 (bitmap_set_subtract_values): Likewise.
29493 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29495 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29496 management with auto_bitmap.
29497 (fix_inter_tick): Likewise.
29498 (fix_recovery_deps): Likewise.
29499 * ira.c (add_store_equivs): Likewise.
29500 (find_moveable_pseudos): Likewise.
29501 (split_live_ranges_for_shrink_wrap): Likewise.
29502 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29503 (rtx_reuse_manager::seen_def_p): Likewise.
29504 (rtx_reuse_manager::set_seen_def): Likewise.
29505 * print-rtl.h (class rtx_reuse_manager): Likewise.
29507 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29509 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29511 (migrate_btr_def): Likewise.
29512 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29513 * df-core.c (loop_post_order_compute): Likewise.
29514 (loop_inverted_post_order_compute): Likewise.
29515 * hsa-common.h: Likewise.
29516 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29517 * init-regs.c (initialize_uninitialized_regs): Likewise.
29518 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29519 (inline_small_functions): Likewise.
29520 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29521 * ira.c (combine_and_move_insns): Likewise.
29522 (build_insn_chain): Likewise.
29523 * loop-invariant.c (find_invariants): Likewise.
29524 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29525 * predict.c (tree_predict_by_opcode): Likewise.
29526 (predict_paths_leading_to): Likewise.
29527 (predict_paths_leading_to_edge): Likewise.
29528 (estimate_loops_at_level): Likewise.
29529 (estimate_loops): Likewise.
29530 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29531 (spread_components): Likewise.
29532 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29533 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29534 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29535 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29536 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29537 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29538 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29539 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29540 (mark_threaded_blocks): Likewise.
29541 (thread_through_all_blocks): Likewise.
29542 * tree-ssa.c (verify_ssa): Likewise.
29543 (execute_update_addresses_taken): Likewise.
29544 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29546 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29548 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29550 (post_order_compute): Likewise.
29551 (inverted_post_order_compute): Likewise.
29552 (pre_and_rev_post_order_compute_fn): Likewise.
29554 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29556 * genrecog.c (int_set::int_set): Explicitly construct our
29557 auto_vec base class.
29558 * vec.h (auto_vec::auto_vec): New constructor.
29560 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29562 * bitmap.h (class auto_bitmap): New constructor taking
29563 bitmap_obstack * argument.
29565 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29567 * bitmap.h (class auto_bitmap): Change type of m_bits to
29568 bitmap_head, and adjust ctor / dtor and member operators.
29570 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29572 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29573 when returned register mode doesn't match original mode.
29575 2017-05-12 Jeff Law <law@redhat.com>
29576 Jakub Jelinek <jakub@redhat.com>
29578 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29579 we look for cc setter after the compare-elim changes.
29580 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29581 within the vector to match what compare-elim now expects.
29582 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29583 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29585 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29586 after the compare-elim changes.
29587 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29588 the vector to match what compare-elim now expects.
29589 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29590 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29591 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29592 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29593 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29595 * config/visium/visium.c (single_set_and_flags): Fix where
29596 we look for cc setter after the compare-elim changes.
29597 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29598 with the vector to match what compare-elim now expects.
29599 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29600 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29601 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29602 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29603 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29604 (neg<mode>2_insn_set_overflow): Likewise.
29606 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29608 PR middle-end/79794
29609 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29610 maybe_expand_insn call, set ops[0].target. If still set after call,
29611 set alt_rtl. Add extra arg to recursive calls.
29612 (extract_bit_field): Add alt_rtl argument. Pass to
29614 * expmed.h (extract_bit_field): Fix prototype.
29615 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29616 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29617 to extract_bit_field_calls.
29618 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29619 Pass alt_rtl to extract_bit_field calls.
29620 * calls.c (store_unaligned_arguments_into_psuedos)
29621 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29622 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29624 * optabs.h (struct expand_operand): Add target bitfield.
29626 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29628 * compare-elim.c (try_eliminate_compare): Canonicalize
29629 operation with embedded compare to
29630 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29631 (set (reg) (operation)].
29633 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29635 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29638 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29639 cost of adding a carry flag for ADC instruction.
29640 [case MINUS]: Ignore the cost of subtracting a carry flag
29641 for SBB instruction.
29643 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29645 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29647 * config/rs6000/bmiintrin.h: New file.
29648 * config/rs6000/bmi2intrin.h: New file.
29649 * config/rs6000/x86intrin.h: New file.
29651 2017-05-12 Jeff Law <law@redhat.com>
29653 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29656 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29658 PR middle-end/80707
29659 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29661 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29663 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29664 early expansion of vector divide builtins.
29665 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29666 builtins identified as having unsigned arguments.
29668 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29670 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29671 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29672 expansion of vector logical operations (and, andc, or, xor,
29675 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29677 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29678 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29680 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29683 early GIMPLE expansion of vector multiplies.
29685 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29687 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29688 TARGET_HAVE_MOVT conditional.
29689 (movt splitter): Likewise.
29691 2017-05-12 Richard Biener <rguenther@suse.de>
29693 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29694 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29695 Fold all stmts not inplace.
29697 2017-05-12 Richard Biener <rguenther@suse.de>
29699 PR tree-optimization/80713
29700 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
29701 inserted_exprs bit for not removed stmts.
29703 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
29705 PR middle-end/69921
29706 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
29707 parallelized" attribute for parallelized OpenACC kernels.
29708 * omp-offload.c (execute_oacc_device_lower): Use it.
29710 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
29711 Set "oacc kernels" attribute.
29712 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
29713 parameter. Adjust all users.
29714 (oacc_fn_attrib_kernels_p): Remove function.
29715 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
29716 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
29717 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29718 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29719 assert "oacc kernels" attribute is set.
29721 2017-05-11 Carl Love <cel@us.ibm.com>
29723 * config/rs6000/rs6000-c: Add support for built-in functions
29724 vector unsigned char vec_popcnt (vector signed char)
29725 vector unsigned char vec_popcnt (vector unsigned char)
29726 vector unsigned short vec_popcnt (vector signed short)
29727 vector unsigned short vec_popcnt (vector unsigned short)
29728 vector unsigned int vec_popcnt (vector signed int)
29729 vector unsigned int vec_popcnt (vector unsigned int)
29730 vector unsigned long long vec_popcnt (vector signed long long)
29731 vector unsigned long long vec_popcnt (vector unsigned long long)
29732 vector signed long long vec_slo (vector signed long long,
29733 vector signed char)
29734 vector signed long long vec_slo (vector signed long long,
29735 vector unsigned char)
29736 vector unsigned long long vec_slo (vector unsigned long long,
29737 vector signed char)
29738 vector unsigned long long vec_slo (vector unsigned long long,
29739 vector unsigned char)
29740 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29741 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29742 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29743 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29744 * doc/extend.texi: Update the built-in documentation file for the
29745 new built-in functions.
29747 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29749 * attribs.h (sorted_attr_string): Move machine independent
29750 functions for target clone support from the i386 port to common
29751 code. Rename ix86_function_versions to common_function_versions.
29752 Rename make_name to make_unique_name.
29753 (common_function_versions): Likewise.
29754 (make_unique_name): Likewise.
29755 (make_dispatcher_decl): Likewise.
29756 (is_function_default_version): Likewise.
29757 * attribs.c (attr_strcmp): Likewise.
29758 (sorted_attr_string): Likewise.
29759 (common_function_versions): Likewise.
29760 (make_unique_name): Likewise.
29761 (make_dispatcher_decl): Likewise.
29762 (is_function_default_version): Likewise.
29763 * config/i386/i386.c (attr_strcmp): Likewise.
29764 (sorted_attr_string): Likewise.
29765 (ix86_function_versions): Likewise.
29766 (make_name): Likewise.
29767 (make_dispatcher_decl): Likewise.
29768 (is_function_default_version): Likewise.
29769 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29771 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29774 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29775 Account for direct move costs for vec_construct of integer
29778 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29781 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29782 (UNSPEC_STX_ATOMIC): Ditto.
29783 (loaddi_via_sse): New insn.
29784 (storedi_via_sse): Ditto.
29785 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29786 Update corresponding peephole2 patterns.
29787 (atomic_storedi_fpu): Ditto.
29789 2017-05-11 Julia Koval <julia.koval@intel.com>
29791 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29792 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29794 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29795 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29796 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29798 2017-05-11 Nathan Sidwell <nathan@acm.org>
29800 * graphite-poly.c: Include dumpfile.h.
29802 * dumpfle.h (dump_function): Declare here ...
29803 * tree-dump.h (dump_function): ... not here.
29804 * dumpfile.c: #include tree-cfg.h.
29805 (dump_function): Move here from ...
29806 * tree-dump.c (dump_function): ... here.
29807 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29808 * graphite-poly.c: Don't include tree-dump.h.
29809 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29810 * print-tree.c: Likewise.
29811 * stor-layout.c: Likewise.
29812 * tree-nested.c: Likewise.
29814 * dumpfile.c (dump_start): Use TDF_FLAGS.
29815 (dump_enable_all): Fix TDF_KIND check thinko.
29817 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29819 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29820 array entries to represent two legal parameterizations of the
29821 overloaded __builtin_cmpb function, as represented by the
29822 P6_OV_BUILTIN_CMPB constant.
29823 (altivec_resolve_overloaded_builtin): Add special case handling
29824 for the __builtin_cmpb function, as represented by the
29825 P6_OV_BUILTIN_CMPB constant.
29826 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29827 (BU_P6_64BIT_2): New macro.
29828 (BU_P6_OVERLOAD_2): New macro
29829 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29830 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29831 (CMPB): Add overload support to represent both 32-bit and 64-bit
29832 compare-bytes function.
29833 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29834 support for TARGET_CMPB.
29835 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29836 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29837 documentation of the __builtin_cmpb overloaded built-in function.
29839 2017-05-11 Richard Biener <rguenther@suse.de>
29841 PR tree-optimization/80705
29842 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29843 bases are not vectorizable.
29845 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29847 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29848 when counting register pressure.
29850 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29852 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29853 (iv_ca_more_deps): Renamed to ...
29854 (iv_ca_compare_deps): ... this.
29855 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29857 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29859 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29861 (determine_group_iv_costs): ... here.
29862 (find_inv_vars_cb): Record inv var if it's not recorded before.
29864 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29866 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29867 (get_shiftadd_cost): Ditto.
29869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29871 * tree-ssa-address.c: Include header file.
29872 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29874 (add_to_parts): Refactor.
29875 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29876 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29881 PR tree-optimization/53090
29882 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29884 (extract_cond_operands): Detect condition with IV on both sides
29885 and return COMP_IV_EXPR_2.
29886 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29887 (rewrite_use_compare): Simplify by removing call to function
29888 extract_cond_operands.
29890 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29892 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29893 (extract_cond_operands): Detect condition comparing against non-
29894 invariant bound and return appropriate enum value.
29895 (find_interesting_uses_cond): Update use of extract_cond_operands.
29896 Handle its return value accordingly.
29897 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29899 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29901 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29902 nonlinear iv_use computation in loop invariant sensitive way.
29904 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29906 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29907 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29909 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29911 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29912 (dump_cand): Support iv_cand.inv_exprs.
29913 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29915 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29920 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29922 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29923 as local function. Include necessary header files.
29924 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29926 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29928 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29932 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29933 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29934 RSHIFT_EXPR and BIT_NOT_EXPR.
29936 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29938 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29939 (adjust_setup_cost): New parameter supporting round up adjustment.
29940 (struct address_cost_data): Delete.
29941 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29942 (split_address_cost, ptr_difference_cost): Delete.
29943 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29944 (struct ainc_cost_data): New struct.
29945 (get_address_cost_ainc): New function.
29946 (get_address_cost, get_computation_cost): Reimplement.
29947 (determine_group_iv_cost_address): Record inv_expr for all uses of
29949 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29950 (iv_ca_has_deps): Reimplemented to ...
29951 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29953 (iv_ca_extend): Call iv_ca_more_deps.
29955 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29957 * tree-ssa-address.c (struct mem_address): Move to header file.
29958 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29959 * tree-ssa-address.h (struct mem_address): Move from C file.
29960 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29962 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29964 * tree-affine.h (aff_combination_type): New interface.
29965 (aff_combination_zero_p): Remove static.
29966 (aff_combination_const_p): New interface.
29967 (aff_combination_singleton_var_p): New interfaces.
29969 2017-05-11 Richard Biener <rguenther@suse.de>
29971 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29972 Skip unreachable blocks and destinations.
29973 (eliminate): Move stmt removal and fixup ...
29974 (fini_eliminate): ... here. Skip inserted exprs.
29975 (pass_pre::execute): Move fini_pre after fini_eliminate.
29976 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29977 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29978 PRE to get rid of dead code that has invalid SSA form and
29979 split critical edges again.
29981 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29983 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29985 2017-05-11 Richard Biener <rguenther@suse.de>
29987 * passes.c (execute_function_todo): Verify loops if they are
29988 said to be up-to-date.
29989 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29990 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29992 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29995 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29996 handle calling assemble_external ourself.
29999 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30000 modes with zero size. Enhance comment.
30002 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30004 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30005 built-ins for vec_xl and vec_xst with short and char pointer
30008 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30010 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30011 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30012 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30013 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30014 (_mm_maskz_min_round_ss): New intrinsics.
30015 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30016 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30017 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30018 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30019 (__builtin_ia32_minss_mask_round): New builtins.
30020 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30021 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30022 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30024 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30025 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30027 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30030 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30032 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30033 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30034 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30035 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30036 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30037 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30038 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30039 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30040 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30041 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30042 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30043 (__builtin_ia32_mulss_mask_round): New builtins.
30044 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30045 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30046 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30048 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30049 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30051 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30054 2017-05-10 Julia Koval <julia.koval@intel.com>
30056 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30057 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30058 (_mm256_setr_m128i): New intrinsics.
30060 2017-05-10 Julia Koval <julia.koval@intel.com>
30062 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30063 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30064 (_mm_maskz_rcp14_ss): New intrinsics.
30065 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30066 (__builtin_ia32_rcp14ss_mask): New builtins.
30067 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30069 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30071 PR tree-optimization/51513
30072 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30073 (assert_unreachable_fallthru_edge_p): Use it.
30074 (group_case_labels_stmt): Likewise.
30075 * tree-cfg.h: Prototype it.
30076 * stmt.c: Include cfghooks.h and tree-cfg.h.
30077 (emit_case_dispatch_table) <gap_label>: New local variable.
30078 Use it to fill dispatch table gaps.
30079 Test for default_label before updating probabilities.
30080 (expand_case) <default_label>: Remove unneeded initialization.
30081 Test for unreachable default case statement and remove its edge.
30082 Set default_label accordingly.
30083 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30085 2017-05-10 Carl Love <cel@us.ibm.com>
30087 * config/rs6000/rs6000-c: Add support for built-in functions
30088 vector signed char vec_neg (vector signed char)
30089 vector signed short int vec_neg (vector short int)
30090 vector signed int vec_neg (vector signed int)
30091 vector signed long long vec_neg (vector signed long long)
30092 vector float vec_neg (vector float)
30093 vector double vec_neg (vector double)
30094 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30096 * config/rs6000/altivec.h: Add define for vec_neg
30097 * doc/extend.texi: Update the built-in documentation for the
30098 new built-in functions.
30100 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30102 PR tree-optimization/77644
30103 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30105 2017-05-10 Nathan Sidwell <nathan@acm.org>
30107 * dumpfile.h (TDI_lang_all): New.
30108 (TDF_KIND): New. Renumber others
30109 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30111 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30113 (get_dump_file_name): Adjust suffix generation.
30114 (dump_enable_all): Use TDF_KIND.
30115 * doc/invoke.texi (-fdump-lang-all): Document.
30117 * dumpfile.h: Tabify.
30119 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30122 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30123 Move member access before delete.
30125 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30127 * tree-inline.c (expand_call_inline): Split block at stmt
30130 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30133 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30134 are now unused after splitting mov{sf,sd}_hardfloat.
30135 (f32_lr2): Likewise.
30136 (f32_lm): Likewise.
30137 (f32_lm2): Likewise.
30138 (f32_li): Likewise.
30139 (f32_li2): Likewise.
30140 (f32_lv): Likewise.
30141 (f32_sr): Likewise.
30142 (f32_sr2): Likewise.
30143 (f32_sm): Likewise.
30144 (f32_sm2): Likewise.
30145 (f32_si): Likewise.
30146 (f32_si2): Likewise.
30147 (f32_sv): Likewise.
30148 (f32_dm): Likewise.
30149 (f32_vsx): Likewise.
30150 (f32_av): Likewise.
30151 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30152 For movsf, order stores so the VSX stores occur before the GPR
30153 store which encourages the register allocator to use a traditional
30154 FPR instead of a GPR. For movsd, order the stores so that the GPR
30155 store comes before the VSX stores to allow the power6 to work.
30156 This is due to the power6 not having a 32-bit integer store
30157 instruction from a FPR.
30158 (movsf_hardfloat): Likewise.
30159 (movsd_hardfloat): Likewise.
30161 2017-05-09 Martin Sebor <msebor@redhat.com>
30163 PR translation/80280
30164 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30167 PR translation/80280
30168 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30169 data member added in r247778.
30170 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30172 2017-05-09 Nathan Sidwell <nathan@acm.org>
30174 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30176 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30179 2017-05-09 Marek Polacek <polacek@redhat.com>
30181 * doc/invoke.texi: Fix typo.
30183 2017-05-09 Richard Biener <rguenther@suse.de>
30185 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30186 (vrp_val_is_min): Likewise.
30187 (set_value_range_to_value): Likewise.
30188 (set_value_range_to_nonnegative): Likewise.
30189 (gimple_assign_nonzero_p): Likewise.
30190 (gimple_stmt_nonzero_p): Likewise.
30191 (vrp_int_const_binop): Likewise. Remove unreachable case.
30192 (adjust_range_with_scev): Adjust comments.
30193 (compare_range_with_value): Likewise.
30194 (extract_range_from_phi_node): Likewise.
30195 (test_for_singularity): Likewise.
30197 2017-05-09 Richard Biener <rguenther@suse.de>
30199 * tree-vrp.c (get_single_symbol): Add assert that we don't
30200 get overflowed constants as invariant part.
30201 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30202 checks. Use wi::cmp instead of recursing for integer constants.
30203 (compare_values): Just ignore whether we assumed undefined
30204 overflow instead of failing the compare.
30205 (extract_range_for_var_from_comparison_expr): Add comment before the
30206 TREE_NO_WARNING sets.
30207 (test_for_singularity): Likewise.
30208 (extract_range_from_comparison): Do not disable optimization
30209 when we assumed undefined overflow.
30210 (extract_range_basic): Remove init of unused var.
30212 2017-05-09 Richard Biener <rguenther@suse.de>
30214 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30215 (extract_range_from_multiplicative_op_1): Adjust.
30216 (extract_range_from_binary_expr_1): Use int_const_binop.
30218 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30221 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30222 rs6000_store_data_bypass_p in seven define_bypass directives and
30223 in several comments.
30224 * config/rs6000/rs6000-protos.h: Add prototype for
30225 rs6000_store_data_bypass_p function.
30226 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30227 function implements slightly different (rs6000-specific) semantics
30228 than store_data_bypass_p, returning false rather than aborting
30229 with assertion error when arguments do not satisfy the
30230 requirements of store data bypass.
30231 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30232 rs6000_store_data_bypass_p.
30234 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30236 * config/xtensa/xtensa-protos.h
30237 (xtensa_initial_elimination_offset): New declaration.
30238 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30239 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30240 macro definition, add case for FRAME_POINTER_REGNUM when
30241 FRAME_GROWS_DOWNWARD.
30242 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30243 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30244 xtensa_initial_elimination_offset.
30246 2017-05-08 Nathan Sidwell <nathan@acm.org>
30248 * doc/invoke.texi: Alphabetize -fdump options.
30250 2017-05-08 Martin Sebor <msebor@redhat.com>
30252 PR translation/80280
30253 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30255 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30257 * target.def (compute_frame_layout): New optional target hook.
30258 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30259 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30260 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30262 * reload1.c (verify_initial_elim_offsets): Likewise.
30263 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30264 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30265 (arm_get_frame_offsets): Split up into this ...
30266 (arm_compute_frame_layout): ... and this function.
30268 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30270 * config/aarch64/constraints.md (Usa): New constraint.
30271 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30273 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30275 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30276 with_multilib_list after it has been checked.
30278 2017-05-08 Richard Biener <rguenther@suse.de>
30280 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30281 (bitmap_set_subtract_values): Likewise.
30283 2017-05-08 Richard Biener <rguenther@suse.de>
30285 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30286 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30288 (gimple_stmt_nonzero_warnv_p): Rename to ...
30289 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30291 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30292 (extract_range_basic): Adjust, do not disable propagation on
30293 strict overflow sensitive simplification.
30294 (vrp_visit_cond_stmt): Likewise.
30296 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30298 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30299 body size unconditionally.
30301 2017-05-07 Jeff Law <law@redhat.com>
30304 2017-05-06 Jeff Law <law@redhat.com>
30305 PR tree-optimization/78496
30306 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30309 PR tree-optimization/78496
30310 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30311 (simplify_stmt_using_ranges): Call it.
30312 (vrp_dom_walker::before_dom_children): Extract equivalences
30313 from an ASSERT_EXPR with an equality comparison against a
30316 2017-05-06 Jeff Law <law@redhat.com>
30318 PR tree-optimization/78496
30319 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30322 PR tree-optimization/78496
30323 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30324 (simplify_stmt_using_ranges): Call it.
30325 (vrp_dom_walker::before_dom_children): Extract equivalences
30326 from an ASSERT_EXPR with an equality comparison against a
30329 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30331 * lra-constraints.c (lra_copy_reg_equiv): New function.
30332 (split_reg): Use it to copy equivalence information from the
30333 original register to the spill register.
30335 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30337 PR rtl-optimization/75964
30338 * simplify-rtx.c (simplify_const_relational_operation): Remove
30339 invalid handling of comparisons of integer ABS.
30341 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30343 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30344 initialize to zero.
30345 (init_regs): Remove declaration.
30346 (function_arg_advance_32): Initialize error_p as boolean variable.
30348 2017-05-05 Nathan Sidwell <nathan@acm.org>
30350 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30351 lines. Use for (;;).
30353 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30355 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30356 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30357 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30358 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30359 VF=2 that require versioning.
30361 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30363 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30366 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30368 * diagnostic.h (diagnostic_override_option_index): Convert from
30369 macro to inline function.
30371 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30373 * diagnostic.c (last_module_changed_p): New function.
30374 (set_last_module): New function.
30375 (diagnostic_report_current_module): Convert macro usage to
30376 the above functions.
30377 * diagnostic.h (diagnostic_context::last_module): Strengthen
30378 from const line_map * to const line_map_ordinary *.
30379 (diagnostic_last_module_changed): Delete macro.
30380 (diagnostic_set_last_module): Delete macro.
30382 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30384 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30385 with diagnostic_report_diagnostic.
30386 (diagnostic_n_impl_richloc): Likewise.
30387 * diagnostic.h (report_diagnostic): Delete macro.
30388 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30389 with diagnostic_report_diagnostic.
30390 * substring-locations.c (format_warning_va): Likewise.
30392 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30394 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30395 save/restor of format_spec. Move option-printing code to...
30396 (print_option_information): ...this new function, and
30397 reimplement by simply printing to the pretty_printer,
30398 rather than appending to the format string.
30400 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30402 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30403 handling logic into...
30404 (update_effective_level_from_pragmas): ...this new function.
30406 2017-05-04 Andrew Waterman <andrew@sifive.com>
30408 * config/riscv/riscv.opt (mstrict-align): New option.
30409 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30410 (SLOW_UNALIGNED_ACCESS): Define.
30411 (riscv_slow_unaligned_access): Declare.
30412 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30414 (riscv_slow_unaligned_access): New variable.
30415 (rocket_tune_info): Set slow_unaligned_access to true.
30416 (optimize_size_tune_info): Set slow_unaligned_access to false.
30417 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30418 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30419 (riscv_option_override): Set riscv_slow_unaligned_access.
30420 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30422 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30424 * config/riscv/riscv.md: Unify indentation.
30426 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30431 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30433 (extendsi<mode>2): Add support for doing sign extension via
30434 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30435 don't have ISA 3.0 instructions.
30436 (extendsi<mode>2 splitter): Likewise.
30437 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30438 generate the normal insns since SImode can now go in vector
30439 registers. Disallow the special UNSPECs needed for previous
30440 machines to hide SImode being used. Add new insns
30441 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30442 (fix_trunc<mode>si2_stfiwx): Likewise.
30443 (fix_trunc<mode>si2_internal): Likewise.
30444 (fixuns_trunc<mode>si2): Likewise.
30445 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30446 (fctiw<u>z_<mode>_smallint): Likewise.
30447 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30448 of floating point to 32-bit integer from doing a direct move to
30449 the GPR registers to do a store.
30450 (fctiwz_<mode>): Break long line.
30452 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30454 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30455 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30456 (addr_list, addr_offset_valid_p): New.
30457 (split_address_groups): Check offset validity with above function.
30458 (gt-tree-ssa-loop-ivopts.h): Include header file.
30460 2017-05-05 Nathan Sidwell <nathan@acm.org>
30462 * config.gcc (arm*-*-*): Add missing 'fi'.
30464 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30466 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30467 included in -fopt-info does not matter.
30468 * doc/optinfo.texi (-fopt-info): Fix description of default
30469 behavour. Explicitly say order of options included in -fopt-info
30472 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30474 * config.gcc: Allow combinations of aprofile and rmprofile values for
30475 --with-multilib-list.
30476 * config/arm/t-multilib: New file.
30477 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30478 variables. Remove setting of ISA and floating-point ABI in
30479 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30480 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30481 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30482 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30484 * config/arm/t-rmprofile: Likewise except for the matches changes.
30485 * doc/install.texi (--with-multilib-list): Document the combination of
30486 aprofile and rmprofile values and warn about pitfalls in doing that.
30488 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30490 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30491 (movdi_aarch64): Likewise.
30493 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30495 PR tree-optimization/80632
30496 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30498 (build_arrays): Initialize it for virtual phis.
30499 (fix_phi_nodes): Use it for virtual phis.
30501 PR tree-optimization/80558
30502 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30503 [x, y] op z into [x op, y op z] for op & or | if conditions
30506 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30507 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30510 * config/arm/arm.md (use_literal_pool): Remove.
30511 (64-bit immediate split): No longer takes cost into consideration
30512 if arm_disable_literal_pool is enabled.
30513 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30514 used when arm_disable_literal_pool is enabled.
30515 (arm_max_const_double_inline_cost): Remove use of
30516 arm_disable_literal_pool.
30517 (push_minipool_fix): Add assert.
30518 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30519 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30520 (no_literal_pool_sf_immediate): New.
30522 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30524 PR tree-optimization/80613
30525 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30526 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30528 2017-05-05 Richard Biener <rguenther@suse.de>
30530 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30532 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30534 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30535 of this flag from insn conditions due to removal from r247495.
30537 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30539 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30541 (arm_early_store_addr_dep_ptr): Likewise.
30542 * config/arm/aarch-common-protos.h
30543 (arm_early_load_addr_dep_ptr): Add prototype.
30544 (arm_early_store_addr_dep_ptr): Likewise.
30545 * config/arm/cortex-a53.md: Add new bypasses.
30547 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30549 * tree.c (next_type_uid): Change type to unsigned.
30550 (type_hash_canon): Decrement back next_type_uid if
30551 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30552 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30555 2017-05-04 Martin Sebor <msebor@redhat.com>
30557 * builtins.c: Fix a trivial typo in a comment.
30559 PR middle-end/79234
30560 * builtins.c (check_sizes): Adjust to handle reading past the end.
30561 Avoid printing excessive upper bound of ranges. Use %E to print
30562 tree nodes instead of converting them to %wu.
30563 (expand_builtin_memchr): New function.
30564 (compute_dest_size): Rename...
30565 (compute_objsize): ...to this.
30566 (expand_builtin_memcpy): Adjust.
30567 (expand_builtin_mempcpy): Adjust.
30568 (expand_builtin_strcat): Adjust.
30569 (expand_builtin_strcpy): Adjust.
30570 (check_strncat_sizes): Adjust.
30571 (expand_builtin_strncat): Adjust.
30572 (expand_builtin_strncpy): Adjust and simplify.
30573 (expand_builtin_memset): Adjust.
30574 (expand_builtin_bzero): Adjust.
30575 (expand_builtin_memcmp): Adjust.
30576 (expand_builtin): Handle memcmp.
30577 (maybe_emit_chk_warning): Check strncat just once.
30579 2017-05-04 Martin Sebor <msebor@redhat.com>
30581 PR preprocessor/79214
30582 PR middle-end/79222
30583 PR middle-end/79223
30584 * builtins.c (check_sizes): Add inlining context and issue
30585 warnings even when -Wno-system-headers is set.
30586 (check_strncat_sizes): Same.
30587 (expand_builtin_strncat): Same.
30588 (expand_builtin_memmove): New function.
30589 (expand_builtin_stpncpy): Same.
30590 (expand_builtin): Handle memmove and stpncpy.
30592 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30594 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30595 which is not used any more.
30597 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30599 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30601 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30603 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30604 (cortexa53_tunings): Likewise.
30605 (cortexa57_tunings): Likewise.
30606 (cortexa72_tunings): Likewise.
30607 (cortexa73_tunings): Likewise.
30609 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30611 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30612 Set loop alignment to 8.
30614 2017-05-04 Martin Sebor <msebor@redhat.com>
30616 PR translation/80280
30617 * builtins.c (expand_builtin_object_size): Add missing quoting to
30618 %D and like directives.
30619 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30620 (hsa_type_for_tree_type): Same.
30621 (verify_function_arguments): Same.
30622 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30623 * varasm.c (get_section): Same.
30626 2017-05-04 Martin Sebor <msebor@redhat.com>
30628 PR translation/80280
30629 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30631 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30633 * config/aarch64/aarch64.c (generic_addrcost_table):
30634 Change HI/TI mode setting.
30636 2017-05-04 Martin Jambor <mjambor@suse.cz>
30638 PR tree-optimization/80622
30639 * tree-sra.c (comes_initialized_p): New function.
30640 (build_accesses_from_assign): Only set write lazily when
30641 comes_initialized_p is false.
30642 (analyze_access_subtree): Use comes_initialized_p.
30643 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30644 instead of testing for PARM_DECL.
30646 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30648 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30649 constraint on operand 0 to allow more general addressing modes.
30650 Adjust output template.
30651 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30653 * config/aarch64/aarch64-protos.h
30654 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30655 * config/aarch64/constraints.md (Dp): New address constraint.
30656 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30659 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30661 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30662 update use of estimate_ipcp_clone_size_and_time.
30663 (estimate_local_effects): Update use of
30664 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30665 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30666 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30667 Return nonspecialized time.
30669 2017-05-04 Richard Biener <rguenther@suse.de>
30671 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30672 for the last VUSE which def dominates the PHI. Directly call
30674 (get_continuation_for_phi_1): Remove.
30676 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30678 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30679 to explain the use of truncating division. Cap the number of
30680 iterations to the maximum given by nb_iterations_upper_bound,
30683 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30685 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30686 * configure: Regenerate.
30687 * config.in: Regenerate.
30688 * config/i386/driver-mingw32.c: new file.
30689 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30690 * config.host: Link driver-mingw32.o on MinGW host.
30691 * doc/install.texi: Document new --enable-mingw-wildcard configure
30694 2017-05-04 Marek Polacek <polacek@redhat.com>
30696 PR tree-optimization/80612
30697 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
30699 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30700 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
30702 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
30703 (movt splitter): Likewise.
30704 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
30705 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
30706 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
30707 block for Thumb-1 with MOVT.
30708 (thumb2_legitimate_address_p): Move code block ...
30709 (can_avoid_literal_pool_for_label_p): ... into this new function.
30710 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
30712 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
30713 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
30714 "M-profile targets with the MOVT instruction".
30716 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30718 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30719 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30720 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30722 2017-05-04 Martin Liska <mliska@suse.cz>
30724 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30725 variable cond_code.
30727 2017-05-04 Richard Biener <rguenther@suse.de>
30729 * tree.c (array_at_struct_end_p): Handle arrays at struct
30730 end with flexarrays more conservatively. Refactor and treat
30731 arrays of arrays or aggregates more strict. Fix
30732 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30733 * tree.c (array_at_struct_end_p): Adjust prototype.
30734 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30735 * gimple-fold.c (get_range_strlen): Likewise.
30736 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30738 2017-05-04 Richard Biener <rguenther@suse.de>
30740 PR tree-optimization/31130
30741 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30743 (supports_overflow_infinity): Likewise.
30744 (is_negative_overflow_infinity): Likewise.
30745 (is_positive_overflow_infinity): Likewise.
30746 (is_overflow_infinity): Likewise.
30747 (stmt_overflow_infinity): Likewise.
30748 (overflow_infinity_range_p): Likewise.
30749 (usable_range_p): Remove as always returning true.
30750 (make_overflow_infinity): Remove.
30751 (negative_overflow_infinity): Likewise.
30752 (positive_overflow_infinity): Likewise.
30753 (avoid_overflow_infinity): Likewise.
30754 (set_value_range): Adjust accordingly.
30755 (set_value_range_to_nonnegative): Likewise, remove now unused
30756 overflow_infinity arg.
30757 (vrp_operand_equal_p): Adjust.
30758 (update_value_range): Likewise.
30759 (range_int_cst_singleton_p): Likewise.
30760 (operand_less_p): Likewise.
30761 (compare_values_warnv): Likewise.
30762 (extract_range_for_var_from_comparison_expr): Likewise.
30763 (vrp_int_const_binop): Likewise.
30764 (zero_nonzero_bits_from_vr): Likewise.
30765 (extract_range_from_multiplicative_op_1): Likewise.
30766 (extract_range_from_binary_expr_1): Likewise.
30767 (extract_range_from_unary_expr): Likewise.
30768 (extract_range_from_comparison): Likewise.
30769 (extract_range_basic): Likewise.
30770 (adjust_range_with_scev): Likewise.
30771 (compare_ranges): Likewise.
30772 (compare_range_with_value): Likewise.
