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 * expmed.h (store_bit_field): Take bitsize and bitnum as
6 poly_uint64s rather than unsigned HOST_WIDE_INTs.
7 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
8 that returns the byte size.
9 (store_bit_field_1): Take bitsize and bitnum as
10 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
11 to simple_mem_bitfield_p. Split the part that can only handle
12 constant bitsize and bitnum out into...
13 (store_integral_bit_field): ...this new function.
14 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
15 than unsigned HOST_WIDE_INTs.
16 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
18 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
19 Alan Hayward <alan.hayward@arm.com>
20 David Sherwood <david.sherwood@arm.com>
22 * lra-int.h (lra_reg): Change offset from int to poly_int64.
23 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
25 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
26 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
27 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
28 offset as a poly_int64 rather than an int.
29 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
30 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
31 * lra-constraints.c (equiv_address_substitution): Track offsets
33 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
34 (curr_insn_transform): Handle the new form of sp_offset.
35 * lra-eliminations.c (lra_elim_table): Change previous_offset
36 and offset from HOST_WIDE_INT to poly_int64.
37 (print_elim_table, update_reg_eliminate): Update accordingly.
38 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
39 (get_elimination): Update accordingly.
40 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
41 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
42 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
43 poly_int64 offsets generally.
44 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
45 (mark_not_eliminable, init_elimination): Update accordingly.
46 (remove_reg_equal_offset_note): Return a bool and pass the new
47 offset back by pointer as a poly_int64.
48 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
49 rather than a HOST_WIDE_INT.
50 (do_remat): Track offsets poly_int64s.
51 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
53 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
54 Alan Hayward <alan.hayward@arm.com>
55 David Sherwood <david.sherwood@arm.com>
57 * rtl.h (mem_attrs): Add a default constructor. Change size and
58 offset from HOST_WIDE_INT to poly_int64.
59 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
60 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
61 (widen_memory_access): Take the sizes and offsets as poly_int64s
62 rather than HOST_WIDE_INTs.
63 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
64 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
66 (adjust_offset_for_component_ref): Change the offset from a
67 HOST_WIDE_INT to a poly_int64.
68 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
69 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
70 * dce.c (find_call_stack_args): Likewise.
71 * dse.c (record_store): Likewise.
72 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
73 * print-rtl.c (rtx_writer::print_rtx): Likewise.
74 * read-rtl-function.c (test_loading_mem): Likewise.
75 * rtlanal.c (may_trap_p_1): Likewise.
76 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
77 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
78 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
79 (mem_attrs::mem_attrs): New function.
80 (set_mem_attributes_minus_bitpos): Change bitpos from a
81 HOST_WIDE_INT to poly_int64.
82 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
83 (clear_mem_offset, clear_mem_size, change_address)
84 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
86 (set_mem_offset, set_mem_size, adjust_address_1)
87 (adjust_automodify_address_1, offset_address, widen_memory_access):
88 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
90 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
91 Alan Hayward <alan.hayward@arm.com>
92 David Sherwood <david.sherwood@arm.com>
94 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
95 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
96 rather than 0 to represent an unknown size. Assert that the size
97 is known when the mode isn't BLKmode.
98 (may_trap_p_1): Use -1 for unknown sizes.
99 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
101 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
102 Alan Hayward <alan.hayward@arm.com>
103 David Sherwood <david.sherwood@arm.com>
105 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
106 to poly_int64. Update commentary for positions_needed.large.
107 (read_info_type): Change offset and width from HOST_WIDE_INT
109 (set_usage_bits): Likewise.
110 (canon_address): Return the offset as a poly_int64 rather than
111 a HOST_WIDE_INT. Use strip_offset_and_add.
112 (set_all_positions_unneeded, any_positions_needed_p): Use
113 positions_needed.large to track stores with non-constant widths.
114 (all_positions_needed_p): Likewise. Take the offset and width
115 as poly_int64s rather than ints. Assert that rhs is nonnull.
116 (record_store): Cope with non-constant offsets and widths.
117 Nullify the rhs of an earlier store if we can't tell which bytes
119 (find_shift_sequence): Take the access_size and shift as poly_int64s
121 (get_stored_val): Take the read_offset and read_width as poly_int64s
122 rather than HOST_WIDE_INTs.
123 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
124 non-constant offsets and widths.
126 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * inchash.h (inchash::hash::add_poly_int): New function.
131 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
132 Use poly_int64 rather than HOST_WIDE_INT.
133 (ao_ref::max_size_known_p): New function.
134 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
135 rather than HOST_WIDE_INT.
136 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
137 to temporaries until its interface is adjusted to match.
138 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
139 (aliasing_component_refs_p, decl_refs_may_alias_p)
140 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
141 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
142 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
144 * alias.c (ao_ref_from_mem): Likewise.
145 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
146 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
147 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
148 (maybe_trim_complex_store, maybe_trim_constructor_store)
149 (live_bytes_read, dse_classify_store): Likewise.
150 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
151 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
152 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
153 (vn_reference_lookup_3): Likewise.
154 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
156 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
157 Alan Hayward <alan.hayward@arm.com>
158 David Sherwood <david.sherwood@arm.com>
160 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
161 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
162 instead of ranges_overlap_p.
164 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
165 Alan Hayward <alan.hayward@arm.com>
166 David Sherwood <david.sherwood@arm.com>
168 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
169 sizes as poly_int64s rather than HOST_WIDE_INTs.
171 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
172 Alan Hayward <alan.hayward@arm.com>
173 David Sherwood <david.sherwood@arm.com>
175 * gimple-fold.h (fold_ctor_reference): Take the offset and size
176 as poly_uint64 rather than unsigned HOST_WIDE_INT.
177 * gimple-fold.c (fold_ctor_reference): Likewise.
179 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
180 Alan Hayward <alan.hayward@arm.com>
181 David Sherwood <david.sherwood@arm.com>
183 * target.def (dwarf_poly_indeterminate_value): New hook.
184 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
185 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
186 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
187 * doc/tm.texi: Regenerate.
188 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
189 offset as a poly_int64.
190 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
191 offset as a poly_int64.
192 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
193 Take the offset as a poly_int64.
194 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
195 (frame_pointer_fb_offset): Change to a poly_int64.
196 (int_loc_descriptor): Take the offset as a poly_int64. Use
197 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
198 (based_loc_descr): Take the offset as a poly_int64.
199 Use strip_offset_and_add to handle (plus X (const)).
200 Use new_reg_loc_descr instead of an open-coded version of the
201 previous implementation.
202 (mem_loc_descriptor): Handle CONST_POLY_INT.
203 (compute_frame_pointer_to_fb_displacement): Take the offset as a
204 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
206 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
207 Alan Hayward <alan.hayward@arm.com>
208 David Sherwood <david.sherwood@arm.com>
210 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
211 (gen_rtx_REG_offset): Take the offset as a poly_int64.
212 * inchash.h (inchash::hash::add_poly_hwi): New function.
213 * gengtype.c (main): Register poly_int64.
214 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
215 offset as a poly_int.
216 (reg_attr_hasher::equal): Use must_eq to compare offsets.
217 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
218 offset as a poly_int64.
219 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
220 * print-rtl.c (print_poly_int): New function.
221 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
223 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
225 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
226 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
227 (vt_get_decl_and_offset): Return the offset as a poly_int64.
228 Enforce track_offset_p for parts of a PARALLEL.
229 (vt_add_function_parameter): Use const_offset for the final
230 offset to track. Use get_tracked_reg_offset for the parts
233 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
234 Alan Hayward <alan.hayward@arm.com>
235 David Sherwood <david.sherwood@arm.com>
237 * target.def (truly_noop_truncation): Take poly_uint64s instead of
238 unsigned ints. Change default to hook_bool_puint64_puint64_true.
239 * doc/tm.texi: Regenerate.
240 * hooks.h (hook_bool_uint_uint_true): Delete.
241 (hook_bool_puint64_puint64_true): Declare.
242 * hooks.c (hook_bool_uint_uint_true): Delete.
243 (hook_bool_puint64_puint64_true): New function.
244 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
245 instead of unsigned ints.
246 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
247 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
249 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
250 Alan Hayward <alan.hayward@arm.com>
251 David Sherwood <david.sherwood@arm.com>
253 * optabs.h (expand_operand): Add an int_value field.
254 (create_expand_operand): Add an int_value parameter and use it
255 to initialize the new expand_operand field.
256 (create_integer_operand): Replace with a declaration of a function
257 that accepts poly_int64s. Move the implementation to...
258 * optabs.c (create_integer_operand): ...here.
259 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
260 the mode preserves the value of int_value, instead of calling
261 const_int_operand on the rtx. Use gen_int_mode to generate
264 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
265 Alan Hayward <alan.hayward@arm.com>
266 David Sherwood <david.sherwood@arm.com>
268 * dumpfile.h (dump_dec): Declare.
269 * dumpfile.c (dump_dec): New function.
270 * pretty-print.h (pp_wide_integer): Turn into a function and
271 declare a poly_int version.
272 * pretty-print.c (pp_wide_integer): New function for poly_ints.
274 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
275 Alan Hayward <alan.hayward@arm.com>
276 David Sherwood <david.sherwood@arm.com>
278 * doc/generic.texi (POLY_INT_CST): Document.
279 * tree.def (POLY_INT_CST): New tree code.
280 * treestruct.def (TS_POLY_INT_CST): New tree layout.
281 * tree-core.h (tree_poly_int_cst): New struct.
282 (tree_node): Add a poly_int_cst field.
283 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
284 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
285 instead of a wide_int_ref.
286 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
288 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
289 instead of an unsigned HOST_WIDE_INT.
290 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
291 (ptrdiff_tree_p): Declare.
292 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
293 extern inline implementations if the target doesn't use POLY_INT_CST.
294 (poly_int_tree_p): New function.
295 (wi::unextended_tree): New class.
296 (wi::int_traits <unextended_tree>): New override.
297 (wi::extended_tree): Add a default constructor.
298 (wi::extended_tree::get_tree): New function.
299 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
300 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
301 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
302 (wi::tree_to_poly_wide_ref): New typedefs.
303 (wi::ints_for): Provide overloads for extended_tree and
305 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
306 (wi::to_wide): New functions.
307 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
308 * tree.c (poly_int_cst_hasher): New struct.
309 (poly_int_cst_hash_table): New variable.
310 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
311 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
313 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
314 (init_ttree): Initialize poly_int_cst_hash_table.
315 (build_int_cst, build_int_cst_type, build_invariant_address): Take
316 a poly_int64 instead of a HOST_WIDE_INT.
317 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
318 instead of an unsigned HOST_WIDE_INT.
319 (wide_int_to_tree): Rename to...
320 (wide_int_to_tree_1): ...this.
321 (build_new_poly_int_cst, build_poly_int_cst): New functions.
322 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
323 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
324 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
325 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
326 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
328 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
329 (streamer_read_tree_body): Likewise.
330 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
331 (streamer_write_tree_body): Likewise.
332 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
333 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
334 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
335 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
336 * gimple-expr.h (is_gimple_constant): Likewise.
337 * gimplify.c (maybe_with_size_expr): Likewise.
338 * print-tree.c (print_node): Likewise.
339 * tree-data-ref.c (data_ref_compare_tree): Likewise.
340 * tree-pretty-print.c (dump_generic_node): Likewise.
341 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
342 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
343 * tree-vrp.c (compare_values_warnv): Likewise.
344 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
345 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
346 (force_expr_to_var_cost): Likewise.
347 * tree-ssa-loop.c (for_each_index): Likewise.
348 * fold-const.h (build_invariant_address, size_int_kind): Take a
349 poly_int64 instead of a HOST_WIDE_INT.
350 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
351 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
353 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
354 (int_const_binop_2): New function, split out from...
355 (int_const_binop_1): ...here. Handle POLY_INT_CST.
356 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
357 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
358 * gimple-ssa-strength-reduction.c (slsr_process_add)
359 (slsr_process_mul): Check for INTEGER_CSTs before using them
361 * stor-layout.c (bits_from_bytes): New function.
362 (bit_from_pos): Use it.
363 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
364 by BITS_PER_UNIT to get the TYPE_SIZE.
365 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
366 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
368 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
369 Alan Hayward <alan.hayward@arm.com>
370 David Sherwood <david.sherwood@arm.com>
372 * doc/rtl.texi (const_poly_int): Document. Also document the
373 rtl sharing behavior.
374 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
375 * rtl.h (const_poly_int_def): New struct.
376 (rtx_def::u): Add a cpi field.
377 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
378 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
379 (wi::rtx_to_poly_wide_ref): New typedef
380 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
381 (poly_int_rtx_p): New functions.
382 (trunc_int_for_mode): Declare a poly_int64 version.
383 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
384 (immed_wide_int_const): Take a poly_wide_int_ref rather than
386 (strip_offset): Declare.
387 (strip_offset_and_add): New function.
388 * rtl.def (CONST_POLY_INT): New rtx code.
389 * rtl.c (rtx_size): Handle CONST_POLY_INT.
390 (shared_const_p): Use poly_int_rtx_p.
391 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
393 (gen_int_shift_amount): Likewise.
394 * emit-rtl.c (const_poly_int_hasher): New class.
395 (const_poly_int_htab): New variable.
396 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
397 (const_poly_int_hasher::hash): New function.
398 (const_poly_int_hasher::equal): Likewise.
399 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
400 (immed_wide_int_const): Rename to...
401 (immed_wide_int_const_1): ...this and make static.
402 (immed_wide_int_const): New function, taking a poly_wide_int_ref
403 instead of a wide_int_ref.
404 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
405 (gen_lowpart_common): Handle CONST_POLY_INT.
406 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
407 * cselib.c (cselib_hash_rtx): Likewise.
408 * dwarf2out.c (const_ok_for_output_1): Likewise.
409 * expr.c (convert_modes): Likewise.
410 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
411 * rtlhash.c (add_rtx): Likewise.
412 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
413 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
414 Handle existing CONST_POLY_INT rtxes.
415 * expmed.h (expand_shift): Take a poly_int64 instead of a
417 * expmed.c (expand_shift): Likewise.
418 * rtlanal.c (strip_offset): New function.
419 (commutative_operand_precedence): Give CONST_POLY_INT the same
420 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
422 * rtl-tests.c (const_poly_int_tests): New struct.
423 (rtl_tests_c_tests): Use it.
424 * simplify-rtx.c (simplify_const_unary_operation): Handle
426 (simplify_const_binary_operation): Likewise.
427 (simplify_binary_operation_1): Fold additions of symbolic constants
429 (simplify_subreg): Handle extensions and truncations of
431 (simplify_const_poly_int_tests): New struct.
432 (simplify_rtx_c_tests): Use it.
433 * wide-int.h (storage_ref): Add default constructor.
434 (wide_int_ref_storage): Likewise.
435 (trailing_wide_ints): Use GTY((user)).
436 (trailing_wide_ints::operator[]): Add a const version.
437 (trailing_wide_ints::get_precision): New function.
438 (trailing_wide_ints::extra_size): Likewise.
440 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
441 Alan Hayward <alan.hayward@arm.com>
442 David Sherwood <david.sherwood@arm.com>
444 * emit-rtl.h (gen_int_shift_amount): Declare.
445 * emit-rtl.c (gen_int_shift_amount): New function.
446 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
448 * calls.c (shift_return_value): Likewise.
449 * cse.c (fold_rtx): Likewise.
450 * dse.c (find_shift_sequence): Likewise.
451 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
452 (expand_shift, expand_smod_pow2): Likewise.
453 * lower-subreg.c (shift_cost): Likewise.
454 * optabs.c (expand_superword_shift, expand_doubleword_mult)
455 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
456 (expand_vec_perm_var): Likewise.
457 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
458 (simplify_binary_operation_1): Likewise.
459 * combine.c (try_combine, find_split_point, force_int_to_mode)
460 (simplify_shift_const_1, simplify_shift_const): Likewise.
461 (change_zero_ext): Likewise. Use simplify_gen_binary.
463 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
465 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
467 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
469 * doc/invoke.texi (ARM Options): Document accepted extension options
470 for -march=armv8.3-a.
472 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
475 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
476 or --with-float=softfp, set the default CPU to arm10e.
478 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
480 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
481 * config/visium/predicates.md (const_shift_operand): Likewise.
482 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
483 (visium_legitimize_reload_address): Likewise.
485 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
488 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
490 2017-12-20 Jakub Jelinek <jakub@redhat.com>
493 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
495 2017-12-20 Julia Koval <julia.koval@intel.com>
497 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
498 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
499 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
500 * config.gcc: Include vpclmulqdqintrin.h.
501 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
502 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
504 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
505 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
506 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
507 * config/i386/i386.c (isa2_opts): Add -mcx16.
508 (isa_opts): Add -mpclmulqdq, remove -mcx16.
509 (ix86_option_override_internal): Move mcx16 to flags2.
510 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
511 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
512 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
513 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
514 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
515 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
516 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
517 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
518 * doc/invoke.texi: Add -mvpclmulqdq.
520 2017-12-20 Tom de Vries <tom@codesourcery.com>
523 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
524 * calls.c (rtx_for_static_chain): ... here. New function.
525 * calls.h (rtx_for_static_chain): Declare.
526 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
527 instead of targetm.calls.static_chain.
528 * df-scan.c (df_get_entry_block_def_set): Same.
530 2017-12-19 Tom de Vries <tom@codesourcery.com>
532 PR tree-optimization/83493
533 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
536 2017-12-19 Jakub Jelinek <jakub@redhat.com>
538 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
539 inform with hardcoded english plural handling.
541 2017-12-18 Jeff Law <law@redhat.com>
543 PR tree-optimization/83477
544 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
545 a non-virtual PHI, always push a new range.
547 2017-12-19 Martin Sebor <msebor@redhat.com>
550 * builtins.c (compute_objsize): Handle non-constant offsets.
552 2017-12-19 Jakub Jelinek <jakub@redhat.com>
554 PR tree-optimization/83444
555 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
556 character load case, if get_stridx on MEM_REF's operand doesn't
557 look usable, retry with get_addr_stridx.
559 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
562 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
563 (variable_tracking_main_1): Keep markers even when VTA fails.
566 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
567 even if there are markers before it.
568 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
570 2017-12-19 Jakub Jelinek <jakub@redhat.com>
572 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
573 typical order conditions.
574 * sel-sched.c (extract_new_fences_from): Likewise.
575 * config/visium/constraints.md (J, K, L): Likewise.
576 * config/visium/predicates.md (const_shift_operand): Likewise.
577 * config/visium/visium.c (visium_legitimize_address,
578 visium_legitimize_reload_address): Likewise.
579 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
580 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
581 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
582 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
583 SET_DUMP_DETAIL): Likewise.
584 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
585 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
586 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
587 avr_set_core_architecture, avr_set_current_function,
588 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
589 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
590 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
591 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
592 avr_map_decompose, avr_fold_builtin): Likewise.
593 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
594 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
595 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
596 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
597 * config/m32c/m32c.c (m32c_conditional_register_usage,
598 m32c_address_cost): Likewise.
599 * config/m32c/predicates.md (shiftcount_operand,
600 longshiftcount_operand): Likewise.
601 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
602 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
603 can_use_cdx_ldstw): Likewise.
604 * config/nios2/nios2.h (CDX_REG_P): Likewise.
605 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
607 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
608 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
609 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
610 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
611 * config/vax/vax.c (adjacent_operands_p): Likewise.
612 * config/ft32/constraints.md (L, b, KA): Likewise.
613 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
615 * cfgexpand.c (expand_stack_alignment): Likewise.
616 * gcse.c (insert_expr_in_table): Likewise.
617 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
618 * cgraphunit.c (cgraph_node::expand): Likewise.
619 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
620 * emit-rtl.c (add_insn): Likewise.
621 * input.c (dump_location_info): Likewise.
622 * passes.c (NEXT_PASS): Likewise.
623 * read-rtl-function.c (parse_note_insn_name,
624 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
626 * sched-rgn.c (sched_rgn_init): Likewise.
627 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
628 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
629 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
630 * explow.c (eliminate_constant_term): Likewise.
631 * final.c (leaf_renumber_regs_insn): Likewise.
632 * cfgrtl.c (print_rtl_with_bb): Likewise.
633 * genhooks.c (emit_init_macros): Likewise.
634 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
635 * tree-data-ref.c (conflict_fn): Likewise.
636 * selftest.c (assert_streq): Likewise.
637 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
638 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
639 fold_binary_loc, multiple_of_p): Likewise.
640 * reload.c (push_reload, find_equiv_reg): Likewise.
641 * et-forest.c (et_nca, et_below): Likewise.
642 * dbxout.c (dbxout_symbol_location): Likewise.
643 * reorg.c (relax_delay_slots): Likewise.
644 * dojump.c (do_compare_rtx_and_jump): Likewise.
645 * gengtype-parse.c (type): Likewise.
646 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
647 simplify_const_relational_operation): Likewise.
648 * reload1.c (do_output_reload): Likewise.
649 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
650 * gengtype.c (type_for_name): Likewise.
651 * gimple-ssa-sprintf.c (format_directive): Likewise.
653 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
656 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
657 accessing it. Adjust comment.
659 2017-12-19 Jakub Jelinek <jakub@redhat.com>
662 * predict.c (zero_one_minusone): New function.
663 (apply_return_prediction): Avoid return prediction for functions
664 returning only -1, 0 and 1 values, unless they only return -1 and 0
667 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
669 * config/arc/arc.c (legitimate_scaled_address_p): Clean
670 fall-through warning.
671 (arc_compute_frame_size): Remove unused variables.
672 (arc_print_operand): Fix fprintif format.
673 (arc_can_follow_jump): Clean fall-through warning.
675 2017-12-19 Marek Polacek <polacek@redhat.com>
678 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
681 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
683 * config/arc/arc.c (overriderregs): New variable.
684 (arc_override_options): Track fixed/call saved/call options.
685 (arc_conditional_register_usage): Check against overrideregs
686 variable whenever we change register properties.
688 2017-12-19 Nathan Sidwell <nathan@acm.org>
690 * opts.c (finish_options): Don't prefix dump_base_name if it
691 already contains directories.
693 2017-12-19 Martin Liska <mliska@suse.cz>
695 PR rtl-optimization/82675
696 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
697 more element in sbitmap.
699 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
701 * gimplify.c (gimplify_expr): Use error_operand_p.
703 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
706 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
709 2017-12-19 Daniel Cederman <cederman@gaisler.com>
711 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
712 the jump is to a label.
714 2017-12-19 Jakub Jelinek <jakub@redhat.com>
716 PR tree-optimization/83444
717 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
722 * ipa-inline.c (flatten_remove_node_hook): New function.
723 (ipa_inline): Keep only nodes with flatten attribute at the end of
724 the array in the order from ipa_reverse_postorder, only walk that
725 portion of array for flattening, if there is more than one such
726 node, temporarily register a removal hook and ignore removed nodes.
728 PR tree-optimization/80631
729 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
730 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
731 IFN_REDUC_MAX or IFN_REDUC_MIN.
733 2017-12-18 Jeff Law <law@redhat.com>
735 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
736 record symbolic equivalences from backedges in the CFG.
739 2017-11-19 Jeff Law <law@redhat.com>
741 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
742 of degenerates resulting from ignoring an edge.
744 2017-12-18 Martin Sebor <msebor@redhat.com>
747 PR tree-optimization/78450
748 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
749 (handle_builtin_strlen): Call it.
751 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
753 PR rtl-optimization/83424
754 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
756 2017-12-18 Marek Polacek <polacek@redhat.com>
759 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
760 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
763 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
765 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
766 (umaddsidi4, umaddsidi_split): Likewise.