30773 (dump_value_range): Likewise.
30774 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30776 (simplify_cond_using_ranges): Adjust.
30778 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30780 * brig-builtins.def: Added a builtin for class_f64.
30781 * builtin-types.def: Added a builtin type needed by class_f64.
30783 2017-05-03 Jason Merrill <jason@redhat.com>
30785 * timevar.def: Add TV_CONSTEXPR.
30787 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30789 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30791 2017-05-03 Martin Jambor <mjambor@suse.cz>
30793 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30794 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30795 * ipa-cp.c (ipcp_propagate_stage): Do not call
30796 ipa_update_after_lto_read.
30797 * ipa-inline.c (ipa_inline): Likewise.
30799 2017-05-03 Martin Jambor <mjambor@suse.cz>
30801 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30802 tag. Added a default constructor and a destructor.
30803 (ipa_edge_args_sum_t): New class;
30804 (ipa_edge_args_sum): Declare.
30805 (ipa_edge_args_vector): Remove declaration.
30806 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30807 (ipa_free_edge_args_substructures): Remove declaration.
30808 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30809 (ipa_edge_args_info_available_for_edge_p): Likewise.
30810 * ipa-prop.c (ipa_edge_args_vector): Removed.
30811 (edge_removal_hook_holder): Likewise.
30812 (edge_duplication_hook_holder): Likewise.
30813 (ipa_edge_args_sum): New variable.
30814 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30815 ipa_edge_args_vector.
30816 (ipa_free_edge_args_substructures): Likewise.
30817 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30818 ipa_edge_args_vector.
30819 (ipa_edge_removal_hook): Turned into method
30820 ipa_edge_args_sum_t::remove.
30821 (ipa_edge_duplication_hook): Turned into method
30822 ipa_edge_args_sum_t::duplicate.
30823 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30824 registering edge hooks.
30825 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30826 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30827 ipa_edge_args_sum instead of ipa_edge_args_vector.
30828 * ipa-profile.c (ipa_profile): Likewise.
30830 2017-05-03 Martin Jambor <mjambor@suse.cz>
30832 * symbol-summary.h (function_summary): New method exists.
30833 (function_summary::symtab_removal): Deallocate through release.
30834 (call_summary): New class.
30835 (gt_ggc_mx): New overload.
30836 (gt_pch_nx): Likewise.
30837 (gt_pch_nx): Likewise.
30839 2017-05-03 Jeff Law <law@redhat.com>
30841 PR tree-optimization/78496
30842 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30843 from simplify_cond_using_ranges. Split off code to walk
30844 backwards through casts into ...
30845 (simplify_cond_using_ranges_2): New function.
30846 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30847 (execute_vrp): After identifying jump threads, call
30848 simplify_cond_using_ranges_2.
30850 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30853 * ipa-inline.h (inline_summary): Add ctor.
30854 (create_ggc): Do not use ggc_cleared_alloc.
30856 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30857 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30859 * gcc.c (handle_braces): Support escaping in switch matching
30861 * doc/invoke.texi (Spec Files): Document it.
30862 Remove superfluous @code markup in items.
30864 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30866 * diagnostic-show-locus.c (struct column_range): New struct.
30867 (get_affected_columns): New function.
30868 (get_printed_columns): New function.
30869 (struct correction): New struct.
30870 (correction::ensure_capacity): New function.
30871 (correction::ensure_terminated): New function.
30872 (struct line_corrections): New struct.
30873 (line_corrections::~line_corrections): New dtor.
30874 (line_corrections::add_hint): New function.
30875 (layout::print_trailing_fixits): Reimplement in terms of the new
30877 (selftest::test_overlapped_fixit_printing): New function.
30878 (selftest::diagnostic_show_locus_c_tests): Call it.
30880 2017-05-03 Nathan Sidwell <nathan@acm.org>
30882 Canonicalize canonical type hashing
30883 * tree.h (type_hash_canon_hash): Declare.
30884 * tree.c (type_hash_list, attribute_hash_list): Move into
30885 type_hash_canon_hash.
30886 (build_type_attribute_qual_variant): Break out hash code calc into
30887 type_hash_canon_hash.
30888 (type_hash_canon_hash): New. Generic type hash computation.
30889 (build_range_type_1, build_array_type_1, build_function_type,
30890 build_method_type_directly, build_offset_type, build_complex_type,
30891 make_vector_type): Call it.
30893 2017-05-03 Richard Biener <rguenther@suse.de>
30895 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30896 When all DRs have unknown misaligned do not always peel
30897 when there is a store but apply the same costing model as if
30898 there were only loads.
30900 2017-05-03 Richard Biener <rguenther@suse.de>
30903 PR tree-optimization/80492
30904 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30905 compare_base_decls returning dont-know properly.
30907 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30909 * config/arm/iterators.md (CCSI): New mode iterator.
30910 (arch): New mode attribute.
30911 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30912 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30913 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30914 code iterator for success result mode.
30915 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30916 the corresponding new insn generators.
30918 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30921 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30922 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30924 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30926 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30927 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30928 (DDR_A): Wrap DDR argument in brackets.
30929 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30930 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30931 (DDR_REVERSED_P): Likewise.
30933 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30935 PR tree-optimization/79472
30936 * tree-switch-conversion.c (struct switch_conv_info): Add
30937 contiguous_range and default_case_nonstandard fields.
30938 (collect_switch_conv_info): Compute contiguous_range and
30939 default_case_nonstandard fields, don't clear final_bb if
30940 contiguous_range and only the default case doesn't have the required
30942 (check_all_empty_except_final): Set default_case_nonstandard instead
30943 of failing if contiguous_range and the default case doesn't have empty
30945 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30946 and only the default case doesn't have the required constants. Skip
30948 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30949 if default_case_nonstandard.
30950 (build_constructors): Build constant 1 just once. Assert that default
30951 values aren't inserted in between cases if contiguous_range. Skip
30953 (build_arrays): Skip virtual phis.
30954 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30955 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30956 Handle virtual phis.
30957 (gen_inbound_check): Handle default_case_nonstandard case.
30958 (process_switch): Adjust check_final_bb caller. Call
30959 gather_default_values with the first non-default case instead of
30960 default case if default_case_nonstandard.
30962 2017-05-02 Nathan Sidwell <nathan@acm.org>
30964 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30965 check. Fix formatting.
30967 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30969 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30970 errors when comparing specialized and unspecialized times.
30972 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30974 * diagnostic-show-locus.c
30975 (layout::should_print_annotation_line_p): Make private.
30976 (layout::print_annotation_line): Make private.
30977 (layout::annotation_line_showed_range_p): Make private.
30978 (layout::show_ruler): Make private.
30979 (layout::print_source_line): Make private. Pass in line and
30980 line_width, rather than calling location_get_source_line. Drop
30982 (layout::print_leading_fixits): New method.
30983 (layout::print_any_fixits): Rename to...
30984 (layout::print_trailing_fixits): ...this, and make private.
30985 Don't print newline fixits.
30986 (diagnostic_show_locus): Move logic for printing one row into...
30987 (layout::print_line): ...this new function. Move the
30988 location_get_source_line call and error-handling from
30989 print_source_line to here. Call print_leading_fixits, and rename
30990 print_any_fixits to print_trailing_fixits.
30991 (selftest::test_fixit_insert_containing_newline): Update now that
30992 newlines are partially supported.
30993 (selftest::test_fixit_insert_containing_newline_2): New test.
30994 (selftest::test_fixit_replace_containing_newline): Update comments.
30995 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30996 * edit-context.c (class added_line): New class.
30997 (class edited_line): Describe newline handling in comment.
30998 (edited_line::actually_edited_p): New method.
30999 (edited_line::print_content): Delete redundant decl.
31000 (edited_line::m_predecessors): New field.
31001 (edited_file::print_content): Call edited_line::print_content.
31002 (edited_file::print_diff): Update to support newlines.
31003 (edited_file::print_diff_hunk): Likewise.
31004 (edited_file::print_run_of_changed_lines): New function.
31005 (edited_file::print_diff_line): Convert to...
31006 (print_diff_line): ...this.
31007 (edited_file::get_effective_line_count): New function.
31008 (edited_line::edited_line): Initialize new field m_predecessors.
31009 (edited_line::~edited_line): Clean up m_predecessors.
31010 (edited_line::apply_fixit): Handle newlines.
31011 (edited_line::get_effective_line_count): New function.
31012 (edited_line::print_content): New function.
31013 (edited_line::print_diff_lines): New function.
31014 (selftest::test_applying_fixits_insert_containing_newline): New
31016 (selftest::test_applying_fixits_replace_containing_newline): New
31018 (selftest::insert_line): New function.
31019 (selftest::test_applying_fixits_multiple_lines): Add example of
31021 (selftest::edit_context_c_tests): Call the new tests.
31023 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31025 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31026 parameter cand. Update dump information.
31027 (get_computation_cost): Update uses.
31029 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31031 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31032 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31033 (get_computation_at, rewrite_use_address): Update use of
31034 get_computation_aff.
31036 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31038 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31039 (get_computation): Delete.
31040 (get_computation_cost): Implement like get_computation_cost_at.
31041 Use get_computation_at.
31042 (get_computation_cost_at): Delete.
31043 (rewrite_use_nonlinear_expr): Use get_computation_at.
31044 (rewrite_use_compare, remove_unused_ivs): Ditto.
31046 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31048 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31050 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31052 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31053 (ivopts_global_cost_for_size): Rename parameter and update uses.
31054 (iv_ca_recount_cost): Update uses.
31055 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31056 candidates seperately in n_invs and n_cands.
31057 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31059 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31061 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31062 (find_inv_vars_cb): New.
31063 (find_depends): Renamed to ...
31064 (find_inv_vars): ... this.
31065 (add_candidate_1, force_var_cost): Call find_inv_vars.
31066 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31068 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31070 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31071 inv_vars. Add inv_exprs.
31072 (struct iv_cand): Rename depends_on to inv_vars.
31073 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31074 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31075 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31076 (dump_cand): Dump inv_vars.
31077 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31078 (record_invariant, find_depends, add_candidate_1): Ditto.
31079 (set_group_iv_cost, force_var_cost): Ditto.
31080 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31081 (get_computation_cost_at, get_computation_cost): Ditto.
31082 (determine_group_iv_cost_generic): Ditto.
31083 (determine_group_iv_cost_address): Ditto.
31084 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31085 (determine_group_iv_costs): Ditto.
31086 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31087 (iv_ca_set_remove_invariants): Renamed to ...
31088 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31089 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31090 (iv_ca_set_add_invariants): Renamed to ...
31091 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31092 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31093 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31094 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31095 (create_new_ivs): Remove useless dump.
31097 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31099 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31101 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31102 (iv_ca_set_no_cp, create_new_iv): Ditto.
31104 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31106 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31108 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31110 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31111 function tree_check2.
31113 2017-05-02 Martin Liska <mliska@suse.cz>
31115 * doc/gcov.texi: Add missing preposition.
31116 * gcov.c (function_info::function_info): Properly fill up
31117 all member variables.
31119 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31121 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31123 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31125 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31127 2017-05-02 Martin Liska <mliska@suse.cz>
31130 * lto-streamer-in.c (lto_read_tree_1): Remove
31131 LTO_STREAMER_DEBUG.
31132 * lto-streamer.c (struct tree_hash_entry): Likewise.
31133 (struct tree_entry_hasher): Likewise.
31134 (tree_entry_hasher::hash): Likewise.
31135 (tree_entry_hasher::equal): Likewise.
31136 (lto_streamer_init): Likewise.
31137 (lto_orig_address_map): Likewise.
31138 (lto_orig_address_get): Likewise.
31139 (lto_orig_address_remove): Likewise.
31140 * lto-streamer.h: Likewise.
31141 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31142 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31144 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31146 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31147 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31148 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31149 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31150 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31151 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31152 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31153 (mm_maskz_sub_ss): New intrinsics.
31154 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31155 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31156 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31157 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31158 (__builtin_ia32_subss_mask_round): New builtins.
31159 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31160 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31161 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31163 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31164 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31166 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31169 2017-05-02 Martin Jambor <mjambor@suse.cz>
31171 PR tree-optimization/78687
31172 * tree-sra.c (access): New field parent.
31173 (process_subtree_disqualification): New function.
31174 (disqualify_candidate): Call it.
31175 (build_accesses_from_assign): Reset write flag if creating an
31177 (build_access_subtree): Fill in parent field and also prpagate
31178 down grp_write flag.
31179 (create_artificial_child_access): New parameter set_grp_write, set
31180 grp_write to its value.
31181 (propagate_subaccesses_across_link): Also propagate grp_write flag
31183 (propagate_all_subaccesses): Push the closest parent back to work
31184 queue if add_access_to_work_queue returned true.
31186 2017-05-02 Richard Biener <rguenther@suse.de>
31188 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31189 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31190 -fstrict-overflow documentation.
31191 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31192 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31193 flag_strict_overflow.
31194 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31195 * lto-opts.c (lto_write_options): Do not stream it.
31196 * lto-wrapper.c (merge_and_complain): Do not handle it.
31197 * opts.c (default_options_table): Do not set -fstrict-overflow.
31198 (finish_options): Likewise do not clear it when sanitizing.
31199 * simplify-rtx.c (simplify_const_relational_operation): Do not
31200 test flag_strict_overflow.
31202 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31204 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31205 using enabled attribute.
31206 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31207 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31208 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31209 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31210 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31211 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31212 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31213 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31214 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31215 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31217 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31219 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31221 2017-05-02 Richard Biener <rguenther@suse.de>
31223 PR tree-optimization/80591
31225 2017-04-10 Richard Biener <rguenther@suse.de>
31227 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31230 2017-05-02 Richard Biener <rguenther@suse.de>
31232 PR tree-optimization/80549
31233 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31234 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31235 headers if they do not have a preheader.
31237 2017-05-02 Martin Liska <mliska@suse.cz>
31240 * common.opt: Fix typo.
31241 * doc/invoke.texi: Likewise.
31243 2017-05-01 Jan Beulich <jbeulich@suse.com>
31245 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31246 swapping, add (x,x,m,x,n) alternative.
31248 2017-05-01 Nathan Sidwell <nathan@acm.org>
31250 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31251 unnecessary unadjusted_alignment check.
31253 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31256 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31258 * gimplify.c (gimplify_cilk_detach): New function.
31259 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31260 * tree-core.h: Document EXPR_CILK_SPAWN.
31261 * tree.h (EXPR_CILK_SPAWN): Define.
31263 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31265 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31266 to use new fixit_hint representation, using the "replace" logic.
31267 (get_line_span_for_fixit_hint): Likewise.
31268 (layout::print_any_fixits): Likewise.
31269 (selftest::test_one_liner_many_fixits): Rename to...
31270 (selftest::test_one_liner_many_fixits_1): ...this, and update
31271 comment and expected output to reflect that the multiple fix-it
31272 hints are now consolidated into one insertion.
31273 (selftest::test_one_liner_many_fixits_2): New test.
31274 (selftest::test_diagnostic_show_locus_one_liner): Update for
31276 (selftest::test_fixit_consolidation): Update for fix-it API
31278 * diagnostic.c (print_parseable_fixits): Likewise.
31279 * edit-context.c (edited_line::m_line_events): Convert from
31280 auto_vec <line_event *> to auto_vec <line_event>.
31281 (class line_event): Convert from abstract base class to a concrete
31282 class, taking over the role of replace_event.
31283 (class insert_event): Delete.
31284 (class replace_event): Rename to class line_event. Convert to
31286 (edit_context::add_fixits): Reimplement.
31287 (edit_context::apply_insert): Delete.
31288 (edit_context::apply_replace): Rename to...
31289 (edit_context::apply_fixit): ...this. Convert to half-open range.
31290 (edited_file::apply_insert): Delete.
31291 (edited_file::apply_replace): Rename to...
31292 (edited_file::apply_fixit): ...this.
31293 (edited_line::~edited_line): Drop deletion of events.
31294 (edited_line::apply_insert): Delete.
31295 (edited_line::apply_replace): Rename to...
31296 (edited_line::apply_fixit): ...this. Convert to half-open range.
31297 Update for change to type of m_line_events.
31298 * edit-context.h (edit_context::apply_insert): Delete.
31299 (edit_context::apply_replace): Rename to...
31300 (edit_context::apply_fixit): ...this.
31302 2017-05-01 Martin Sebor <msebor@redhat.com>
31304 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31307 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31310 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31311 __get_cpuid_max returns 0.
31312 (__get_cpuid_count): Ditto.
31314 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31316 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31317 replacement expression is another instance of one of its arguments.
31319 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31322 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31323 check for stack push/pop autoinc.
31324 * config/i386/i386.c (ix86_agi_dependent): Return false
31325 if the only reason why modified_in_p returned true is that
31326 addr is SP based and set_insn is a push or pop.
31328 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31330 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31333 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31336 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31337 (account_size_time): Use two predicates - exec_pred and
31339 (evaluate_conditions_for_known_args): Compute both clause and
31341 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31342 (inline_summary_t::duplicate): Update.
31343 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31345 (compute_inline_parameters): Likewise.
31346 (estimate_edge_size_and_time): Update caluclation of time.
31347 (estimate_node_size_and_time): Compute both time and nonspecialized
31349 (estimate_ipcp_clone_size_and_time): Update.
31350 (inline_merge_summary): Update.
31351 (do_estimate_edge_time): Update.
31352 (do_estimate_edge_size): Update.
31353 (do_estimate_edge_hints): Update.
31354 (inline_read_section, inline_write_summary): Stream both new predicates.
31355 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31357 (compute_inlined_call_time): Cleanup.
31358 (big_speedup_p): Update.
31359 (edge_badness): Update.
31360 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31361 (size_time_entry): Replace predicate by exec_predicate and
31362 nonconst_predicate.
31363 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31364 (estimate_edge_time): Return also nonspec_time.
31365 (reset_edge_growth_cache): Update.
31367 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31369 PR rtl-optimization/80491
31370 * ifcvt.c (noce_process_if_block): When looking for x setter
31371 with missing else_bb, don't check only the insn right before
31372 cond_earliest, but look for the last insn that x is modified in
31373 within the same bb.
31375 PR rtl-optimization/80491
31376 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31378 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31380 PR tree-optimization/80487
31381 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31383 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31385 PR tree-optimization/79697
31386 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31387 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31388 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31390 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31391 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31393 2017-04-28 Martin Sebor <msebor@redhat.com>
31395 PR tree-optimization/80523
31396 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31397 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31399 (maybe_warn, format_directive, parse_directive): Use new functions.
31400 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31402 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31404 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31406 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31408 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31409 target_header_dir): Set correctly.
31410 * configure: Regenerated.
31411 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31412 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31413 instead of SYSTEM_HEADER_DIR.
31415 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31417 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31418 (estimate_local_effects): Likewise.
31419 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31420 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31421 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31422 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31423 do_estimate_edge_time, estimate_edge_time): Likewise.
31424 * ipa-inline-analysis.c (estimate_node_size_and_time,
31425 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31426 (estimate_time_after_inlining): Remove.
31428 2017-04-28 Martin Liska <mliska@suse.cz>
31430 * doc/gcov.texi: Enhance documentation of gcov.
31432 2017-04-28 Martin Liska <mliska@suse.cz>
31434 * doc/gcov.texi: Sort options in alphabetic order.
31435 * doc/gcov-dump.texi: Likewise.
31436 * doc/gcov-tool.texi: Likewise.
31437 * gcov.c (print_usage): Likewise.
31438 * gcov-dump.c (print_usage): Likewise.
31439 * gcov-tool.c (print_merge_usage_message): Likewise.
31440 (print_rewrite_usage_message): Likewise.
31441 (print_overlap_usage_message): Likewise.
31443 2017-04-28 Martin Liska <mliska@suse.cz>
31445 PR gcov-profile/53915
31446 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31448 2017-04-28 Martin Liska <mliska@suse.cz>
31450 PR gcov-profile/79891
31451 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31452 is marked by compiler as living on a line.
31453 (get_cycles_count): Remove usage of the union.
31454 (output_intermediate_file): Likewise.
31455 (find_source): Fix GNU coding style.
31456 (accumulate_line_counts): Remove old non-all block mode.
31457 (output_lines): Remove usage of the union.
31458 * profile.c (output_location): Include all BBs, even if
31459 belonging to a same line (and file) as a previous BB.
31461 2017-04-28 Martin Liska <mliska@suse.cz>
31463 * gcov.c (process_args): Handle new argument 'w'.
31464 (read_graph_file): Assign ID to BBs.
31465 (output_branch_count): Display BB # if verbose flag is set.
31466 (output_lines): Likewise for arcs.
31467 (print_usage): Add '--verbose' option help.
31468 * doc/gcov.texi: Document --verbose (-w) option.
31470 2017-04-28 Martin Liska <mliska@suse.cz>
31472 * gcov.c (struct block_location_info): New struct.
31473 (process_file): Fill up the new structure.
31474 (read_graph_file): Replace usage of encoding by the newly added
31476 (add_line_counts): Likewise.
31477 (accumulate_line_counts): Remove usage of the union.
31478 (function_info::function_info): New function.
31479 (function_info::~function_info): Likewise.
31480 (process_file): Call delete instead of release_function.
31481 (release_function): Release the function.
31482 (release_structures): Call delete instead of release_function.
31483 (solve_flow_graph): Replace usage of num_blocks.
31484 (find_exception_blocks): Likewise.
31485 (output_lines): Fix GNU coding style.
31487 2017-04-28 Martin Liska <mliska@suse.cz>
31490 * coverage.c (coverage_remove_note_file): New function.
31491 * coverage.h: Declare the function.
31492 * toplev.c (finalize): Clean if an error has been seen.
31494 2017-04-28 Martin Liska <mliska@suse.cz>
31496 PR gcov-profile/80031
31497 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31498 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31499 * gcov.c (read_graph_file): Read just number of blocks.
31500 * profile.c (branch_prob): Do not stream 0 flags per a basic
31503 2017-04-28 Martin Liska <mliska@suse.cz>
31505 * gcov-dump.c (tag_*): Add new argument to declarations.
31506 (dump_gcov_file): Likewise.
31507 (tag_blocks): Add and use new argument depth.
31508 (tag_arcs): Likewise.
31509 (tag_lines): Likewise.
31510 (tag_counters): Likewise.
31511 (tag_summary): Likewise.
31512 (dump_working_sets): Use depth to do a proper indentation.
31514 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31517 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31518 * symtab.c (symtab_node::debug_symtab): Move definition here.
31520 2017-04-28 Richard Biener <rguenther@suse.de>
31522 * lto-streamer.h (LTO_major_version): Bump to 7.
31524 2017-04-28 Richard Biener <rguenther@suse.de>
31526 * tree-vrp.c (assert_info): New struct.
31527 (add_assert_info): New helper.
31528 (register_edge_assert_for_2): Refactor to add asserts to a vector
31530 (register_edge_assert_for_1): Likewise.
31531 (register_edge_assert_for): Likewise.
31532 (finish_register_edge_assert_for): New helper actually registering
31533 asserts where live on edge.
31534 (find_conditional_asserts): Adjust.
31535 (find_switch_asserts): Likewise.
31536 (evrp_dom_walker::try_find_new_range): Generalize.
31537 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31539 2017-04-27 Marek Polacek <polacek@redhat.com>
31542 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31543 arg10 and arg11 to itype.
31545 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31547 * doc/extend.texi (Object Size Checking): Improve grammar.
31549 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31552 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31553 that the logic for permitting reciprocal estimates matches that
31556 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31559 * tree.c (type_cache_hasher::equal): Only compare
31560 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31561 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31562 non-aggregate element types.
31563 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31564 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31566 2017-04-27 Richard Biener <rguenther@suse.de>
31568 PR middle-end/80533
31569 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31570 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31571 keeping a reference to a trailing array.
31573 2017-04-27 Richard Biener <rguenther@suse.de>
31575 PR middle-end/80539
31576 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31577 being in loop-closed SSA form conservatively.
31578 (chrec_fold_multiply_poly_poly): Likewise.
31580 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31582 PR middle-end/79665
31583 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31584 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31586 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31589 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31590 (aarch64_function_arg_alignment): Return unsigned int again, but still
31591 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31592 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31593 Don't emit -Wpsabi note.
31594 (aarch64_function_arg_boundary): Likewise.
31595 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31598 2017-04-26 Nathan Sidwell <nathan@acm.org>
31600 * tree.h (crc32_unsigned_n): Declare.
31601 (crc32_unsigned, crc32_unsigned): Make inline.
31602 * tree.c (crc32_unsigned_bits): Replace with ...
31603 (crc32_unsigned_n): ... this.
31604 (crc32_unsigned, crc32_byte): Remove.
31605 (crc32_string): Remove unnecessary braces.
31607 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31609 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31610 * ipa-inline-analysis.c (MAX_TIME): Remove.
31611 (account_size_time): Use sreal for time.
31612 (dump_inline_summary): Update.
31613 (estimate_function_body_sizes): Update.
31614 (estimate_edge_size_and_time): Update.
31615 (estimate_calls_size_and_time): Update.
31616 (estimate_node_size_and_time): Update.
31617 (inline_merge_summary): Update.
31618 (inline_update_overall_summary): Update.
31619 (estimate_time_after_inlining): Update.
31620 (inline_read_section): Update.
31621 (inline_write_summary): Update.
31622 * ipa-inline.c (compute_uninlined_call_time): Update.
31623 (compute_inlined_call_time): Update.
31624 (recursive_inlining): Update.
31625 (inline_small_functions): Update.
31626 (dump_overall_stats): Update.
31627 * ipa-inline.h: Include sreal.h.
31628 (size_time_entry): Turn time to sreal.
31629 (inline_summary): Turn self_time nad time to sreal.
31631 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31633 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31635 (sreal::stream_out, sreal::stream_in): New.
31636 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31638 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31640 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31643 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31646 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31648 (dimode_scalar_chain::compute_convert_gain): Ditto.
31649 (dimode_scalar_chain::make_vector_copies): Ditto.
31650 (dimode_scalar_chain::convert_reg): Ditto.
31651 (dimode_scalar_chain::convert_insn): Ditto.
31652 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31653 (VI248_AVX512BW_1): New mode iterator.
31654 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31655 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31658 2017-04-25 Martin Sebor <msebor@redhat.com>
31660 PR tree-optimization/80497
31661 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31662 constants are representable in HOST_WIDE_INT.
31663 (parse_directive): Ditto.
31665 2017-04-25 Martin Sebor <msebor@redhat.com>
31668 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31669 (new_zero_array): Adjust signature.
31670 (dom_info::dom_init): Used unsigned rather that size_t.
31671 (dom_info::dom_info): Same.
31673 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31674 Jakub Jelinek <jakub@redhat.com>
31677 * config/arm/arm.c: Include gimple.h.
31678 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31679 returns negative, increment ncrn only if it returned positive.
31680 (arm_needs_doubleword_align): Return int instead of bool,
31681 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31682 members, but if there is any such non-FIELD_DECL
31683 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31684 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31685 returns negative, increment nregs only if it returned positive.
31686 (arm_setup_incoming_varargs): Likewise.
31687 (arm_function_arg_boundary): Emit -Wpsabi note if
31688 arm_needs_doubleword_align returns negative, return
31689 DOUBLEWORD_ALIGNMENT only if it returned positive.
31691 2017-04-25 Marek Polacek <polacek@redhat.com>
31694 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31695 first argument to type.
31697 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
31700 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
31701 type checks to test for compatibility instead of equality.
31703 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31704 Jakub Jelinek <jakub@redhat.com>
31707 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
31709 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
31710 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
31711 the alignment computation, but return their maximum in warn_alignment.
31712 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31713 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
31715 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
31716 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31719 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31721 * config/arc/simdext.md (dmpyh): Fix typo.
31723 2017-04-25 Richard Biener <rguenther@suse.de>
31725 PR tree-optimization/80492
31726 * alias.c (compare_base_decls): Handle registers with asm
31727 specification conservatively.
31728 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31729 compare_base_decls returning dont-know properly.
31731 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31733 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31734 (legitimate_offset_address_p): New function.
31735 (arc_legitimate_address_p): Use above function.
31737 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31739 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31741 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31743 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31744 ACCH registers whenever they are available.
31746 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31748 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31749 double regs fix when not used.
31751 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31753 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31755 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31756 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31758 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31760 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31761 when emitting short ADD instructions.
31763 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31765 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31767 (cmpsi_cc_c_insn): Likewise.
31768 (cbranchsi4_scratch): Compute proper instruction length using
31769 compact_hreg_operand.
31770 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31772 2017-04-25 Richard Biener <rguenther@suse.de>
31774 PR middle-end/80509
31775 * passes.c (pass_manager::pass_manager): Initialize
31776 m_name_to_pass_map.
31778 2017-04-25 Richard Biener <rguenther@suse.de>
31780 PR tree-optimization/79201
31781 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31783 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31786 * config/s390/vector.md: Split MEM->GPR vector moves for
31787 non-s_operand addresses.
31789 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31792 * config/s390/predicates.md (reload_const_wide_int_operand): New
31794 * config/s390/s390.md ("movti"): Remove d/P alternative.
31795 ("movti_bigconst"): New pattern definition.
31797 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31800 * s390-protos.h (s390_expand_cs_hqi): Removed.
31801 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31802 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31803 modes as well as CCZ1mode and CCZmode.
31804 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31805 signature of s390_emit_compare_and_swap.
31806 (s390_expand_cs_hqi): Likewise, make static.
31807 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31808 compare-and-swap, in some cases.
31809 (s390_expand_cs): Wrapper function.
31810 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31812 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31813 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31814 patterns for small and large integers. Forbid symref memory operands.
31815 Move expander to s390.c. Require cc register.
31816 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31817 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31818 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31819 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31820 symref memory operands. Remove CC mode and call s390_match_ccmode
31822 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31824 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31826 * config/s390/s390.md (define_peephole2): New peephole to help
31827 combining the load-and-test pattern with volatile memory.
31829 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31831 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31832 with CCZmode for TARGET_Z196.
31834 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31836 PR rtl-optimization/80501
31837 * combine.c (make_compound_operation_int): Set subreg_code to SET
31838 even for AND with mask of the sign bit of mode.
31840 PR rtl-optimization/80500
31841 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31842 sum's initial value.
31844 2017-04-25 Julian Brown <julian@codesourcery.com>
31845 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31847 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31849 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31851 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31853 2017-04-25 Julian Brown <julian@codesourcery.com>
31854 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31856 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31857 (thunderx2t99_sha): New Reservation.
31859 2017-04-25 Julian Brown <julian@codesourcery.com>
31860 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31862 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31863 type for 1-element load.
31865 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31867 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31869 2017-04-24 Martin Jambor <mjambor@suse.cz>
31871 PR tree-optimization/80293
31872 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31873 char arrays not totally scalarizable if it is false.
31874 (analyze_all_variable_accesses): Pass correct value in the new
31875 parameter. Add a statistics counter.
31877 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31879 PR middle-end/79931
31880 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31882 2017-04-24 Richard Biener <rguenther@suse.de>
31884 PR tree-optimization/80494
31885 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31886 out for complex types.
31888 2017-04-24 Richard Biener <rguenther@suse.de>
31890 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31891 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31892 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31893 (DFS): Adjust and never fail.
31894 (sccvn_dom_walker::fail): Remove.
31895 (sccvn_dom_walker::before_dom_children): Adjust.
31896 (run_scc_vn): Likewise and never fail.
31897 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31898 (pass_fre::execute): Likewise.
31900 2017-04-24 Richard Biener <rguenther@suse.de>
31902 PR tree-optimization/79725
31903 * tree-ssa-sink.c (statement_sink_location): Return whether
31904 failure reason was zero uses. Move that check later.
31905 (sink_code_in_bb): Deal with zero uses by removing the stmt
31908 2017-04-24 Richard Biener <rguenther@suse.de>
31911 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31912 pointer-based references.
31914 2017-04-24 Richard Biener <rguenther@suse.de>
31917 * pass_manager.h (pass_manager::operator new): Remove.
31918 (pass_manager::operator delete): Likewise.
31919 * passes.c (pass_manager::operator new): Remove.
31920 (pass_manager::operator delete): Likewise.
31921 (pass_manager::pass_manager): Zero individual pass members.
31923 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31926 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31927 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31928 Check "XEXP (src, 1)" operand here.
31929 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31930 Check "XEXP (src, 1)" operand here.
31931 (dimode_scalar_chain::make_vector_copies): Detect count register
31932 of a shift instruction. Zero extend count register from QImode
31933 to DImode to satisfy vector shift pattern count operand predicate.
31934 Substitute vector shift count operand with a DImode copy.
31935 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31938 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31940 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31941 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31942 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31943 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31944 (UNSPEC_NOREX_MEM): Remove definition.
31946 2017-04-21 Richard Biener <rguenther@suse.de>
31948 PR tree-optimization/79547
31949 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31950 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31951 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31952 without any constraints.
31954 2017-04-21 Richard Biener <rguenther@suse.de>
31956 PR tree-optimization/78847
31957 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31959 2017-04-21 Richard Biener <rguenther@suse.de>
31961 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31962 (build_distinct_type_copy): Likewise.
31963 (build_variant_type_copy): Likewise.
31964 * tree.c (build_qualified_type): Pass down mem-stat info.
31965 (build_distinct_type_copy): Likewise.
31966 (build_variant_type_copy): Likewise.
31968 2017-04-21 Richard Biener <rguenther@suse.de>
31970 PR tree-optimization/80237
31971 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31973 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31974 for a simplified result.
31976 2016-04-21 Richard Biener <rguenther@suse.de>
31978 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31979 sth as strict as a simple_iv but a chrec without symbols and an
31980 operand defined in the loop we are peeling (and not some subloop).
31981 (propagate_constants_for_unrolling): Propagate all constants.
31983 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31986 * config/i386/i386.c (print_reg): Remove assert for disalowed
31987 regno values, call output_operand_lossage instead.
31989 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31992 * config/i386/constraints.md (Yc): New register constraint.
31993 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31994 Use Yc constraint for alternative 2 of operand 0. Remove
31995 preferred_for_speed attribute.
31997 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31999 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32000 lastprivate clauses in SIMT case.