768 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
770 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
773 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
776 * config/s390/s390.c (s390_option_override): Avoid strncpy.
778 2017-12-18 Richard Biener <rguenther@suse.de>
780 PR tree-optimization/81877
781 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
782 (outermost_indep_loop): Adjust.
783 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
784 (can_sm_ref_p): Adjust.
786 2017-12-18 Richard Biener <rguenther@suse.de>
789 * tree.c (array_at_struct_end_p): Return true if the underlying
790 object has space for at least one element in excess of what
791 the array domain specifies.
793 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
795 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
796 example to avoid overfull hbox.
797 * doc/invoke.texi (Option Summary): Add missing @gol.
798 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
801 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
802 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
804 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
805 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
806 (AArch64 Options): Document -mverbose-cost-dump.
807 (ARM Options): Likewise, plus -mflip-thumb.
809 2017-12-17 Martin Sebor <msebor@redhat.com>
812 * gimple-ssa-warn-restrict.c
813 (builtin_memref::offset_out_of_bounds): Correct the handling of
816 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
818 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
820 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
822 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
824 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
826 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
829 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
831 * doc/invoke.texi: Fix some typos.
833 2017-12-16 Martin Sebor <msebor@redhat.com>
835 PR tree-optimization/78918
836 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
837 * builtins.c (check_sizes): Rename...
838 (check_access): ...to this. Rename function arguments for clarity.
839 (check_memop_sizes): Adjust names.
840 (expand_builtin_memchr, expand_builtin_memcpy): Same.
841 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
842 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
843 (check_strncat_sizes, expand_builtin_strncat): Same.
844 (expand_builtin_strncpy, expand_builtin_memset): Same.
845 (expand_builtin_bzero, expand_builtin_memcmp): Same.
846 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
847 (maybe_emit_sprintf_chk_warning): Same.
848 (expand_builtin_strcpy): Adjust.
849 (expand_builtin_stpcpy): Same.
850 (expand_builtin_with_bounds): Detect out-of-bounds accesses
851 in pointer-checking forms of memcpy, memmove, and mempcpy.
852 (gcall_to_tree_minimal, max_object_size): Define new functions.
853 * builtins.h (max_object_size): Declare.
854 * calls.c (alloc_max_size): Call max_object_size instead of
855 hardcoding ssizetype limit.
856 (get_size_range): Handle new argument.
857 * calls.h (get_size_range): Add a new argument.
858 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
859 * doc/invoke.texi (-Wrestrict): Adjust, add example.
860 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
862 (gimple_fold_builtin_memory_chk): Same.
863 (gimple_fold_builtin_stxcpy_chk): New function.
864 * gimple-ssa-warn-restrict.c: New source.
865 * gimple-ssa-warn-restrict.h: New header.
866 * gimple.c (gimple_build_call_from_tree): Propagate location.
867 * passes.def (pass_warn_restrict): Add new pass.
868 * tree-pass.h (make_pass_warn_restrict): Declare.
869 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
871 (handle_builtin_strcat): Same.
872 (strlen_optimize_stmt): Rename...
873 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
874 stpncpy, strncpy, and their checking forms.
876 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
878 PR rtl-optimization/82849
879 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
880 and get_max_loop_iterations_int.
882 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
883 Alan Hayward <alan.hayward@arm.com>
884 David Sherwood <david.sherwood@arm.com>
886 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
887 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
888 size as a poly_uint64.
889 (mode_for_vector, mode_for_int_vector): Take the number of vector
890 elements as a poly_uint64.
891 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
892 size as a poly_uint64.
893 (mode_for_vector, mode_for_int_vector): Take the number of vector
894 elements as a poly_uint64.
896 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
897 Alan Hayward <alan.hayward@arm.com>
898 David Sherwood <david.sherwood@arm.com>
900 * machmode.h (MACRO_MODE): New macro.
901 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
902 * targhooks.c (default_libcall_value, default_secondary_reload)
903 (default_memory_move_cost, default_register_move_cost)
904 (default_class_max_nregs): Likewise.
906 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
907 Alan Hayward <alan.hayward@arm.com>
908 David Sherwood <david.sherwood@arm.com>
910 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
911 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
913 * genautomata.c (main): Likewise.
914 * genconditions.c (write_header): Likewise.
915 * genemit.c (main): Likewise.
916 * genextract.c (print_header): Likewise.
917 * genopinit.c (main): Likewise.
918 * genoutput.c (output_prologue): Likewise.
919 * genpeep.c (main): Likewise.
920 * genpreds.c (write_insn_preds_c): Likewise.
921 * genrecog.c (writer_header): Likewise.
922 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
923 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
924 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
925 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
926 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
927 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
928 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
929 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
930 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
931 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
932 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
933 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
934 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
935 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
936 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
937 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
938 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
939 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
940 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
941 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
942 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
943 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
944 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
945 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
946 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
947 * config/darwin.c (IN_TARGET_CODE): Likewise.
948 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
949 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
950 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
951 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
952 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
953 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
954 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
955 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
956 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
957 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
958 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
959 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
960 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
961 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
962 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
963 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
964 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
965 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
966 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
967 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
968 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
969 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
970 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
971 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
972 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
973 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
974 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
975 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
976 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
977 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
978 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
979 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
980 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
981 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
982 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
983 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
984 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
985 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
986 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
987 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
988 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
989 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
990 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
991 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
992 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
993 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
994 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
995 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
996 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
997 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
998 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
999 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1000 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1001 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1002 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1003 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1004 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1005 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1006 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1007 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1008 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1009 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1010 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1011 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1012 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1013 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1014 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1015 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1016 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1017 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1018 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1019 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1020 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1021 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1022 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1023 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1024 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1025 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1026 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1027 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1028 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1029 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1030 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1031 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1032 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1033 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1034 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1035 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1036 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1037 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1038 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1039 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1040 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1041 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1042 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1043 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1044 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1045 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1046 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1047 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1048 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1049 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1050 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1051 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1052 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1053 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1055 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1056 Alan Hayward <alan.hayward@arm.com>
1057 David Sherwood <david.sherwood@arm.com>
1059 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1062 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1063 Alan Hayward <alan.hayward@arm.com>
1064 David Sherwood <david.sherwood@arm.com>
1066 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1067 * doc/md.texi (vec_series@var{m}): Document.
1068 * tree.def (VEC_SERIES_EXPR): New tree code.
1069 * tree.h (build_vec_series): Declare.
1070 * tree.c (build_vec_series): New function.
1071 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1072 * tree-pretty-print.c (dump_generic_node): Likewise.
1073 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1074 * tree-inline.c (estimate_operator_cost): Likewise.
1075 * expr.c (expand_expr_real_2): Likewise.
1076 * optabs-tree.c (optab_for_tree_code): Likewise.
1077 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1078 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1079 * expmed.c (make_tree): Handle VEC_SERIES.
1080 * optabs.def (vec_series_optab): New optab.
1081 * optabs.h (expand_vec_series_expr): Declare.
1082 * optabs.c (expand_vec_series_expr): New function.
1083 * tree-vect-generic.c (expand_vector_operations_1): Check that
1084 the operands also have vector type.
1086 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1087 Alan Hayward <alan.hawyard@arm.com>
1088 David Sherwood <david.sherwood@arm.com>
1090 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1091 (VEC_COND_EXPR): Add missing @tindex.
1092 * doc/md.texi (vec_duplicate@var{m}): Document.
1093 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1094 * tree.c (build_vector_from_val): Add stubbed-out handling of
1095 variable-length vectors, using VEC_DUPLICATE_EXPR.
1096 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1097 * cfgexpand.c (expand_debug_expr): Likewise.
1098 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1099 * tree-inline.c (estimate_operator_cost): Likewise.
1100 * tree-pretty-print.c (dump_generic_node): Likewise.
1101 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1102 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1103 (test_vec_duplicate_folding): New function.
1104 (fold_const_c_tests): Call it.
1105 * optabs.def (vec_duplicate_optab): New optab.
1106 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1107 * optabs.h (expand_vector_broadcast): Declare.
1108 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1109 vec_duplicate_optab.
1110 * expr.c (store_constructor): Try using vec_duplicate_optab for
1112 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1114 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1117 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1118 div/mod latencies a bit.
1120 2017-12-15 Jeff Law <law@redhat.com>
1122 PR tree-optimization/36550
1123 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1124 (mark_threaded_blocks): Rewrite code to avoid block copying when
1125 optimizing for size. Don't pessimize blocks which will be
1126 copied, but all the statements will be dead.
1128 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1130 PR tree-optimization/81165
1131 * tree-ssa-threadupdate.c (uses_in_bb): New.
1132 (estimate_threading_killed_stmts): New.
1133 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1134 * tree-ssa-threadedge.c
1135 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1138 2017-12-15 Jeff Law <law@redhat.com>
1140 PR tree-optimization/83410
1141 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1142 threads when parallelizing loops.
1144 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1146 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1148 * config/alpha/alpha.c (vms_attribute_table): Swap
1149 affects_type_identity and handler fields, adjust comments.
1150 * config/mips/mips.c (mips_attribute_table): Likewise.
1151 * config/visium/visium.c (visium_attribute_table): Likewise.
1152 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1153 * config/microblaze/microblaze.c (microblaze_attribute_table):
1155 * config/spu/spu.c (spu_attribute_table): Likewise.
1156 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1157 * config/arc/arc.c (arc_attribute_table): Likewise.
1158 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1159 * config/v850/v850.c (v850_handle_interrupt_attribute,
1160 v850_handle_data_area_attribute): Formatting fixes.
1161 (v850_attribute_table): Swap affects_type_identity and handler
1162 fields, adjust comments.
1163 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1164 * config/arm/arm.c (arm_attribute_table): Likewise.
1165 * config/avr/avr.c (avr_attribute_table): Likewise.
1166 * config/s390/s390.c (s390_attribute_table): Likewise.
1167 * config/sh/sh.c (sh_attribute_table): Likewise.
1168 * config/i386/i386.c (ix86_handle_cconv_attribute,
1169 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1170 (ix86_attribute_table): Swap affects_type_identity and handler
1171 fields, adjust comments.
1172 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1173 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1174 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1175 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1176 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1177 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1178 * config/rx/rx.c (rx_attribute_table): Likewise.
1179 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1180 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1181 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1182 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1183 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1184 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1185 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1186 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1187 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1188 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1189 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1191 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1193 (register_scoped_attributes, decl_attributes): Formatting fixes.
1195 PR tree-optimization/83269
1196 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1197 subtraction in arg0's type if type is signed and arg0 is unsigned.
1201 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1202 convert? on @0 instead of convert. Check type of @1, not @0.
1203 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1204 convert? on @0 instead of convert. Check type of @1, not @0.
1205 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1206 only optimize if either both @1 and @2 types are narrower
1207 precision, or both are wider or equal precision, and in the former
1208 case only if both have undefined overflow.
1210 2017-12-15 Richard Biener <rguenther@suse.de>
1213 * internal-fn.def (IFN_NOP): Add.
1214 * internal-fn.c (expand_NOP): Do nothing.
1215 * lto-streamer-in.c (input_function): Instead of removing
1216 sanitizer calls replace them with IFN_NOP calls.
1218 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1219 Alan Hayward <alan.hayward@arm.com>
1220 David Sherwood <david.sherwood@arm.com>
1222 * dse.c (store_info, read_info_type): Replace begin and end with
1224 (print_range): New function.
1225 (set_all_positions_unneeded, any_positions_needed_p)
1226 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1228 (record_store): Likewise. Optimize the case in which all positions
1230 (get_stored_val): Replace read_begin and read_end with read_offset
1232 (replace_read): Update call accordingly.
1234 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1236 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1237 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1238 (loop_cand::loop_cand): Initialize above members.
1239 (loop_cand::supported_operations): Delete.
1240 (loop_cand::can_interchange_p): Inline above function.
1241 (loop_cand::classify_simple_reduction): Record number of constant
1242 initialized simple reductions.