32002 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32004 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32006 2017-04-20 Richard Biener <rguenther@suse.de>
32008 PR tree-optimization/57796
32009 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32010 as N scalar stores.
32011 (vect_model_load_cost): Cost gathers as N scalar loads.
32013 2017-04-20 Richard Biener <rguenther@suse.de>
32015 * ggc-page.c (ggc_allocated_p): Rename to ...
32016 (safe_lookup_page_table_entry): ... this and return the lookup
32018 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32020 2017-04-20 Richard Biener <rguenther@suse.de>
32022 PR tree-optimization/80453
32023 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32024 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32025 from the conditions.
32026 (vn_phi_eq): Pass them down.
32027 (vn_phi_lookup): Record them.
32028 (vn_phi_insert): Likewise.
32030 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32032 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32033 uninitialized variable warning to avoid buffer overrun.
32035 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32038 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32039 is suppressed for '{ 0 }' in C.
32041 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32043 * BASE-VER: Set to 8.0.0.
32045 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32047 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32048 priority .init_array and .fini_array section with SECTION_NOTYPE
32051 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32053 PR middle-end/80423
32054 * tree.h (build_array_type): Add typeless_storage default argument.
32055 * tree.c (type_cache_hasher::equal): Also compare
32056 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32057 (build_array_type): Add typeless_storage argument, set
32058 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32060 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32061 (build_array_type): Likewise. Add typeless_storage argument.
32063 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32064 Jakub Jelinek <jakub@redhat.com>
32066 PR tree-optimization/80426
32067 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32068 operation on symbolic operands, also compute the overflow for the
32069 invariant part when the operation degenerates into a negation.
32071 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32074 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32075 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32078 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32080 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32083 * config/avr/avr.c (tree.h): Include it.
32084 (cgraph.h): Include it.
32085 (avr_encode_section_info): Don't warn for uninitialized progmem
32086 variable if it's just an alias.
32088 2017-04-19 Richard Biener <rguenther@suse.de>
32091 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32092 when needed by AutoPGO.
32094 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32097 * doc/lto.texi: Remove an extra 'that'.
32099 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32101 PR rtl-optimization/80429
32102 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32103 are only used in debug insns.
32105 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32106 Vladimir Makarov <vmakarov@redhat.com>
32108 * config/sparc/predicates.md (input_operand): Add comment. Return
32109 true for any memory operand when LRA is in progress.
32110 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32112 2017-04-18 Jeff Law <law@redhat.com>
32115 * mips.md ({return,simple_return}_internal): Do not overwrite
32118 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32120 PR tree-optimization/80443
32121 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32122 instead of adding 1, subtract -1 and similarly instead of subtracting
32125 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32127 PR rtl-optimization/80357
32128 * haifa-sched.c (tmp_bitmap): New variable.
32129 (model_recompute): Handle duplicate use records.
32130 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32131 (free_global_sched_pressure_data): Free it.
32133 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32136 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32137 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32138 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32139 instead of SYSTEM_HEADER_DIR.
32141 2017-04-18 Jeff Law <law@redhat.com>
32143 PR middle-end/80422
32144 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32145 predecessors after walking up the insn chain.
32147 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32150 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32151 sizetype type into debug info.
32153 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32156 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32157 unneeded test for TARGET_UPPER_REGS_SF.
32158 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32160 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32163 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32164 instead of gsi_after_labels.
32166 2017-04-18 Jeff Law <law@redhat.com>
32168 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32172 2017-04-13 Jeff Law <law@redhat.com>
32173 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32174 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32176 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32179 * config/avr/avr.c (intl.h): Include it.
32180 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32182 2017-04-18 Martin Liska <mliska@suse.cz>
32184 PR gcov-profile/78783
32185 * gcov-tool.c (gcov_output_files): Validate that destination
32186 file is either removed by the tool or by a user.
32188 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32189 Guy Benyei <guybe@mellanox.com>
32191 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32192 block, and do not negate it, the stored id is already negative.
32194 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32196 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32198 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32201 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32202 masks of options that should be turned off if the VSX vector
32203 options are turned off.
32204 (OTHER_P8_VECTOR_MASKS): Likewise.
32205 (OTHER_VSX_VECTOR_MASKS): Likewise.
32206 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32207 rs6000_disable_incompatible_switches to validate no type switches
32209 (rs6000_incompatible_switch): New function to disallow turning on
32210 other vector options if -mno-vsx, -mno-power8-vector, or
32211 -mno-power9-vector are specified.
32213 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32215 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32217 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32219 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32220 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32221 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32222 (ARG_POINTER_CFA_OFFSET): Likewise.
32224 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32226 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32227 conditions to take advantage of various optimizations.
32229 2017-04-13 Jeff Law <law@redhat.com>
32231 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32232 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32233 (zero_extendsidi2_dext): Likewise.
32235 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32238 * fold-const.c (fold_ternary_loc): Revert
32239 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32242 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32244 PR rtl-optimization/80343
32245 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32246 new scratch pseudo.
32248 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32251 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32252 to ubsan_encode_value.
32254 2017-04-13 Jeff Law <law@redhat.com>
32256 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32257 appearing in DEBUG_INSNs.
32259 2017-04-13 Martin Liska <mliska@suse.cz>
32261 PR gcov-profile/80413
32262 * gcov-io.c (gcov_write_string): Copy to buffer just when
32263 allocated size is greater than zero.
32265 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32268 * dwarf2out.c (decls_for_scope): Ignore declarations of
32269 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32271 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32274 * ipa-visibility.c (non_local_p): Fix typos.
32275 (localize_node): When localizing symbol in same comdat group,
32276 dissolve the group only when we know external symbols are going
32278 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32280 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32282 PR tree-optimization/79390
32283 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32284 order does not result in usable sequence, retry with reversed operand
32290 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32291 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32292 op0 instead of fold_convert_loc (loc, type, arg0).
32294 2017-04-12 Jeff Law <law@redhat.com>
32296 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32297 has a delay slot in the generated code.
32299 * config/cris/cris.md (cris_preferred_reload_class): Return
32300 GENNONACR_REGS rather than GENERAL_REGS.
32302 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32305 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32306 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32307 signedness of the result type.
32309 2017-04-12 Richard Biener <rguenther@suse.de>
32310 Jeff Law <law@redhat.com>
32312 PR tree-optimization/80359
32313 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32314 trim stores to TARGET_MEM_REFs.
32316 2017-04-12 Richard Biener <rguenther@suse.de>
32318 PR tree-optimization/79390
32319 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32320 threading case even more.
32322 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32325 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32326 for quad_address_p for TImode, instead of just not indexed_address.
32328 2017-04-12 Richard Biener <rguenther@suse.de>
32329 Bernd Edlinger <bernd.edlinger@hotmail.de>
32331 PR middle-end/79671
32332 * alias.c (component_uses_parent_alias_set_from): Handle
32333 TYPE_TYPELESS_STORAGE.
32334 (get_alias_set): Likewise.
32335 * tree-core.h (tree_type_common): Add typeless_storage flag.
32336 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32337 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32338 for types containing members with TYPE_TYPELESS_STORAGE.
32339 (place_field): Likewise.
32340 (layout_type): Likewise for ARRAY_TYPE.
32341 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32342 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32343 TYPE_TYPELESS_STORAGE.
32344 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32346 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32349 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32350 first argument to type.
32352 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32356 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32357 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32358 (rs6000_expand_binop_builtin): Likewise.
32359 (rs6000_expand_ternop_builtin): Likewise; also add missing
32360 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32361 vshasigma built-ins.
32362 * doc/extend.texi: Document that vec_xxpermdi's third argument
32363 must be a constant.
32365 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32367 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32368 Use shift_const cost parameter when calculating gain of STV shifts.
32370 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32372 PR rtl-optimization/70478
32373 * lra-constraints.c (process_alt_operands): Check memory for
32374 disfavoring memory insn operand.
32376 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32378 PR middle-end/80100
32379 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32380 left shift in unsigned HOST_WIDE_INT type.
32382 PR rtl-optimization/80385
32383 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32384 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32387 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32388 if they have any depend clauses.
32390 2017-04-11 Martin Liska <mliska@suse.cz>
32393 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32394 * ipa-split.c (split_function): Create a local comdat symbol
32395 if caller is in a comdat group.
32397 2017-04-11 Martin Liska <mliska@suse.cz>
32400 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32403 2017-04-11 Martin Sebor <msebor@redhat.com>
32405 PR middle-end/80364
32406 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32407 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32409 (directive::set_width, directive::set_precision, format_character):
32411 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32414 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32417 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32418 conflict, set target->arch_name instead of target->cpu_name.
32420 2017-04-11 Richard Biener <rguenther@suse.de>
32422 PR tree-optimization/80374
32423 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32424 build_zero_cst, remove fold_convertible_p check again.
32426 2017-04-11 Martin Liska <mliska@suse.cz>
32429 * ubsan.c (instrument_object_size): Do not instrument register
32432 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32435 * config/i386/i386-builtin-types.def
32436 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32437 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32438 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32439 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32440 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32441 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32442 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32443 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32444 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32445 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32446 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32447 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32448 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32449 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32450 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32451 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32452 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32453 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32454 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32455 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32456 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32457 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32458 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32459 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32460 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32461 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32462 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32463 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32464 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32465 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32466 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32467 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32468 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32469 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32470 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32471 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32472 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32473 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32474 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32475 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32476 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32477 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32478 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32479 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32480 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32481 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32483 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32484 flag to second_arg_count, handle 4 argument function type _COUNT
32485 aliases, handle second_arg_count on second argument rather than last.
32487 2017-04-10 Jeff Law <law@redhat.com>
32489 PR tree-optimization/80374
32490 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32491 record anything if we can not convert integer_zero_node to the
32494 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32497 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32498 Enhance special handling given to the TARGET_P9_MINMAX option in
32499 relation to certain other options.
32501 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32503 PR tree-optimization/80153
32504 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32505 remove POINTER_PLUS_EXPR's base part directly, rather than through
32508 2017-04-10 Richard Biener <rguenther@suse.de>
32509 Bin Cheng <bin.cheng@arm.com>
32511 PR tree-optimization/80153
32512 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32513 the first element of pointer type aff_tree. Build result expr in
32515 (add_elt_to_tree): Convert to type unconditionally. Remove other
32516 fold_convert calls.
32517 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32518 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32520 2017-04-10 Richard Biener <rguenther@suse.de>
32522 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32525 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32527 PR rtl-optimization/70478
32528 * lra-constraints.c (curr_small_class_check): New.
32529 (update_and_check_small_class_inputs): New.
32530 (process_alt_operands): Update curr_small_class_check. Disfavor
32531 alternative insn memory operands. Check available regs for small
32534 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32537 * config/mips/mips.opt (-mvirt): Update description.
32538 * doc/invoke.texi (-mvirt): Likewise.
32540 2017-04-10 Richard Biener <rguenther@suse.de>
32542 PR middle-end/80362
32543 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32544 looking for NEGATE_EXPR in -A / -B to A / B folding.
32546 2017-04-10 Martin Liska <mliska@suse.cz>
32548 PR gcov-profile/80224
32549 * gcov.c (print_usage): Fix usage string.
32550 (get_gcov_intermediate_filename): Remove.
32551 (output_gcov_file): Use both for normal and intermediate format.
32552 (generate_results): Do not initialize special file for
32553 intermediate format.
32555 2017-04-10 Richard Biener <rguenther@suse.de>
32557 PR tree-optimization/80304
32558 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32561 2017-04-10 Nathan Sidwell <nathan@acm.org>
32564 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32565 (rs6000_init_builtins): Use it.
32567 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32569 * config/arm/arm.md (<mrc>): Add mode to SET source.
32570 (<mrrc>): Likewise.
32572 2017-04-10 Richard Biener <rguenther@suse.de>
32574 PR middle-end/80344
32575 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32577 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32580 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32581 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32582 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32583 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32584 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32585 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32586 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32587 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32588 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32589 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32590 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32591 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32592 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32593 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32594 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32595 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32596 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32597 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32598 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32599 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32600 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32601 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32602 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32604 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32606 PR rtl-optimization/70478
32607 * lra-constraints.c: Reverse the last patch.
32609 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32611 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32612 Add comment for WCHAR_T.
32614 2017-04-08 Martin Liska <mliska@suse.cz>
32617 2017-04-07 Martin Liska <mliska@suse.cz>
32620 * ipa-split.c (split_function): Add function part to a same comdat
32623 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32626 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32628 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32630 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32631 (vec_pairing): ...to this.
32632 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32633 (rs6000_sched_init): Adjust for name change.
32634 (struct rs6000_sched_context): Likewise.
32635 (rs6000_init_sched_context): Likewise.
32636 (rs6000_set_sched_context): Likewise.
32638 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32644 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32646 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32647 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32648 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32650 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32652 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32654 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32656 PR rtl-optimization/70703
32657 * ira-color.c (update_conflict_hard_regno_costs): Use
32658 int64_t instead of HOST_WIDE_INT.
32660 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32662 PR rtl-optimization/70478
32663 * lra-constraints.c (process_alt_operands): Disfavor alternative
32664 insn memory operands.
32666 2017-04-07 Jeff Law <law@redhat.com>
32668 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32669 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32671 2017-04-07 Martin Liska <mliska@suse.cz>
32674 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32675 Show error message instead of an ICE.
32677 2017-04-07 Martin Liska <mliska@suse.cz>
32680 * ipa-split.c (split_function): Add function part to a same comdat
32683 2017-04-07 Richard Biener <rguenther@suse.de>
32685 PR middle-end/80341
32686 * tree.c (get_unwidened): Also handle ! for_type case for
32688 * convert.c (do_narrow): Split out from ...
32689 (convert_to_integer_1): ... here. Do not pass final truncation
32690 type to get_unwidened for TRUNC_DIV_EXPR.
32692 2017-04-07 Richard Biener <rguenther@suse.de>
32694 * tree-affine.c (wide_int_ext_for_comb): Take type rather
32696 (aff_combination_const): Adjust.
32697 (aff_combination_scale): Likewise.
32698 (aff_combination_add_elt): Likewise.
32699 (aff_combination_add_cst): Likewise.
32700 (aff_combination_convert): Likewise.
32701 (add_elt_to_tree): Likewise. Remove unused argument.
32702 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
32704 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
32706 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
32708 * config/arm/arm.c (arm_default_short_enums): Use
32709 ARM_DEFAULT_SHORT_ENUMS.
32710 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
32712 2017-04-06 Jakub Jelinek <jakub@redhat.com>
32715 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
32716 members with redundant out-of-class redeclaration.
32718 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32721 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32722 * config/i386/i386.md (*zero_extendsidi2):
32723 Add (?*x,*x) and (?*v,*v) alternatives.
32725 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32728 * config/i386/i386.c (ix86_expand_builtin)
32729 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32730 mode from insn data. Convert operands to insn operand mode.
32731 Copy operands that don't satisfy insn predicate to a register.
32733 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32735 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32738 2017-04-06 Richard Biener <rguenther@suse.de>
32740 PR tree-optimization/80334
32741 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32742 preserve alignment of accesses.
32744 2017-04-06 Richard Biener <rguenther@suse.de>
32746 PR tree-optimization/80262
32747 * tree-sra.c (build_ref_for_offset): Preserve address-space
32749 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32750 Drop useless address-space information on MEM_REF offsets.
32752 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32754 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32756 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32758 PR rtl-optimization/70703
32759 * ira-color.c (update_conflict_hard_regno_costs): Use
32760 HOST_WIDE_INT instead of long.
32762 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32765 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32766 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32768 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32769 for x86_64 target. Handle -m3dnowa option.
32771 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32773 PR rtl-optimization/70703
32774 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32775 (update_conflict_hard_regno_costs): Use long instead of unsigned
32776 arithmetic for cost calculation.
32778 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32779 Bernd Edlinger <bernd.edlinger@hotmail.de>
32782 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32785 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32788 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32789 ptr_mode with Pmode throughout.
32790 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32791 into probe_stack_range and use DImode.
32793 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32796 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32797 call_eh_return is true.
32799 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32801 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32802 Initialize last_match_fntype_index.
32804 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32807 * tree-nvr.c: Include internal-fn.h.
32808 (pass_return_slot::execute): Ignore internal calls without
32811 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32812 Richard Biener <rguenther@suse.de>
32815 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32816 captures used multiple times, except for the last use.
32817 * generic-match-head.c: Include gimplify.h.
32819 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32821 PR tree-optimization/79390
32822 * target.h (struct noce_if_info): Declare.
32823 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32824 * target.def (noce_conversion_profitable_p): New target hook.
32825 * ifcvt.h (struct noce_if_info): New type, moved from ...
32826 * ifcvt.c (struct noce_if_info): ... here.
32827 (noce_conversion_profitable_p): Renamed to ...
32828 (default_noce_conversion_profitable_p): ... this. No longer
32830 (noce_try_store_flag_constants, noce_try_addcc,
32831 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32832 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32833 instead of noce_conversion_profitable_p.
32834 * config/i386/i386.c: Include ifcvt.h.
32835 (ix86_option_override_internal): Don't override
32836 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32837 (ix86_noce_conversion_profitable_p): New function.
32838 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32839 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32840 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32841 * doc/tm.texi: Regenerated.
32843 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32845 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32848 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32851 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32852 instructions for small multiply cores.
32854 2017-04-04 Jeff Law <law@redhat.com>
32856 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32858 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32859 that are not set by the loop over the elements.
32861 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32864 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32865 int mode, convert_modes it to mode as unsigned, otherwise use
32866 lowpart_subreg to mode rather than SImode.
32867 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32868 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32869 Use DImode instead of SImode for the shift count operand.
32870 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32873 2017-04-04 Richard Biener <rguenther@suse.de>
32875 PR middle-end/80281
32876 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32877 arithmetic done for the negate or the plus. Simplify.
32878 (A - (-B) -> A + B): Likewise.
32879 * fold-const.c (split_tree): Make sure to not negate pointers.
32881 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32883 PR rtl-optimization/60818
32884 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32885 a compare of comparisons with the thing compared if this results
32886 in a different machine mode.
32888 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32890 * alias.c (base_alias_check): Fix typo in comment.
32891 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32892 * cgraphunit.c (symbol_table::compile): Likewise.
32893 * collect2.c (maybe_run_lto_and_relink): Likewise.
32894 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32895 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32896 * config/avr/avr.c (avr_map_op_t): Likewise.
32897 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32898 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32899 * config/epiphany/epiphany.md (movcc): Likewise.
32900 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32901 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32903 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32904 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32905 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32906 * config/sh/sh.c (sh_rtx_costs): Likewise.
32907 * fold-const.c (fold_truth_andor): Likewise.
32908 * genautomata.c (collapse_flag): Likewise.
32909 * gengtype.h (struct type::u::s): Likewise.
32910 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32911 * input.c (FORMAT_AMOUNT): Likewise.
32912 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32913 (known_aggs_to_agg_replacement_list): Likewise.
32914 * ipa-inline-analysis.c: Likewise.
32915 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32916 * ipa-polymorphic-call.c
32917 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32918 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32919 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32921 * modulo-sched.c (apply_reg_moves): Likewise.
32922 * omp-expand.c (build_omp_regions_1): Likewise.
32923 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32924 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32925 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32926 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32927 * value-prof.c: Likewise.
32928 * var-tracking.c (val_reset): Likewise.
32930 2017-04-03 Richard Biener <rguenther@suse.de>
32932 PR tree-optimization/80275
32933 * fold-const.c (split_address_to_core_and_offset): Handle
32936 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32938 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32939 descriptors is at least equal to that of functions.
32941 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32943 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32945 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32948 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32949 (mov<IMOD4:mode>): New expander.
32950 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32952 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32954 PR rtl-optimization/79405
32955 * fwprop.c (propagations_left): New variable.
32956 (forward_propagate_into): Decrement it.
32957 (fwprop_init): Initialize it.
32958 (fw_prop): If the variable has reached zero, stop propagating.
32959 (fwprop_addr): Ditto.
32961 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32964 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32965 a FUNCTION_DECL, pass it as decl instead of origin to
32968 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32970 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32973 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32976 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32977 TARGET_VSX_SMALL_INTEGER.
32979 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32981 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32982 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32984 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32986 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32987 extraction from odd-numbered MSA register.
32989 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32991 PR middle-end/80173
32992 * expmed.c (store_bit_field_1): Don't attempt to create
32993 a word subreg out of hard registers wider than word if they
32994 have HARD_REGNO_NREGS of 1 for their mode.
32996 PR middle-end/80163
32997 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32998 conversions to integer types wider than word and pointer.
33001 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33002 (rtx_equal_for_cselib_p): Pass 0 to it.
33003 * cselib.c (cselib_hasher::equal): Likewise.
33004 (rtx_equal_for_cselib_1): Add depth argument. If depth
33005 is 128, don't look up VALUE locs and punt. Increment
33006 depth in recursive calls when walking VALUE locs.
33008 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33010 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33011 (make_gcov_file_name): Use the canonical path name for generating
33013 (read_line): Fix handling of files with ascii null bytes.
33015 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33017 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33018 to initialise a vector register instead
33019 of using a const_int.
33021 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33023 PR translation/80189
33024 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33025 diagnostic messages.
33027 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33030 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33031 (dfp_diex_<mode>): Update mode of operand 1.
33032 * doc/extend.texi (dxex, dxexq): Document change to return type.
33033 (diex, diexq): Document change to argument type.
33035 2017-03-30 Martin Jambor <mjambor@suse.cz>
33038 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33039 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33040 it reflects the signature changes performed at the callee side.
33041 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33042 to cgraph_build_function_type_skip_args.
33043 (build_function_decl_skip_args): Adjust call to the above function.
33045 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33048 * config/i386/sse.md
33049 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33050 register as dest whenever it is a MEM not rtx_equal_p to the
33051 corresponding dup operand, and when forcing into reg move the
33052 reg into the memory afterwards.
33053 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33054 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33055 for the force_reg mode.
33056 (avx512vl_vextractf128<mode>): Use register as dest either
33057 always when a MEM, or when it is a MEM not rtx_equal_p to the
33058 corresponding dup operand, or even not when it is a CONST_VECTOR
33059 depending on the mode and lo vs. hi.
33060 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33062 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33063 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33064 Likewise. Require that operands[2] is even.
33065 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33066 Remove extraneous parens. Require that operands[2] is a multiple
33068 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33069 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33070 disallow memory then.
33072 2017-03-30 Richard Biener <rguenther@suse.de>
33074 PR tree-optimization/77498
33075 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33076 to non-constants over backedges.
33078 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33080 PR rtl-optimization/80233
33081 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33082 as last_combined_insn. Do not test for BARRIER_P separately.
33084 2017-03-29 Andreas Schwab <schwab@suse.de>
33087 * calls.c (prepare_call_address): Convert funexp to Pmode before
33088 copying to temp reg.
33090 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33092 PR tree-optimization/80158
33093 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33094 Handle possible future case of more than one alternate
33096 (replace_rhs_if_not_dup): Likewise.
33097 (replace_one_candidate): Likewise.
33099 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33101 PR rtl-optimization/80193
33102 * ira.c (ira): Do not check allocation for LRA.
33104 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33106 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33107 (nvptx_output_simt_exit): Declare.
33108 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33109 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33110 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33111 (nvptx_declare_function_name): ...here. Emit declaration of local
33112 memory space buffer for omp_simt_enter insn.
33113 (nvptx_output_unisimt_switch): New.
33114 (nvptx_output_softstack_switch): New.
33115 (nvptx_output_simt_enter): New.
33116 (nvptx_output_simt_exit): New.
33117 * config/nvptx/nvptx.h (struct machine_function): New fields
33118 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33119 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33120 (UNSPECV_SIMT_EXIT): Ditto.
33121 (omp_simt_enter_insn): New insn.
33122 (omp_simt_enter): New expansion.
33123 (omp_simt_exit): New insn.
33124 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33126 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33127 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33128 (expand_GOMP_SIMT_EXIT): New.
33129 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33130 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33131 (GOMP_SIMT_EXIT): Ditto.
33132 * target-insns.def (omp_simt_enter): New insn.
33133 (omp_simt_exit): Ditto.
33134 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33136 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33137 (lower_rec_input_clauses): Likewise.
33138 (lower_lastprivate_clauses): Handle SIMT privatization.
33140 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33141 (ompdevlow_adjust_simt_enter): New.
33142 (find_simtpriv_var_op): New.
33143 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33144 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33146 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33147 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33148 (copy_decl_for_dup_finish): Ditto.
33150 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33152 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33155 * config/i386/i386.c (ix86_option_override_internal): Always
33156 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33158 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33160 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33162 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33164 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33165 mark new edge's irreducible flag accordign to it.
33166 (vect_do_peeling): Check loop preheader edge's irreducible flag
33167 and pass it to function slpeel_add_loop_guard.
33169 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33171 PR tree-optimization/80218
33172 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33173 Update block frequencies and counts.
33175 2017-03-28 Richard Biener <rguenther@suse.de>
33177 PR tree-optimization/78644
33178 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33179 of a simplification result we may not use it at all.
33181 2017-03-28 Richard Biener <rguenther@suse.de>
33184 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33185 without arguments, generate default definition of a SSA name.
33187 2017-03-28 Richard Biener <rguenther@suse.de>
33189 PR middle-end/80222
33190 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33191 TYPE_REF_CAN_ALIAS_ALL references.
33192 * fold-const.c (fold_indirect_ref_1): Likewise.
33194 2017-03-28 Martin Liska <mliska@suse.cz>
33197 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33198 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33200 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33201 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33203 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33204 (EXTRA_SPECS): Define.
33205 (SUBTARGET_EXTRA_SPECS): Likewise.
33206 (SUBTARGET_CPP_SPEC): Likewise.
33207 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33208 SUBTARGET_EXTRA_SPECS.
33209 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33211 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33213 * config/arc/simdext.md (vst64_insn): Update pattern.
33214 (vld32wh_insn): Likewise.
33215 (vld32wl_insn): Likewise.
33216 (vld64_insn): Likewise.
33217 (vld32_insn): Likewise.
33219 2017-03-28 Marek Polacek <polacek@redhat.com>
33222 * fold-const.c (fold_comparison): Use protected_set_expr_location
33223 instead of SET_EXPR_LOCATION.
33225 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33227 * tree.c (add_expr): Avoid name lookup warning.
33229 2017-03-27 Jeff Law <law@redhat.com>
33231 PR tree-optimization/80216
33232 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33233 function name. Limit recursion depth.
33234 (record_temporary_equivalences): Corresponding changes.
33236 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33238 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33241 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33244 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33246 * cfgcleanup.c (reg_note_cfa_p): New array.
33247 (insns_have_identical_cfa_notes): New function.
33248 (old_insns_match_p): Don't cross-jump in between /f
33249 and non-/f instructions. If both i1 and i2 are frame related,
33250 verify all CFA notes, their order and content.
33252 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33255 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33256 HImode and SImode with zero extend to DImode to one insn.
33257 (bswap<mode>2_extenddi): Likewise.
33258 (bswapsi2_extenddi): Likewise.
33259 (bswaphi2_extendsi): Likewise.
33260 (bswaphi2): Combine bswap HImode and SImode into one insn.
33261 Separate memory insns from swapping register.
33262 (bswapsi2): Likewise.
33263 (bswap<mode>2): Likewise.
33264 (bswaphi2_internal): Delete, no longer used.
33265 (bswapsi2_internal): Likewise.
33266 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33267 store, and gpr<-gpr swap insns.
33268 (bswap<mode>2_store): Likewise.
33269 (bswaphi2_reg): Register only splitter, combine with the splitter.
33270 (bswaphi2 splitter): Likewise.
33271 (bswapsi2_reg): Likewise.
33272 (bswapsi2 splitter): Likewise.
33273 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33274 the insns into load, store, and register/register insns.
33275 (bswapdi2_ldbrx): Likewise.
33276 (bswapdi2_load): Likewise.
33277 (bswapdi2_store): Likewise.
33278 (bswapdi2_reg): Likewise.
33280 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33282 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33283 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33285 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33288 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33290 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33291 special handling for target option conflicts between dform
33292 options (-mpower9-dform, -mpower9-dform-vector,
33293 -mpower9-dform-scalar) and -mno-direct-move.
33295 2017-03-27 Richard Biener <rguenther@suse.de>
33297 PR tree-optimization/80181
33298 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33300 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33302 * config/arc/predicates.md (move_double_src_operand): Replace the
33303 call to move_double_src_operand with a call to address_operand.
33305 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33307 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33308 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33309 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33311 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33313 * config/arc/predicates.md (long_immediate_loadstore_operand):
33314 Consider scaled addresses cases.
33316 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33318 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33319 restored when in interrupt.
33320 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33321 doesn't have delay slot.
33323 2017-03-27 Richard Biener <rguenther@suse.de>
33326 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33327 inlined thunk clones.
33329 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33332 * asan.c (instrument_derefs): Copy over last operand from
33333 original COMPONENT_REF to the new COMPONENT_REF with
33334 DECL_BIT_FIELD_REPRESENTATIVE.
33335 * ubsan.c (instrument_object_size): Likewise.
33337 2017-03-27 Richard Biener <rguenther@suse.de>
33339 PR tree-optimization/80170
33340 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33341 sure DR/SCEV didnt fold in constants we do not see when looking
33342 at the reference base alignment.
33344 2017-03-27 Richard Biener <rguenther@suse.de>
33346 PR middle-end/80171
33347 * gimple-fold.c (fold_ctor_reference): Properly guard against
33348 NULL return value from canonicalize_constructor_val.
33350 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33353 * config/i386/i386.c (ix86_expand_builtin)
33354 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33355 flags reg setting and flags reg using instructions.
33356 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33357 clobbering instructions to zero extend op2.
33359 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33361 * doc/install.texi (Configuration) <--with-aix-soname>:
33362 Update link to AIX ld.
33364 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33366 PR rtl-optimization/80160
33367 PR rtl-optimization/80159
33368 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33369 reg_alternate_class into account.
33371 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33374 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33375 to consider in curr_insn_transform.
33377 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33379 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33380 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33381 and emit_mode_inner.
33383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33385 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33386 argument to the overloaded builtin variants. Use the new flag to
33387 deprecate certain builtin variants.
33388 * config/s390/s390-builtin-types.def: Add new builtin types.
33389 * config/s390/s390-builtins.h: Support new flags field for
33390 overloaded builtins.
33391 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33392 (s390_macro_to_expand): Enable vector float data type.
33393 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33394 builtins by incrementing the __VEC__ version number.
33395 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33397 (s390_resolve_overloaded_builtin): Emit error messages depending
33398 on the builtin flags.
33399 * config/s390/s390.c (s390_expand_builtin): Support additional
33400 flags argument. Change error message to match the messages
33401 emitted in s390-c.c.
33402 * config/s390/s390.md: New UNSPEC_* constants.
33403 (op_type): Add new instruction types.
33404 * config/s390/vecintrin.h: Add new builtins and test data class
33406 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33407 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33408 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33409 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33410 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33411 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33413 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33414 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33415 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33416 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33418 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33419 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33420 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33421 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33423 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33424 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33425 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33426 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33427 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33428 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33429 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33431 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33432 ("vec_scatter_element<V_HW_4:mode>_DI")
33433 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33434 ("vec_fpint<mode>", "vflls")
33435 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33436 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33437 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33438 ("*vec_cmphe<mode>_cc"): ... these.
33440 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33441 mode constant instead of magic value.
33443 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33445 * config/s390/s390.c (s390_expand_vec_compare): Support other
33446 vector floating point modes than just V2DF.
33447 (s390_expand_vcond): Likewise.
33448 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33449 (s390_cannot_change_mode_class): Prevent mode changes between TF
33450 and V1TF in vector registers.
33451 * config/s390/s390.md (DF, SF): New mode attributes.
33452 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33453 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33454 SFmode support for VRs.
33455 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33457 (VFT, VF_HW): New mode iterators.
33458 (vw, sdx): New mode attributes.
33459 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33460 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33461 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33462 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33463 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33464 also the new vector floating point modes. Renaming to ...
33466 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33467 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33468 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33469 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33470 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33471 ("vec_unordered<mode>"): ... these.
33473 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33474 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33475 ("*vec_extendv2df"): New insn definitions.
33477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33479 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33480 ("mulditi3_2", "*muldi3_sign"): New patterns.
33481 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33482 rename the pattern definition.
33484 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33486 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33488 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33490 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33492 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33493 instruction if possible.
33494 * config/s390/vector.md (vec_halfnumelts): New mode
33496 ("*vec_vllezlf<mode>"): New pattern.
33498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33500 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33501 ("popcountv4si2", "popcountv2di2"): Rename to ...
33502 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33503 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33505 ("popcount<mode>2_vxe"): New pattern.
33507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33509 * common/config/s390/s390-common.c (processor_flags_table): Add
33511 * config.gcc: Add arch12.
33512 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33513 Default to arch12 for unknown CPU model numbers.
33514 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33515 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33516 PROCESSOR_max sanity check.
33517 * config/s390/s390-opts.h (enum processor_type): Add
33519 * config/s390/s390.c (processor_table): Add arch12.
33520 (s390_expand_builtin): Add check for B_VXE flag.
33521 (s390_issue_rate): Add PROCESSOR_ARCH12.
33522 (s390_get_sched_attrmask): Likewise.
33523 (s390_get_unit_mask): Likewise.
33524 (s390_sched_score): Enable z13 scheduling for arch12.
33525 (s390_sched_reorder): Likewise.
33526 (s390_sched_variable_issue): Likewise.
33527 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33529 (s390_tune_attr): Use z13 scheduling also for arch12.
33530 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33531 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33532 (TARGET_VXE_P): New macros.
33533 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33534 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33535 * config/s390/s390.opt: Add arch12 as processor_type.
33537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33539 * config/s390/s390.md
33540 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33541 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33542 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33544 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33545 Rename expanders to ...
33547 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33548 ("fixuns_truncdddi2_emu"): ... these.
33550 ("fixuns_trunc<mode>si2_emu"): New expander.
33552 ("*fixuns_truncdfdi2_z13"): Rename to ...
33553 ("*fixuns_truncdfdi2_vx"): ... this.
33555 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33557 * config/s390/2964.md: Remove the single element vector compare
33558 instructions which are no longer used.
33559 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33561 (s390_canonicalize_comparison): Remove handling of DFmode
33563 (s390_expand_vec_compare_scalar): Remove function.
33564 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33565 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33567 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33569 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33571 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33573 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33574 will anyway by matched by mov<mode>_64dfp.
33576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33578 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33579 vlef/vstef. Add missing operand to vleif.
33581 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33583 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33584 pair for all vector types with 64 bit elements.
33585 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33586 * config/s390/vector.md (V_HW_64): ... here.
33587 (V_128_NOSINGLE): New mode iterator.
33588 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33589 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33590 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33591 ("*vec_load_pairv2di"): Change to ...
33592 ("*vec_load_pair<mode>"): ... this one.
33594 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33596 * config/s390/constraints.md: Add comments.
33597 (jKK): Reject element sizes > 8 bytes.
33598 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33600 * config/s390/s390.md: Add the s_operand checks formerly in
33601 s390_split_ok_p to various splitters where they are still
33603 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33604 for 128 bit vectors. Plus two splitters.
33606 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33608 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33611 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33614 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33615 error if the boundary argument is not constant.
33617 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33619 PR rtl-optimization/80112
33620 * loop-doloop.c (doloop_condition_get): Don't check condition
33621 if cmp isn't SET with IF_THEN_ELSE src.
33623 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33625 PR tree-optimization/80158
33626 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33627 replacing a candidate statement, also replace it for the
33628 candidate's alternate interpretation.
33629 (replace_rhs_if_not_dup): Likewise.
33630 (replace_one_candidate): Likewise.
33632 2017-03-24 Richard Biener <rguenther@suse.de>
33634 PR tree-optimization/80167
33635 * graphite-isl-ast-to-gimple.c
33636 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33638 (translate_isl_ast_to_gimple::get_rename): Likewise.
33640 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33642 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33643 handling of certain combinations of target options, including the
33644 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33645 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33647 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33650 * config/arm/arm.md (*load_multiple): Add reload_completed to
33651 matching condition.
33653 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33654 Richard Biener <rguenth@suse.de>
33656 PR tree-optimization/79908
33657 PR tree-optimization/80136
33658 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33659 been cast away, gimplify_and_add suffices.
33661 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33663 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33665 2017-03-23 Richard Biener <rguenther@suse.de>
33667 PR tree-optimization/80032
33668 * gimplify.c (gimple_push_cleanup): Forced unconditional
33669 cleanups still have to go to the conditional_cleanups
33672 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33674 PR tree-optimization/80072
33675 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33677 (next_operand_entry_id): Change type to unsigned int.
33678 (sort_by_operand_rank): Make sure to return the right return value
33679 even if unsigned fields are bigger than INT_MAX.
33680 (struct oecount): Change cnt and id type to unsigned int.
33681 (oecount_hasher::equal): Formatting fix.
33682 (oecount_cmp): Make sure to return the right return value
33683 even if unsigned fields are bigger than INT_MAX.
33684 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33687 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33688 TREE_READONLY on result if writing it more than once.
33691 * doc/invoke.texi (-fsanitize=thread): Document that with
33692 -fnon-call-exceptions atomics are not able to throw
33696 * tsan.c: Include tree-eh.h.
33697 (instrument_builtin_call): Call maybe_clean_eh_stmt or
33698 maybe_clean_or_replace_eh_stmt where needed.
33699 (instrument_memory_accesses): Add cfg_changed argument.
33700 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
33701 if it returned true.
33702 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
33704 PR rtl-optimization/63191
33705 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
33706 wrapper function, moved the whole old content into ...
33707 (ix86_delegitimize_address_1): ... this. New inline function.
33708 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
33709 true as last argument instead of ix86_delegitimize_address.
33711 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33713 * config/aarch64/aarch64.c (generic_branch_cost): Copy
33714 cortexa57_branch_cost.
33716 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33718 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33720 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33723 * doc/md.texi (Constraints): Document wA constraint.
33724 * config/rs6000/constraints.md (wA): New.
33725 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33726 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33727 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33728 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33730 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33733 * gimplify.c (is_oacc_declared): New function.
33734 (oacc_default_clause): Use it to set default flags for acc declared
33735 variables inside parallel regions.
33736 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33737 declared variables.
33738 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33739 declare attribute to any decl as necessary.
33741 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33744 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33745 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33746 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33747 (arm_arch_lpae): This.
33748 * config/arm/arm.c (arm_arch7ve): Rename into ...
33749 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33750 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33753 2017-03-22 Martin Liska <mliska@suse.cz>
33756 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33757 error message instead of an ICE.
33759 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33761 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33763 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33765 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33767 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33770 2017-03-21 Martin Sebor <msebor@redhat.com>
33772 * doc/extend.texi: Use "cannot" instead of "can't."
33773 * doc/hostconfig.texi: Same.
33774 * doc/install.texi: Same.
33775 * doc/invoke.texi: Same.
33776 * doc/loop.texi: Same.
33777 * doc/md.texi: Same.
33778 * doc/objc.texi: Same.
33779 * doc/rtl.texi: Same.
33780 * doc/tm.texi: Same.
33781 * doc/tm.texi.in: Same.
33782 * doc/trouble.texi: Same.
33784 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33787 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33788 (collect_checksum_attributes): Set it.
33789 (die_checksum_ordered): Use it.
33791 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33793 PR tree-optimization/79908
33794 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33795 change: For a VA_ARG whose LHS has been cast away, use
33796 force_gimple_operand to construct the side effects.
33798 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33800 PR translation/80001
33801 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33802 more amenable to translation.
33803 (oacc_loop_auto_partitions): Likewise.
33805 2017-03-21 Marek Polacek <polacek@redhat.com>
33806 Martin Sebor <msebor@redhat.com>
33808 PR tree-optimization/80109
33809 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33810 on INTEGRAL_TYPE_P.
33812 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33813 Segher Boessenkool <segher@kernel.crashing.org>
33816 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33817 check reg_used_between_p between insn and one of succ or succ2
33818 depending on if succ is artificial insn not inserted into insn
33821 2017-03-21 Martin Liska <mliska@suse.cz>
33823 PR gcov-profile/80081
33824 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33825 * doc/gcc.texi: Include gcov-dump stuff.
33826 * doc/gcov-dump.texi: New file.
33828 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33830 PR rtl-optimization/79150
33831 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33832 conditional jump, if the jump is the last insn of the loop.
33834 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33835 Richard Biener <rguenth@suse.de>
33837 PR tree-optimization/79908
33838 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33839 been cast away, use force_gimple_operand to construct the side
33842 2017-03-21 Martin Liska <mliska@suse.cz>
33844 PR libfortran/79956
33845 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33848 2017-03-21 Brad Spengler <spender@grsecurity.net>
33851 * plugin.c (htab_hash_plugin): New function.
33852 (add_new_plugin): Use it and adjust.
33853 (parse_plugin_arg_opt): Adjust.
33854 (init_one_plugin): Likewise.
33856 2017-03-21 Richard Biener <rguenther@suse.de>
33858 PR tree-optimization/80032
33859 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33860 if set force the cleanup to happen unconditionally.
33861 (gimplify_target_expr): Push inserted clobbers with force_uncond
33862 to avoid them being removed by control-dependent DCE.
33864 2017-03-21 Richard Biener <rguenther@suse.de>
33866 PR tree-optimization/80122
33867 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33868 va_arg_pack_len when the inlined call stmt requires pack
33870 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33872 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33875 * tsan.c (instrument_builtin_call): If the memory model argument
33876 is not a constant, assume it is valid.
33879 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33882 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33884 PR rtl-optimization/79910
33885 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33886 if its dest is used by an insn before I2 (other than the combined
33887 insns themselves, which are properly handled already).
33889 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33892 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33894 * combine.c (record_used_regs): New static function.
33895 (try_combine): Handle situations where there is an additional
33896 instruction between I2 and I3 which needs to have a LOG_LINK
33900 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33902 * combine.c (try_combine): Delete redundant i1 test. Call
33903 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33905 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33908 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33909 alternatives 13/14.
33911 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33913 PR tree-optimization/80054
33914 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33915 the optimization if a PHI or any of its arguments is not dominated
33916 by the candidate's basis. Use gphi* rather than gimple* as
33918 (replace_profitable_candidates): Clean up a gimple* variable that
33919 should be a gphi* variable.
33921 2017-03-20 Martin Sebor <msebor@redhat.com>
33924 * doc/extend.texi (attribute constructor): Document present limitation.
33926 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33929 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33930 __POWER9_VECTOR__ #ifdef control, change template definition to
33931 use Power9-specific built-in function.
33932 (vec_any_eq): Likewise.
33933 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33934 to control outcomes from this test.
33935 (vector_ae_<mode>p): For VEC_F modes, likewise.
33937 2017-03-20 Ian Lance Taylor <iant@google.com>
33939 * config/i386/i386.c (ix86_function_regparm): Save an extra
33940 register for -fsplit-stack with DECL_STATIC_CHAIN.
33942 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33945 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33946 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33948 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33950 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33952 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33955 2017-03-20 Marek Polacek <polacek@redhat.com>
33958 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33960 2017-03-20 Richard Biener <rguenther@suse.de>
33962 PR tree-optimization/80113
33963 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33964 allocate extra SSA name for PHI def.
33965 (add_close_phis_to_outer_loops): Likewise.
33966 (add_close_phis_to_merge_points): Likewise.
33967 (copy_loop_close_phi_args): Likewise.
33968 (copy_cond_phi_nodes): Likewise.
33970 2017-03-20 Martin Liska <mliska@suse.cz>
33972 PR middle-end/79753
33973 * tree-chkp.c (chkp_build_returned_bound): Do not build
33974 returned bounds for a LHS that's not a BOUNDED_P type.
33976 2017-03-20 Martin Liska <mliska@suse.cz>
33980 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33981 COMPLEX_CST and VECTOR_CST.
33983 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33986 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33987 target operand. A new splitter adds the clobber statement in case
33988 the target operand is dead anyway.
33990 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33992 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33993 to age-old versions of binutils and glibc.
33995 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33997 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33999 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34001 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34003 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34005 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34006 requirement for binutils 2.13.
34008 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34010 * combine.c (try_combine): Delete redundant i1 test. Call
34011 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34013 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34015 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34016 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34018 <riscv64-*-elf>: Re-arrange section
34019 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34020 <riscv32-*-linux>: Likewise.
34021 <riscv64-*-elf>: Likewise
34022 <riscv64-*-linux>: Likewise.
34024 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34027 * aarch64.opt(verbose-cost-dump): Fix typo.
34029 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34032 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34033 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34035 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34037 * reload.c (find_reloads): When reloading a nonoffsettable address,
34038 use RELOAD_OTHER for it and its address reloads.
34040 PR rtl-optimization/79910
34041 * combine.c (record_used_regs): New static function.
34042 (try_combine): Handle situations where there is an additional
34043 instruction between I2 and I3 which needs to have a LOG_LINK
34046 2017-03-17 Jeff Law <law@redhat.com>
34048 PR tree-optimization/71437
34049 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34050 conditional in the hash table first.
34051 (vrp_dom_walker::before_dom_children): Extract condition from
34052 ASSERT_EXPR. Record condition, its inverion and any implied
34053 conditions as well.
34055 2017-03-17 Marek Polacek <polacek@redhat.com>
34056 Markus Trippelsdorf <markus@trippelsdorf.de>
34058 PR tree-optimization/80079
34059 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34062 2017-03-17 Richard Biener <rguenther@suse.de>
34064 PR middle-end/80075
34065 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34066 Properly verify the LHS before the RHS possibly claims to be
34068 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34071 2017-03-17 Martin Jambor <mjambor@suse.cz>
34073 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34074 (List of -O2 options): Likewise.
34075 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34078 2017-03-17 Tom de Vries <tom@codesourcery.com>
34080 * gcov-dump.c (print_usage): Print bug_report_url.
34082 2017-03-17 Richard Biener <rguenther@suse.de>
34084 PR middle-end/80050
34085 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34086 (parser::peek): Likewise.
34088 2017-03-17 Richard Biener <rguenther@suse.de>
34090 PR tree-optimization/80048
34091 * sese.c (free_sese_info): Properly release rename_map and
34092 copied_bb_map elements.
34094 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34096 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34097 Add linked-list forward and backlinks. Insert on
34098 construction, remove on destruction.
34099 (class pass_store_merging): Add m_stores_head field.
34100 (pass_store_merging::terminate_and_process_all_chains):
34101 Iterate over m_stores_head list.
34102 (pass_store_merging::terminate_all_aliasing_chains):
34104 (pass_store_merging::execute): Check for debug stmts first.
34105 Push new chains onto the m_stores_head stack.
34107 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34110 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34111 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34112 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34114 2017-03-16 Jeff Law <law@redhat.com>
34116 PR tree-optimization/71437
34117 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34118 member function. Implementation moved into after_dom_children
34119 member function and into the threader's thread_outgoing_edges
34121 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34122 some code into new thread_outgoing_edges.
34123 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34124 definition. Simplify marker handling (do it here). Assume we always
34125 have the available expression and the const/copies tables.
34126 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34128 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34129 * tree-vrp.c (equiv_stack): No longer file scoped.
34130 (vrp_dom_walker): New class.
34131 (vrp_dom_walker::before_dom_children): New member function.
34132 (vrp_dom_walker::after_dom_children): Likewise.
34133 (identify_jump_threads): Setup domwalker. Use it rather than
34134 walking edges in a random order by hand. Simplify setup/finalization.
34135 (finalize_jump_threads): Remove.
34136 (vrp_finalize): Do not call identify_jump_threads here.
34137 (execute_vrp): Do it here instead and call thread_through_all_blocks
34140 PR tree-optimization/71437
34141 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34143 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34145 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34146 (dom_opt_dom_walker::thread_across_edge): Remove
34147 handle_dominating_asserts argument. All callers changed.
34148 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34149 changes. Remove calls to lhs_of_dominating_assert. Other
34150 uses of handle_dominating_asserts turn into unconditional code
34151 (simplify_control_stmt_condition_1): Likewise.
34152 (simplify_control_stmt_condition): Likewise.
34153 (thread_through_normal_block, thread_across_edge): Likewise.
34154 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34155 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34156 object if it is not an SSA_NAME.
34157 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34158 before calling into the VRP specific simplifiers.
34159 (identify_jump_threads): Remove handle_dominating_asserts
34162 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34165 * tree-diagnostic.c (default_tree_printer): No longer static.
34166 * tree-diagnostic.h (default_tree_printer): New prototype.
34168 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34170 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34171 Change ins into fmov.
34173 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34175 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34176 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34177 Use h_con constraint for operand 1.
34178 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34179 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34181 2017-03-15 Jeff Law <law@redhat.com>
34183 PR tree-optimization/71437
34184 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34185 (record_temporary_equivalences): Use it.
34187 PR tree-optimization/71437
34188 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34189 tree-ssa-scopedtables.
34190 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34191 (record_conditions, record_cond, vuse_eq): Likewise.
34192 (record_edge_info): Adjust to API tweak of record_conditions.
34193 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34194 (record_temporary_equivalences, optimize_stmt): Likewise.
34195 (eliminate_redundant_computations): Likewise.
34196 (record_equivalences_from_stmt): Likewise.
34197 * tree-ssa-scopedtables.c: Include options.h and params.h.
34198 (vuse_eq): New function, moved from tree-ssa-dom.c
34199 (build_and_record_new_cond): Likewise.
34200 (record_conditions): Likewise. Accept vector of conditions rather
34201 than edge_equivalence structure for first argument.
34202 for the first argument.
34203 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34204 from tree-ssa-dom.c.
34205 (avail_exprs_stack::record_cond): Likewise.
34206 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34207 from tree-ssa-dom.c.
34208 (avail_exprs_stack): Add new member functions lookup_avail_expr
34210 (record_conditions): Declare.
34212 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34215 * lra-constraints.c (process_alt_operands): Increase reject for
34216 reloading an input/output operand.
34218 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34221 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34222 insns to convert from signed/unsigned char/short to IEEE 128-bit
34224 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34226 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34229 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34230 subreg of inner mode for values already in registers.
34232 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34234 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34235 iteration reg is used after the loop.
34237 2017-03-14 Martin Sebor <msebor@redhat.com>
34239 PR tree-optimization/79800
34240 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34241 precision in negative-positive range.
34242 (format_floating): Call non-const overload with adjusted precision.
34244 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34247 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34250 2017-03-14 Martin Sebor <msebor@redhat.com>
34252 PR middle-end/80020
34253 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34254 * builtins.def (aligned_alloc): Use it.
34257 * Makefile.in (GTFILES): Add calls.c.
34258 * calls.c: Include "gt-calls.h".
34260 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34262 PR rtl-optimization/79728
34263 * regs.h (struct target_regs): New field
34264 x_contains_allocatable_regs_of_mode.
34265 (contains_allocatable_regs_of_mode): New macro.
34266 * reginfo.c (init_reg_sets_1): Initialize it, and change
34267 contains_reg_of_mode so it includes global regs as well.
34268 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34269 rather than contains_regs_of_mode.
34271 2017-03-14 Martin Liska <mliska@suse.cz>
34273 * doc/invoke.texi: Document options that can't be combined with
34274 -fcheck-pointer-bounds.
34276 2017-03-14 Martin Liska <mliska@suse.cz>
34278 PR middle-end/79831
34279 * doc/invoke.texi (-Wchkp): Document the option.
34281 2017-03-14 Martin Liska <mliska@suse.cz>
34283 * Makefile.in: Install gcov-dump.
34285 2017-03-14 Martin Liska <mliska@suse.cz>
34287 * multiple_target.c (expand_target_clones): Bail out for
34288 an invalid attribute.
34290 2017-03-14 Richard Biener <rguenther@suse.de>
34292 * alias.c (struct alias_set_entry): Pack properly.
34293 * cfgloop.h (struct loop): Likewise.
34294 * cse.c (struct set): Likewise.
34295 * ipa-utils.c (struct searchc_env): Likewise.
34296 * loop-invariant.c (struct invariant): Likewise.
34297 * lra-remat.c (struct cand): Likewise.
34298 * recog.c (struct change_t): Likewise.
34299 * rtl.h (struct address_info): Likewise.
34300 * symbol-summary.h (function_summary): Likewise.
34301 * tree-loop-distribution.c (struct partition): Likewise.
34302 * tree-object-size.c (struct object_size_info): Likewise.
34303 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34304 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34305 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34306 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34307 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34308 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34309 (struct _stmt_vec_info): Likewise.
34311 2017-03-14 Martin Liska <mliska@suse.cz>
34314 * multiple_target.c (create_dispatcher_calls): Check that
34315 a target can create a function dispatcher.
34317 2017-03-14 Martin Liska <mliska@suse.cz>
34320 * multiple_target.c (expand_target_clones): Drop local.local
34321 flag for default implementation.
34323 2017-03-14 Richard Biener <rguenther@suse.de>
34325 PR tree-optimization/80030
34326 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34328 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34330 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34331 gcc_fallthrough() instead of __attribute__((fallthrough));
34333 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34335 * doc/gcc.texi: Remove "up" link to (DIR).
34336 * doc/gccint.texi: Ditto.
34338 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34340 * doc/install.texi (Specific) <avr>: Remove reference to
34343 2017-03-13 Jeff Law <law@redhat.com>
34345 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34346 attribute rather than comments.
34348 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34349 match_scratch operand is highest.
34351 2017-03-13 Martin Liska <mliska@suse.cz>
34353 PR middle-end/78339
34354 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34355 is a CHKP clone, use original declaration.
34357 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34359 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34360 (arc_conditional_register_usage): Use a different allocation order
34361 when optimizing for size.
34362 * common/config/arc/arc-common.c (arc_option_optimization_table):
34363 Section anchors default on when optimizing for size.
34365 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34367 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34369 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34371 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34372 * config/arc/arc.md (cpu_facility): Add cd variant.
34373 (*movqi_insn): Add code density variant.
34374 (*movhi_insn): Likewise.
34375 (*movqi_insn): Likewise.
34376 (*addsi3_mixed): Likewise.
34377 (subsi3_insn): Likewise.
34379 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34381 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34383 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34385 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34386 expressions with MINUS and UNARY ops.
34388 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34391 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34393 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34394 between vec_select and vector argument.
34395 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34396 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34397 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34398 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34399 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34400 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34402 2017-03-13 Richard Biener <rguenther@suse.de>
34405 * params.def (vect-max-peeling-for-alignment): Fix typo.
34407 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34409 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34410 issue that only occurred with binutils below 2.18.
34412 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34414 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34415 refer to binutils 2.11/2.12 minimum.
34417 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34419 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34420 ftp.kernel.org and simplify binutils requirement.
34422 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34424 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34426 (Optimize Options): Ditto. Also remove redundancy.
34428 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34430 PR translation/79848
34431 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34433 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34434 to G_ to avoid double translation.
34436 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34438 PR translation/79923
34439 * auto-profile.c (get_combined_location): Convert leading
34440 character of diagnostics to lower case and remove trailing period.
34441 (read_profile): Likewise for various diagnostics.
34442 * config/arm/arm.c (arm_option_override): Remove trailing period
34443 from various diagnostics.
34444 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34445 (msp430_expand_delay_cycles): Likewise.
34447 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34450 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34451 full command-line argument, rather than just "str".
34452 (aarch64_validate_march): Likewise.
34453 (aarch64_validate_mtune): Likewise.
34455 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34457 PR rtl-optimization/78911
34458 * lra-assigns.c (must_not_spill_p): New function.
34459 (spill_for): Use it.
34461 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34463 PR tree-optimization/79981
34464 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34465 ATOMIC_COMPARE_EXCHANGE ifn result.
34466 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34467 IFN_ATOMIC_COMPARE_EXCHANGE.
34469 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34472 * opts.c (parse_sanitizer_options): Add missing question mark to
34473 "did you mean" message.
34475 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34477 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34479 (VMULEUH_UNS): Likewise.
34480 (VMULOUB_UNS): Likewise.
34481 (VMULOUH_UNS): Likewise.
34482 * config/rs6000/rs6000.c (builtin_function_type): Remove
34483 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34485 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34488 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34489 x with result of extra_parsing_for_operand_code_0.
34490 (function_reader::extra_parsing_for_operand_code_0): Convert
34491 return type from void to rtx, returning x. When reading
34492 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34493 larger size containing struct block_symbol.
34495 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34497 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34498 -mfloat128-hardware without -m64.
34500 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34503 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34504 entries to the case statement that marks unsigned arguments to
34505 overloaded functions.
34507 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34509 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34510 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34512 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34515 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34516 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34518 2017-03-10 Martin Liska <mliska@suse.cz>
34522 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34523 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34526 2017-03-10 Olivier Hainque <hainque@adacore.com>
34528 * tree-switch-conversion (array_value_type): Start by resetting
34529 candidate type to it's main variant.
34531 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34533 PR rtl-optimization/79909
34534 * combine.c (try_combine): Use simplify_replace_rtx on individual
34535 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34536 of the whole CALL_INSN_FUNCTION_USAGE.
34538 PR tree-optimization/79972
34539 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34540 get_range_info on SSA_NAMEs. Formatting fixes.
34542 2017-03-10 Richard Biener <rguenther@suse.de>
34543 Jakub Jelinek <jakub@redhat.com>
34545 PR tree-optimization/77975
34546 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34547 edge to be constant.
34548 (get_val_for): For constant x return it. Formatting fix.
34549 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34550 would use the same bases as the current one.
34552 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34554 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34555 instead of vec_select for V1TImode.
34556 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34558 (VSX_LE_128): Add V1TI to this mode iterator.
34559 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34560 (*vsx_le_perm_store_<mode>): Likewise.
34561 (pre-reload splitter for VSX stores): Likewise.
34562 (post-reload splitter for VSX stores): Likewise.
34563 (*vsx_xxpermdi2_le_<mode>): Likewise.
34564 (*vsx_lxvd2x2_le_<mode>): Likewise.
34565 (*vsx_stxvd2x2_le_<mode>): Likewise.
34567 2017-03-09 Michael Eager <eager@eagercon.com>
34569 Correct failures with --enable-checking=yes,rtl.
34571 * config/microblaze/microblaze.c (microblaze_expand_shift):
34572 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34573 test for const0_rtx.
34574 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34575 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34577 2017-03-09 Richard Biener <rguenther@suse.de>
34579 PR tree-optimization/79977
34580 * graphite-scop-detection.c (scop_detection::merge_sese):
34581 Handle the case of extra exits to blocks dominating the entry.
34583 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34585 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34588 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34590 PR rtl-optimization/79949
34591 * lra-constraints.c (process_alt_operands): Check memory when
34592 trying to predict a cycle. Print about the overall increase.
34594 2017-03-09 Richard Biener <rguenther@suse.de>
34596 PR middle-end/79971
34597 * gimple-expr.c (useless_type_conversion_p): Preserve
34598 TYPE_SATURATING for fixed-point types.
34600 2017-03-09 Richard Biener <rguenther@suse.de>
34603 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34604 alignment of BLKmode params.
34606 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34609 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34610 (VALL_NO_V2Q): Likewise.
34612 * config/aarch64/aarch64-simd.md
34613 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34615 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34616 VALL_NO_V2Q mode iterator.
34617 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34619 2017-03-09 Martin Liska <mliska@suse.cz>
34621 PR tree-optimization/79631
34622 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34623 tree_int_cst_sign_bit just for INTEGER constants.
34625 2017-03-09 Martin Liska <mliska@suse.cz>
34629 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34632 2017-03-09 Marek Polacek <polacek@redhat.com>
34635 * tree.c (inchash::add_expr): Handle TREE_VEC.
34637 2017-03-09 Martin Liska <mliska@suse.cz>
34640 (chkp_narrow_size_and_offset): New function.
34641 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34642 (void chkp_parse_bit_field_ref): New function.
34643 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34644 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34646 2017-03-09 Martin Liska <mliska@suse.cz>
34649 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34650 (chkp_find_bounds_1): Remove gcc_unreachable.
34652 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34655 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34656 BUILT_IN_SYNC*, determine the access type from the size suffix and
34657 always build a MEM_REF with that type. Handle forgotten
34658 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34661 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34662 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34663 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34664 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34665 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34666 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34667 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34668 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34669 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34670 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34671 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34672 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34673 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34674 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34675 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34676 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34677 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34678 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34679 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34680 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34681 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34682 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34683 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34684 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34685 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34686 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34687 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34688 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34689 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34690 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34691 definitions outside of __OPTIMIZE__ guarded section.
34694 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34695 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
34696 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
34697 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
34700 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34702 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
34703 ("vfenez<mode>"): Add missing constraints.
34705 2017-03-08 Martin Sebor <msebor@redhat.com>
34708 * config/nds32/nds32.c (nds32_option_override):
34709 Fix misspelled diagnostic.
34711 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34714 * gimplify.c (gimplify_omp_for): Replace index var in outer
34715 taskloop statement with an artificial variable and add
34716 OMP_CLAUSE_PRIVATE clause for it.
34718 2017-03-08 Richard Biener <rguenther@suse.de>
34720 PR tree-optimization/79955
34721 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34722 for accesses that are completely outside of the variable.
34724 2017-03-08 Andrew Haley <aph@redhat.com>
34726 PR tree-optimization/79943
34727 * tree-ssa-loop-split.c (compute_new_first_bound): When
34728 calculating the new upper bound, (END-BEG) should be added, not
34731 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34733 * config/avr/avr.md (setmemhi): Make sure match_dup
34734 operand number comes before match_scratch.
34736 2017-03-08 Richard Biener <rguenther@suse.de>
34738 PR tree-optimization/79920
34739 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34740 with ncopies == 1 to ...
34741 (vect_transform_slp_perm_load): ... here. Properly compute
34742 all element loads by iterating VF times over the group. Do
34743 not handle ncopies (computed in a broken way) in
34744 vect_create_mask_and_perm.
34746 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34749 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34750 is a uniform vector, use uniform_vector_p return value instead of
34751 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34753 2017-03-07 Marek Polacek <polacek@redhat.com>
34755 PR middle-end/79809
34756 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34757 (alloca_call_type): Likewise.
34759 2017-03-07 Martin Liska <mliska@suse.cz>
34761 * gcov.c (process_args): Put comment to correct location.
34763 2017-03-07 Martin Liska <mliska@suse.cz>
34765 PR middle-end/68270
34766 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34767 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34768 (chkp_narrow_bounds_for_field): Likewise.
34769 (chkp_parse_array_and_component_ref): Pass one more argument to
34772 2017-03-07 Richard Biener <rguenther@suse.de>
34774 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34777 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34779 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34780 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34782 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34785 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34786 to end of description.
34787 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34789 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34791 PR rtl-optimization/79901
34792 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34794 (*avx512f_<code><mode>3<mask_name>): ... this.
34795 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34796 iterator instead of VI8_AVX2_AVX512BW.
34798 PR rtl-optimization/79901
34799 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34800 min/max expander, expand it using expand_vec_cond_expr.
34803 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34806 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34809 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34810 to void * for PCH reasons.
34811 * dwarf2out.c (output_loc_operands, output_die): Cast
34812 v.val_vec.array to unsigned char *.
34814 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34817 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34820 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34822 PR rtl-optimization/79571
34823 * lra-constraints.c (process_alt_operands): Calculate static
34824 reject and subtract it from overall when only addresses will be
34827 2017-03-06 Julia Koval <julia.koval@intel.com>
34830 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34831 incoming stack boundary to 128 for 64-bit targets.
34833 2017-03-06 Richard Biener <rguenther@suse.de>
34835 PR tree-optimization/79894
34836 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34837 to NULL after folding it.
34839 2017-03-06 Richard Biener <rguenther@suse.de>
34841 PR tree-optimization/79824
34842 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34843 check disabling peeling for gaps.
34845 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34847 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34848 attributes): Document gettimeofday.
34850 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34852 * config/s390/s390.c (s390_option_override_internal): Set
34853 PARAM_MIN_VECT_LOOP_BOUND
34855 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34857 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34858 * config/s390/s390.md: Likewise.
34860 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34863 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34864 (<avx2_avx512>_perm<mode>): Rename to ...
34865 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34867 (<avx512>_perm<mode>_mask): Rename to ...
34868 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34870 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34871 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34872 instead of VI8F_256_512.
34873 (avx512f_perm<mode>): New define_expand.
34874 (avx512f_perm<mode>_mask): Likewise.
34875 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34876 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34878 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34880 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34881 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34883 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34885 2017-03-06 Martin Liska <mliska@suse.cz>
34888 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34889 when having a SSA NAME w/o VAR_DECL assigned to it.
34891 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34893 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34894 msa_dpsub_<su>_d): Fix MODE for vec_select.
34896 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34898 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34900 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34902 2017-03-06 Richard Biener <rguenther@suse.de>
34904 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34905 * plugin.c (register_plugin_info): Likewise.
34906 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34908 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34910 * config/i386/sse.md (sse_storehps, sse_storelps,
34911 avx_<castmode><avxsizesuffix>_<castmode>,
34912 avx512f_<castmode><avxsizesuffix>_<castmode>,
34913 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34914 in condition that at least one operand is not a MEM.
34916 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34918 PR middle-end/79805
34919 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34920 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34922 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34923 gimple_call_nothrow_p flag based on whether original builtin can throw.
34924 If it can, emit following stmts on the fallthrough edge.
34925 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34926 don't create new bb if inserting just debug stmts on the edge, try to
34927 insert them on the fallthru bb or just reset debug stmts.
34929 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34932 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34933 restore recog_data (including the operand rtxes inside it) around
34934 the call to get_insn_template.
34936 2017-03-03 Martin Sebor <msebor@redhat.com>
34938 PR tree-optimization/79699
34939 * context.c (context::~context): Free MPFR caches to avoid
34940 a memory leak on program exit.
34942 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34944 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34945 Use wide_int::ulow () instead of .elt (0).
34947 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34949 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34950 (*pushxf): Limit oF constraint to 32bit targets and add oC
34951 constraint for 64bit targets.
34952 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34953 (*pushdf): Change rmF constraint to rmC.
34955 2017-03-03 Martin Liska <mliska@suse.cz>
34957 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34958 Remove unused variable.
34960 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34963 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34964 is a memory operand, increase num_memory.
34965 (ix86_expand_args_builtin): Likewise.
34967 2017-03-03 Jan Hubicka <jh@suse.cz>
34970 * ipa-devirt.c (maybe_record_node): Properly handle
34971 __cxa_pure_virtual visibility.
34973 2017-03-03 Martin Liska <mliska@suse.cz>
34975 PR tree-optimization/79803
34976 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34978 (pass_loop_prefetch::execute): Disabled optimization if an
34979 assumption about L1 cache size is not met.
34981 2017-03-03 Martin Liska <mliska@suse.cz>
34983 PR rtl-optimization/79574
34984 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34985 (hash_scan_set): Likewise.
34986 (dump_hash_table): Likewise.
34987 (hoist_code): Likewise.
34989 2017-03-03 Richard Biener <rguenther@suse.de>
34991 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34992 in place of uhigh ().
34993 (fixed_convert_from_real): Likewise.
34995 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34998 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35000 2017-03-03 Richard Biener <rguenther@suse.de>
35002 PR middle-end/79818
35003 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35004 TYPE_OVERFLOW_UNDEFINED check.
35006 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35008 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35010 (vector_ae_<mode>_p): Likewise.
35011 (vector_nez_<mode>_p): Likewise.
35012 (vector_ne_v2di_p): Likewise.
35013 (vector_ae_v2di_p): Likewise.
35014 (vector_ne_<mode>_p): Likewise.
35015 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35017 (vsx_tsqrt<mode>2_fe): Likewise.
35019 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35022 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35024 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35026 PR rtl-optimization/79780
35027 * cprop.c (one_cprop_pass): When second and further conditional trap
35028 in a single basic block is turned into an unconditional trap, turn it
35029 into a deleted note to avoid RTL verification failures.
35031 2017-03-02 Richard Biener <rguenther@suse.de>
35033 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35035 2017-03-02 Richard Biener <rguenther@suse.de>
35037 PR tree-optimization/79345
35039 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35040 param and abort the walk, returning -1 if it is hit.
35041 (walk_aliased_vdefs): Take a limit param and pass it on.
35042 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35043 defaulting to 0 and return a signed int.
35044 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35045 (check_defs): New helper.