1243 (should_interchange_loops): New parameters. Check stmt cost of loops
1245 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1246 Update call to should_interchange_loops.
1247 (should_interchange_loop_nest): Update call to
1248 should_interchange_loops.
1250 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1253 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1254 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1255 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1257 2017-12-15 Julia Koval <julia.koval@intel.com>
1259 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1260 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1261 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1262 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1263 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1265 2017-12-15 Julia Koval <julia.koval@intel.com>
1267 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1268 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1269 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1270 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1271 _mm_aesenc_epi128): New intrinsics.
1273 2017-12-15 Julia Koval <julia.koval@intel.com>
1275 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1276 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1277 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1278 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1279 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1281 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1283 * gimple-ssa-strength-reduction.c (analyze_increments):
1284 Distinguish replacement costs for constant strides from those for
1287 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1289 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1291 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1293 * doc/invoke.texi: Document -Wcast-function-type.
1294 * recog.h (stored_funcptr): Change signature.
1295 * tree-dump.c (dump_node): Avoid warning.
1296 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1298 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1301 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1304 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1306 PR tree-optimization/83312
1307 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1308 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1309 "val" param, and to cope with arbitrary basic blocks.
1310 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1311 handle NULL_TREE for "val", dropping "bb" param.
1312 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1313 drop "bb" param. Handle NULL_TREE for "val".
1314 (find_case_label_for_value): Make "switch_stmt" param const.
1315 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1317 (vrp_prop::check_all_array_refs): Reimplement as...
1318 (check_array_bounds_dom_walker::before_dom_children): ...this new
1319 vfunc. Replace linear search through BB block list, excluding
1320 those with non-executable in-edges via dominator walk.
1322 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1324 * config/arm/arm.opt (mverbose-cost-dump): New option.
1325 * config/arm/arm.c (arm_rtx_costs): Use it.
1327 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1330 * reload1.c (emit_input_reload_insns): Skip debug markers.
1332 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1334 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1338 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1339 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1340 * config/sh/sh.c: Skip debug insns besides notes.
1341 * config/sh/sh.md: Likewise.
1342 * config/sh/sh_treg_combine.cc: Likewise.
1343 * config/sh/sync.md: Likewise.
1345 2017-12-14 Tom de Vries <tom@codesourcery.com>
1347 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1348 Add item for weak_undefined.
1350 2017-12-14 Richard Biener <rguenther@suse.de>
1352 PR tree-optimization/67842
1353 * sese.h (bb_in_region): Remove #if 0'ed code.
1355 2017-12-14 Richard Biener <rguenther@suse.de>
1357 PR tree-optimization/83326
1358 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1359 may_be_zero parameter and handle it by not marking the first
1360 peeled copy as not exiting the loop.
1361 (try_peel_loop): Likewise.
1362 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1363 to handle the case of constant or zero iterations and perform
1364 loop header copying on-the-fly.
1366 2017-12-14 Richard Biener <rguenther@suse.de>
1368 PR tree-optimization/83418
1369 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1370 Instead of asserting we don't get unfolded comparisons deal with
1373 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1376 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1377 basic blocks. Assert debug bind insns don't appear outside of bbs,
1378 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1381 PR tree-optimization/83198
1382 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1383 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1384 value if arg is a REAL_CST with incompatible type.
1386 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1387 Bin Cheng <bin.cheng@arm.com>
1390 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1392 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1395 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1397 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1398 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1400 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1401 Alan Hayward <alan.hayward@arm.com>
1402 David Sherwood <david.sherwood@arm.com>
1404 * poly-int.h: New file.
1405 * poly-int-types.h: Likewise.
1406 * coretypes.h: Include them.
1407 (POLY_INT_CONVERSION): Define.
1408 * target.def (estimated_poly_value): New hook.
1409 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1410 * doc/tm.texi: Regenerate.
1411 * doc/poly-int.texi: New file.
1412 * doc/gccint.texi: Include it.
1413 * doc/rtl.texi: Describe restrictions on subreg modes.
1414 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1415 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1416 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1417 * targhooks.h (default_estimated_poly_value): Declare.
1418 * targhooks.c (default_estimated_poly_value): New function.
1419 * target.h (estimated_poly_value): Likewise.
1420 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1421 (wi::unary_traits): Delete.
1422 (wi::binary_traits::signed_shift_result_type): Define for
1423 offset_int << HOST_WIDE_INT, etc.
1424 (generic_wide_int::operator <<=): Define for all types and use
1425 wi::lshift instead of <<.
1426 (wi::hwi_with_prec): Add a default constructor.
1427 (wi::ints_for): New class.
1428 (operator <<): Define for all wide-int types.
1429 (operator /): New function.
1430 (operator %): Likewise.
1431 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1432 (ASSERT_MAYBE_NE_AT): New macros.
1434 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1435 Dominik Vogt <vogt@linux.vnet.ibm.com>
1438 * emit-rtl.c (init_emit): Remove ??? comment.
1439 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1440 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1441 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1442 alignment of 3 virtual registers to BITS_PER_WORD.
1444 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1446 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1448 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1449 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1451 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1454 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1456 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1458 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1459 with the exception of debug begin stmt markers appear before
1463 * final.c (rest_of_handle_final): Call variable_tracking_main only
1464 if !flag_var_tracking.
1466 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1467 Jakub Jelinek <jakub@redhat.com>
1471 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1472 labels debug stmts that can only appear after labels.
1474 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1476 PR rtl-optimization/82398
1477 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1478 EXPR_USEFULNESS in priority comparison.
1480 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1482 PR rtl-optimization/83393
1483 * combine.c (move_deaths): If reg_stat points to a too new insn in
1484 last_death, do not use it: find the proper insn instead.
1486 2017-12-12 Jeff Law <law@redhat.com>
1488 PR tree-optimization/83298
1489 PR tree-optimization/83362
1490 PR tree-optimization/83383
1491 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1492 push_value_range a public interface. Add new argument to
1493 record_ranges_from_stmt.
1494 * gimple-ssa-evrp-analyze.c
1495 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1496 Update comments. Handle recording temporary equivalences.
1497 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1498 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1499 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1500 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1501 gimple-ssa-evrp-analyze.h.
1502 (record_temporary_equivalences_from_phis): Add new argument. When
1503 the PHI arg is an SSA_NAME, set the result's range to the range
1505 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1506 from statements too.
1507 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1508 Pass it down to children as needed.
1509 (thread_outgoing_edges): Likewise.
1510 (thread_across_edge): Likewise. Push/pop range state as needed.
1511 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1513 2017-12-12 Julia Koval <julia.koval@intel.com>
1515 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1516 (PTA_CANNONLAKE): Remove PTA_CLWB.
1518 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1521 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1522 (vcondv2div2df): Likewise.
1523 (vconduv2dfv2di): Likewise.
1525 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1527 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1528 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1529 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1530 ATTR_MATHFN_FPROUNDING_ERRNO.
1532 2017-12-12 Richard Biener <rguenther@suse.de>
1534 PR tree-optimization/83385
1535 * graphite-scop-detection.c (get_order, order): Remove.
1536 (bb_to_rpo): New global.
1538 (build_scops): Sort pbbs in RPO order.
1540 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1542 * combine.c (simplify_set): Do not transform subregs to zero_extends
1543 if the destination mode is a vector mode.
1545 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1547 PR tree-optimization/83359
1548 * tree-cfg.h (fold_loop_internal_call): Declare.
1549 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1550 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1551 (find_loop_dist_alias): New function.
1552 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1553 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1556 PR tree-optimization/80631
1557 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1558 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1559 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1560 hardcoding zero as the value if COND_EXPR is never true. For
1561 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1562 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1563 hardcoding MAX_EXPR as the reduction operation.
1564 (is_nonwrapping_integer_induction): Allow negative step.
1565 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1566 vect_create_epilog_for_reduction, if no value is suitable, don't
1567 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1569 2017-12-12 Richard Biener <rguenther@suse.de>
1571 PR tree-optimization/81889
1572 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1573 range info from the non-wrapping IV instead of just the range
1576 2017-12-12 Julia Koval <julia.koval@intel.com>
1578 * config.gcc: Add vaesintrin.h.
1579 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1581 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1582 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1584 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1585 * config/i386/immintrin.h: Include vaesintrin.h.
1586 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1587 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1588 _mm_aesdec_epi128): New intrinsics.
1590 2017-12-12 Julia Koval <julia.koval@intel.com>
1592 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1593 OPTION_MASK_ISA_VAES_UNSET): New.
1594 (ix86_handle_option): Handle -mvaes.
1595 * config/i386/cpuid.h: Define bit_VAES.
1596 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1597 * config/i386/i386-c.c (__VAES__): New.
1598 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1599 (ix86_valid_target_attribute_inner_p): Ditto.
1600 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1601 * config/i386/i386.opt: Add -mvaes.
1602 * doc/invoke.texi: Ditto.
1604 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1606 * debug.h (gcc_debug_hooks): Add inline_entry.
1607 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1608 * debug.c (do_nothing_debug_hooks): Likewise.
1609 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1610 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1611 (dwarf2_lineno_debug_hooks): Likewise.
1613 * common.opt (gstatement-frontiers): New, setting
1614 debug_nonbind_markers_p.
1615 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1616 * toplev.c (process_options): Autodetect value for debug statement
1618 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1619 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1621 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1622 markers. Integrate source bind into debug stmt expand loop.
1623 (pass_expand::execute): Check debug marker limit. Avoid deep
1624 TER and expand debug locations for debug bind insns only.
1625 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1627 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1628 * final.c (reemit_insn_block_notes): Take current block from
1629 nonbind markers. Declare note where it's first set.
1630 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1631 begin stmt markers if enabled.
1632 (notice_source_line): Handle nonbind markers. Fail if their
1633 location is unknown or that of builtins.
1634 (rest_of_handle_final): Convert begin stmt markers to notes if
1635 var-tracking didn't run.
1636 (rest_of_clean_state): Skip begin stmt markers.
1637 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1639 * function.c (allocate_struct_function): Set begin_stmt_markers.
1640 * function.h (struct function): Add debug_marker_count counter
1641 and debug_nonbind_markers flag.
1642 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1643 * gimple-low.c (lower_function_body): Adjust
1644 debug_nonbind_markers.
1645 (lower_stmt): Drop or skip gimple debug stmts.
1646 (lower_try_catch): Skip debug stmts.
1647 * gimple.c (gimple_build_debug_begin_stmt): New.
1648 (gimple_copy): Increment debug_marker_count if copying one.
1649 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1650 * gimplify.c (rexpr_location): New.
1651 (rexpr_has_location): New.
1652 (warn_switch_unreachable_r): Handle gimple debug stmts.
1653 (shortcut_cond_r): Call expr_location.
1655 (find_goto_label): New.
1656 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1658 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1660 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1661 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1662 (LANG_HOOKS_INITIALIZER): ... this.
1663 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1664 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1666 (update_ebb_live_info): Skip debug insn markers.
1667 * lra.c (debug_insn_static_data): Rename to...
1668 (debug_bind_static_data): ... this.
1669 (debug_marker_static_data): New.
1670 (lra_set_insn_recog_data): Select one of the above depending
1672 (lra_update_isn_regno_info): Don't assume debug insns have
1674 (push_insns): Skip debug insns.
1675 * lto-streamer-in.c (input_function): Drop debug stmts
1676 depending on active options. Adjust debug_nonbind_markers.
1677 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1678 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1679 begin stmt marker notes.
1680 (print_insn): Likewise.
1681 * recog.c (extract_insn): Recognize rtl for debug markers.
1682 * rtl.def (DEBUG_MARKER): New.
1683 * tree-inline.c: Include params.h.
1684 (remap_gimple_stmt): Handle nonbind markers.
1685 (maybe_move_debug_stmts_to_successors): Likewise.
1686 (copy_debug_stmt): Likewise.
1687 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1688 markers regardless of no side effects.