35046 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35047 about uninitialized memory.
35048 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35049 bogus uninitialized warning.
35050 (fixed_convert_from_real): Likewise.
35052 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35054 PR tree-optimization/66768
35055 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35056 iv_use if base object can't be determined.
35058 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35060 PR tree-optimization/79345
35061 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35062 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35063 (get_pattern_stats): Initialize it.
35064 * genemit.c (gen_expand): Verify match_scratch numbers come after
35065 match_operand/match_dup numbers.
35066 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35067 match_scratch numbers.
35068 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35070 * config/s390/s390.md (trunctdsd2): Likewise.
35072 2017-03-02 Richard Biener <rguenther@suse.de>
35074 * wide-int.h (wide_int_storage::operator=): Implement in terms
35077 2017-03-02 Richard Biener <rguenther@suse.de>
35079 PR tree-optimization/79777
35080 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35081 the to insert expression to sth existing.
35083 2017-03-01 Martin Sebor <msebor@redhat.com>
35085 PR middle-end/79692
35086 * gimple-ssa-sprintf.c
35087 (directive::known_width_and_precision): New function.
35088 (format_integer): Use it.
35089 (get_mpfr_format_length): Consider the full range of precision
35090 when computing %g output with the # flag. Set the likely byte
35091 count to 3 rather than 1 when precision is indeterminate.
35092 (format_floating): Correct the lower bound of precision.
35094 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35096 * doc/invoke.texi: Document default code model for 64-bit Linux.
35098 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35101 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35102 udiv rather than div since input pattern is unsigned.
35104 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35106 * config/i386/i386.c (print_reg): Warn for values of
35107 unsupported size in integer register.
35109 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35112 * config/rs6000/predicates.md (current_file_function_operand): Do
35113 not allow self calls to be local if the function is replaceable.
35115 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35118 * config/rs6000/altivec.h (vec_ctz and others): Change the
35119 preprocessor macro that controls conditional compilation from
35120 _ARCH_PWR9 to __POWER9_VECTOR__.
35121 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35122 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35123 control (instead of _ARCH_PWR9 control) so that template
35124 definition uses power9-specific function.
35125 (vec_any_eq): Likewise.
35126 (vec_all_ne): Change macro definition to use a power9-specific
35127 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35128 _ARCH_PWR9 control).
35129 (vec_any_eq) Likewise.
35130 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35131 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35132 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35133 support for xvcmpnedp instruction.
35134 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35135 macro expansion so that Power9 implementation of vec_all_ne does
35136 not use the AltiVec predicate framework.
35137 (VCMPNEH_P): Likewise.
35138 (VCMPNEW_P): Likewise.
35139 (VCMPNED_P): Likewise.
35140 (VCMPNEFP_P): Likewise.
35141 (VCMPNEDP_P): Likewise.
35142 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35143 implementation of vec_any_eq to not use AltiVec predicate
35145 (VCMPAEH_P): Likewise.
35146 (VCMPAEW_P): Likewise.
35147 (VCMPAED_P): Likewise.
35148 (VCMPAEFP_P): Likewise.
35149 (VCMPAEDP_P): Likewise.
35150 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35151 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35152 not use the AltiVec predicate framework.
35153 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35154 of vec_any_eq to not use AltiVec predicate framework.
35155 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35156 support for predefined __POWER9_VECTOR__ macro to indicate that
35157 Power9 instruction selection is enabled.
35158 (altivec_overloaded_builtins): Remove extraneous
35159 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35160 function argument types RS6000_BTI_bool_V16QI and
35161 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35162 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35163 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35164 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35165 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35166 Power9 for implementations of vec_cmpne. Change the signature for
35167 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35168 (representing vec_all_ne) to remove the previously described first
35169 argument of type RS6000_BTI_INTSI, as this was an artifact of
35170 reliance on the AltiVec predicate framework, which is no longer
35171 used in the implementation of these functions. Add
35172 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35173 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35174 since, unlike the AltiVec predicate framework implementation, we
35175 do not share function descriptors between vec_alle and vec_anyeq.
35176 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35177 set of modes that receive special treatment even when
35178 TARGET_P9_VECTOR is true. The special treatment emits code that
35179 does not depend on Power9 instructions.
35180 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35181 define_expand to not rely on AltiVec predicate framework.
35182 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35184 (vector_ne_v2di_p): Change this define_expand to not rely on
35185 AltiVec predicate framework.
35186 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35188 (vector_ne_<mode>_p): Change this define_expand to not rely on
35189 AltiVec predicate framework.
35190 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35192 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35193 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35194 define_insn pattern.
35195 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35196 define_insn pattern because the xvcmpne<VSs>. instruction is not
35198 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35199 instruction is not supported.
35201 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35203 * config/nvptx/nvptx.c: Include intl.h.
35205 2017-03-01 Martin Jambor <mjambor@suse.cz>
35208 * ipa-prop.h (ipa_bits): Removed field known.
35209 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35210 to pointers. Adjusted their comments to warn about their sharing.
35211 (ipcp_transformation_summary): Change bits to a vector of pointers.
35212 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35213 (ipa_get_ipa_bits_for_value): Declare.
35214 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35215 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35216 (ipa_bits_hash_table): Likewise.
35217 (ipa_vr_ggc_hash_traits): Likewise.
35218 (ipa_vr_hash_table): Likewise.
35219 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35220 being pointers and vr_known being removed.
35221 (ipa_set_jf_unknown): Likewise.
35222 (ipa_get_ipa_bits_for_value): New function.
35223 (ipa_set_jfunc_bits): Likewise.
35224 (ipa_get_value_range): New overloaded functions.
35225 (ipa_set_jfunc_vr): Likewise.
35226 (ipa_compute_jump_functions_for_edge): Use the above functions to
35227 construct bits and vr parts of jump functions.
35228 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35229 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35231 (ipcp_grow_transformations_if_necessary): Also allocate
35232 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35234 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35235 them. Fix too long lines.
35236 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35237 vr_known being removed.
35238 (ipa_read_jump_function): Use new setter functions to construct bits
35239 and vr parts of jump functions or set them to NULL.
35240 (write_ipcp_transformation_info): Adjust for bits being pointers.
35241 (read_ipcp_transformation_info): Likewise.
35242 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35244 Include gt-ipa-prop.h.
35245 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35247 (ipcp_store_bits_results): Likewise.
35248 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35249 Do not write to existing jump functions but use a temporary instead.
35251 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35254 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35255 attempt to use its first operand as BIT_FIELD_REF base.
35257 2017-03-01 Richard Biener <rguenther@suse.de>
35259 PR middle-end/79721
35260 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35261 interpolating formula in wrapping arithmetic.
35262 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35264 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35266 PR tree-optimization/79734
35267 * tree-vect-generic.c (expand_vector_condition): Optimize
35268 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35269 Handle VEC_COND_EXPR where comparison has different inner width from
35270 type's inner width.
35272 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35274 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35275 markup, and similar issues. Remove @opindex entries for things
35276 that aren't options. Add missing -mmpy-option entries.
35278 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35280 PR tree-optimization/79737
35281 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35282 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35283 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35284 instead of byte_size. Formatting fix.
35285 (shift_bytes_in_array_right): Formatting fix.
35287 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35290 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35291 condition on optimize for the leaf function test.
35293 2017-02-28 Martin Liska <mliska@suse.cz>
35296 * read-rtl-function.c (function_reader::handle_unknown_directive):
35297 Bail out when one uses -flto.
35299 2017-02-28 Martin Liska <mliska@suse.cz>
35301 * common.opt: Replace space with tabular for options of <number>
35303 * config/i386/i386.opt: Show <number> value for
35304 -mlarge-data-threshold.
35305 * opts.c (print_filtered_help): Do not display number in hexadecimal
35308 2017-02-28 Martin Liska <mliska@suse.cz>
35310 * common.opt: Fix --help=option -Q for options which are of
35313 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35315 * config/i386/i386.c (print_reg): Error out for values
35316 of 8-bit size in invalid integer register.
35318 2017-02-28 Martin Sebor <msebor@redhat.com>
35320 PR tree-optimization/79691
35321 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35323 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35326 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35327 gcc_unreachable with output_operand_lossage.
35329 2017-02-28 Richard Biener <rguenther@suse.de>
35331 PR tree-optimization/79740
35332 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35334 (visit_nary_op): Insert the nary into the hashtable if we
35335 pattern-matched sth.
35336 * tree-ssa-pre.c (eliminate_insert): Robustify.
35338 2017-02-28 Richard Biener <rguenther@suse.de>
35340 PR middle-end/79731
35341 * fold-const.c (decode_field_reference): Reject out-of-bound
35344 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35346 * config/i386/i386.c: Include intl.h.
35347 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35348 instead of just cond ? "..." : "...".
35349 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35350 * coverage.c (read_counts_file): Likewise.
35351 * omp-offload.c: Include intl.h.
35352 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35353 of just cond ? "..." : "...".
35354 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35355 of just cond ? "..." : "...".
35357 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35360 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35362 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35363 'tune for' CPU name.
35364 * config/arm/arm-cpu-data.h: Regenerated.
35366 2017-02-28 Richard Biener <rguenther@suse.de>
35368 PR tree-optimization/79732
35369 * tree-inline.c (expand_call_inline): Do not shadow var.
35371 2017-02-28 Richard Biener <rguenther@suse.de>
35373 PR tree-optimization/79723
35374 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35375 address-space properly.
35377 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35379 * doc/optinfo.texi (Optimization groups): Fix option used for
35381 * doc/invoke.texi (-fopt-info): Document "omp".
35382 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35383 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35384 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35385 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35386 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35388 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35390 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35391 OPTGROUP_OMP with "omp".
35393 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35396 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35397 for arithmetic shift of unsigned V2DI.
35399 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35401 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35402 arc/linux.h headers.
35403 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35404 (LINK_SPEC): Likewise.
35405 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35406 (EXTRA_SPECS): Likewise.
35407 (STARTFILE_SPEC): Likewise.
35408 (ENDFILE_SPEC): Likewise.
35409 (LIB_SPEC): Likewise.
35410 (TARGET_SDATA_DEFAULT): Likewise.
35411 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35412 (MULTILIB_DEFAULTS): Likewise.
35413 (DWARF2_UNWIND_INFO): Likewise.
35414 * config/arc/big.h: New file.
35415 * config/arc/elf.h: Likewise.
35416 * config/arc/linux.h: Likewise.
35417 * config/arc/t-uClibc: Remove.
35419 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35421 PR tree-optimization/77536
35422 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35423 (tree_transform_and_unroll_loop): Use above function to compute the
35424 estimated niter of unrolled loop and use it when scaling profile.
35425 Also use count info rather than frequency if it's non-zero.
35426 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35427 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35428 (vect_transform_loop): Call above function.
35430 2017-02-27 Richard Biener <rguenther@suse.de>
35432 PR tree-optimization/45397
35433 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35434 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35435 (visit_nary_op): Add pattern matching for CSEing sign-changed
35436 or truncated operations with wider ones.
35438 2017-02-27 Richard Biener <rguenther@suse.de>
35440 PR tree-optimization/79690
35441 * tree-vect-stmts.c (vectorizable_store): Use vector type
35442 built from the DR with address-space.
35444 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35446 * doc/invoke.texi (Optimize Options): Refine the description
35447 of asan-use-after-return.
35449 2017-02-25 Alan Modra <amodra@gmail.com>
35451 PR rtl-optimization/79584
35452 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35453 base, not ad->base_term, the reg within base. Remove assertion
35454 that ad->base == ad->base_term. Replace gen_int_mode using
35455 bogus mode with const0_rtx.
35457 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35459 PR middle-end/79396
35460 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35461 FMA_EXPR like tcc_binary or tcc_unary.
35463 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35466 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35468 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35469 (output_loc_operands): Handle DW_OP_call_ref and
35470 DW_OP_GNU_variable_value.
35471 (struct variable_value_struct): New type.
35472 (struct variable_value_hasher): Likewise.
35473 (variable_value_hash): New variable.
35474 (string_types): Remove.
35475 (copy_loc_descr): New function.
35476 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35477 (prepend_loc_descr_to_each): New function.
35478 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35479 instead of add_loc_descr_to_each if the first argument is single
35480 location list and the second has multiple.
35481 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35482 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35483 when looking for variable value which doesn't have other location info.
35484 (loc_list_from_tree): Formatting fix.
35485 (gen_array_type_die): Simplify DW_AT_string_length handling.
35486 (adjust_string_types): Remove.
35487 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35488 string_types. Call resolve_variable_values.
35489 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35490 (resolve_addr_in_expr): Likewise. Add A argument.
35491 (copy_deref_exprloc): Remove deref argument. Adjust for the
35492 original expression being DW_OP_GNU_variable_value with optionally
35493 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35494 optionally after it.
35495 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35496 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35497 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35498 (variable_value_hasher::hash, variable_value_hasher::equal): New
35500 (resolve_variable_value_in_expr, resolve_variable_value,
35501 resolve_variable_values, note_variable_value_in_expr,
35502 note_variable_value): New functions.
35503 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35506 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35509 * opts.h (handle_generated_option): Add GENERATED_P argument.
35510 * opts-common.c (handle_option): Adjust function comment.
35511 (handle_generated_option): Add GENERATED_P argument, pass it to
35513 (control_warning_option): Pass false to handle_generated_option
35515 * opts.c (maybe_default_option): Pass true to handle_generated_option
35517 * optc-gen.awk: Likewise.
35519 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35521 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35522 a REG, look at the REG it is a SUBREG of.
35523 (splitter for cmpeqsi_t): Ditto.
35525 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35527 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35528 the special USEs with the pattern of the insn, not the insn itself.
35530 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35533 * doc/invoke.texi: Document -mload-store-pairs.
35535 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35536 Sandra Loosemore <sandra@codesourcery.com>
35538 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35539 argument isn't a CONST_INT.
35540 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35541 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35542 (nios2_validate_compare): Bypass alternate compare logic if *op2
35543 is not a CONST_INT.
35544 (ldstwm_operation_p): Return false if first_base is not a REG or
35545 if first_offset is not a CONST_INT.
35547 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35549 * config/cris/cris.md: Use correct operand in a define_peephole2.
35551 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35553 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35555 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35557 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35558 this_insn if it is an INSN or JUMP_INSN.
35559 (force_offsettable): Look at base, not at addr.
35560 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35561 on things that aren't necessarily CONST_INTs.
35563 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35565 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35566 -mfpmath=sse is the default also for x86-32 targets with SSE2
35567 instruction set when @option{-ffast-math} is enabled
35569 2017-02-24 Jeff Law <law@redhat.com>
35571 PR rtl-optimizatoin/79286
35572 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35575 2017-02-24 Richard Biener <rguenther@suse.de>
35577 PR tree-optimization/79389
35578 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35581 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35583 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35584 function comment to reflect reality.
35585 (loop_exits_before_overflow): Fix typo in function description.
35587 2017-02-24 Richard Biener <rguenther@suse.de>
35589 PR tree-optimization/79389
35590 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35591 properly that a threading opportunity exists. Detect conditional
35592 copy/constant propagation opportunities.
35594 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35596 * config/visium/visium.md (type): Add trap.
35597 (b): New mode attribute.
35598 (*btst): Rename into...
35599 (*btst<mode>): ...this and adjust.
35600 (*cbranchsi4_btst_insn): Rename into...
35601 (*cbranch<mode>4_btst_insn): ...this and adjust.
35602 (trap): New define_insn.
35604 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35606 PR tree-optimization/79389
35607 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35608 (noce_reversed_cond_code): New function.
35609 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35610 reversed_comparison_code. Formatting fix.
35611 (noce_try_store_flag): Test rev_cond != NULL in addition to
35612 reversed_comparison_code.
35613 (noce_try_store_flag_constants): Likewise.
35614 (noce_try_store_flag_mask): Likewise.
35615 (noce_try_addcc): Use rev_cond if non-NULL instead of
35616 reversed_comparison_code.
35617 (noce_try_cmove_arith): Likewise. Formatting fixes.
35618 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35619 (noce_find_if_block): Initialize rev_cond.
35620 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35621 instead of false as last argument never attempt to reverse it
35624 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35626 PR tree-optimization/79663
35627 * tree-predcom.c (combine_chains): Process refs in reverse order
35628 only for ZERO length chains, and add explaining comment.
35630 2017-02-23 Jeff Law <law@redhat.com>
35632 PR tree-optimization/79578
35633 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35634 in call to operand_equal_p.
35636 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35639 * config/i386/cpuid.h: Fix another undefined behavior.
35641 2017-02-23 Richard Biener <rguenther@suse.de>
35643 PR tree-optimization/79683
35644 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35645 vector types for data-refs.
35647 2017-02-23 Martin Liska <mliska@suse.cz>
35649 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35651 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35653 PR middle-end/79665
35654 * internal-fn.c (get_range_pos_neg): Moved to ...
35655 * tree.c (get_range_pos_neg): ... here. No longer static.
35656 * tree.h (get_range_pos_neg): New prototype.
35657 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35658 are known to be in between 0 and signed maximum inclusive, try to
35659 expand both unsigned and signed divmod and use the cheaper one from
35662 2017-02-22 Jeff Law <law@redhat.com>
35664 PR tree-optimization/79578
35665 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35666 to compare base operands.
35668 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35671 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35672 gpc_reg_operand instead of fpr_reg_operand.
35674 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35676 * config/mips/mips.c (mips_return_in_memory): Force FP
35677 vector types to be returned in memory for o32 ABI.
35679 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35681 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35682 instead of DW_TAG_member for static data member declarations and don't
35683 set no_linkage_name for static inline data members.
35684 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35687 2017-02-22 Martin Liska <mliska@suse.cz>
35689 * doc/invoke.texi: Replace inequality signs with square brackets
35692 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35695 * lra-constraints.c (simplify_operand_subreg): Handle
35696 WORD_REGISTER_OPERATIONS targets.
35698 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35701 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
35702 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
35703 elimination by swapping fld*.
35705 2017-02-22 Richard Biener <rguenther@suse.de>
35707 PR tree-optimization/79673
35708 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
35709 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
35710 irrelevant address-space qualifiers and avoiding a
35711 ADDR_SPACE_CONVERT_EXPR from fold_convert.
35713 2017-02-22 Richard Biener <rguenther@suse.de>
35715 PR tree-optimization/79666
35716 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
35717 to not symbolically negate if that may introduce undefined
35720 2017-02-22 Martin Liska <mliska@suse.cz>
35723 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35724 * data-streamer-out.c (streamer_write_gcov_count_stream):
35726 * value-prof.c (stream_out_histogram_value): Make assert more
35727 precise based on type of counter.
35729 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35732 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35733 Use nonimmediate_operand instead of memory_operand for operand 1.
35734 (float-extend standard_x87sse_constant_load splitter): Ditto.
35736 2017-02-21 Jeff Law <law@redhat.com>
35738 PR tree-optimization/79621
35739 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35740 blocks with edges to themselves.
35742 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35745 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35746 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35747 Use gimple_call_builtin_p.
35750 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35751 on temporarily removed DEBUG_INSNs.
35753 PR tree-optimization/79649
35754 * tree-loop-distribution.c (classify_partition): Give up on
35755 non-generic address space loads/stores.
35757 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35759 * doc/loop.texi (Loop manipulation): Remove nonexistent
35760 tree_ssa_loop_version from the documentation.
35761 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35763 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35766 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35767 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35768 * config/rs6000/rs6000.c: Include except.h.
35769 (rs6000_expand_split_stack_prologue): Call
35770 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35772 2017-02-21 Martin Jambor <mjambor@suse.cz>
35775 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35776 have been analyzed.
35778 2017-02-21 Martin Jambor <mjambor@suse.cz>
35780 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35781 for backward compatibility only.
35782 * doc/invoke.texi (Option Summary): Remove all references to
35783 -fipa-cp-alignment.
35785 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35789 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35791 * lra-constraints.c (curr_insn_transform): Handle
35792 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35794 2017-02-21 Martin Liska <mliska@suse.cz>
35796 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35798 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35800 PR translation/79638
35801 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35803 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35806 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35807 (arm_function_ok_for_sibcall): Return false for an indirect call by
35808 descriptor if all the argument registers are used.
35809 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35810 alignment of the function.
35812 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35814 PR tree-optimization/61441
35815 * simplify-rtx.c (simplify_const_unary_operation): For
35816 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35817 the sNaN unmodified.
35819 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35821 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35822 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35823 instead of SYSTEM_HEADER_DIR.
35825 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35826 Martin Liška <mliska@suse.cz>
35828 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35829 Fix typos and grammar, use active voice, and clarify.
35831 2017-02-20 Marek Polacek <polacek@redhat.com>
35833 PR middle-end/79537
35834 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35837 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35839 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35842 * config/i386/i386.c (ix86_expand_builtin): Handle
35843 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35844 ix86_builtins_isa[fcode].isa as a requirement of those
35845 flags and any other flag in the bitmask.
35846 (ix86_init_mmx_sse_builtins): Use 0 instead of
35847 ~OPTION_MASK_ISA_64BIT as mask.
35848 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35849 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35850 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35851 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35853 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35856 * lra-constraints.c (split_reg): Check requested split mode
35857 is supported by the register.
35859 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35861 * lra-constraints.c (simplify_operand_subreg): Remove early
35864 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35867 * lra-constraints.c (curr_insn_transform): Tighten condition
35868 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35870 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35873 * lra-constraints.c (curr_insn_transform): Handle
35874 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35876 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35879 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35881 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35883 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35886 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35889 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35891 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35894 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35896 * final.c (last_columnnum, override_columnnum): New variables.
35897 (final_start_function): Set last_columnnum, pass it to begin_prologue
35898 hook and pass 0 to dwarf2out_begin_prologue.
35899 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35900 to source_line debug hook.
35901 (notice_source_line): Compute last_columnnum and for debug_column_info
35902 return true on column changes.
35903 * debug.h (struct gcc_debug_hooks): Add column argument to
35904 source_line and begin_prologue hooks.
35905 (debug_nothing_int_charstar_int_bool): Remove prototype.
35906 (debug_nothing_int_int_charstar,
35907 debug_nothing_int_int_charstar_int_bool): New prototypes.
35908 (dwarf2out_begin_prologue): Add column argument.
35909 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35910 begin_prologue hooks.
35911 (debug_nothing_int_charstar_int_bool): Remove.
35912 (debug_nothing_int_int_charstar,
35913 debug_nothing_int_int_charstar_int_bool): New functions.
35914 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35915 through to dwarf2out_source_line.
35916 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35917 (dwarf2out_source_line): Add column argument, emit it if requested.
35918 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35920 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35921 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35922 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35923 through to dwarf2out_begin_prologue.
35924 (vmsdbgout_source_line): Add column argument, pass it through to
35925 dwarf2out_source_line.
35926 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35927 dbxout_source_line caller.
35928 (dbxout_source_line): Add column argument.
35930 * common.opt (gno-column-info, gcolumn-info): New options.
35931 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35932 (check_die): Also test for multiple DW_AT_decl_column attributes.
35933 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35934 DW_AT_decl_column if requested.
35935 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35937 (gen_variable_die): Likewise.
35938 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35939 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35942 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35943 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35944 (ix86_handle_option): Handle OPT_m3dnowa.
35945 * doc/invoke.texi (-m3dnowa): Document.
35946 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35947 -m3dnowa instead of -m3dnow -march=athlon.
35950 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35951 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35953 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35956 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35957 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35958 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35959 generator for vsx_xxpermdi_<mode>_be.
35960 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35961 force big-endian semantics.
35962 (vsx_xxpermdi_<mode>_be): New define_expand with same
35963 implementation as previous version of vsx_xxpermdi_<mode>.
35965 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35967 PR tree-optimization/79327
35968 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35969 variable, its initialization and use.
35971 2017-02-17 Julia Koval <julia.koval@intel.com>
35973 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35974 (OPTION_MASK_ISA_PKU_UNSET): New.
35975 (ix86_handle_option): Handle -mrdpid.
35976 * config/i386/cpuid.h (bit_RDPID): New.
35977 * config/i386/driver-i386.c (host_detect_local_cpu):
35978 Detect RDPID feature.
35979 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35980 * config/i386/i386-c.c (ix86_target_macros_internal):
35982 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35983 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35984 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35985 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35986 * config/i386/i386.md (define_insn "rdpid"): New.
35987 * config/i386/i386.opt Add -mrdpid.
35988 * config/i386/immintrin.h (_rdpid_u32): New.
35990 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35992 PR rtl-optimization/79541
35993 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35994 instead of transforming it into USE.
35996 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35998 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35999 If HONOR_SNANS (SFmode) force the input to a register.
36000 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36001 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36002 an frsp or similar insn.
36004 2017-02-17 Martin Liska <mliska@suse.cz>
36006 PR rtl-optimization/79577
36007 * params.def (selsched-max-sched-times): Increase minimum to 1.
36009 2017-02-17 Martin Liska <mliska@suse.cz>
36011 PR rtl-optimization/79574
36012 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36014 2017-02-17 Martin Liska <mliska@suse.cz>
36016 PR tree-optimization/79529
36017 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36018 ssa_defined_default_def_p to handle cases which are implicitly
36020 * tree-ssa.c (ssa_defined_default_def_p): New function.
36021 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36022 which are implicitly defined.
36023 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36025 2017-02-17 Richard Biener <rguenther@suse.de>
36027 PR middle-end/79576
36028 * params.def (max-ssa-name-query-depth): Limit to 10.
36030 2017-02-17 Richard Biener <rguenther@suse.de>
36032 PR tree-optimization/79552
36033 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36036 2017-02-17 Richard Biener <rguenther@suse.de>
36039 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36041 2017-02-17 Marek Polacek <polacek@redhat.com>
36043 PR middle-end/79536
36044 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36045 (fold_negate_expr): New wrapper.
36047 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36049 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36050 Correct terminology and de-emphasize pre-standard behavior.
36052 2017-02-16 Alan Modra <amodra@gmail.com>
36054 PR rtl-optimization/79286
36055 * ira.c (def_dominates_uses): New function.
36056 (update_equiv_regs): Don't create an equivalence for insns that
36057 may trap where the register def does not dominate the use.
36059 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36061 PR rtl-optimization/78127
36062 * lra.c (lra): Call lra_eliminate before finish the loop after
36065 2017-02-16 Richard Biener <rguenther@suse.de>
36067 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36069 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36070 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36071 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36072 (isl_val_int_from_wi): New function.
36073 (extract_affine_gmp): Rename to ...
36074 (extract_affine_wi): ... this, take a widest_int.
36075 (extract_affine_int): Just wrap extract_affine_wi.
36076 (add_param_constraints): Use isl_val_int_from_wi.
36077 (add_loop_constraints): Likewise, and extract_affine_wi.
36079 2017-02-15 Jeff Law <law@redhat.com>
36081 PR middle-end/79521
36082 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36083 ira_init_register_move_cost_if_necessary.
36085 2017-02-15 Martin Sebor <msebor@redhat.com>
36087 PR middle-end/32003
36088 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36089 removed in a prior commit.
36091 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36093 PR tree-optimization/79347
36094 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36095 counters during peeling.
36097 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36099 * Makefile.in (site.exp): Remove "set ISLVER".
36101 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36104 * real.c (real_from_integer): Call real_convert even for decimal.
36106 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36109 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36111 2017-02-14 Andrew Pinski <apinski@cavium.com>
36113 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36114 cores and change the partno/implementer to be correct.
36115 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36116 the 'B" as the implementer.
36117 * config/aarch64/aarch64-tune.md: Regenerate.
36119 2017-02-14 Carl Love <cel@us.ibm.com>
36121 * config/rs6000/rs6000.c: Add case statement entry to make the
36122 xvcvuxdsp built-in argument unsigned.
36123 * config/rs6000/vsx.md: Fix the source and return operand types so they
36124 match the instruction definitions from the ISA document. Fix typo
36125 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36128 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36131 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36132 member early_clobber_alts.
36133 * lra-lives.c (reg_early_clobber_p): New.
36134 (process_bb_lives): Use it.
36135 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36136 (debug_operand_data): Initialize early_clobber_alts.
36137 (setup_operand_alternative): Set up early_clobber_alts.
36138 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36139 alternatives to new_insn_reg.
36140 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36142 (lra_update_insn_regno_info): Pass the new arg.
36144 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36146 PR middle-end/79505
36147 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36148 (new_oacc_loop_raw): Don't clear already cleared fields.
36151 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36152 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36153 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36155 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36158 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36160 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36163 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36164 the extra instruction to the right place to store 128-bit constant
36167 2017-02-14 Martin Sebor <msebor@redhat.com>
36169 PR middle-end/79448
36170 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36171 warning for strings of unknown length.
36173 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36175 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36177 2017-02-14 Jeff Law <law@redhat.com>
36180 * ira-costs.c (scan_one_insn): Initialize register move costs
36181 for pseudos seen in USE/CLOBBER insns.
36183 PR tree-optimization/79095
36184 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36185 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36186 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36187 if the operands are known to be not equal, then the resulting range
36189 (intersect_ranges): If the new range is ~[0,0] and the old range is
36190 wide, then prefer ~[0,0].
36191 * tree-vrp.c (overflow_comparison_p_1): New function.
36192 (overflow_comparison_p): New function.
36193 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36194 if NAME is used in an overflow test.
36195 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36196 overflow check that can be expressed as an equality test, then adjust
36197 ops to be that equality test.
36199 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36201 * config/s390/s390-builtin-types.def: Remove flags argument.
36202 * config/s390/s390.c (s390_init_builtins): Likewise.
36204 2017-02-14 Martin Liska <mliska@suse.cz>
36206 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36207 vector. Fix trailing white spaces.
36209 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36211 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36214 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36216 PR rtl-optimization/68664
36217 * config/arm/arm.c (arm_sched_can_speculate_insn):
36218 New function. Declare prototype.
36219 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36221 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36223 PR rtl-optimization/68664
36224 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36226 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36228 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36230 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36231 max skip bytes for function, loop and jump.
36233 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36235 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36236 ABS_EXPR for gimple dump.
36238 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36241 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36243 PR tree-optimization/79408
36244 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36245 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36246 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36247 also if rhs1 is INTEGER_CST.
36249 2017-02-14 Richard Biener <rguenther@suse.de>
36251 PR middle-end/79432
36252 * tree-into-ssa.c (insert_phi_nodes): When the function can
36253 have abnormal edges rewrite SSA names with broken use-def
36254 dominance out of SSA and register them for PHI insertion.
36256 2017-02-13 Martin Sebor <msebor@redhat.com>
36258 PR middle-end/79496
36259 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36260 clearing info.nowrite flag when snprintf size argument is a range.
36262 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36264 * cprop.c (cprop_jump): Add missing space in string literal.
36265 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36266 (get_constraint_for_component_ref): Likewise.
36267 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36268 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36269 * lra-constraints.c (process_alt_operands): Likewise.
36270 * ipa-inline.c (inline_small_functions): Likewise.
36271 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36272 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36273 * trans-mem.c (diagnose_tm_1_op): Likewise.
36274 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36275 (grid_parallel_clauses_gridifiable): Likewise.
36277 * config/nvptx/mkoffload.c (process): Add space in between
36280 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36281 "MOD4_SSE_REGS" and "ALL_REGS".
36283 * spellcheck.c (test_data): Add , in between "foo" and "food".
36285 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36288 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36289 boundary crossing check and subsequent code generation agree.
36291 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36293 * config/aarch64/aarch64.c (has_memory_op): Delete.
36294 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36297 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36299 PR rtl-optimization/79388
36300 PR rtl-optimization/79450
36301 * combine.c (distribute_notes): When removing TEM_INSN for which
36302 corresponding dest has last value recorded, invalidate that last
36305 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36307 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36308 of explicit '@'. Add missing assembly comment marker on branch costs
36311 2017-02-13 Nathan Sidwell <nathan@acm.org>
36313 * gengtype-lex.l (<in_struct>): Add '/'.
36315 2017-02-13 Martin Liska <mliska@suse.cz>
36318 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36320 2017-02-13 Richard Biener <rguenther@suse.de>
36322 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36324 * configure: Re-generate.
36325 * config.in: Likewise.
36326 * graphite-dependences.c: Simplify as if
36327 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36328 * graphite-isl-ast-to-gimple.c: Likewise.
36329 * graphite-optimize-isl.c: Likewise.
36330 * graphite-poly.c: Likewise.
36331 * graphite-sese-to-poly.c: Likewise.
36332 * graphite.h: Likewise.
36333 * toplev.c: Include isl/version.h and use isl_version () for
36334 printing the ISL version.
36335 * doc/install.texi: Update ISL requirement.
36337 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36339 * doc/standards.texi (Standards): Update reference to
36342 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36344 * doc/extend.texi (Named Address Spaces): sourceware.org now
36346 * doc/install.texi (Binaries): Ditto.
36349 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36351 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36352 standard terminology "stringize"/"stringizing" throughout.
36353 * doc/cppinternals.texi: Likewise.
36355 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36357 * doc/extend.texi: Fix some spelling mistakes and typos.
36358 * doc/invoke.texi: Likewise.
36360 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36363 * params.def (inline-min-speedup) Change from 10 to 8.
36365 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36367 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36370 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36373 * ipa-inline-analysis.c (get_minimal_bb): New function.
36374 (record_modified): Use it.
36375 (remap_edge_change_prob): Handle also ancestor functions.
36377 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36379 * doc/contrib.texi (Contributors): Remove broken link into
36380 the Mauve CVS repository.
36382 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36384 PR middle-end/79454
36385 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36386 result computation whenever lhs doesn't have vector mode, not
36387 just when it has BLKmode.
36389 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36391 * doc/makefile.texi (profiledbootstrap): Refer to the
36392 installation instructions only in textual form.
36394 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36397 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36399 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36401 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36402 (Specific): Update mingw-w64 reference.
36404 (Specific): Remove broken link to Renesas RX processor.
36406 2017-02-10 Richard Biener <rguenther@suse.de>
36408 * toplev.c (process_options): Do not mention obsolete graphite
36409 options when printing sorry message about missing graphite support.
36410 Mention -floop-nest-optimize.
36412 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36414 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36415 (vtst_p16): Likewise.
36416 (vtstq_p8): Likewise.
36417 (vtstq_p16): Likewise.