1689 (tsi_link_before): Don't update container's side effects when adding
1690 a begin stmt marker.
1691 (tsi_link_after): Likewise.
1692 (expr_first): Skip begin stmt markers.
1693 (expr_last): Likewise.
1694 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1695 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1696 Disregard nonbind markers.
1697 * tree.c (make_node_stat): Don't set side effects for begin stmt
1699 (build1_stat): Likewise.
1700 * tree.def (DEBUG_BEGIN_STMT): New.
1701 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1702 * var-tracking.c (delete_debug_insns): Renamed to...
1703 (delete_vta_debug_insns): ... this.
1704 (reemit_marker_as_note): New.
1705 (vt_initialize): Reemit markers.
1706 (delete_vta_debug_insns): Likewise.
1707 (vt_debug_insns_local): Reemit or delete markers.
1708 (variable_tracking_main_1): Likewise.
1709 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1710 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1711 (gimple_debug_nonbind_marker_p): New.
1712 (gimple_build_debug_bind): Adjust.
1713 (gimple_build_debug_begin_stmt): New.
1714 * doc/invoke.texi (max-debug-marker-count): New param.
1715 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
1716 (debug_parameter_ref, debug_marker): New.
1717 (NOTE_INSN_BEGIN_STMT): New.
1718 (DEBUG_INSN): Describe begin stmt markers.
1720 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
1721 without debug insns, we wouldn't, but clean up debug insns
1722 after a control flow insn nevertheless.
1724 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
1725 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
1726 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
1727 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
1728 (rtl_tidy_fallthru_edge): Likewise.
1729 (rtl_verify_fallthru): Likewise.
1730 (rtl_verify_bb_layout): Likewise.
1731 (skip_insns_after_block): Likewise.
1732 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
1733 * dwarf2out.c: Include print-rtl.h.
1734 (dwarf2out_next_real_insn): New.
1735 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1736 Dump debug binds in asm comments.
1737 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1738 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1739 callers to use gsi_start_nondebug_bb instead.
1740 (gsi_after_labels): Skip gimple debug stmts.
1741 (gsi_start_nondebug): New.
1742 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1743 (proper_loop_form_for_interchange): Adjust.
1744 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1745 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1746 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1747 (collect_fallthrough_labels): Likewise.
1748 (should_warn_for_implicit_fallthrough): Likewise.
1749 (warn_implicit_fallthrough_r): Likewise.
1750 (expand_FALLTHROUGH_r): Likewise.
1751 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1752 (graphite_copy_stmts_from_block): Skip nonbind markers.
1753 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1754 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1755 * jump.c (clean_barriers): Skip debug insns.
1756 * omp-expand.c (expand_parallel_call): Skip debug insns.
1757 (expand_task_call): Likewise.
1758 (remove_exit_barrier): Likewise.
1759 (expand_omp_taskreg): Likewise.
1760 (expand_omp_for_init_counts): Likewise.
1761 (expand_omp_for_generic): Likewise.
1762 (expand_omp_for_static_nochunk): Likewise.
1763 (expand_omp_for_static_chunk): Likewise.
1764 (expand_omp_simd): Likewise.
1765 (expand_omp_taskloop_for_outer): Likewise.
1766 (expand_omp_taskloop_for_inner): Likewise.
1767 (expand_oacc_for): Likewise.
1768 (expand_omp_sections): Likewise.
1769 (expand_omp_single): Likewise.
1770 (expand_omp_synch): Likewise.
1771 (expand_omp_atomic_load): Likewise.
1772 (expand_omp_atomic_store): Likewise.
1773 (expand_omp_atomic_fetch_op): Likewise.
1774 (expand_omp_atomic_pipeline): Likewise.
1775 (expand_omp_atomic_mutex): Likewise.
1776 (expand_omp_target): Likewise.
1777 (grid_expand_omp_for_loop): Likewise.
1778 (grid_expand_target_grid_body): Likewise.
1779 (build_omp_regions_1): Likewise.
1780 * omp-low.c (check_combined_parallel): Skip debug stmts.
1781 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1782 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1783 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1785 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1786 (make_edges): Likewise.
1787 (cleanup_dead_labels): Likewise.
1788 (gimple_can_merge_blocks_p): Likewise.
1789 (stmt_starts_bb_p): Likewise.
1790 (gimple_block_label): Likewise.
1791 (gimple_redirect_edge_and_branch): Likewise.
1792 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1794 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1796 * tree-pretty-print (print_declaration): Omit initializer in slim
1798 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1800 (eliminate_unnecessary_stmts): Stabilize block removal order.
1801 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1802 * var-tracking.c (get_first_insn): New.
1803 (vt_emit_notes): Call it.
1804 (vt_initialize): Walk any insns before the first BB.
1805 (delete_debug_insns): Likewise.
1807 * gimple.h (enum gimple_debug_subcode): Add
1808 GIMPLE_DEBUG_BEGIN_STMT.
1809 (gimple_debug_begin_stmt_p): New.
1810 (gimple_debug_nonbind_marker_p): New.
1811 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1812 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1813 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1814 * insn-notes.def (BEGIN_STMT): New.
1815 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1816 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1817 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1818 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1819 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1820 (INSN_DEBUG_MARKER_KIND): New.
1821 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1822 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1823 (INSN_VAR_LOCATION_PTR): New.
1824 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1826 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1827 deps only if debug bind insns are enabled.
1828 (pass_expand::execute): Avoid deep TER and expand
1829 debug locations for debug bind insns only.
1830 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1831 debug stmts special handling down to debug bind stmts.
1832 * combine.c (try_combine): Narrow debug insns special handling
1833 down to debug bind insns.
1834 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1835 Narrow debug insns preexisting special handling down to debug
1837 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1838 handling down to debug bind insns.
1839 * function.c (instantiate_virtual_regs): Skip debug markers,
1840 adjust handling of debug binds.
1841 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1842 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1843 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1844 insns to debug bind insns.
1845 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1846 special handling of debug stmts to debug bind stmts.
1847 * ipa-split.c (split_function): Likewise.
1848 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1849 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1851 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1852 * regrename.c (build_def_use): Likewise.
1853 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1854 (pass_cprop_hardreg): Narrow special casing of debug insns to
1856 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1857 * reload1.c (reload): Likewise.
1858 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1859 casing of debug stmts to debug bind stmts.
1860 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1861 * ssa-iterators.h (num_imm_uses): Likewise.
1862 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1863 debug stmts to debug bind stmts.
1864 * tree-inline.c (tree_function_versioning): Narrow special casing
1865 of debug stmts to debug bind stmts.
1866 * tree-loop-distribution.c (generate_loops_for_partition):
1867 Narrow special casing of debug stmts to debug bind stmts.
1868 * tree-sra.c (analyze_access_subtree): Narrow special casing
1869 of debug stmts to debug bind stmts.
1870 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1871 stmts to debug bind stmts.
1872 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1873 casing of debug stmts to debug bind stmts.
1874 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1875 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1876 casing of debug stmts to debug bind stmts.
1877 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1879 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1880 debug stmts to debug bind stmts.
1881 (gimple_replace_ssa_lhs): Likewise.
1882 (insert_debug_temp_for_var_def): Likewise.
1883 (insert_debug_temps_for_defs): Likewise.
1884 (reset_debug_uses): Likewise.
1885 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1886 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1887 (adjust_debug_stmts): Likewise.
1888 (adjust_phi_and_debug_stmts): Likewise.
1889 (vect_do_peeling): Likewise.
1890 * tree-vect-loop.c (vect_transform_loop): Likewise.
1891 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1892 * var-tracking.c (adjust_mems): Narrow special casing of debug
1893 insns to debug bind insns.
1894 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1895 (compute_bb_dataflow, vt_find_locations): Likewise.
1896 (vt_expand_loc, emit_notes_for_changes): Likewise.
1897 (vt_init_cfa_base): Likewise.
1898 (vt_emit_notes): Likewise.
1899 (vt_initialize): Likewise.
1900 (vt_finalize): Likewise.
1902 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1903 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1904 (next_nonnote_nondebug_insn_bb): New.
1905 (prev_nonnote_nondebug_insn_bb): New.
1906 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1907 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1908 (prev_nonnote_nondebug_insn_bb): Declare.
1909 (next_nonnote_nondebug_insn_bb): Declare.
1910 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1911 * cfgrtl.c (get_last_bb_insn): Likewise.
1912 * lra.c (push_insns): Likewise.
1914 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1917 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1919 * selftest.c (selftest::test_runner::test_runner): New ctor.
1920 (selftest::test_runner::~test_runner): New dtor.
1921 * selftest.h (class selftest::test_runner): New class.
1923 2017-12-11 Carl Love <cel@us.ibm.com>
1925 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1926 vec_extract_fp32_from_shortl]): Add #defines.
1927 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1928 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1929 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1930 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1931 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1932 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1933 * doc/extend.texi: Add documentation for the added builtins.
1935 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1937 PR rtl-optimization/80693
1938 PR rtl-optimization/81019
1939 PR rtl-optimization/81020
1940 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1941 are not mentioned in i3. Place the REG_UNUSED note on i2,
1942 possibly modified to REG_DEAD, if it did not originate in i3.
1944 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1946 * recog.c (store_data_bypass_p_1): New function.
1947 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1948 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1950 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1952 PR rtl-optimization/83361
1953 * ifcvt.c (if_convert): Call fixup_partitions.
1955 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1957 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1958 early folding of splat_u{8,16,32}.
1960 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1962 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1963 output_operand_lossage first argument with capital letter.
1964 (aarch64_override_options): Don't start error and sorry first argument
1965 with capital letter.
1967 2017-12-11 Andi Kleen <ak@linux.intel.com>
1969 PR gcov-profile/83355
1970 * auto-profile.c (string_table::get_index_by_decl): Don't
1971 recurse when abstract origin points to itself.
1973 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1975 PR tree-optimization/83320
1976 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1977 (prune_datarefs_not_in_loop): Ditto.
1979 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1981 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1983 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1985 PR tree-optimization/83337
1986 * gimple-loop-interchange.cc (compute_access_stride): Handle
1987 bitfield DRs properly.
1989 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1991 PR tree-optimization/83338
1992 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1993 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1994 vector of unsigned integers to vector of signed integers.
1996 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1998 PR rtl-optimization/83317
1999 * lra-constraints.c (process_address_1): Add insn code check.
2001 2017-12-08 Michael Matz <matz@suse.de>
2003 Fix PR tree-optimization/83323
2004 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2005 head-controlled loops and loop BBs.
2006 * common.opt (funroll-and-jam): Remove, instead ...
2007 (floop-unroll-and-jam): ... reuse this option.
2008 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2009 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2010 (-floop-unroll-and-jam): ... this option.
2012 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2014 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2015 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2016 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2018 2017-12-08 Richard Biener <rguenther@suse.de>
2021 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2022 handle accesses outside of zero-sized vars.
2024 2017-12-08 Martin Jambor <mjambor@suse.cz>
2026 PR tree-optimization/83141
2027 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2028 test for MEM_REFs implicitely changing types with padding. Remove
2030 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2032 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2034 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2036 * config/avr/avr.c (avr_attribute_table): Likewise.
2037 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2038 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2039 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2040 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2041 of struct attribute_spec.
2042 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2044 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2047 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2049 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2052 * profile-count.c (profile_count::from_gcov_type): Move from
2053 profile-count.h; handle overflow.
2054 * profile-count.h (profile_count::from_gcov_type): Move offline.
2056 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2058 PR rtl-optimization/83304
2059 * combine.c (move_deaths): If we do not know where a register died,
2062 2017-12-08 Richard Biener <rguenther@suse.de>
2064 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2065 Provide -fopt-info-loop feedback when we interchange in a nest.
2067 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2069 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2070 for armv6 ARM CPU IDs.
2072 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2074 * common/config/arm/arm-common.c: Include <algorithm>.
2075 (INCLUDE_VECTOR): Define.
2076 (compare_opt_names): New function.
2077 (arm_rewrite_selected_arch): Only strip out extensions that can be
2078 expressed through -mfpu. Sort the remaining extensions
2081 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2083 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2084 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2085 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2086 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2088 2017-06-08 Tristan Gingold <gindold@adacore.com>
2091 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2092 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2094 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2096 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2097 of struct attribute_spec.
2099 2017-12-08 Julia Koval <julia.koval@intel.com>
2101 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2102 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2103 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2104 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2105 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2106 _mm_maskz_dpwssds_epi32): Ditto.
2108 2017-12-08 Richard Biener <rguenther@suse.de>
2110 PR tree-optimization/81303
2111 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2112 conditions try to create a comparison vector type matching
2113 the data vector type.
2114 (vectorizable_condition): Adjust.
2115 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2116 Leave invariant conditions alone in case we can vectorize those.
2118 2017-12-08 Julia Koval <julia.koval@intel.com>
2120 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2121 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2122 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2123 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2124 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2126 2017-12-08 Julia Koval <julia.koval@intel.com>
2128 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2129 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2130 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2131 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2132 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2133 _mm_maskz_dpbusds_epi32): New intrinsics.
2135 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2137 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2139 (ld<bh>io_signed): Likewise.
2140 (st<bhw>io): Likewise.
2141 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2142 SMALL_INT12 constant integer operand.
2144 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2146 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2148 (nios2_large_constant_allowed): ...this. Adjust uses.
2149 (nios2_plus_symbolic_constant_p): Rename to...
2150 (nios2_plus_large_constant_p): ...this. Adjust uses.
2151 (nios2_legitimate_address_p): Correct CONST_INT handling.
2152 (nios2_symbolic_memory_operand_p): Rename to...
2153 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2154 (nios2_large_constant_p): Check for large constant integers too.
2155 (nios2_split_large_constant): Handle constant integers.
2156 (nios2_split_symbolic_memory_operand): Rename to...
2157 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2158 (nios2_legitimize_constant_address): Handle constant integers.
2159 (r0rel_constant_p): Handle small constant integers.
2160 (nios2_print_operand_address): Handle r0-relative integer addresses.
2161 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2162 * config/nios2/nios2.md: Adjust for renamed functions.
2164 2017-12-07 Andrew Waterman <andrew@sifive.com>
2166 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2167 (TARGET_HAVE_SRODATA_SECTION): New define.
2168 (riscv_select_section): New function.
2170 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2171 Alexander Monakov <amonakov@ispras.ru>
2172 Jakub Jelinek <jakub@redhat.com>
2175 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2177 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2179 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2180 a bool success value. Don't call output_operand_lossage here.
2181 (aarch64_print_ldpstp_address): Return a bool success value.
2182 (aarch64_print_operand_address): Call output_addr_const if
2183 aarch64_print_address_internal fails.
2184 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2185 'y'; call output_operand_lossage instead. Call output_operand_lossage
2186 if aarch64_print_ldpstp_address fails.
2188 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2190 * tree-vector-builder.h
2191 (tree_vector_builder::binary_encoded_nelts): Declare.
2192 * tree-vector-builder.c
2193 (tree_vector_builder::binary_encoded_nelts): New function.
2194 * fold-const.c (negate_expr_p): Likewise.
2195 (operand_equal_p, fold_checksum_tree): Likewise.
2196 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2197 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2198 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2199 (uniform_vector_p): Likewise.
2200 * varasm.c (const_hash_1, compare_constant): Likewise.
2201 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2202 (valid_lattice_transition): Operate directly on the VECTOR_CST
2204 * ipa-icf.c: Include tree-vector-builder.h.
2205 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2206 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2208 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2210 * tree.c (build_vector): Delete.
2211 * tree.h (build_vector): Make static and move into the self-testing
2214 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2216 * vector-builder.h (vector_builder::derived): New const overload.
2217 (vector_builder::elt): New function.
2218 * tree-vector-builder.h (tree_vector_builder::type): New function.
2219 (tree_vector_builder::apply_step): Declare.
2220 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2222 * gimple-fold.h (tree_vector_builder): Declare.
2223 (gimple_build_vector): Take a tree_vector_builder instead of a
2224 type and vector of elements.
2225 * gimple-fold.c (gimple_build_vector): Likewise.
2226 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2228 (get_initial_defs_for_reduction): Likewise.
2229 (vectorizable_induction): Likewise.
2231 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2233 * tree-vector-builder.h
2234 (tree_vector_builder::new_binary_operation): Declare.
2235 * tree-vector-builder.c
2236 (tree_vector_builder::new_binary_operation): New function.
2237 * fold-const.c (fold_relational_const): Use it.
2238 (const_binop): Likewise. Check that both input vectors have
2239 the same number of elements, thus excluding things like WIDEN_SUM.
2240 Check whether it is possible to operate directly on the encodings
2243 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2245 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2246 new_unary_operation, operating only on the encoded elements.
2247 (const_unop): Likewise.
2248 (exact_inverse): Likewise.
2249 (distributes_over_addition_p): New function.
2250 (const_binop): Use tree_vector_builder and new_unary_operation
2251 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2252 on the encoded elements unless the encoding is strided and the
2253 operation does not distribute over addition.
2254 (fold_convert_const): Use tree_vector_builder and
2255 new_unary_operation. Operate only on the encoded elements
2256 for truncating integer conversions, or for non-stepped encodings.
2258 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2260 * config/sparc/sparc.c: Include tree-vector-builder.h.
2261 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2262 * expmed.c: Include tree-vector-builder.h.
2263 (make_tree): Use tree_vector_builder instead of build_vector.
2264 * fold-const.c: Include tree-vector-builder.h.
2265 (const_binop): Use tree_vector_builder instead of build_vector.
2266 (const_unop): Likewise.
2267 (native_interpret_vector): Likewise.
2268 (fold_vec_perm): Likewise.
2269 (fold_ternary_loc): Likewise.
2270 * gimple-fold.c: Include tree-vector-builder.h.
2271 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2273 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2274 (simplify_vector_constructor): Use tree_vector_builder instead
2276 * tree-vect-generic.c: Include tree-vector-builder.h.
2277 (add_rshift): Use tree_vector_builder instead of build_vector.
2278 (expand_vector_divmod): Likewise.
2279 (optimize_vector_constructor): Likewise.
2280 * tree-vect-loop.c: Include tree-vector-builder.h.
2281 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2282 of build_vector. Explicitly use a stepped encoding for
2284 * tree-vect-slp.c: Include tree-vector-builder.h.
2285 (vect_get_constant_vectors): Use tree_vector_builder instead
2287 (vect_transform_slp_perm_load): Likewise.
2288 (vect_schedule_slp_instance): Likewise.
2289 * tree-vect-stmts.c: Include tree-vector-builder.h.
2290 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2291 (vect_gen_perm_mask_any): Likewise.
2292 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2293 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2295 (build_vector_from_val): Likewise. Explicitly use a duplicate
2298 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2300 * doc/generic.texi (VECTOR_CST): Describe new representation of
2302 * vector-builder.h: New file.
2303 * tree-vector-builder.h: Likewise.
2304 * tree-vector-builder.c: Likewise.
2305 * Makefile.in (OBJS): Add tree-vector-builder.o.
2306 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2307 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2308 (tree_vector): Change the number of elements to
2309 vector_cst_encoded_nelts.
2310 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2311 (VECTOR_CST_ELTS): Delete.
2312 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2313 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2314 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2315 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2316 (VECTOR_CST_ENCODED_ELT): Likewise.
2317 (vector_cst_encoded_nelts): New function.
2318 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2319 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2320 (vector_cst_int_elt, vector_cst_elt): Declare.
2321 * tree.c: Include tree-vector-builder.h.
2322 (tree_code_size): Abort if passed VECTOR_CST.
2323 (tree_size): Update for new VECTOR_CST layout.
2324 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2325 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2326 (build_vector): Use tree_vector_builder.
2327 (vector_cst_int_elt, vector_cst_elt): New functions.
2328 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2329 encoded elements and then create the vector in the canonical form.
2330 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2331 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2332 (tree_c_tests): Call test_vector_cst_patterns.
2333 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2335 (hash_tree): Likewise.
2336 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2337 (streamer_write_tree_header): Likewise.
2338 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2339 (streamer_alloc_tree): Likewise. Update call to make_vector.
2340 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2342 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2344 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2345 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2346 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2348 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2349 Richard Biener <rguenther@suse.de>
2351 PR tree-optimization/81303
2352 * Makefile.in (gimple-loop-interchange.o): New object file.
2353 * common.opt (floop-interchange): Reuse the option from graphite.
2354 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2355 -floop-interchange and mention it for -O3.
2356 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2357 * gimple-loop-interchange.cc: New file.
2358 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2359 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2360 * passes.def (pass_linterchange): New pass.
2361 * timevar.def (TV_LINTERCHANGE): New time var.
2362 * tree-pass.h (make_pass_linterchange): New declaration.
2363 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2364 interchange. Record IV before/after increment in new parameters.
2365 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2366 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2368 (check_reduction_path): ...New function here.
2369 * tree-vectorizer.h (check_reduction_path): New declaration.
2371 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2374 PR rtl-optimization/80818
2375 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2376 always early clobbered.
2377 * lra-lives.c (process_bb_lives): Check input hard regs for early
2378 clobbered non-operand hard reg.
2380 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2383 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2384 types_compatible_p, just that TYPE_MODE is the same.
2386 2017-12-07 Martin Sebor <msebor@redhat.com>
2389 * attribs.c (empty_attribute_table): Initialize new member of
2390 struct attribute_spec.
2391 (decl_attributes): Add argument. Handle mutually exclusive
2392 combinations of attributes.
2393 (selftests::test_attribute_exclusions): New function.
2394 (selftests::attribute_c_tests): Ditto.
2395 * attribs.h (decl_attributes): Add default argument.
2396 * selftest.h (attribute_c_tests): Declare.
2397 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2398 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2400 * doc/extend.texi (Common Function Attributes): Update const and pure.
2401 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2402 of struct attribute_spec.
2403 * config/arc/arc.c (arc_attribute_table): Same.
2404 * config/arm/arm.c (arm_attribute_table): Same.
2405 * config/avr/avr.c ( avr_attribute_table): Same.
2406 * config/bfin/bfin.c (bfin_attribute_table): Same.
2407 * config/cr16/cr16.c (cr16_attribute_table): Same.
2408 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2409 * config/h8300/h8300.c (h8300_attribute_table): Same.
2410 * config/i386/i386.c (ix86_attribute_table): Same.
2411 * config/ia64/ia64.c (ia64_attribute_table): Same.
2412 * config/m32c/m32c.c (m32c_attribute_table): Same.
2413 * config/m32r/m32r.c (m32r_attribute_table): Same.
2414 * config/m68k/m68k.c (m68k_attribute_table): Same.
2415 * config/mcore/mcore.c (mcore_attribute_table): Same.
2416 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2417 * config/mips/mips.c (mips_attribute_table): Same.
2418 * config/msp430/msp430.c (msp430_attribute_table): Same.
2419 * config/nds32/nds32.c (nds32_attribute_table): Same.
2420 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2421 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2422 * config/rl78/rl78.c (rl78__attribute_table): Same.
2423 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2424 * onfig/rx/rx.c (rx_attribute_table): Same.
2425 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2426 * config/sh/sh.c (sh_attribute_table): Same.
2427 * config/sparc/sparc.c (sparc_attribute_table): Same.
2428 * config/spu/spu.c (spu_attribute_table): Same.
2429 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2430 * config/v850/v850.c (v850_attribute_table): Same.
2431 * config/visium/visium.c (visium_attribute_table): Same.
2433 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2436 * config/arm/arm.c (INCLUDE_STRING): Define.
2437 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2438 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2441 2017-12-07 Michael Matz <matz@suse.de>
2443 Add unroll and jam pass
2445 * gimple-loop-jam.c: New file.