36419 (vtstq_p64): Likewise.
36420 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36421 (vset_lane_p64): New.
36422 (vsetq_lane_p64): New.
36424 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36426 PR tree-optimization/79411
36427 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36428 stmt operands are SSA_NAMEs used in abnormal phis.
36429 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36432 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36435 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36438 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36440 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36442 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36444 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36448 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36449 not arbitrary TREE_CONSTANT.
36452 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36453 "omp declare target link" attribute unless is_global_var.
36454 * omp-offload.c (find_link_var_op): Likewise.
36456 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36457 Chung-Lin Tang <cltang@codesourcery.com>
36459 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36461 (gimplify_adjust_omp_clauses): Don't delete TILE.
36462 (gimplify_omp_for): Deal with TILE.
36463 * internal-fn.c (expand_GOACC_TILE): New function.
36464 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36466 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36467 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36469 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36470 avoid DIV for outermost collapse var.
36471 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36472 Remove out of date comments, fix whitespace.
36473 * omp-general.c (omp_extract_for_data): Deal with tiling.
36474 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36475 adjust OLF_DIM_BASE value.
36476 (struct omp_for_data): Add tiling field.
36477 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36478 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36479 for auto loops. Remove default auto determining, moved to
36480 oacc_loop_fixed_partitions.
36481 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36482 stmts, add e_mask field.
36483 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36484 (oacc_thread_numbers): Use oacc_dim_call.
36485 (oacc_xform_tile): New.
36486 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36487 (finish_oacc_loop): Adjust for ifns vector.
36488 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36489 add case for GOACC_TILE fns.
36490 (oacc_loop_xform_loop): Delete.
36491 (oacc_loop_process): Iterate over call list directly, and add
36492 handling for GOACC_TILE fns.
36493 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36495 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36496 vector partitioning to outer loops. Assign 2 partitions to loops
36497 when available. Add TILE handling.
36498 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36499 (execite_oacc_device_lower): Process GOACC_TILE fns,
36500 ignore unknown specs.
36501 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36502 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36503 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36505 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36507 * configure.ac (ACX_BUGURL): Update.
36508 * configure: Regenerate.
36510 2017-02-09 Richard Biener <rguenther@suse.de>
36512 PR tree-optimization/69823
36513 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36514 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36516 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36518 * config/arc/arc-c.def: Add __NPS400__ definition.
36519 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36520 (TARGET_NPS400): Define.
36522 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36524 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36526 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36527 pointer, arch_info.
36528 (arc_cpu_types): Fill the arch_info field with a pointer into the
36529 arc_arch_types table.
36530 (arc_selected_cpu): Declare.
36531 * config/arc/arc.c (arc_selected_cpu): Make global.
36532 (arc_selected_arch): Delete.
36533 (arc_base_cpu): Delete.
36534 (arc_override_options): Remove references to deleted variables,
36535 update access to arch information.
36536 (ARC_OPT): Update access to arch information.
36537 (ARC_OPTX): Likewise.
36538 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36539 (TARGET_ARC600): Update access to arch information.
36540 (TARGET_ARC601): Likewise.
36541 (TARGET_ARC700): Likewise.
36542 (TARGET_EM): Likewise.
36543 (TARGET_HS): Likewise.
36544 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36547 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36550 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36551 condition/operands for integer GE/LE/GEU/LEU operations.
36553 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36555 PR translation/79397
36556 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36559 2017-02-08 Martin Jambor <mjambor@suse.cz>
36562 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36563 whether allocation happened.
36564 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36565 nothing was allocated.
36567 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36569 PR tree-optimization/79408
36570 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36571 constant, but SSA_NAME with a known integer range, use the minimum
36572 of that range instead of op1 to determine if modulo can be replaced
36573 with its first operand.
36575 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36577 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36579 2017-02-08 Richard Biener <rguenther@suse.de>
36581 PR tree-optimization/71824
36582 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36583 Check all loops contained in the merged region.
36585 2017-02-07 Andrew Pinski <apinski@cavium.com>
36587 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36589 2017-02-07 Andrew Pinski <apinski@cavium.com>
36591 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36592 (thunderxt88): Likewise.
36593 (thunderxt81): Disable LSE and change v8.1 to v8.
36594 (thunderxt83): Likewise.
36596 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36597 Richard Biener <rguenther@suse.de>
36599 PR middle-end/79399
36600 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36601 type from int to size_t.
36602 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36604 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36606 PR rtl-optimization/79386
36607 * cprop.c (bypass_conditional_jumps): Initialize
36608 bypass_last_basic_block already before splitting bbs after
36609 unconditional traps...
36610 (bypass_conditional_jumps): ... rather than here.
36613 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36614 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36615 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36616 fix -masm=intel patterns.
36618 2017-02-07 Richard Biener <rguenther@suse.de>
36620 PR tree-optimization/79256
36621 PR middle-end/79278
36622 * builtins.c (get_object_alignment_2): Use min_align_of_type
36623 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36624 and ADJUST_FIELD_ALIGN.
36626 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36628 * doc/tm.texi: Regenerate.
36629 * stor-layout.c (layout_decl): Adjust.
36630 (update_alignment_for_field): Likewise.
36631 (place_field): Likewise.
36632 (min_align_of_type): Likewise.
36633 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36634 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36635 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36636 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36637 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36638 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36639 * config/i386/i386.c (x86_field_alignment): Likewise.
36640 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36641 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36642 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36643 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36644 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36645 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36649 2017-01-30 Richard Biener <rguenther@suse.de>
36651 PR tree-optimization/79256
36652 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36653 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36656 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36658 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36659 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36660 builtins to SImode and emit a zero-extend, if necessary.
36662 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36664 * docs/invoke.texi (RISC-V Options): Alphabetize.
36666 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36668 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36671 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36673 * config/riscv/riscv.c: New file.
36674 * common/config/riscv/riscv-common.c: Likewise.
36675 * config.gcc: Likewise.
36676 * config/riscv/constraints.md: Likewise.
36677 * config/riscv/elf.h: Likewise.
36678 * config/riscv/generic.md: Likewise.
36679 * config/riscv/linux.h: Likewise.
36680 * config/riscv/multilib-generator: Likewise.
36681 * config/riscv/peephole.md: Likewise.
36682 * config/riscv/pic.md: Likewise.
36683 * config/riscv/predicates.md: Likewise.
36684 * config/riscv/riscv-builtins.c: Likewise.
36685 * config/riscv/riscv-c.c: Likewise.
36686 * config/riscv/riscv-ftypes.def: Likewise.
36687 * config/riscv/riscv-modes.def: Likewise.
36688 * config/riscv/riscv-opts.h: Likewise.
36689 * config/riscv/riscv-protos.h: Likewise.
36690 * config/riscv/riscv.h: Likewise.
36691 * config/riscv/riscv.md: Likewise.
36692 * config/riscv/riscv.opt: Likewise.
36693 * config/riscv/sync.md: Likewise.
36694 * config/riscv/t-elf-multilib: Likewise.
36695 * config/riscv/t-linux: Likewise.
36696 * config/riscv/t-linux-multilib: Likewise.
36697 * config/riscv/t-riscv: Likewise.
36698 * configure.ac: Likewise.
36699 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
36700 Waterman as RISC-V maintainers.
36701 * doc/install.texi: Add RISC-V entries.
36702 * doc/invoke.texi: Add RISC-V options section.
36703 * doc/md.texi: Add RISC-V constraints section.
36704 * configure: Regenerated.
36706 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
36709 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
36710 false values to be constant vectors with all 0 or all 1 bits set.
36711 (vcondu<mode><mode>): Likewise.
36712 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
36714 (fpmask_comparison_operator): Update comment.
36715 (vecint_comparison_operator): New predicate.
36716 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
36717 vector conditionals when the true and false values are constant
36718 vectors with all 0 bits or all 1 bits set.
36720 2017-02-06 Martin Sebor <msebor@redhat.com>
36722 PR tree-optimization/79376
36723 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36725 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36727 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36728 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36729 to simplify split condition.
36731 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36733 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36734 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36737 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36739 PR rtl-optimization/68664
36740 * target.def (can_speculate_insn): New hook.
36741 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36742 * doc/tm.texi: Regenerate.
36743 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36744 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36745 (rs6000_sched_can_speculate_insn): New function.
36747 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36749 PR tree-optimization/79284
36750 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36751 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36752 vectorizable_mask_load_store, vectorizable_operation,
36753 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36754 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36755 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36756 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36757 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36758 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36759 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36760 is_gimple_assign (stmt). Replace another such test with
36761 is_gimple_assign (stmt).
36763 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36766 * config/avr/avr.c (rtl-iter.h): Include it.
36767 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36768 (avr_legitimate_combined_insn): ...and implementation.
36770 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36772 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36773 * config/s390/s390.c (s390_const_operand_ok)
36774 (s390_canonicalize_comparison, s390_extract_part)
36775 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36776 (s390_contiguous_bitmask_p, s390_rtx_costs)
36777 (legitimize_pic_address): Likewise.
36778 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36779 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36780 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36781 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36782 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36784 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36786 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36787 REGNO($0) == REGNO($1).
36789 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36791 * config/s390/linux.h(SIZE_TYPE): Add comment.
36793 2017-02-06 Julian Brown <julian@codesourcery.com>
36794 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36795 Virendra Pathak <virendra.pathak@broadcom.com>
36797 * config/aarch64/aarch64-cores.def: Change the scheduler
36799 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36800 * config/aarch64/thunderx2t99.md: New file.
36802 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36804 * doc/standards.texi (Go Language): Update link to language
36807 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36809 * tree-eh.c (lower_resx): Sanitize profile.
36810 (cleanup_empty_eh_move_lp): Likewise.
36812 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36815 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36817 * cfgloopmanip.h (loop_version): Update prototype.
36818 * modulo-sched.c (sms_schedule): Update call of loop_version.
36819 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36820 * tree-parloops.c (gen_parallel_loop): Likewise.
36821 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36822 * tree-ssa-loop-split.c (split_loop): Likewise.
36823 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36824 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36826 2017-02-05 Martin Liska <mliska@suse.cz>
36829 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36831 (print_operand_address): Initialize a struct to zero.
36833 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36835 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36836 garbage collector only in textual form.
36838 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36840 * doc/extend.texi (x86 specific memory model extensions for
36841 transactional memory): Simplify a phrase.
36843 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36846 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36847 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36848 (atomic_storedi_1): Likewise.
36850 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36852 PR tree-optimization/79338
36853 * tree-parloops.c (gather_scalar_reductions): Don't call
36854 vect_analyze_loop_form for loop->inner before destroying loop's
36857 2017-02-03 Martin Sebor <msebor@redhat.com>
36859 PR tree-optimization/79327
36860 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36861 when precision has resulted in leading zeros.
36862 (format_integer): Adjust the likely counter to assume an unknown
36863 argument that may be zero is non-zero.
36865 2017-02-03 Jason Merrill <jason@redhat.com>
36868 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36869 avoid copying non-taken branch.
36871 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36873 PR tree-optimization/79340
36874 * tree-vect-loop.c (vectorizable_reduction): Release
36875 vec_defs elements after safe_splicing them into other vectors.
36878 PR tree-optimization/79327
36879 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36880 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36882 (format_integer): Use wide_int_to_tree instead of build_int_cst
36883 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36884 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36885 of shortest and longest sequence.
36887 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36889 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36890 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36892 2017-02-03 Walter Lee <walt@tilera.com>
36895 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36896 after initial stackframe link reg save.
36897 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36899 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36902 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36903 wu for stxssp alternative.
36905 2017-02-03 Martin Sebor <msebor@redhat.com>
36907 PR tree-optimization/79352
36908 * gimple-fold.c (get_range_strlen): Add argument.
36909 (get_range_strlen): Change return type to bool.
36910 (get_maxval_strlen): Pass in a dummy argument.
36911 * gimple-fold.h (get_range_strlen): Change return type to bool.
36912 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36913 * tree.h (array_at_struct_end_p): Add argument.
36914 * tree.c (array_at_struct_end_p): Handle it.
36916 2017-02-03 Martin Liska <mliska@suse.cz>
36919 * multiple_target.c (create_dispatcher_calls): Redirect edge
36920 from a caller of a dispatcher.
36921 (expand_target_clones): Make the clones local.
36922 (ipa_target_clone): Do both target clones and resolvers.
36923 (ipa_dispatcher_calls): Remove the pass.
36924 (pass_dispatcher_calls::gate): Likewise.
36925 (make_pass_dispatcher_calls): Likewise.
36926 * passes.def (pass_target_clone): Put as very first IPA early
36929 2017-02-03 Martin Liska <mliska@suse.cz>
36931 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36932 in case of a function with ifunc attribute.
36934 2017-02-03 Martin Liska <mliska@suse.cz>
36936 * cgraph.c (cgraph_node::dump): Dump function version info.
36937 * symtab.c (symtab_node::dump_base): Add missing new line.
36939 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36941 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36942 (ifcombine_ifandif): Use it.
36944 2017-02-03 Martin Liska <mliska@suse.cz>
36946 * doc/invoke.texi: Document default value for
36947 use-after-scope-direct-emission-threshold.
36949 2017-02-03 Martin Liska <mliska@suse.cz>
36951 PR tree-optimization/79339
36952 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36953 (format_floating): Likewise.
36955 2017-02-03 Martin Liska <mliska@suse.cz>
36958 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36960 (ipa_node_params_t::remove): Likewise.
36961 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36962 initialization from removed ipa_node_params_t::insert.
36963 (ipa_node_params::~ipa_node_params): Move from removed
36964 ipa_node_params_t::release.
36965 * symbol-summary.h (symbol_summary::m_released): New member.
36966 Do not release a summary twice. Do not allow to call finalizer
36967 for types of a summary that live in GGC memory.
36969 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36971 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36974 2017-02-02 Martin Sebor <msebor@redhat.com>
36976 PR middle-end/79275
36977 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36978 (format_string): Tighten up the range of output for non-constant
36979 strings and correct the expected range for wide non-constant strings.
36981 2017-02-02 Martin Sebor <msebor@redhat.com>
36983 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36985 PR middle-end/32003
36986 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36988 (-fdump-tree-@var): Add to index and document how to come up
36989 with pass-specific option and dump file names.
36990 (-fdump-passes): Clarify where to look for output.
36992 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36994 PR middle-end/77445
36995 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36996 statistics of the analyzed path; allow threading for speed when
36997 any of BBs along the path are optimized for speed.
36999 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37001 PR middle-end/78468
37002 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37003 settings of the virtual registers.
37006 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37008 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37009 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37012 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37014 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37015 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37017 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37019 * config/s390/s390.md: Add missing comments with the expanded
37021 * config/s390/vector.md: Likewise.
37022 * config/s390/vx-builtins.md: Likewise.
37024 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37027 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37028 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37029 conditions on a single line.
37031 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37033 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37034 __S390_VX__ to __VX__.
37036 2017-02-01 Andrew Pinski <apinski@cavium.com>
37038 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37039 stmt_info to record_stmt_cost.
37040 (vect_get_known_peeling_cost): Pass stmt_info if known to
37042 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37043 cpu_vector_cost field into
37044 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37045 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37046 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37047 splitting of scalar_stmt_cost and vec_stmt_cost.
37048 (thunderx_vector_cost): Likewise.
37049 (cortexa57_vector_cost): LIkewise.
37050 (exynosm1_vector_cost): Likewise.
37051 (xgene1_vector_cost): Likewise.
37052 (thunderx2t99_vector_cost): Improve after the splitting of the two
37054 (aarch64_builtin_vectorization_cost): Update for the splitting of
37055 scalar_stmt_cost and vec_stmt_cost.
37057 2017-02-01 Torvald Riegel <triegel@redhat.com>
37058 Richard Henderson <rth@redhat.com>
37060 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37061 conditional on existance of a fast atomic load.
37062 * optabs-query.c (can_atomic_load_p): New function.
37063 * optabs-query.h (can_atomic_load_p): Declare it.
37064 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37065 no fast atomic load is available for the particular size of access.
37066 (expand_atomic_compare_and_swap): Likewise.
37067 (expand_atomic_load): Likewise.
37068 (expand_atomic_store): Likewise.
37069 (expand_atomic_fetch_op): Likewise.
37070 * testsuite/lib/target-supports.exp
37071 (check_effective_target_sync_int_128): Remove x86 because it provides
37072 no fast atomic load.
37073 (check_effective_target_sync_int_128_runtime): Likewise.
37075 2017-02-01 Richard Biener <rguenther@suse.de>
37077 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37078 (graphite_transform_loops): Provide opt-info for optimized nests.
37079 * tree-parloop.c (parallelize_loops): Provide opt-info for
37080 parallelized loops.
37082 2017-02-01 Richard Biener <rguenther@suse.de>
37084 PR middle-end/79315
37085 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37086 was not set before.
37088 2017-02-01 Richard Biener <rguenther@suse.de>
37090 PR tree-optimization/71824
37091 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37092 Verify the loops are valid in the merged SESE region.
37093 (scop_detection::can_represent_loop_1): Check analyzing the
37094 evolution of the number of iterations in the region succeeds.
37096 2017-01-31 Ian Lance Taylor <iant@golang.org>
37098 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37099 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37101 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37103 PR preprocessor/79210
37104 * input.c (get_substring_ranges_for_loc): Replace line_width
37105 assertion with error-handling.
37107 2017-01-31 Richard Biener <rguenther@suse.de>
37109 PR tree-optimization/77318
37110 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37111 (create_pw_aff_from_tree): Take loop parameter.
37112 (add_condition_to_pbb): Pass loop of the condition to
37113 create_pw_aff_from_tree.
37115 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37117 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37118 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37120 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37124 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37126 (convert_int_to_float128): Likewise.
37127 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37128 (convert_int_to_float128): Likewise.
37129 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37130 (UNSPEC_IEEE128_CONVERT): Likewise.
37131 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37132 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37133 Use local variables for IBM extended format.
37134 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37135 (fix_trunc<mode>si2_fprs): Likewise.
37136 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37137 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37138 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37139 to know that we can now have integers of all sizes in vector
37141 (fix<uns>_<mode>di2_hw): Likewise.
37142 (float<uns>_<mode>si2_hw): Likewise.
37143 (fix_<mode>si2_hw): Likewise.
37144 (fixuns_<mode>si2_hw): Likewise.
37145 (float<uns>_<mode>di2_hw): Likewise.
37146 (float_<mode>di2_hw): Likewise.
37147 (float_<mode>si2_hw): Likewise.
37148 (floatuns_<mode>di2_hw): Likewise.
37149 (floatuns_<mode>si2_hw): Likewise.
37150 (xscvqp<su>wz_<mode>): Delete, no longer used.
37151 (xscvqp<su>dz_<mode>): Likewise.
37152 (xscv<su>dqp_<mode>): Likewise.
37153 (ieee128_mfvsrd_64bit): Likewise.
37154 (ieee128_mfvsrd_32bit): Likewise.
37155 (ieee128_mfvsrwz): Likewise.
37156 (ieee128_mtvsrw): Likewise.
37157 (ieee128_mtvsrd_64bit): Likewise.
37158 (ieee128_mtvsrd_32bit): Likewise.
37160 2017-01-31 Martin Liska <mliska@suse.cz>
37163 * ipa-prop.c (ipa_free_all_node_params): Call release method
37164 instead of ~sumbol_summary to not to trigger double times
37167 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37169 PR tree-optimization/71691
37170 * bitmap.h (class auto_bitmap): New.
37171 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37172 is_maybe_undefined instead of ssa_undefined_value_p.
37174 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37176 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37177 __S390_ARCH_LEVEL__ to __ARCH__.
37179 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37181 PR tree-optimization/79267
37182 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37183 if should_remove_lhs_p is true.
37185 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37188 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37189 (add_alignment_attribute): New.
37190 (base_type_die): Add alignment attribute.
37191 (subrange_type_die): Likewise.
37192 (modified_type_die): Likewise.
37193 (gen_array_type_die): Likewise.
37194 (gen_descr_array_type_die: Likewise.
37195 (gen_enumeration_type_die): Likewise.
37196 (gen_subprogram_die): Likewise.
37197 (gen_variable_die): Likewise.
37198 (gen_field_die): Likewise.
37199 (gen_ptr_to_mbr_type_die): Likewise.
37200 (gen_struct_or_union_type_die): Likewise.
37201 (gen_subroutine_type_die): Likewise.
37202 (gen_typedef_die): Likewise.
37203 (base_type_cmp): Compare alignment attribute.
37205 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37208 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37209 (setb_unsigned) New pattern for setb with CCUNS.
37210 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37211 subfc./subfe sequence to avoid overflow problems. Generate a
37212 shorter sequence with cmpld/setb for power9.
37213 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37214 for generating subfc. instruction.
37215 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37216 now uses this instruction.
37218 2017-01-30 Ian Lance Taylor <iant@google.com>
37221 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37222 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37224 2017-01-30 Martin Sebor <msebor@redhat.com>
37226 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37227 Move constant to the right of a relational operator.
37228 (get_mpfr_format_length, format_character, format_string): Ditto.
37229 (should_warn_p, maybe_warn): Same.
37231 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37233 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37236 * asan.c (get_translation_unit_decl): Remove function.
37237 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37239 2017-01-30 Martin Liska <mliska@suse.cz>
37241 PR gcov-profile/79259
37242 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37243 -fprofile-generate.
37245 2017-01-30 Martin Liska <mliska@suse.cz>
37248 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37249 Initialize variables with NULL value.
37251 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37254 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37256 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37258 2017-01-30 Richard Biener <rguenther@suse.de>
37260 PR tree-optimization/79276
37261 * tree-vrp.c (process_assert_insertions): Properly adjust common
37262 when removing a duplicate.
37264 2017-01-30 Richard Biener <rguenther@suse.de>
37266 PR tree-optimization/79256
37267 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37268 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37270 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37272 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37275 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37276 ("*r<noxa>sbg_<mode>_sll_bitmask")
37277 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37278 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37279 Use contiguous_bitmask_nowrap_operand.
37281 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37284 * config/rs6000/altivec.h (vec_xl): Revise #define.
37285 (vec_xst): Likewise.
37287 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37289 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37291 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37293 PR rtl-optimization/79194
37294 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37295 traps before call to bypass_conditional_jumps.
37297 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37299 PR tree-optimization/71374
37300 * lra-constraints.c (check_conflict_input_operands): New.
37301 (match_reload): Use it.
37303 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37306 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37307 account to calculate conflict_set.
37309 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37311 PR rtl-optimization/78559
37312 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37313 other_insn in combine.
37315 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37317 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37318 uint16_type_node for BT_UINT16.
37320 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37322 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37323 "RTL Tests" to menu.
37324 (GIMPLE Tests): New node.
37325 (RTL Tests): New node.
37327 2017-01-27 Richard Biener <rguenther@suse.de>
37329 PR tree-optimization/79245
37330 * tree-loop-distribution.c (distribute_loop): Apply cost
37331 modeling also to detected patterns.
37333 2017-01-27 Richard Biener <rguenther@suse.de>
37335 PR tree-optimization/71433
37336 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37337 (compare_assert_loc): New function.
37338 (process_assert_insertions): Sort and optimize assert locations
37339 to remove duplicates and push down identical assertions on
37340 edges to their destination block.
37342 2017-01-27 Richard Biener <rguenther@suse.de>
37344 PR tree-optimization/79244
37345 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37346 out SSA names even if abnormal.
37348 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37350 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37351 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37352 instead of MPFR_RNDN.
37354 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37357 * arm.c (arm_option_override): Don't call build_target_option_node
37358 until after doing all option overrides.
37359 (arm_valid_target_attribute_tree): Likewise.
37361 2017-01-27 Martin Liska <mliska@suse.cz>
37363 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37364 for {cd}tors and C++ {cd}tors.
37366 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37368 * config/s390/s390.md ("*setmem_long_and")
37369 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37371 2017-01-26 Martin Sebor <msebor@redhat.com>
37373 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37376 2017-01-26 Martin Sebor <msebor@redhat.com>
37378 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37379 HAVE_DFmode before using XFmode or DFmode.
37380 (parse_directive): Avoid using the z length modifier to avoid
37381 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37383 PR middle-end/78703
37384 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37385 to accept adjustment as an array.
37386 (get_int_range): New function.
37387 (struct directive): Make width and prec arrays.
37388 (directive::set_width, directive::set_precision): Call get_int_range.
37389 (format_integer, format_floating): Handle width and precision ranges.
37390 (format_string, parse_directive): Same.
37392 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37395 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37396 comdat_type_p set, just clone them, but keep the children in the
37400 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37401 which have direct callers with -fvar-tracking-assignments enabled
37403 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37404 inside of type units.
37406 2017-01-26 Martin Sebor <msebor@redhat.com>
37408 PR middle-end/78703
37409 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37411 (struct format_result): Replace number_chars, number_chars_min,
37412 and number_chars_max with a single member of struct result_range.
37414 (format_result::operator+=): Adjust.
37415 (struct fmtresult): Remove bounded. Handle likely and unlikely
37417 (fmtresult::adjust_for_width_or_precision): New function.
37418 (fmtresult:type_max_digits): New function.
37419 (bytes_remaining): Handle likely and unlikely counters.
37420 (min_bytes_remaining): Remove.
37421 (format_percent): Simplify.
37422 (format_integer, format_floating): Set likely and unlikely counters.
37423 (get_string_length, format_character, format_string): Same.
37424 (format_plain, should_warn_p): New function.
37425 (maybe_warn): Call should_warn_p. Update diagnostic messages
37426 and handle those for all directives, including plain strings.
37427 (format_directive): Handle likely and unlikely counters.
37428 Remove unnecessary quoting from diagnostics. Add an informational
37430 (add_bytes): Remove.
37431 (pass_sprintf_length::compute_format_length): Simplify.
37432 (try_substitute_return_value): Handle likely and unlikely counters.
37434 2017-01-26 Carl Love <cel@us.ibm.com>
37436 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37437 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37439 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37442 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37443 endianess for subregs into account.
37444 * lra-constraints.c (lra_constraints): Do risky transformations
37445 always on the first iteration.
37446 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37447 last_call_used_reg_set.
37448 (process_bb_lives): Define and use last_call_used_reg_set.
37449 * lra.c (lra): Always continue after lra_constraints on the first
37452 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37454 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37456 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37458 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37460 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37461 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37462 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37463 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37464 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37465 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37466 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37467 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37468 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37470 2017-01-26 Marek Polacek <polacek@redhat.com>
37473 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37474 for the third operand.
37476 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37478 PR middle-end/79236
37479 * omp-low.c (struct omp_context): Add simt_stmt field.
37480 (scan_omp_for): Return omp_context *.
37481 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37482 context to the _simt_ SIMD stmt.
37483 (lower_omp_for): For combined SIMD with sibling _simt_
37484 SIMD, make sure to use the same decls in _looptemp_
37485 clauses as in the sibling.
37487 2017-01-26 David Sherwood <david.sherwood@arm.com>
37489 PR middle-end/79212
37490 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37493 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37496 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37497 emit fld b; fld a; if possible.
37499 * brig-builtins.def: Update copyright years.
37500 * config/arm/arm_acle_builtins.def: Update copyright years.
37502 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37505 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37506 constraint instead of o for the stxsd instruction.
37508 2017-01-25 Carl Love <cel@us.ibm.com>
37510 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37511 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37513 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37515 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37517 2017-01-25 Richard Biener <rguenther@suse.de>
37519 PR tree-optimization/69264
37520 * target.def (vector_alignment_reachable): Improve documentation.
37521 * doc/tm.texi: Regenerate.
37522 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37524 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37525 earlier changes with respect to TYPE_USER_ALIGN.
37526 (vector_alignment_reachable_p): Likewise. Improve dumping.
37528 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37531 * config/arm/arm.md (xordi3): Force constant operand into a register
37534 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37536 * doc/invoke.texi (-fstore-merging): Correct default optimization
37537 levels at which it is enabled.
37538 (-O): Move -fstore-merging from list to...
37541 2017-01-25 Richard Biener <rguenther@suse.de>
37544 * omp-expand.c: Include debug.h.
37545 (expand_omp_taskreg): Make sure to generate early debug before
37546 outlining anything from a function.
37547 (expand_omp_target): Likewise.
37548 (grid_expand_target_grid_body): Likewise.
37550 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37553 * asan.c (get_translation_unit_decl): New function.
37554 (asan_add_global): Extract modules file name from globals
37555 TRANSLATION_UNIT_DECL name.
37557 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37560 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37561 for long calls with APCS frame and VFP.
37563 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37565 * cfg.c (original_copy_tables_initialized_p): New function.
37566 * cfg.h (original_copy_tables_initialized_p): New decl.
37567 * cfgrtl.c (relink_block_chain): Guard the call to
37568 free_original_copy_tables with a call to
37569 original_copy_tables_initialized_p.
37570 * cgraph.h (symtab_node::native_rtl_p): New decl.
37571 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37572 (symtab_node::needed_p): Don't assert for early assembly output
37573 for __RTL functions.
37574 (cgraph_node::finalize_function): Set "force_output" for __RTL
37576 (cgraph_node::analyze): Bail out early for __RTL functions.
37577 (analyze_functions): Update assertion to support __RTL functions.
37578 (cgraph_node::expand): Bail out early for __RTL functions.
37579 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37581 * function.h (struct function): Update comment for field
37583 * gimple-expr.c: Include "tree-pass.h".
37584 (gimple_has_body_p): Return false for __RTL functions.
37585 * Makefile.in (OBJS): Add run-rtl-passes.o.
37586 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37588 (gcc::pass_manager::get_clean_slate): New accessor.
37589 * passes.c: Include "insn-addr.h".
37590 (should_skip_pass_p): Add logging. Update logic for running
37591 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37592 property-provider override so it is only done for gimple passes.
37594 (skip_pass): New function.
37595 (execute_one_pass): Call skip_pass when skipping passes.
37596 * read-md.c (md_reader::read_char): Support filtering
37597 the input to a subset of line numbers.
37598 (md_reader::md_reader): Initialize fields
37599 m_first_line and m_last_line.
37600 (md_reader::read_file_fragment): New function.
37601 * read-md.h (md_reader::read_file_fragment): New decl.
37602 (md_reader::m_first_line): New field.
37603 (md_reader::m_last_line): New field.
37604 * read-rtl-function.c (function_reader::create_function): Only
37605 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37606 curr_properties. Set DECL_INITIAL to a dummy block.
37607 (read_rtl_function_body_from_file_range): New function.
37608 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37610 * run-rtl-passes.c: New file.
37611 * run-rtl-passes.h: New file.
37613 2017-01-24 Jeff Law <law@redhat.com>
37615 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37618 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37620 PR tree-optimization/79159
37621 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37622 (record_nonwrapping_iv): Improve boundary using above function if no
37623 value range information.
37625 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37626 Martin Jambor <mjambor@suse.cz>
37628 * brig-builtins.def: New file.
37629 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37630 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37631 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37632 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37633 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37634 * builtin-types.def (BT_INT8): New.
37635 (BT_INT16): Likewise.
37636 (BT_UINT8): Likewise.
37637 (BT_UINT16): Likewise.
37638 (BT_FN_ULONG): Likewise.
37639 (BT_FN_UINT_INT): Likewise.
37640 (BT_FN_UINT_ULONG): Likewise.
37641 (BT_FN_UINT_LONG): Likewise.
37642 (BT_FN_UINT_PTR): Likewise.
37643 (BT_FN_ULONG_PTR): Likewise.
37644 (BT_FN_INT8_FLOAT): Likewise.
37645 (BT_FN_INT16_FLOAT): Likewise.
37646 (BT_FN_UINT32_FLOAT): Likewise.
37647 (BT_FN_UINT16_FLOAT): Likewise.
37648 (BT_FN_UINT8_FLOAT): Likewise.
37649 (BT_FN_UINT64_FLOAT): Likewise.
37650 (BT_FN_UINT16_UINT32): Likewise.
37651 (BT_FN_UINT32_UINT16): Likewise.
37652 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37653 (BT_FN_INT_PTR_INT): Likewise.
37654 (BT_FN_UINT_PTR_UINT): Likewise.
37655 (BT_FN_LONG_PTR_LONG): Likewise.
37656 (BT_FN_ULONG_PTR_ULONG): Likewise.
37657 (BT_FN_VOID_UINT64_UINT64): Likewise.
37658 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37659 (BT_FN_INT8_INT8_INT8): Likewise.
37660 (BT_FN_INT16_INT16_INT16): Likewise.
37661 (BT_FN_INT_INT_INT): Likewise.
37662 (BT_FN_UINT_FLOAT_UINT): Likewise.
37663 (BT_FN_FLOAT_UINT_UINT): Likewise.
37664 (BT_FN_ULONG_UINT_UINT): Likewise.
37665 (BT_FN_ULONG_UINT_PTR): Likewise.
37666 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37667 (BT_FN_UINT_UINT_UINT): Likewise.
37668 (BT_FN_VOID_UINT_PTR): Likewise.
37669 (BT_FN_UINT_UINT_PTR: Likewise.
37670 (BT_FN_UINT32_UINT64_PTR): Likewise.
37671 (BT_FN_INT_INT_UINT_UINT): Likewise.
37672 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37673 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37674 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37675 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37676 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37677 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37678 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37679 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37680 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37681 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37682 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37683 * doc/frontends.texi: List BRIG FE.
37684 * doc/install.texi (Testing): Add BRIG tesring requirements.
37685 * doc/invoke.texi (Overall Options): Mention BRIG.
37686 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37688 2017-01-24 Richard Biener <rguenther@suse.de>
37690 PR translation/79208
37691 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37693 2017-01-24 Martin Jambor <mjambor@suse.cz>
37696 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
37697 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
37698 and known_contexts.
37700 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
37702 PR middle-end/79123
37703 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
37704 casts from signed to unsigned really don't have a range.
37706 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
37708 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
37709 GMP_RNDx for compatiblity.
37711 2017-01-24 Martin Liska <mliska@suse.cz>
37714 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
37715 that would prevent us to call alloca with -1 as argument.
37717 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37719 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37720 Avoid -Wformat-security warning.
37722 2017-01-23 Andrew Pinski <apinski@cavium.com>
37724 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37727 2017-01-23 Martin Sebor <msebor@redhat.com>
37729 PR middle-end/78703
37730 * gimple-ssa-sprintf.c (warn_level): New global.