2446 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2447 * common.opt (funroll-and-jam): New option.
2448 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2449 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2450 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2451 * passes.def: Add pass_loop_jam.
2452 * timevar.def (TV_LOOP_JAM): Add.
2453 * tree-pass.h (make_pass_loop_jam): Declare.
2454 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2455 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2456 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2457 to flow_loop_tree_node_add.
2458 (duplicate_subloops, copy_loops_to): Append to sibling list.
2459 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2460 * doc/invoke.texi (-funroll-and-jam): Document new option.
2461 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2463 2017-12-07 Richard Biener <rguenther@suse.de>
2465 PR tree-optimization/83296
2466 PR tree-optimization/67769
2467 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2468 flow sensitive info in an unrelated BB.
2469 (value_replacement): Use reset_flow_sensitive_info.
2470 (minmax_replacement): Reset flow sensitive info on the def
2471 we move. Do not reset flow sensitive info in the whole BB
2472 we move the stmt to.
2473 (abs_replacement): Likewise.
2475 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2478 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2479 rs6000_cpu to the given -mcpu=, or to the default processor.
2481 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2483 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2484 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2485 instead of rs6000_cpu_attr.
2486 (rs6000_adjust_cost): Ditto.
2487 (is_microcoded_insn): Ditto.
2488 (rs6000_adjust_priority): Ditto.
2489 (rs6000_issue_rate): Ditto.
2490 (rs6000_use_sched_lookahead): Ditto.
2491 (rs6000_use_sched_lookahead_guard): Ditto.
2492 (rs6000_sched_reorder): Ditto.
2493 (force_new_group): Ditto.
2494 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2495 (group_ending_nop): Ditto.
2497 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2499 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2500 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2501 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2502 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2504 (rs6000_reassociation_width): Ditto.
2505 (rs6000_emit_epilogue): Ditto.
2506 (rs6000_adjust_cost): Ditto.
2507 (is_microcoded_insn): Ditto.
2508 (is_cracked_insn): Ditto.
2509 (rs6000_adjust_priority): Ditto.
2510 (rs6000_sched_reorder): Ditto.
2511 (rs6000_sched_reorder2): Ditto.
2512 (insn_must_be_first_in_group): Ditto.
2513 (insn_must_be_last_in_group): Ditto.
2514 (rs6000_register_move_cost): Ditto.
2515 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2518 2017-12-07 Julia Koval <julia.koval@intel.com>
2520 * config.gcc: Add vaesintrin.h.
2521 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2523 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2524 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2526 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2527 * config/i386/immintrin.h: Include vaesintrin.h.
2528 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2529 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2530 _mm_aesdec_epi128): New intrinsics.
2532 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2534 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2535 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2536 c-family/c-spellcheck.cc.
2537 (best_macro_match::best_macro_match): Likewise.
2539 (struct edit_distance_traits<cpp_hashnode *>): Move to
2540 c-family/c-spellcheck.h.
2541 (class best_macro_match): Likewise.
2543 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2545 PR tree-optimization/83293
2546 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2547 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2548 might insert into empty bb.
2551 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2552 simplify for plus with :c added, and pointer_plus without that.
2553 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2554 with undefined overflow and the conversion is not widening,
2555 perform negation in utype and only convert to type afterwards.
2556 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2557 simplify for plus with :c added, and pointer_plus without that.
2558 If type is integral with undefined overflow and the conversion is
2559 not widening, perform minus in utype and only convert to type
2560 afterwards. Move the last pointer_diff_expr simplify into the
2563 2017-12-06 Martin Sebor <msebor@redhat.com>
2565 PR tree-optimization/82646
2566 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2567 strncpy, not maxlen.
2569 2017-12-06 Martin Sebor <msebor@redhat.com>
2571 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2574 PR tree-optimization/83075
2575 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2576 strncat/strncpy don't change length of source string.
2578 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2581 2017-11-29 Martin Aberg <maberg@gaisler.com>
2583 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2584 to prevent b2bst errata sequence.
2585 (sqrtdf2_fix): Likewise.
2587 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2589 PR tree-optimization/81945
2590 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2591 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2592 to dest_cfun has orig_loop_num set, either remap it to the new
2593 loop number if the loop got moved too, or clear it.
2595 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2597 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2599 (thunderx2t99_nothing): New insn reservation.
2600 (thunderx2t99_mrs): New insn reservation.
2601 (thunderx2t99_multiple): New insn reservation.
2602 (thunderx2t99_alu_basi): Add bfx to reservation.
2603 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2605 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2608 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2610 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2612 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2613 tree-ssa-pre.c::remove_dead_inserted_code.
2614 * tree-ssa-dce.h: New file.
2615 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2616 (remove_dead_inserted_code): Move and rename to function
2617 tree-ssa-dce.c::simple_dce_from_worklist.
2618 (pass_pre::execute): Update use.
2620 2017-12-05 Richard Biener <rguenther@suse.de>
2622 PR tree-optimization/83277
2623 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2624 to code-gen liveout vars.
2626 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2628 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2629 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2630 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2633 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2635 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2637 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2638 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2641 2017-12-05 Richard Biener <rguenther@suse.de>
2643 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2644 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2645 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2646 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2648 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2650 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2652 2017-12-05 Martin Liska <mliska@suse.cz>
2653 Jakub Jelinek <jakub@redhat.com>
2655 * doc/invoke.texi: Document the options.
2656 * flag-types.h (enum sanitize_code): Add
2657 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2658 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2659 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2660 * opts.c: Define new sanitizer options.
2661 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2662 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2664 2017-12-05 Julia Koval <julia.koval@intel.com>
2666 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2667 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2668 (ix86_handle_option): Handle -mavx512vnni.
2669 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2670 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2671 * config/i386/i386-c (__AVX512VNNI__): New.
2672 * config/i386/i386.c (ix86_target_string): Handle new option.
2673 (ix86_valid_target_attribute_inner_p): Handle new option.
2674 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2675 * config/i386/i386.opt (mavx512vnni): New option.
2677 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2680 * athlon.md: Disable for generic.
2681 * haswell.md: Enable for generic.
2682 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2683 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2685 (ix86_adjust_cost): Move generic to haswell path.
2687 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2689 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2690 instead of MEM_P in a couple more places. Fix formatting issues.
2692 2017-12-04 Jim Wilson <jimw@sifive.com>
2694 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2695 instead of GP_REG_LAST-1.
2696 (riscv_adjust_libcall_cfi_prologue): Likewise.
2697 (riscv_adjust_libcall_cri_epilogue): Likewise.
2698 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2701 2017-12-04 Luis Machado <luis.machado@linaro.org>
2703 * ipa-pure-const.c (check_decl): Add missing newline.
2704 (state_from_flags): Likewise.
2706 2017-12-04 Jeff Law <law@redhat.com>
2708 PR tree-optimizatin/78496
2709 * gimple-ssa-evrp-analyze.h
2710 (evrp_range_analyzer::get_vr_values): Simplify.
2711 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2712 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
2713 and gimple-ssa-evrp-analyze.h.
2714 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
2715 (simplify_stmt_for_jump_threading): Copy a blob of code from
2716 tree-vrp.c to use ranges to simplify statements.
2717 (dom_opt_dom_walker::before_dom_children): Call
2718 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
2719 (dom_opt_dom_walker::after_dom_children): Similarly for
2720 evrp_range_analyzer::leave.
2721 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
2724 * gimple-ssa-evrp-analyze.c
2725 (evrp_range_analyzer::extract_range_from_stmt): Always use
2726 vr_values::update_value_range so preexisting range info is
2727 medged with new range info, even if the new range is VR_VARYING.
2729 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2731 * combine.c: Adjust comment.
2732 (use_crosses_set_p): Delete.
2733 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
2734 (try_combine): Ditto.
2736 2017-12-04 Richard Biener <rguenther@suse.de>
2738 PR tree-optimization/83255
2739 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2740 Re-add zero-iteration check.
2742 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2744 PR rtl-optimization/83245
2745 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2746 hard registers as earlyclobber, also if not in an asm.
2748 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2752 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2755 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2756 rs6000_cpu based on cpu_index, not tune_index.
2758 2017-12-04 Richard Biener <rguenther@suse.de>
2760 PR tree-optimization/83238
2761 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2762 code match comment, rejecting invalid SESE regions.
2764 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2766 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2767 require base operand is a REG_POINTER prior to reload on targets
2768 with non-equivalent space registers.
2770 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2772 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2773 (update_specialized_profile): Fix updating of counts.
2774 (perhaps_add_new_callers): Likewise.
2776 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2779 * x86-tune.def: Remove obsolette FIXMEs.
2780 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2781 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2782 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2784 (X86_TUNE_PAD_RETURNS): Disable for generic.
2786 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2788 PR tree-optimization/83170
2789 PR tree-optimization/83241
2790 * gimple-ssa-store-merging.c
2791 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2792 gimple_vuse (ins_stmt) in case it has changed.
2793 (imm_store_chain_info::output_merged_store): Likewise.
2795 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2799 * tree-cfg.c (pass_warn_function_return::execute): Handle
2800 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2801 with BUILTINS_LOCATION.
2805 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2806 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2807 instead of DECL_MODE.
2809 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2810 last argument to rtx pointer.
2811 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2812 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2813 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2814 depending on the chosen ISAs.
2815 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2816 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2817 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2819 * config/i386/sse.md (mov<mode>_internal): Likewise.
2820 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2822 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2824 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2826 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2828 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2829 parameters from prototype.
2830 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2831 parameters. Don't print a comment.
2832 (emit_fusion_gpr_load): Adjust.
2833 (emit_fusion_load_store): Adjust.
2834 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2835 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2836 comment on the second line.
2838 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2842 rs6000_cpu based on cpu_index, not tune_index.
2844 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2846 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2847 which_alternative instead of which_alternative + 1.
2848 (output_asm_insn): Print an extra tab if the template is short.
2850 2017-12-01 Jim Wilson <jimw@sifive.com>
2852 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2854 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2855 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2857 * doc/tm.texi: Regenerate.
2859 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2862 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2863 whether we can allocate pseudos before trying to fix an address.
2864 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2865 memory address is indexed or indirect.
2866 (floatuns_<mode>si2_hw2): Likewise.
2868 2017-12-01 Jason Merrill <jason@redhat.com>
2870 * Makefile.in (TAGS): Add c-family/*.cc.
2872 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2874 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2875 (call_value_insn): Likewise.
2876 (sibcall_insn): Likewise.
2877 (sibcall_value_insn): Likewise.
2878 (movsi_aarch64): Likewise.
2879 (movdi_aarch64): Likewise.
2880 (add_losym_): Likewise.
2881 (ldr_got_small_): Likewise.
2882 (ldr_got_small_sidi): Likewise.
2883 (ldr_got_small_28k_): Likewise.
2884 (ldr_got_small_28k_sidi): Likewise.
2885 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2886 Move output_addr_const to symbolic case. Add error check.
2888 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2890 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2891 * config/s390/s390.md: Disable MVC merging peephole if it would
2892 disable operand forwarding.
2893 (new peephole2): Split MVCs if it would turn them into up to 2
2896 2017-12-01 Richard Biener <rguenther@suse.de>
2898 PR tree-optimization/83232
2899 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2900 detection of same access. Instead of breaking the group here
2901 do not consider the duplicate. Add comment explaining real fix.
2903 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2905 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2907 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2909 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2911 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2913 * function.h (struct function): Remove cilk_frame_decl,
2914 is_cilk_function and calls_cilk_spawn fields.
2915 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2917 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2918 cilk_elemental field.
2919 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2920 * target.def: Adjust comment.
2921 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2922 Don't test cilk_elemental.
2924 PR tree-optimization/83233
2925 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2926 bswap_stat name for the struct.
2929 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2930 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2931 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2932 SWITCH_BREAK_LABEL_P set on the label.
2933 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2934 added for default case if it was missing and not all cases covered.