37731 (format_integer): Use it here and throughout the rest of the file.
37732 Use the same switch to compute sign as base.
37733 (maybe_warn): New function.
37734 (format_directive): Factor out warnings into maybe_warn.
37735 Add debugging output. Use warn_level.
37736 (add_bytes): Use warn_level.
37737 (pass_sprintf_length::compute_format_length): Add debugging output.
37738 (try_substitute_return_value): Same.
37739 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37741 PR middle-end/78703
37742 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37743 (struct fmtresult, format_integer, format_floating): Adjust.
37744 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37745 (get_string_length, format_string,format_directive): Same.
37746 (pass_sprintf_length::compute_format_length): Same.
37747 (try_substitute_return_value): Simplify slightly.
37749 PR middle-end/78703
37750 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37751 (fmtresult::operator+=): Outlined.
37752 (struct fmtresult): Add ctors.
37753 (struct conversion_spec): Rename...
37754 (struct directive): ...to this. Add and remove data members.
37755 (directive::set_width, directive::set_precision): New functions.
37756 (format_percent): Use fmtresult ctor.
37757 (get_width_and_precision): Remove.
37758 (format_integer): Make naming changes. Avoid computing width and
37760 (format_floating): Same. Adjust indentation.
37761 (format_character, format_none): New functions.
37762 (format_string): Moved character handling to format_character.
37763 (format_directive): Remove arguments, change return type.
37764 (parse_directive): New function.
37765 (pass_sprintf_length::compute_format_length): Move directive
37766 parsing to parse_directive.
37768 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37770 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37771 (assign_assembler_name_if_needed): ... this.
37772 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37773 (assign_assembler_name_if_needed): ... this.
37774 (free_lang_data_in_cgraph): Adjust callers.
37775 * cgraphunit.c (cgraph_node::analyze): Likewise.
37776 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37779 2017-01-23 Richard Biener <rguenther@suse.de>
37781 PR tree-optimization/79088
37782 PR tree-optimization/79188
37783 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37784 resetting loop bounds after last path deletion. Reset loop
37785 bounds of the target loop, make code match the comments.
37786 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37787 Make sure loops need no fixups.
37789 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37791 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37792 exponent support with double type for first argument.
37793 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37794 type returned by __builtin_vec_extract_sig,
37795 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37796 functions from "vector int" to "vector unsigned int" or from
37797 "vector long long int" to "vector unsigned long long int".
37798 Changed type returned by __builtin_vec_extract_exp,
37799 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37800 functions from "vector int" to "vector unsigned int" or from
37801 "vector long long int" to "vector unsigned long long int".
37802 Changed return type of __builtin_vec_test_data_class,
37803 __builtin_vec_test_data_class_sp, and
37804 __builtin_vec_test_data_class_dp from "vector int" to
37805 "vector bool int" or from "vector long long int" to "vector bool
37806 long long int" and changed second argument type from "unsigned
37807 int" to "int". Added new overloaded function forms "vector float
37808 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37809 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37810 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37811 double, vector unsigned long long int)" and "vector double
37812 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37813 long int)". Changed return type of
37814 __builtin_scalar_test_data_class and
37815 __builtin_scalar_test_data_class_sp and
37816 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37817 int" and changed second argument from "unsigned int" to "int".
37818 Changed type returned by __builtin_scalar_test_neg,
37819 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37820 from "int" to "bool int". Added new overloaded function form
37821 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37822 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37823 exponent double-precision with floating point first argument.
37824 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37825 documentation of scalar_test_data_class, scalar_test_neg,
37826 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37827 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37828 vec_test_data_class built-in functions to reflect refinements in
37829 their type signatures.
37831 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37833 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37835 (aarch64_elf_asm_destructor): Likewise.
37837 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37839 PR rtl-optimization/78634
37840 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37841 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37842 * ifcvt.c (noce_try_cmove): Add missing cost check.
37844 PR rtl-optimization/71724
37845 * combine.c (if_then_else_cond): Look for situations where it is
37846 beneficial to undo the work of one of the recursive calls.
37848 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37850 PR tree-optimization/70754
37851 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37852 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37853 combined stmt before it if not NULL.
37854 (combine_chains): Process refs reversely and compute dominance point
37857 2017-01-23 Martin Liska <mliska@suse.cz>
37859 PR tree-optimization/79196
37860 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37861 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37863 (strlen_optimize_stmt): Call the renamed function.
37865 2017-01-23 Michael Matz <matz@suse.de>
37867 PR tree-optimization/78384
37868 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37870 2017-01-23 Richard Biener <rguenther@suse.de>
37872 PR tree-optimization/79186
37873 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37874 both incoming edges before moving an assert.
37876 2017-01-23 Martin Jambor <mjambor@suse.cz>
37878 * ipa-prop.c (load_from_param_1): Removed.
37879 (load_from_unmodified_param): Bits from load_from_param_1 put back
37881 (load_from_param): Removed.
37882 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37883 with stmt. Reverted back to use of load_from_unmodified_param.
37885 2017-01-23 Martin Jambor <mjambor@suse.cz>
37888 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37889 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37890 field a pointer to garbage collected vector, mark lattices and
37891 ipcp_orig_node with GTY((skip)).
37892 (ipa_get_param_count): Adjust to descriptors being a pointer.
37893 (ipa_get_param): Likewise.
37894 (ipa_get_type): Likewise.
37895 (ipa_get_param_move_cost): Likewise.
37896 (ipa_set_param_used): Likewise.
37897 (ipa_get_controlled_uses): Likewise.
37898 (ipa_set_controlled_uses): Likewise.
37899 (ipa_is_param_used): Likewise.
37900 (ipa_node_params_t): Move into garbage collector. New methods insert
37902 (ipa_node_params_sum): Annotate wth GTY(()).
37903 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37905 (ipa_load_from_parm_agg): Adjust declaration.
37906 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37907 * ipa-profile.c (ipa_profile): Likewise.
37908 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37909 (ipa_populate_param_decls): Make descriptors parameter garbage
37911 (ipa_dump_param): Adjust to descriptors being a pointer.
37912 (ipa_alloc_node_params): Likewise.
37913 (ipa_initialize_node_params): Likewise.
37914 (load_from_param_1): Make descriptors parameter garbage collected.
37915 (load_from_unmodified_param): Likewise.
37916 (load_from_param): Likewise.
37917 (ipa_load_from_parm_agg): Likewise.
37918 (ipa_node_params::~ipa_node_params): Removed.
37919 (ipa_free_all_node_params): Remove call to delete operator.
37920 (ipa_node_params_t::insert): New.
37921 (ipa_node_params_t::remove): Likewise.
37922 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37923 copy known_csts and known_contexts vectors.
37924 (ipa_read_node_info): Adjust to descriptors being a pointer.
37925 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37927 (ipcp_transform_function): Make descriptors variable garbage
37930 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37932 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37933 * config/i386/avx512dqintrin.h: Ditto.
37934 * config/i386/avx512fintrin.h: Ditto.
37935 * config/i386/i386.c: Handle new builtins.
37936 * config/i386/i386-builtin.def: Add new builtins.
37937 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37938 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37940 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37941 Martin Liska <mliska@suse.cz>
37943 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37944 * asan.c (asan_expand_poison_ifn): Support stores and use
37945 appropriate ASAN report function.
37946 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37947 * internal-fn.def (ASAN_POISON_USE): Declare.
37948 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37949 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37950 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37951 ASAN_POISON calls w/o LHS.
37952 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37953 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37954 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37955 * gimplify.c (asan_poison_variables): Add attribute
37956 use_after_scope_memory to variables that really needs to live
37958 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37959 having the attribute.
37961 2017-01-23 Martin Liska <mliska@suse.cz>
37963 * asan.c (create_asan_shadow_var): New function.
37964 (asan_expand_poison_ifn): Likewise.
37965 * asan.h (asan_expand_poison_ifn): New declaration.
37966 * internal-fn.c (expand_ASAN_POISON): Likewise.
37967 * internal-fn.def (ASAN_POISON): New builtin.
37968 * sanopt.c (pass_sanopt::execute): Expand
37969 asan_expand_poison_ifn.
37970 * tree-inline.c (copy_decl_for_dup_finish): Make function
37972 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37973 * tree-ssa.c (is_asan_mark_p): New function.
37974 (execute_update_addresses_taken): Rewrite local variables
37975 (identified just by use-after-scope as addressable) into SSA.
37977 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37979 * doc/install.texi (Specific): opensource.apple.com uses https
37980 now. Remove trailing slash.
37982 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37984 * README.Portability: Remove note on an Irix compatibility issue.
37986 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37988 * gcov.c (INCLUDE_ALGORITHM): Define.
37989 (INCLUDE_VECTOR): Define.
37990 No longer include <vector> and <algorithm> directly.
37992 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37994 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37996 * doc/invoke.texi (Code Gen Options): Ditto.
37998 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38001 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38003 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38005 rtl-optimization/79125
38006 * cprop.c (local_cprop_pass): Handle cases where we make an
38007 unconditional trap.
38009 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38013 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38014 read from, for big endian.
38016 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38018 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38019 register pauth builtins for LP64 only.
38021 2017-01-20 Marek Polacek <polacek@redhat.com>
38024 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38027 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38029 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38031 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38032 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38033 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38035 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38038 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38039 in big-endian mode when they are not a single duplicated value.
38041 2017-01-20 Richard Biener <rguenther@suse.de>
38043 * BASE-VER: Bump to 7.0.1.
38045 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38047 * omp-low.c (omplow_simd_context): New struct. Use it...
38048 (lower_rec_simd_input_clauses): ...here and...
38049 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38050 references to idx, lane, max_vf, is_simt.
38052 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38054 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38057 2017-01-20 Martin Jambor <mjambor@suse.cz>
38059 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38060 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38062 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38063 (GTFILES): Rename hsa.c to hsa-common.c.
38064 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38065 * hsa-dump.c: Likewise.
38066 * hsa-gen.c: Likewise.
38067 * hsa-regalloc.c: Likewise.
38068 * ipa-hsa.c: Likewise.
38069 * omp-expand.c: Likewise.
38070 * omp-low.c: Likewise.
38071 * toplev.c: Likewise.
38073 2017-01-20 Marek Polacek <polacek@redhat.com>
38076 * doc/invoke.texi: Document -Wduplicated-branches.
38077 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38078 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38079 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38080 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38081 return 0 only when not OEP_LEXICOGRAPHIC.
38082 (fold_build_cleanup_point_expr): Use the expression
38083 location when building CLEANUP_POINT_EXPR.
38084 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38085 * tree.c (add_expr): Handle error_mark_node.
38087 2017-01-20 Martin Liska <mliska@suse.cz>
38090 * tree-profile.c (init_ic_make_global_vars): Do not call
38092 (gimple_init_gcov_profiler): Likewise.
38094 2017-01-20 Martin Liska <mliska@suse.cz>
38097 * cgraph.h (maybe_create_reference): Remove argument and
38099 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38101 * ipa-cp.c (create_specialized_node): Likewise.
38102 * symtab.c (symtab_node::maybe_create_reference): Handle
38103 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38105 2017-01-20 Martin Liska <mliska@suse.cz>
38107 * read-rtl-function.c (function_reader::create_function): Use
38108 build_decl instread of build_decl_stat.
38110 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38112 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38113 * config/i386/avx512dqintrin.h: Ditto.
38114 * config/i386/avx512fintrin.h: Ditto.
38115 * config/i386/i386-builtin-types.def: Add new types.
38116 * config/i386/i386.c: Handle new types.
38117 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38118 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38119 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38120 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38121 (__builtin_ia32_kshiftridi): New.
38122 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38124 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38128 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38129 define to rs6000_init_stack_protect_guard.
38130 (rs6000_init_stack_protect_guard): New function.
38132 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38133 Yunqiang Su <yunqiang.su@imgtec.com>
38135 * config.gcc (supported_defaults): Add madd4.
38136 (with_madd4): Add validation.
38137 (all_defaults): Add madd4.
38138 * config/mips/mips.opt (mmadd4): New option.
38139 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38141 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38143 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38144 (ISA_HAS_FUSED_MADD4): Likewise.
38145 * doc/invoke.texi (-mmadd4): Document the new option.
38146 * doc/install.texi (--with-madd4): Document the new option.
38148 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38150 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38151 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38152 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38153 (aarch64_init_pauth_hint_builtins): New.
38154 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38155 (aarch64_expand_builtin): Expand new builtins.
38157 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38159 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38160 * combine-stack-adj.c (no_unhandled_cfa): Handle
38161 REG_CFA_TOGGLE_RA_MANGLE.
38162 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38163 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38164 info for return address signing.
38165 (aarch64_expand_epilogue): Likewise.
38167 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38169 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38170 * config/aarch64/aarch64-protos.h
38171 (aarch64_return_address_signing_enabled): New declaration.
38172 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38174 (aarch64_expand_prologue): Sign return address before it's pushed onto
38176 (aarch64_expand_epilogue): Authenticate return address fetched from
38178 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38179 (aarch64_attributes): New function attributes for "sign-return-address".
38180 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38181 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38182 ("*do_return"): Generate combined instructions according to key index.
38183 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38184 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38186 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38187 * config/aarch64/aarch64.opt (msign-return-address=): New.
38188 * doc/extend.texi (AArch64 Function Attributes): Documents
38189 "sign-return-address=".
38190 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38192 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38194 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38195 overall option summary.
38197 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38199 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38200 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38201 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38202 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38204 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38206 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38207 -mpower9-minmax by default for -mcpu=power9.
38208 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38209 128-bit floating point.
38211 2017-01-20 Alan Modra <amodra@gmail.com>
38213 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38214 optimizing for size.
38216 2017-01-20 Alan Modra <amodra@gmail.com>
38219 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38220 for strcmp and strncmp from corresponding builtin decl.
38222 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38224 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38225 instead of i386/rtems-64.h.
38226 * config/i386/rtems-64.h: Remove.
38228 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38232 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38234 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38236 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38238 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38239 Change int to HOST_WIDE_INT.
38240 * config/aarch64/aarch64-protos.h
38241 (aarch64_simd_gen_const_vector_dup): Likewise.
38242 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38244 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38246 * langhooks-def.h (lhd_type_for_size): New decl.
38247 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38248 * langhooks.c (lhd_type_for_size): New function, taken from
38251 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38253 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38254 define_bypass for CR latency.
38255 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38256 (power9-alu2): Add define_bypass for CR latency.
38258 (power9-mul): Update insn latency.
38259 (power9-mul-compare): Update insn latency, bypass latency and remove
38262 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38264 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38266 * config/aarch64/aarch64.md
38267 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38268 aarch64_nopcrelative_literal_loads.
38269 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38271 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38273 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38274 TARGET_LOONGSON_3A.
38275 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38277 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38280 * config.gcc (supported_defaults): Add lxc1-sxc1.
38281 (with_lxc1_sxc1): Add validation.
38282 (all_defaults): Add lxc1-sxc1.
38283 * config/mips/mips.opt (mlxc1-sxc1): New option.
38284 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38286 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38287 __mips_no_lxc1_sxc1.
38288 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38289 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38290 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38292 2017-01-19 Richard Biener <rguenther@suse.de>
38294 PR tree-optimization/72488
38295 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38296 sure to restore SSA info.
38297 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38299 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38301 PR rtl-optimization/79121
38302 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38303 of the inner type when shifting an extended value.
38305 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38308 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38309 interposable aliases.
38311 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38314 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38315 Use the evmergelohi instruction.
38316 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38317 (mov_sitf_e500_subreg8_2_be): Likewise.
38318 (mov_sitf_e500_subreg12_2_le): Likewise.
38319 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38320 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38321 (mov_sitf_e500_subreg8_2_le): Likewise.
38322 (mov_sitf_e500_subreg12_2_be): Likewise.
38324 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38326 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38327 attribute from vecsimple to vecperm.
38328 (altivec_vbpermq2): Likewise.
38330 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38333 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38335 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38336 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38337 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38338 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38339 case where N arg is SIZE_MAX.
38340 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38341 (cmpstrsi): Add pattern.
38343 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38345 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38346 __builtin_vec_revb builtins.
38347 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38348 built-in functions to support generation of the ISA 3.0 XXBR<x>
38349 vector byte reverse instructions.
38350 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38351 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38352 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38353 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38354 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38355 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38356 (P9V_BUILTIN_VEC_REVB): Likewise.
38357 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38358 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38359 (p9_xxbrq_v16qi): Likewise.
38360 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38361 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38362 (p9_xxbrh_v8hi): Likewise.
38363 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38364 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38365 vec_revb built-in functions.
38367 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38369 PR rtl-optimization/78952
38370 * config/i386/i386.md (any_extract): New code iterator.
38371 (*insvqi_2): Use any_extract for source operand.
38372 (*insvqi_3): Use any_shiftrt for source operand.
38374 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38376 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38378 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38380 2017-01-18 Matthias Klose <doko@ubuntu.com>
38382 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38384 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38386 * config/rs6000/altivec.h (vec_bperm): Change #define.
38387 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38388 (altivec_vbpermq2): New define_insn.
38389 (altivec_vbpermd): Likewise.
38390 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38391 function interface.
38392 (VBPERMD): Likewise.
38393 (VBPERM): New polymorphic function interface.
38394 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38395 Add entries for P9V_BUILTIN_VEC_VBPERM.
38396 * doc/extend.texi: Add interfaces for vec_bperm.
38398 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38400 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38401 first letter of error messages.
38402 (s390_resolve_overloaded_builtin): Likewise.
38403 * config/s390/s390.c (s390_expand_builtin): Likewise.
38404 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38405 (s390_valid_target_attribute_inner_p): Likewise.
38406 * config/s390/s390.md ("tabort"): Likewise.
38408 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38410 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38411 (ISA_AVOID_DIV_HILO): New macro.
38412 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38413 (ISA_HAS_DDIV): Likewise.
38415 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38417 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38419 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38421 * doc/invoke.texi (fabi-version): Spelling fix.
38423 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38426 * doc/invoke.texi (fabi-version): Mention mangling fix for
38429 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38432 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38434 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38437 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38438 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38440 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38442 (stack_protect_test): Ditto.
38443 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38444 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38446 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38447 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38448 -mstack-protector-guard-offset=.
38449 (RS/6000 and PowerPC Options): Ditto.
38451 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38453 * config/i386/i386.h (MASK_CLASS_P): New define.
38454 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38455 there are no registers from different register sets also when
38456 mask registers are used. Update function comment.
38457 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38458 to (*k/*r) and (*k/*km) alternatives.
38460 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38462 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38463 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38464 (EH_RETURN_HANDLER_RTX): New define.
38465 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38466 Force frame pointer in EH return functions.
38467 (aarch64_expand_epilogue): Add barrier for eh_return.
38468 (aarch64_final_eh_return_addr): Remove.
38469 (aarch64_eh_return_handler_rtx): New function.
38470 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38472 (aarch64_eh_return_handler_rtx): New prototype.
38474 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38476 * config/rs6000/altivec.h (vec_rlmi): New #define.
38477 (vec_vrlnm): Likewise.
38478 (vec_rlnm): Likewise.
38479 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38480 (UNSPEC_VRLNM): Likewise.
38481 (VIlong): New mode iterator.
38482 (altivec_vrl<VI_char>mi): New define_insn.
38483 (altivec_vrl<VI_char>nm): Likewise.
38484 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38486 (VRLDNM): Likewise.
38487 (RLNM): New polymorphic function entry.
38488 (VRLWMI): New monomorphic function entry.
38489 (VRLDMI): Likewise.
38490 (RLMI): New polymorphic function entry.
38491 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38492 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38493 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38496 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38499 * dwarf2out.c (field_byte_offset): Restore the
38500 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38501 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38502 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38505 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38508 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38510 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38513 * dwarf2out.c (add_data_member_location_attribute): For constant
38514 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38515 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38516 DW_AT_byte_size attributes.
38518 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38520 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38521 after forcing to constant memory when the code model is medium.
38523 2017-01-17 Julia Koval <julia.koval@intel.com>
38526 * config/i386/avx512fintrin.h
38527 (_mm512_i32gather_ps): Change __addr type to void const*.
38528 (_mm512_mask_i32gather_ps): Ditto.
38529 (_mm512_i32gather_pd): Ditto.
38530 (_mm512_mask_i32gather_pd): Ditto.
38531 (_mm512_i64gather_ps): Ditto.
38532 (_mm512_mask_i64gather_ps): Ditto.
38533 (_mm512_i64gather_pd): Ditto.
38534 (_mm512_mask_i64gather_pd): Ditto.
38535 (_mm512_i32gather_epi32): Ditto.
38536 (_mm512_mask_i32gather_epi32): Ditto.
38537 (_mm512_i32gather_epi64): Ditto.
38538 (_mm512_mask_i32gather_epi64): Ditto.
38539 (_mm512_i64gather_epi32): Ditto.
38540 (_mm512_mask_i64gather_epi32): Ditto.
38541 (_mm512_i64gather_epi64): Ditto.
38542 (_mm512_mask_i64gather_epi64): Ditto.
38543 (_mm512_i32scatter_ps): Change __addr type to void*.
38544 (_mm512_mask_i32scatter_ps): Ditto.
38545 (_mm512_i32scatter_pd): Ditto.
38546 (_mm512_mask_i32scatter_pd): Ditto.
38547 (_mm512_i64scatter_ps): Ditto.
38548 (_mm512_mask_i64scatter_ps): Ditto.
38549 (_mm512_i64scatter_pd): Ditto.
38550 (_mm512_mask_i64scatter_pd): Ditto.
38551 (_mm512_i32scatter_epi32): Ditto.
38552 (_mm512_mask_i32scatter_epi32): Ditto.
38553 (_mm512_i32scatter_epi64): Ditto.
38554 (_mm512_mask_i32scatter_epi64): Ditto.
38555 (_mm512_i64scatter_epi32): Ditto.
38556 (_mm512_mask_i64scatter_epi32): Ditto.
38557 (_mm512_i64scatter_epi64): Ditto.
38558 (_mm512_mask_i64scatter_epi64): Ditto.
38559 * config/i386/avx512pfintrin.h
38560 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38561 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38562 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38563 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38564 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38565 (_mm512_prefetch_i32scatter_ps): Ditto.
38566 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38567 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38568 (_mm512_prefetch_i64scatter_pd): Ditto.
38569 (_mm512_prefetch_i64scatter_ps): Ditto.
38570 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38571 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38572 * config/i386/avx512vlintrin.h
38573 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38574 (_mm_mmask_i32gather_ps): Ditto.
38575 (_mm256_mmask_i32gather_pd): Ditto.
38576 (_mm_mmask_i32gather_pd): Ditto.
38577 (_mm256_mmask_i64gather_ps): Ditto.
38578 (_mm_mmask_i64gather_ps): Ditto.
38579 (_mm256_mmask_i64gather_pd): Ditto.
38580 (_mm_mmask_i64gather_pd): Ditto.
38581 (_mm256_mmask_i32gather_epi32): Ditto.
38582 (_mm_mmask_i32gather_epi32): Ditto.
38583 (_mm256_mmask_i32gather_epi64): Ditto.
38584 (_mm_mmask_i32gather_epi64): Ditto.
38585 (_mm256_mmask_i64gather_epi32): Ditto.
38586 (_mm_mmask_i64gather_epi32): Ditto.
38587 (_mm256_mmask_i64gather_epi64): Ditto.
38588 (_mm_mmask_i64gather_epi64): Ditto.
38589 (_mm256_i32scatter_ps): Change __addr type to void*.
38590 (_mm256_mask_i32scatter_ps): Ditto.
38591 (_mm_i32scatter_ps): Ditto.
38592 (_mm_mask_i32scatter_ps): Ditto.
38593 (_mm256_i32scatter_pd): Ditto.
38594 (_mm256_mask_i32scatter_pd): Ditto.
38595 (_mm_i32scatter_pd): Ditto.
38596 (_mm_mask_i32scatter_pd): Ditto.
38597 (_mm256_i64scatter_ps): Ditto.
38598 (_mm256_mask_i64scatter_ps): Ditto.
38599 (_mm_i64scatter_ps): Ditto.
38600 (_mm_mask_i64scatter_ps): Ditto.
38601 (_mm256_i64scatter_pd): Ditto.
38602 (_mm256_mask_i64scatter_pd): Ditto.
38603 (_mm_i64scatter_pd): Ditto.
38604 (_mm_mask_i64scatter_pd): Ditto.
38605 (_mm256_i32scatter_epi32): Ditto.
38606 (_mm256_mask_i32scatter_epi32): Ditto.
38607 (_mm_i32scatter_epi32): Ditto.
38608 (_mm_mask_i32scatter_epi32): Ditto.
38609 (_mm256_i32scatter_epi64): Ditto.
38610 (_mm256_mask_i32scatter_epi64): Ditto.
38611 (_mm_i32scatter_epi64): Ditto.
38612 (_mm_mask_i32scatter_epi64): Ditto.
38613 (_mm256_i64scatter_epi32): Ditto.
38614 (_mm256_mask_i64scatter_epi32): Ditto.
38615 (_mm_i64scatter_epi32): Ditto.
38616 (_mm_mask_i64scatter_epi32): Ditto.
38617 (_mm256_i64scatter_epi64): Ditto.
38618 (_mm256_mask_i64scatter_epi64): Ditto.
38619 (_mm_i64scatter_epi64): Ditto.
38620 (_mm_mask_i64scatter_epi64): Ditto.
38621 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38622 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38623 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38624 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38625 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38626 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38627 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38628 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38629 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38630 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38631 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38632 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38633 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38634 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38635 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38636 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38637 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38638 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38639 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38640 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38641 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38642 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38643 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38644 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38645 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38646 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38647 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38648 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38649 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38650 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38651 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38652 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38653 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38654 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38655 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38656 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38657 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38658 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38659 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38660 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38661 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38662 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38663 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38664 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38665 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38666 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38667 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38668 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38669 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38670 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38671 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38672 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38673 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38674 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38675 definitions accordingly.
38677 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38678 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38681 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38684 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38687 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38688 hard regs for inner regno.
38690 2017-01-17 Martin Liska <mliska@suse.cz>
38693 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38694 assumption and add comment.
38696 2017-01-17 Nathan Sidwell <nathan@acm.org>
38698 * ipa-visibility.c (localize_node): New function, broken out of ...
38699 (function_and_variable_visibility): ... here. Call it.
38701 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38703 PR middle-end/77445
38704 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
38705 correctly set frequency of oudgoing edge.
38706 (duplicate_thread_path): Fix profile updating.
38708 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38711 * configure.ac: Add GCC_BASE_VER.
38712 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
38713 version from BASE-VER file.
38714 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
38715 (gcc.o): Depend on $(BASEVER).
38716 * common.opt (dumpfullversion): New option.
38717 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38718 * doc/invoke.texi: Document -dumpfullversion.
38719 * doc/install.texi: Document --with-gcc-major-version-only.
38720 * configure: Regenerated.
38722 2017-01-17 Richard Biener <rguenther@suse.de>
38724 PR tree-optimization/71433
38725 * tree-vrp.c (register_new_assert_for): Merge same asserts
38726 on all incoming edges.
38727 (process_assert_insertions_for): Handle insertions at the
38730 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38732 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38733 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38735 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38738 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38741 2017-01-17 Alan Modra <amodra@gmail.com>
38744 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38745 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38746 symbolic stack limit when pic.
38748 2017-01-16 Martin Sebor <msebor@redhat.com>
38750 PR tree-optimization/78608
38751 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38753 2017-01-16 Jeff Law <law@redhat.com>
38756 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38757 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38758 for several include directories that may be relative to sysroot.
38759 * config/i386/x-mingw32 (gplus_includedir): Define.
38760 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38761 (native_system_includedir): Likewise.
38762 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38763 override if TARGET_SYSTEM_ROOT is defined.
38764 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38766 PR tree-optimization/79090
38767 PR tree-optimization/33562
38768 PR tree-optimization/61912
38769 PR tree-optimization/77485
38770 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38771 and computed trims into the dump file.
38773 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38775 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38777 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38780 * gimplify.c (gimplify_init_constructor): If want_value and
38781 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38785 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38786 sequence. Formatting fixes.
38787 (doloop_optimize): Formatting fixes.
38790 * gcc.c (debug_level_greater_than_spec_func): New function.
38791 (static_spec_functions): Add debug-level-gt spec function.
38792 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38794 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38795 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38796 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38797 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38798 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38799 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38801 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38803 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38804 QImode fixups to general and mask registers only.
38806 2017-01-16 Carl Love <cel@us.ibm.com>
38808 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38809 for built-in functions
38810 vector signed char vec_nabs (vector signed char)
38811 vector signed short vec_nabs (vector signed short)
38812 vector signed int vec_nabs (vector signed int)
38813 vector signed long long vec_nabs (vector signed long long)
38814 vector float vec_nabs (vector float)
38815 vector double vec_nabs (vector double)
38816 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38818 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38819 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38820 * doc/extend.texi: Update the documentation file for the new built-in
38823 2017-01-16 Martin Sebor <msebor@redhat.com>
38825 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38828 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38830 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38831 UNSPEC_VSX__XXSPLTD to require special splat handling.
38833 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38836 * system.h: Poison strndup.
38838 2017-01-16 Alan Modra <amodra@gmail.com>
38841 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38844 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38846 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38848 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38850 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38851 call recog here. Assert that INSN_CODE (insn) is non-negative.
38853 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38856 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38858 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38859 in the currently scheduled RTL fragment.
38861 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38863 PR rtl-optimization/78751
38864 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38867 2017-01-14 Jeff Law <law@redhat.com>
38869 PR tree-optimization/79090
38870 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38871 variable length stores.
38872 (compute_trims): Delete dead assignment to *trim_tail.
38873 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38876 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38878 PR rtl-optimization/78626
38879 PR rtl-optimization/78727
38880 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38881 of a block, and split such blocks after everything else is finished.
38883 2017-01-14 Alan Modra <amodra@gmail.com>
38886 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38887 target legitimate_combined_insn.
38888 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38889 (rs6000_legitimate_combined_insn): New function.
38890 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38892 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38893 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38894 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38896 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38898 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38900 2017-01-13 Jeff Law <law@redhat.com>
38902 PR tree-optimization/33562
38903 PR tree-optimization/61912
38904 PR tree-optimization/77485
38905 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38907 (delete_dead_assignment): Likewise.
38908 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38909 statement to delete_dead_call and delete_dead_assignment.
38911 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38914 * substring-locations.c (format_warning_va): Strengthen case 1 so
38915 that both endpoints of the substring must be within the format
38916 range for just the substring to be printed.
38918 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38920 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38921 * config/i386/i386.c (ix86_target_string): Add missing options
38922 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38923 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38924 flags_other and ix86_target_other to flags2_other. Display unknown
38926 (ix86_valid_target_attribute_inner_p): Add missing options and
38927 reorder options by implied ISAs, as in ix86_target_string.
38929 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38931 * hash-table.h (hash_table::too_empty_p): New function.
38932 (hash_table::expand): Use it.
38933 (hash_table::traverse): Likewise.
38934 (hash_table::empty_slot): Use sizeof (value_type) instead of
38935 sizeof (PTR) to convert bytes to elements. Shrink the table
38936 if the current size is excessive for the current number of
38939 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38941 * ira-costs.c (record_reg_classes): Break from the inner loop
38942 early once alt_fail is known to be true. Update outer loop
38943 handling accordingly.
38945 2017-01-13 Jeff Law <law@redhat.com>
38947 * tree-ssa-dse.c (decrement_count): New function.
38948 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38949 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38950 when we know the partially dead statement is a mem* function.
38952 PR tree-optimization/61912
38953 PR tree-optimization/77485
38954 * tree-ssa-dse.c: Include expr.h.
38955 (maybe_trim_constructor_store): New function.
38956 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38958 PR tree-optimization/33562
38959 PR tree-optimization/61912
38960 PR tree-optimization/77485
38961 * doc/invoke.texi: Document new dse-max-object-size param.
38962 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38963 * tree-ssa-dse.c: Include params.h.
38964 (dse_store_status): New enum.
38965 (initialize_ao_ref_for_dse): New, partially extracted from
38967 (valid_ao_ref_for_dse, normalize_ref): New.
38968 (setup_live_bytes_from_ref, compute_trims): Likewise.
38969 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38970 (maybe_trim_partially_dead_store): Likewise.
38971 (maybe_trim_complex_store): Likewise.
38972 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38973 Track what bytes live from the original store. Return tri-state
38974 for dead, partially dead or live.
38975 (dse_dom_walker): Add constructor, destructor and new private members.
38976 (delete_dead_call, delete_dead_assignment): New extracted from
38978 (dse_optimize_stmt): Make a member of dse_dom_walker.
38979 Use initialize_ao_ref_for_dse.
38981 PR tree-optimization/33562
38982 PR tree-optimization/61912
38983 PR tree-optimization/77485
38984 * sbitmap.h (bitmap_count_bits): Prototype.
38985 (bitmap_clear_range, bitmap_set_range): Likewise.
38986 * sbitmap.c (bitmap_clear_range): New function.
38987 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38989 2017-01-13 Martin Liska <mliska@suse.cz>
38992 * function.c (set_cfun): Add new argument force.
38993 * function.h (set_cfun): Likewise.
38994 * ipa-inline-transform.c (inline_call): Use the function when
38995 strict alising from is dropped for function we inline to.
38997 2017-01-13 Richard Biener <rguenther@suse.de>
38999 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39000 for dumping GIMPLE INTEGER_CSTs.
39002 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39004 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39005 to 201112L since C++17.
39007 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39010 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39011 if -fsanitize=kernel-address is present.
39013 2017-01-13 Richard Biener <rguenther@suse.de>
39015 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39016 as _Literal ( type ) number in case usual suffixes do not
39017 preserve all information.
39019 2017-01-13 Richard Biener <rguenther@suse.de>
39021 PR tree-optimization/77283
39022 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39023 and ssa-iterators.h.
39024 (is_feasible_trace): Implement a cost model based on joiner
39027 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39030 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39031 char or short to __float128/_Float128 directly.
39033 2017-01-12 Martin Sebor <msebor@redhat.com>
39035 to -Wformat-overflow.
39036 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39037 (min_bytes_remaining): Same.
39038 (get_string_length): Same.
39039 (format_string): Same.
39040 (format_directive): Same.
39042 (pass_sprintf_length::handle_gimple_call): Same.
39044 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39046 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39047 info.nowrite calls with no lhs that can't throw. Return bool
39048 whether gsi_remove has been called or not.
39049 (pass_sprintf_length::handle_gimple_call): Return bool whether
39050 try_substitute_return_value called gsi_remove. Formatting fix.
39051 (pass_sprintf_length::execute): Don't use gsi_remove if
39052 handle_gimple_call returned true.
39055 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39056 be removed due to side-effects, don't remove following barrier nor
39057 turn the successor edge into fallthru edge.
39059 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39062 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39063 element-reversing loads and stores as not swappable.
39065 2017-01-12 Nathan Sidwell <nathan@acm.org>
39066 Nicolai Stange <nicstange@gmail.com>
39068 * combine.c (try_combine): Don't ignore result of overlap checking
39069 loop. Combine overlap & asm check into single loop.
39071 2017-01-12 Richard Biener <rguenther@suse.de>
39073 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39074 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39076 2017-01-12 Richard Biener <rguenther@suse.de>
39078 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39079 and TS_TARGET_OPTION directly derive from TS_BASE.
39080 * tree-core.h (tree_optimization_option): Derive from tree_base.
39081 (tree_target_option): Likewise.
39083 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39085 * config/i386/i386.c (memory_address_length): Increase len
39086 only when rip_relative_addr_p returns false.
39088 2017-01-11 Julia Koval <julia.koval@intel.com>
39090 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39091 (OPTION_MASK_ISA_SGX_SET): New.
39092 (ix86_handle_option): Handle OPT_msgx.
39093 * config.gcc: Added sgxintrin.h.
39094 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39095 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39096 * config/i386/i386.c (ix86_target_string): Add -msgx.
39098 (ix86_option_override_internal): Handle new options.
39099 (ix86_valid_target_attribute_inner_p): Add sgx.
39100 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39101 * config/i386/i386.opt: Add msgx.
39102 * config/i386/sgxintrin.h: New file.
39103 * config/i386/x86intrin.h: Add sgxintrin.h.
39105 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39108 * fold-const.c (maybe_nonzero_address): Return 1 for function
39110 (tree_single_nonzero_warnv_p): Don't handle function local objects
39114 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39117 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39120 * opts.c: Include "spellcheck.h"
39121 (struct string_fragment): New struct.
39122 (struct edit_distance_traits<const string_fragment &>): New
39124 (get_closest_sanitizer_option): New function.
39125 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39127 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39129 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39131 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39132 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39133 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39134 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39135 for initial die_offset if dwarf_split_debug_info.
39136 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39137 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39139 (output_skeleton_debug_sections): Formatting fix. Use
39140 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39141 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39143 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39145 * config/arm/cortex-a53.md: Add bypasses for
39146 cortex_a53_r2f_cvt.
39147 (cortex_a53_r2f): Only use for transfers.
39148 (cortex_a53_f2r): Likewise.
39149 (cortex_a53_r2f_cvt): Add reservation for conversions.
39150 (cortex_a53_f2r_cvt): Likewise.
39152 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39154 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39155 to all inlined functions, change static to extern.
39157 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39160 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39162 (arm_assemble_integer): Likewise.
39164 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39166 * config.gcc: Use new awk script to check CPU, FPU and architecture
39167 parameters for --with-... options.
39168 * config/arm/parsecpu.awk: New file
39169 * config/arm/arm-cpus.in: New file.
39170 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39172 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39174 * config/arm/t-arm: Update dependency rules.
39175 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39176 of processing .def files.
39177 * config/arm/genopt.sh: Deleted.
39178 * config/arm/gentune.sh: Deleted.
39179 * config/arm/arm-cores.def: Deleted.
39180 * config/arm/arm-arches.def: Deleted.
39181 * config/arm/arm-fpus.def: Deleted.
39182 * config/arm/arm-tune.md: Regenerated.
39183 * config/arm/arm-tables.opt: Regenerated.
39184 * config/arm/arm-cpu.h: New generated file.
39185 * config/arm/arm-cpu-data.h: New generated file.
39186 * config/arm/arm-cpu-cdata.h: New generated file.
39188 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39191 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39193 (input_varpool_node): Unpack dynamically_initialized bit.
39195 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39197 PR rtl-optimization/79032
39198 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39199 the alignment of the adjusted memory reference against that of MODE,
39200 instead of the alignment of the original memory reference.
39202 2017-01-11 Martin Jambor <mjambor@suse.cz>
39204 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39206 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39207 decorated functions.
39209 2017-01-11 Richard Biener <rguenther@suse.de>
39211 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39212 set range/nonnull info for PHI results. Do not set it on
39213 stmts marked for removal.
39215 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39217 * expr.c (store_field): In the bitfield case, fetch the return value
39218 from the registers before applying a single big-endian adjustment.
39219 Always do a final load for a BLKmode value not larger than a word.
39221 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39224 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39225 that we correctly handle column numbers greater than
39226 LINE_MAP_MAX_COLUMN_NUMBER.
39228 2017-01-10 Martin Sebor <msebor@redhat.com>
39230 PR middle-end/78245
39231 * gimple-ssa-sprintf.c (get_destination_size): Call
39232 {init,fini}object_sizes.
39233 * tree-object-size.c (addr_object_size): Adjust.
39234 (pass_through_call): Adjust.
39235 (pass_object_sizes::execute): Adjust.
39236 * tree-object-size.h (fini_object_sizes): Declare.
39238 2017-01-10 Martin Sebor <msebor@redhat.com>
39240 PR tree-optimization/78775
39241 * builtins.c (get_size_range): Move...
39242 * calls.c: ...to here.
39243 (alloc_max_size): Accept zero argument.
39244 (operand_signed_p): Remove.
39245 (maybe_warn_alloc_args_overflow): Call get_size_range.
39246 * calls.h (get_size_range): Declare.
39248 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39250 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39251 from TI's devices.csv file as of September 2016.
39252 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39254 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39256 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39257 * doc/invoke.texi: Likewise.
39258 * doc/md.texi: Likewise.
39259 * doc/objc.texi: Likewise.
39261 2017-01-10 Joshua Conner <joshconner@google.com>
39263 * config/arm/fuchsia-elf.h: New file.
39264 * config/fuchsia.h: New file.
39265 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39266 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39268 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39270 2016-01-10 Richard Biener <rguenther@suse.de>
39272 PR tree-optimization/79034
39273 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39274 Propagate out degenerate PHIs in the joiner.
39276 2017-01-10 Martin Liska <mliska@suse.cz>
39278 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39279 (sort_congruence_classes_by_decl_uid): Likewise.
39280 (sort_congruence_class_groups_by_decl_uid): Likewise.
39281 (sem_item_optimizer::merge_classes): Sort class, groups in these
39282 classes and members in the groups by DECL_UID of declarations.
39283 This would make merge operations stable.
39285 2017-01-10 Martin Liska <mliska@suse.cz>
39287 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39288 usage of m_classes_vec.
39289 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39290 (sem_item_optimizer::get_group_by_hash): Likewise.
39291 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39292 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39293 (sem_item_optimizer::verify_classes): Likewise.
39294 (sem_item_optimizer::process_cong_reduction): Likewise.
39295 (sem_item_optimizer::dump_cong_classes): Likewise.
39296 (sem_item_optimizer::merge_classes): Likewise.
39297 * ipa-icf.h (congruence_class_hash): Rename from
39298 congruence_class_group_hash. Remove declaration of m_classes_vec.
39300 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39302 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39303 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39304 * config.gcc: Add avx512vpopcntdqintrin.h.
39305 * config/i386/avx512vpopcntdqintrin.h: New.
39306 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39307 * config/i386/i386-builtin-types.def: Add new types.
39308 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39309 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39310 __builtin_ia32_vpopcountq_v8di_mask): New.
39311 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39312 __AVX512VPOPCNTDQ__.
39313 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39314 (PTA_AVX512VPOPCNTDQ): Define.
39315 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39316 TARGET_AVX512VPOPCNTDQ_P): Define.
39317 * config/i386/i386.opt: Add mavx512vpopcntdq.
39318 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39319 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39321 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39323 PR middle-end/77484
39324 * predict.def (PRED_CALL): Set to 67.
39326 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39328 * expr.c (store_field): In the bitfield case, if the value comes from
39329 a function call and is of an aggregate type returned in registers, do
39330 not modify the field mode; extract the value in all cases if the mode
39331 is BLKmode and the size is not larger than a word.
39333 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39336 * config/i386/cpuid.h: Fix undefined behavior.
39338 2017-01-04 Jeff Law <law@redhat.com>
39340 PR tree-optimization/79007
39341 PR tree-optimization/67955
39342 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39343 conservative for pt.null when flag_non_call_exceptions is on.
39345 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39347 PR translation/79019
39348 PR translation/79020
39349 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39350 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39351 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39353 * config/avr/avr.opt (maccumulate-args): Likewise.
39354 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39355 * common.opt (freport-bug): Likewise.
39356 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39357 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39358 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39359 translatable string.
39360 * config/i386/i386.c (function_value_32): Likewise.
39361 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39362 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39364 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39365 * common/config/msp430/msp430-common.c (msp430_handle_option):
39367 * symtab.c (symtab_node::verify_base): Likewise.
39368 * opts.c (set_debug_level): Likewise.
39369 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39370 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39371 missing whitespace to translatable strings.
39372 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39373 * config/sh/superh.h: Likewise.
39374 * config/i386/xopintrin.h: Likewise.
39375 * config/i386/znver1.md: Likewise.
39376 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39377 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39378 * double-int.h (struct double_int): Likewise.
39379 * double-int.c (div_and_round_double): Likewise.
39380 * wide-int.cc: Likewise.
39381 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39382 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39383 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39384 (crossjumps_occurred): ... this.
39385 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39388 PR tree-optimization/78899
39389 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39390 returning bool return struct loop *, NULL for failure and the new
39392 (versionable_outer_loop_p): Don't version outer loop if it has
39393 dont_vectorized bit set.
39394 (tree_if_conversion): When versioning outer loop, ensure
39395 tree_if_conversion is performed also on the inner loop of the
39396 non-vectorizable outer loop copy.
39397 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39398 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39399 prevent vectorization of it.
39400 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39401 the outer loop vectorization of the non-scalar version is attempted
39402 before vectorization of the inner loop in scalar version. If
39403 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39404 vectorization of its inner loop.
39405 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39406 has 2 inner loops, rename also on edges from bb whose single pred
39407 is outer_loop->header. Fix typo in function comment.
39409 2017-01-09 Martin Sebor <msebor@redhat.com>
39412 * asan.c (asan_emit_stack_protection): Increase local buffer size
39413 to avoid snprintf truncation warning.
39415 2017-01-09 Andrew Pinski <apinski@cavium.com>
39417 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39418 to reference thunderx2t99 for the tuning structure
39419 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39421 (thunderx2t99_extra_costs): This.
39422 * config/aarch64/aarch64-tune.md: Regenerate.
39423 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39424 (vulcan_addrcost_table): This.
39425 (vulcan_regmove_cost): Rename to ...
39426 (thunderx2t99_regmove_cost): This.
39427 (vulcan_vector_cost): Rename to ...
39428 (thunderx2t99_vector_cost): this.
39429 (vulcan_branch_cost): Rename to ...
39430 (thunderx2t99_branch_cost): This.
39431 (vulcan_tunings): Rename to ...
39432 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39433 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39435 2017-01-09 Martin Jambor <mjambor@suse.cz>
39439 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39440 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39441 (propagate_vr_accross_jump_function): Use the above function for all
39442 value range computations for pass-through jump functions and type
39443 converasion from explicit value range values.
39444 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39445 parameters from TYPE_ARG_TYPES.
39446 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39447 (ipa_write_node_info): Stream type of the actual argument.
39448 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39450 2017-01-09 Martin Liska <mliska@suse.cz>
39453 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39454 (lookup_compiler): Do not show error message with have_E.
39456 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39458 PR tree-optimization/78938
39459 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39460 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39461 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39462 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39465 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39467 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39470 2017-01-09 Richard Biener <rguenther@suse.de>
39472 PR tree-optimization/78997
39473 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39474 name condition properly.
39476 2017-01-09 Richard Biener <rguenther@suse.de>
39479 * dwarf2out.c (is_cxx): New overload with context.
39480 (is_naming_typedef_decl): Use it.
39482 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39484 * invoke.texi (Option Summary): Correct spacing in option lists
39485 and add line breaks to fix over-long lines.
39487 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39489 PR middle-end/17660
39491 * extend.texi (Common Variable Attributes): Add xref to GCC
39492 Internals manual to explain mode attribute keywords.
39494 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39497 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39498 and Preprocessor Options.
39499 (Options for Linking): Document -pthread here....
39500 (RS/6000 and PowerPC Options): ...not here.
39501 (Solaris 2 Options): ...or here.
39502 * doc/cppopts.texi: Document -pthread.
39504 2017-01-08 Martin Sebor <msebor@redhat.com>
39506 PR middle-end/77708
39507 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39508 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39509 New member functions.
39510 (format_directive): Used them.
39512 (pass_sprintf_length::handle_gimple_call): Same.
39513 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39514 to avoid truncation for any argument.
39515 (extract_affine_mul): Same.
39516 * tree.c (get_file_function_name): Same.
39518 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39520 PR middle-end/77484
39521 * predict.def (PRED_INDIR_CALL): Set to 86.
39523 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39525 PR preprocessor/54124
39526 * doc/cppopts.texi: Reformat -d subtable to list the full name
39527 of the options. Add cross-reference to the docs for the general
39528 compiler -d options.
39529 * doc/invoke.texi (Developer Options): Add cross-reference to the
39530 preprocessor-specific -d option documentation.
39532 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39534 PR preprocessor/13498
39535 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39536 redudant material, and reflect new command-line options.
39537 (System Headers): Likewise.
39539 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39541 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39542 -isystem, and -idirafter. Copy-edit.
39543 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39544 default for -ftrack-macro-expansion. Delete obsolete and
39545 badly-formatted implementation details about -fdebug-cpp output.
39546 * doc/cppwarnopts.texi: Copy-edit.
39548 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39551 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39552 that the transition from a max line width >= 1<<10 to narrower
39553 lines works correctly.
39555 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39557 * doc/options.texi (PerFunction): New.
39558 * opt-functions.awk (switch_flags): Map both Optimization and
39559 PerFunction to CL_OPTIMIZATION.
39560 * opth-gen.awk: Test for PerFunction flag along with
39562 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39563 it only when the latter is present. Skip those that don't in
39564 the hash function generator.
39565 * common.opt (fvar-tracking): Mark as PerFunction instead of
39567 (fvar-tracking-assignments): Likewise.
39568 (fvar-tracking-assignments-toggle): Likewise.
39569 (fvar-tracking-uninit): Likewise.
39571 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39573 PR translation/79018
39574 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39577 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39580 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39581 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39582 TARGET_LONG_JUMP_TABLE_OFFSETS.
39583 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39584 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39585 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39586 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39587 * config/m68k/m68k.md (tablejump expander): Likewise.
39588 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39589 TARGET_LONG_JUMP_TABLE_OFFSETS.
39590 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39591 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39593 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39594 David Holsgrove <david.holsgrove@xilinx.com>
39596 * common/config/microblaze/microblaze-common.c
39597 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39598 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39600 * config/microblaze/microblaze.c (microblaze_must_save_register)
39601 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39603 (microblaze_eh_return): New function.
39604 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39605 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39606 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39607 * config/microblaze/microblaze.md (eh_return): New pattern.
39609 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39611 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39612 GCC_DIAGNOSTIC_STRINGIFY): Define.
39614 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39616 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39618 * config/arm/arm.md (<mcrr>): New.
39620 * config/arm/arm.c (arm_arch5te): New.
39621 (arm_option_override): Set arm_arch5te.
39622 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39624 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39625 (arm_mcrr_qualifiers): ... this. New.
39626 (MRRC_QUALIFIERS): Define to...
39627 (arm_mrrc_qualifiers): ... this. New.
39628 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39630 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39631 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39632 (MRRCI, mrrc, MRRC): New.
39633 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39634 VUNSPEC_MRRC2): New.
39636 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39638 * config/arm/arm.md (<mcr>): New.
39640 * config/arm/arm.c (arm_coproc_builtin_available): Add
39641 support for mcr, mrc, mcr2 and mrc2.
39642 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39643 (arm_mcr_qualifiers): ... this. New.
39644 (MRC_QUALIFIERS): Define to ...
39645 (arm_mrc_qualifiers): ... this. New.
39646 (MCR_QUALIFIERS): Define to ...
39647 (arm_mcr_qualifiers): ... this. New.
39648 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39650 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39651 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39652 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39653 VUNSPEC_MRC2): New.
39655 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39657 * config/arm/arm.md (*ldc): New.
39661 * config/arm/arm.c (arm_coproc_builtin_available): Add
39662 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39663 (arm_coproc_ldc_stc_legitimate_address): New.
39664 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39665 'qualifier_const_pointer'.
39666 (LDC_QUALIFIERS): Define to...
39667 (arm_ldc_qualifiers): ... this. New.
39668 (STC_QUALIFIERS): Define to...
39669 (arm_stc_qualifiers): ... this. New.
39670 * config/arm/arm-protos.h
39671 (arm_coproc_ldc_stc_legitimate_address): New.
39672 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39673 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39674 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39675 stc2, stcl, stc2l): New.
39676 * config/arm/constraints.md (Uz): New.
39677 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39678 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39679 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39680 VUNSPEC_STC2L): New.
39682 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39684 * config/arm/arm.md (<cdp>): New.
39685 * config/arm/arm.c (neon_const_bounds): Rename this ...
39686 (arm_const_bounds): ... this.
39687 (arm_coproc_builtin_available): New.
39688 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39689 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39690 (CDP_QUALIFIERS): Define to...
39691 (arm_cdp_qualifiers): ... this. New.
39693 (arm_expand_builtin_args): Add case for 6 arguments.
39694 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39695 (arm_const_bounds): ... this.
39696 (arm_coproc_builtin_available): New.
39697 * config/arm/arm_acle.h (__arm_cdp): New.
39699 * config/arm/arm_acle_builtins.def (cdp): New.
39701 * config/arm/iterators.md (CDPI,CDP,cdp): New.
39702 * config/arm/neon.md: Rename all 'neon_const_bounds' to
39703 'arm_const_bounds'.
39704 * config/arm/types.md (coproc): New.
39705 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
39706 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
39707 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
39708 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
39710 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39712 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
39713 (UBINOP_QUALIFIERS): New.
39715 (acle_builtin_data): New. Change comment.
39716 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
39717 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39718 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39719 arm_acle_builtins.def.
39720 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39721 (arm_init_acle_builtins): New.
39722 (CRC32_BUILTIN): Remove.
39723 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39724 crc32cb, crc32ch and crc32cw.
39725 (arm_init_crc32_builtins): Remove.
39726 (arm_init_builtins): Use arm_init_acle_builtins rather
39727 than arm_init_crc32_builtins.
39728 (arm_expand_acle_builtin): New.
39729 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39730 * config/arm/arm_acle_builtins.def: New.
39732 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39734 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39735 (arm_builtin_datum): ... this.
39736 (arm_init_neon_builtin): Rename to ...
39737 (arm_init_builtin): ... this. Add a new parameters PREFIX
39738 and USE_SIG_IN_NAME.
39739 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39740 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39741 'arm_builtin_datum'.
39742 (arm_init_vfp_builtins): Likewise.
39743 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39744 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39745 (arm_expand_neon_args): Rename to ...
39746 (arm_expand_builtin_args): ... this. Rename builtin_arg
39747 enum values and differentiate between ARG_BUILTIN_MEMORY
39748 and ARG_BUILTIN_NEON_MEMORY.
39749 (arm_expand_neon_builtin_1): Rename to ...
39750 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39751 values, arm_expand_builtin_args and add bool parameter NEON.
39752 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39753 (arm_expand_vfp_builtin): Likewise.
39754 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39756 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39758 PR middle-end/77484
39759 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39760 * predict.c (tree_estimate_probability_bb): Reverse direction of
39761 polymorphic call predictor.
39763 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39765 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39766 (determine_pass_name_match): ...this new function and...
39767 (should_skip_pass_p): ...this new function.
39769 2017-01-06 Nathan Sidwell <nathan@acm.org>
39771 * ipa-visibility.c (function_and_variable_visibility): Reformat
39772 comments and long lines. Remove extrneous if.
39773 * symtab.c (symtab_node::make_decl_local): Fix code format.
39774 (symtab_node::set_section_for_node): Fix comment typo.
39776 2017-01-06 Martin Liska <mliska@suse.cz>
39779 * lra-constraints.c: Rename invariant to lra_invariant.
39780 * predict.c (set_even_probabilities): Initialize e to NULL.
39782 2017-01-05 Martin Sebor <msebor@redhat.com>
39784 PR tree-optimization/78910
39785 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39786 (format_integer): Correct off-by-one error in the handling
39787 of precision with negative numbers in signed conversions..
39789 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39791 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39793 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39795 PR tree-optimization/71016
39796 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39797 factor_out_conditional_conversion. Formatting fix.
39798 (factor_out_conditional_conversion): Add cond_stmt argument.
39799 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39800 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39803 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39805 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39806 read-rtl-function.o, and selftest-rtl.o.
39807 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39808 (selftest::aarch64_test_loading_full_dump): New function.
39809 (selftest::aarch64_run_selftests): New function.
39810 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39811 selftest::aarch64_run_selftests.
39812 * config/i386/i386.c
39813 (selftest::ix86_test_loading_dump_fragment_1): New function.
39814 (selftest::ix86_test_loading_call_insn): New function.
39815 (selftest::ix86_test_loading_full_dump): New function.
39816 (selftest::ix86_test_loading_unspec): New function.
39817 (selftest::ix86_run_selftests): Call the new functions.
39818 * emit-rtl.c (maybe_set_max_label_num): New function.
39819 * emit-rtl.h (maybe_set_max_label_num): New decl.
39820 * function.c (instantiate_decls): Guard call to
39821 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39822 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39824 * gensupport.c (gen_reader::gen_reader): Pass "false"
39825 for new "compact" param of rtx_reader.
39826 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39827 rather than an empty string for NULL strings.
39828 * read-md.c: Potentially include config.h rather than bconfig.h.
39829 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39830 (have_error): New global, copied from errors.c.
39831 (md_reader::read_name): Rename to...
39832 (md_reader::read_name_1): ...this, adding "out_loc" param,
39833 and converting "missing name or number" to returning false, rather
39835 (md_reader::read_name): Reimplement in terms of read_name_1.
39836 (md_reader::read_name_or_nil): New function.
39837 (md_reader::read_string): Handle "(nil)" by returning NULL.
39838 (md_reader::md_reader): Add new param "compact".
39839 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39840 (md_reader::read_file): New method.
39841 * read-md.h (md_reader::md_reader): Add new param "compact".
39842 (md_reader::read_file): New method.
39843 (md_reader::is_compact): New accessor.
39844 (md_reader::read_name): Convert return type from void to file_location.
39845 (md_reader::read_name_or_nil): New decl.
39846 (md_reader::read_name_1): New decl.
39847 (md_reader::m_compact): New field.
39848 (noop_reader::noop_reader): Pass "false" for new "compact" param
39850 (rtx_reader::rtx_reader): Add new "compact" param.
39851 (rtx_reader::read_rtx_operand): Make virtual and convert return
39852 type from void to rtx.
39853 (rtx_reader::read_until): New decl.
39854 (rtx_reader::handle_any_trailing_information): New virtual function.
39855 (rtx_reader::postprocess): New virtual function.
39856 (rtx_reader::finalize_string): New virtual function.
39857 (rtx_reader::m_in_call_function_usage): New field.
39858 (rtx_reader::m_reuse_rtx_by_id): New field.
39859 * read-rtl-function.c: New file.
39860 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39861 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39862 (selftest::verify_three_block_rtl_cfg): New decl.
39863 * read-rtl-function.h: New file.
39864 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39865 For host, include function.h, memmodel.h, and emit-rtl.h.
39866 (one_time_initialization): New function.
39867 (struct compact_insn_name): New struct.
39868 (compact_insn_names): New array.
39869 (find_code): Handle insn codes in compact dumps.
39870 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39871 (bind_subst_iter_and_attr): Likewise.
39872 (add_condition_to_string): Likewise.
39873 (add_condition_to_rtx): Likewise.
39874 (apply_attribute_uses): Likewise.
39875 (add_current_iterators): Likewise.
39876 (apply_iterators): Likewise.
39877 (initialize_iterators): Guard usage of apply_subst_iterator with
39878 #ifdef GENERATOR_FILE.
39879 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39880 (md_reader::read_mapping): Likewise.
39881 (add_define_attr_for_define_subst): Likewise.
39882 (add_define_subst_attr): Likewise.
39883 (read_subst_mapping): Likewise.
39884 (check_code_iterator): Likewise.
39885 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39887 (one_time_initialization): New function.
39888 (rtx_reader::read_until): New method.
39889 (read_flags): New function.
39890 (parse_reg_note_name): New function.
39891 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39892 Handle reuse_rtx ids.
39893 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39894 Add parsing support for RTL dumps, mirroring the special-cases in
39895 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39896 values, and calling handle_any_trailing_information.
39897 (rtx_reader::read_rtx_operand): Convert return type from void
39898 to rtx, returning return_rtx. Handle case 'e'. Call
39899 finalize_string on XSTR and XTMPL fields.
39900 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39901 "(nil)" values were omitted. Call the postprocess vfunc on the
39903 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39904 class ctor. Initialize m_in_call_function_usage. Call
39905 one_time_initialization.
39906 * rtl-tests.c (selftest::test_uncond_jump): Call
39907 set_new_first_and_last_insn.
39908 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39909 * selftest-rtl.c: New file.
39910 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39911 (selftest::get_insn_by_uid): New decl.
39912 * selftest-run-tests.c (selftest::run_tests): Call
39913 read_rtl_function_c_tests.
39914 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39915 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39918 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39920 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39921 operands in a special way. Assert that pos+len <= mode precision.
39923 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39925 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39926 3 argument Alias with unlimited for the negative form.
39927 (fno-vect-cost-model): Removed.
39929 2017-01-05 Martin Liska <mliska@suse.cz>
39931 * hsa-gen.c (gen_hsa_divmod): New function.
39932 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39934 2017-01-05 Martin Liska <mliska@suse.cz>
39937 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39940 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39942 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39943 small constant length operands.
39945 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39947 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39948 between loop iterations.
39950 2017-01-05 Martin Liska <mliska@suse.cz>
39953 * gimplify.c (gimplify_decl_expr): Compare to
39954 asan_poisoned_variables instread of checking flags.
39955 (gimplify_target_expr): Likewise.
39956 (gimplify_expr): Likewise.
39957 (gimplify_function_tree): Conditionally initialize
39958 asan_poisoned_variables.
39960 2017-01-04 Jeff Law <law@redhat.com>
39962 PR tree-optimizatin/78812
39963 * rtl.h (contains_mem_rtx_p): Prototype.
39964 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39965 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39966 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39967 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39968 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39970 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39972 * input.c (assert_char_at_range): Default-initialize actual_range.
39974 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39976 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39977 to match the caller.
39979 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39981 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39982 insns after final jump in test to emit dummy move.
39984 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39986 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39987 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39989 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39991 * multiple_target.c (create_dispatcher_calls): Init e_next.
39992 * tree-ssa-loop-split.c (split_loop): Init border.
39993 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39996 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40001 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40002 (altivec_register_operand): Do not return true if the operand
40003 contains a SUBREG mixing SImode and SFmode.
40004 (vsx_register_operand): Likewise.
40005 (vsx_reg_sfsubreg_ok): New predicate.
40006 (vfloat_operand): Do not return true if the operand contains a
40007 SUBREG mixing SImode and SFmode.
40008 (vint_operand): Likewise.
40009 (vlogical_operand): Likewise.
40010 (gpc_reg_operand): Likewise.
40011 (int_reg_operand): Likewise.
40012 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40013 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40014 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40016 (rs6000_emit_move_si_sf_subreg): New helper function.
40017 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40018 fixup SUBREGs involving SImode and SFmode.
40019 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40020 numbers for the new peephole2 optimization.
40021 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40022 the GLIBC math library that do AND/IOR/XOR operations on single
40023 precision floating point.
40024 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40025 target macros to say whether we need to avoid SUBREGs mixing
40027 (TARGET_ALLOW_SF_SUBREG): Likewise.
40028 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40029 (UNSPEC_SI_FROM_SF): Likewise.
40030 (iorxor): Change spacing.
40031 (and_ior_xor): New iterator for AND, IOR, and XOR.
40032 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40033 (movdi_from_sf_zero_ext): Likewise.
40034 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40035 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40036 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40037 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40038 (fms<mode>4): Likewise.
40039 (fnma<mode>4): Likewise.
40040 (fnms<mode>4): Likewise.
40041 (nfma<mode>4): Likewise.
40042 (nfms<mode>4): Likewise.
40044 2017-01-04 Marek Polacek <polacek@redhat.com>
40047 * doc/invoke.texi: Document -Wpointer-compare.
40049 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40051 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40054 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40055 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40058 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40061 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40062 documentation of the powerpc_popcntb_ok attribute.
40063 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40064 code to issue warning messages if a requested CPU configuration is
40065 not supported by the binary (assembler and loader) toolchain.
40066 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40067 made to define a built-in function that has been disabled.
40068 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40069 made to define a built-in function that has been disabled.
40070 (altivec_init_builtins): Add comment explaining why definition
40071 of the DST built-in functions is not preceded by an assertion
40072 check. Add assertions to prevent ICE if attempts are made to
40073 define an altivec predicate or an abs* built-in function that has
40075 (htm_init_builtins): Add comment explaining why definition of the
40076 htm built-in functions is not preceded by an assertion check.
40078 2017-01-04 Jeff Law <law@redhat.com>
40080 PR tree-optimizatin/67955
40081 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40082 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40083 the points-to solution does not include pt_null. Use DECL_PT_UID
40086 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40088 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40089 Use gen_int_mode instead of gen_lopwart for const_int operands.
40091 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40093 PR tree-optimization/71563
40094 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40095 out of range value - has low bits known to be zero.
40097 2017-01-04 Alan Modra <amodra@gmail.com>
40099 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40100 * configure: Regenerate.
40101 * config.in: Regenerate.
40103 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40106 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40107 a substring of the message, but strcmp with the whole message. Ifdef
40108 ENABLE_NLS, translate the message first using dgettext.
40110 2017-01-03 Jeff Law <law@redhat.com>
40112 PR tree-optimizatin/78856
40113 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40114 (mark_threaded_blocks): Remove code to truncate thread paths that
40115 cross multiple loop headers. Instead invalidate the cached loop
40116 iteration information and handle case of a thread path walking
40117 into an irreducible region.
40119 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40122 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40123 assertions. Add support for doing the signbit if the IEEE 128-bit
40124 floating point value is in a GPR.
40125 * config/rs6000/rs6000.md (Fsignbit): Delete.
40126 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40127 Update the length attribute if the value is in a GPR.
40128 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40129 the sign or zero extension instruction, since the value is always 0/1.
40130 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40133 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40134 extracting SImode to a GPR register so that we can generate a
40135 store, limit the vector to be in a traditional Altivec register
40136 for the vextuwrx instruction.
40138 2017-01-03 Ian Lance Taylor <iant@google.com>
40140 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40142 2017-01-03 Martin Sebor <msebor@redhat.com>
40144 PR tree-optimization/78696
40145 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40146 precision. Use MPFR for %f for greater fidelity. Correct handling
40148 (pass_sprintf_length::compute_format_length): Set width and precision
40149 specified by asrerisk to void_node for vararg functions.
40150 (try_substitute_return_value): Adjust dump output.
40152 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40154 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40156 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40158 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40159 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40160 -mlra/-mno-lra was passed to the compiler.
40162 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40164 PR rtl-optimization/65618
40165 * emit-rtl.c (try_split): Move initialization of "before" and
40166 "after" to just before the call to emit_insn_after_setloc.
40168 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40170 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40172 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40174 * dwarf2out.c (gen_enumeration_type_die): When
40175 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40177 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40179 PR tree-optimization/78965
40180 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40181 Change first argument from const call_info & to call_info &. For %n
40182 set info.nowrite to false.
40184 PR middle-end/78901
40185 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40186 possibly throwing calls.
40188 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40189 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40190 and fns handling, rather than in a separate case SSA_NAME.
40192 2017-01-02 Jeff Law <law@redhat.com>
40194 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40195 fixes for first_period and second_period variables.
40197 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40200 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40201 (*insvqi_1): New insn pattern.
40202 (*insvqi_1_mem_rex64): Ditto.
40203 (*insvqi_2): Ditto.
40204 (*insvqi_3): Rename from *insvqi.
40206 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40208 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40210 * doc/cfg.texi (Edges): Remove reference to Java.
40211 (Maintaining the CFG): Ditto.
40213 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40215 PR middle-end/77674
40216 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40217 transparent aliases.
40219 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40221 PR middle-end/77484
40222 * predict.def (PRED_CALL): Update hitrate.
40223 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40224 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40225 into direct/indirect/polymorphic variants.
40227 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40229 Update copyright years.
40231 * gcc.c (process_command): Update copyright notice dates.
40232 * gcov-dump.c (print_version): Ditto.
40233 * gcov.c (print_version): Ditto.
40234 * gcov-tool.c (print_version): Ditto.
40235 * gengtype.c (create_file): Ditto.
40236 * doc/cpp.texi: Bump @copying's copyright year.
40237 * doc/cppinternals.texi: Ditto.
40238 * doc/gcc.texi: Ditto.
40239 * doc/gccint.texi: Ditto.
40240 * doc/gcov.texi: Ditto.
40241 * doc/install.texi: Ditto.
40242 * doc/invoke.texi: Ditto.
40244 Copyright (C) 2017 Free Software Foundation, Inc.
40246 Copying and distribution of this file, with or without modification,
40247 are permitted in any medium without royalty provided the copyright
40248 notice and this notice are preserved.