2935 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2936 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2938 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2939 to avoid -Wimplicit-fallthrough warning.
2940 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2941 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2943 PR tree-optimization/83221
2944 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2946 (init_reassoc): Formatting fix.
2949 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2950 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2954 * tree-cfg.c: Include asan.h.
2955 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2957 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2959 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2961 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2963 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2964 vpcmpeqd instruction.
2966 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2968 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2969 registers type for 128bit mode.
2971 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2973 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2975 * gengtype-state.c (read_state_pair): Likewise.
2976 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2977 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2978 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2979 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2980 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2981 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2982 * builtins.c (fold_builtin_3): Likewise.
2983 * graphite-scop-detection.c
2984 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2985 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2988 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2991 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2992 parameter. Return true if ignore_decl_rtl_set_p is true and other
2993 conditions are satisfied.
2994 * asan.h (asan_protect_global): Add new parameter.
2995 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2996 to asan_protect_global calls.
2998 2017-11-30 Jim Wilson <jimw@sifive.com>
3000 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3001 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3002 -mstrict-align, add info on default value. Delete redundant lines for
3003 -mabi. Add missing -mexplicit-relocs docs.
3005 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3007 * config/arc/arc.md (trap): New pattern.
3009 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3011 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3012 instruction to end into a delay slot.
3013 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3014 can be placed into a delay slot against reg_note.
3016 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3018 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3019 labels number of usages.
3021 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3023 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3025 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3027 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3030 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3031 multiplication by power of 2 constant into two shifts + comparison.
3033 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3036 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3038 2017-11-30 Richard Biener <rguenther@suse.de>
3040 PR tree-optimization/83202
3041 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3042 allow_peel argument and guard peeling.
3043 (canonicalize_loop_induction_variables): Likewise.
3044 (canonicalize_induction_variables): Pass false.
3045 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3046 peeling from cunrolli.
3048 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3050 * combine.c (try_combine): Print a message to dump file whenever
3051 I0, I1, or I2 cannot be combined into I3.
3053 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3055 PR rtl-optimization/83156
3056 PR rtl-optimization/82621
3057 * combine.c (try_combine): Don't split an I2 if one of the dests is
3058 set again before I3. Allow unused dests.
3060 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3062 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3064 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3066 PR rtl-optimization/80818
3067 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3068 recursively. Use insn code for clobber.
3069 (lra_set_insn_recog_data): Pass the new arg to
3070 collect_non_operand_hard_regs.
3071 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3073 (lra_update_insn_regno_info): Pass insn to
3074 add_regs_to_insn_regno_info.
3076 2017-11-29 Jim Wilson <jimw@sifive.com>
3077 Andrew Waterman <andrew@sifive.com>
3079 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3080 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3081 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3082 use SINGLE_SHIFT_COST.
3083 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3084 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3086 2017-11-29 Julia Koval <julia.koval@intel.com>
3088 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3089 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3090 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3091 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3092 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3093 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3094 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3095 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3096 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3097 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3098 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3099 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3100 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3101 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3102 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3103 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3104 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3105 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3106 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3107 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3108 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3109 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3110 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3111 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3112 __builtin_ia32_vpshldv_v2di_mask,
3113 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3114 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3115 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3117 2017-11-29 Julia Koval <julia.koval@intel.com>
3119 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3120 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3121 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3122 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3123 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3124 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3125 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3126 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3127 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3128 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3129 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3130 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3131 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3132 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3133 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3134 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3135 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3136 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3137 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3138 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3139 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3141 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3143 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3144 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3147 2017-11-29 Martin Aberg <maberg@gaisler.com>
3149 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3150 to prevent b2bst errata sequence.
3151 (sqrtdf2_fix): Likewise.
3153 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3155 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3156 (div_sqrt_insn_p): New function.
3157 (sparc_do_work_around_errata): Insert NOP instructions to
3158 prevent sequences that could trigger the TN-0013 errata for
3159 certain LEON3 processors.
3160 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3161 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3162 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3163 (in_branch_delay): Prevent div and sqrt in delay slot if
3165 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3167 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3169 * config/sparc/sparc.c (atomic_insn_p): New function.
3170 (sparc_do_work_around_errata): Insert NOP instructions to
3171 prevent sequences that could trigger the TN-0010 errata for
3173 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3174 instruction referable in atomic_insns_p.
3176 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3178 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3179 (atomic_compare_and_swap_leon3_1): Likewise.
3182 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3184 * config/sparc/sparc.c (fpop_insn_p): New function.
3185 (sparc_do_work_around_errata): Insert NOP instructions to
3186 prevent sequences that could trigger the TN-0012 errata for
3188 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3189 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3190 (in_branch_annul_delay): Prevent floating-point instructions
3191 in delay slot of annulled integer branch.
3193 2017-11-29 Richard Biener <rguenther@suse.de>
3195 PR tree-optimization/83202
3196 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3198 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3199 nodes vectorized to the same stmts multiple times.
3200 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3201 (vect_analyze_slp_instance): Adjust.
3202 (scalar_stmts_to_slp_tree_map_t): New typedef.
3203 (vect_schedule_slp_instance): Add a map recording the SLP node
3204 representing the vectorized stmts for a set of scalar stmts.
3205 Avoid code-generating redundancies.
3206 (vect_schedule_slp): Allocate map and pass it down.
3208 2017-11-29 Nathan Sidwell <nathan@acm.org>
3211 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3214 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3216 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3217 cases for printing LDP/STP memory addresses.
3218 (aarch64_print_address_internal): Renamed from
3219 aarch64_print_operand_address, added parameter, add Pmode check.
3220 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3221 (aarch64_print_operand_address): Indirect to
3222 aarch64_print_address_internal.
3223 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3224 'y' operand output specifier.
3226 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3229 * tree.c (build_simple_mem_ref_loc): Handle
3230 get_addr_base_and_unit_offset returning a MEM_REF.
3233 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3236 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3239 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3241 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3243 (layout::move_to_column): Likewise.
3245 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3248 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3249 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3251 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3252 Martin Jambor <mjambor@suse.cz>
3255 * tree.h (expr_type_first_operand_type_p): Declare
3256 * tree.c (expr_type_first_operand_type_p): New function.
3257 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3258 (ipa_value_from_jfunc): Adjust declaration.
3259 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3260 Use it as result type for arithmetics, unless it is NULL in which case
3261 be more conservative.
3262 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3263 ipa_get_jf_pass_through_result.
3264 (propagate_vals_across_pass_through): Likewise.
3265 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3266 is to propagate_vals_across_pass_through.
3267 (propagate_constants_across_call): Pass PARM_TYPE to
3268 propagate_scalar_across_jump_function.
3269 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3270 ipa_value_from_jfunc.
3271 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3272 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3273 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3274 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3275 target_type, pass it to ipa_value_from_jfunc.
3276 (update_indirect_edges_after_inlining): Pass parameter type to
3277 try_make_edge_direct_simple_call.
3279 2017-11-28 Jeff Law <law@redhat.com>
3281 * gimple-ssa-evrp-analyze.c
3282 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3283 refine ranges if scev_initialized_p returns true.
3284 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3286 2017-11-28 Julia Koval <julia.koval@intel.com>
3288 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3289 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3290 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3291 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3292 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3293 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3294 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3295 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3296 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3297 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3298 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3299 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3300 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3301 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3302 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3303 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3304 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3305 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3306 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3307 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3308 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3309 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3311 2017-11-28 Julia Koval <julia.koval@intel.com>
3313 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3314 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3315 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3316 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3317 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3318 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3319 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3320 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3321 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3322 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3323 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3324 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3325 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3326 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3327 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3328 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3329 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3330 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3331 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3332 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3333 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3334 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3335 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3336 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3337 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3338 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3339 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3340 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3341 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3342 __builtin_ia32_vpshld_v2di_mask): New builtins.
3343 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3344 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3346 2017-11-28 Richard Biener <rguenther@suse.de>
3348 PR tree-optimization/80776
3349 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3351 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3353 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3354 If the incoming edge is an effective fallthru because the other
3355 edge only reaches a __builtin_unreachable () then record ranges
3356 derived from the controlling condition in SSA info.
3357 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3358 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3360 2017-11-28 Olivier Hainque <hainque@adacore.com>
3362 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3363 on mingw build hosts.
3365 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3368 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3370 2017-11-28 Richard Biener <rguenther@suse.de>
3373 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3374 copies generated from memcpy use a character array as reference
3377 2017-11-28 Julia Koval <julia.koval@intel.com>
3378 Sebastian Peryt <sebastian.peryt@intel.com>
3380 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3381 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3382 cilk-common.o, cilk.h, cilk-common.c): Remove.
3384 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3385 * builtins.c (is_builtin_name): Remove cilkplus condition.
3386 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3387 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3388 cilk-builtins.def, cilkplus.def): Remove.
3389 * cif-code.def (CILK_SPAWN): Remove.
3390 * cilk-builtins.def: Delete.
3391 * cilk-common.c: Ditto.
3393 * cilkplus.def: Ditto.
3394 * config/darwin.h (fcilkplus): Delete.
3395 * cppbuiltin.c: Ditto.
3396 * doc/extend.texi: Remove cilkplus doc.
3397 * doc/generic.texi: Ditto.
3398 * doc/invoke.texi: Ditto.
3399 * doc/passes.texi: Ditto.
3400 * gcc.c (fcilkplus): Remove.
3401 * gengtype.c (cilk.h): Remove.
3402 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3404 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3406 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3408 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3409 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3410 cilkplus conditions.
3411 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3412 inline_read_section): Ditto.
3413 * ipa-inline-analysis.c (cilk.h): Remove.
3414 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3415 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3416 append_linker_options): Remove condition for fcilkplus.
3417 * lto/lto-lang.c (cilk.h): Remove.
3418 (lto_init): Remove condition for fcilkplus.
3419 * omp-expand.c (expand_cilk_for_call): Delete.
3420 (expand_omp_taskreg, expand_omp_for_static_chunk,
3421 expand_omp_for): Remove cilkplus
3423 (expand_cilk_for): Delete.
3424 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3425 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3426 execute_lower_omp, diagnose_sb_0): Ditto.
3427 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3428 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3429 * tree-nested.c: Ditto.
3430 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3431 (dump_generic_node): Ditto.
3432 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3433 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3435 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3437 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3439 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3440 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3441 for vector mode and !TARGET_SIMD.
3443 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3445 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3447 * tree.h (SWITCH_LABELS): Remove.
3448 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3449 assert SWITCH_BODY is non-NULL.
3450 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3452 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3454 PR tree-optimization/80788
3455 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3456 has TREE_OVERFLOW set, call drop_tree_overflow.
3458 2017-11-28 Richard Biener <rguenther@suse.de>
3460 PR tree-optimization/83158
3461 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3463 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3466 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3467 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3469 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3471 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3472 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3474 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3477 * varasm.c (compare_constant): Take the mode of the constants into
3478 account when comparing floating point constants.
3480 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3482 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3483 from explicit instantiation of debug_helper.
3484 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3486 2017-11-27 Richard Biener <rguenther@suse.de>
3488 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3491 2017-11-27 Richard Biener <rguenther@suse.de>
3493 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3494 (type_hash_canon): Also clear int_cst_hash_table entry for
3496 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3498 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3500 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3501 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3502 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3504 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3506 * hash-map.h (gt_cleare_cache): Avoid UB.
3508 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3510 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3511 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3512 (decide_unroll_runtime_iterations): Likewise.
3513 (decide_unroll_stupid): Likewise.
3515 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3518 * config/i386/i386.md: Add a loop with incssp.
3520 2017-11-27 Martin Jambor <mjambor@suse.cz>
3522 PR tree-optimization/81248
3523 * tree-sra.c (splice_param_accesses): Remove size check.
3524 (decide_one_param_reduction): Fix size check.
3525 * gimple-pretty-print.c (dump_profile): Silence warning.
3526 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3528 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3531 * dbxout.c (lastlineno): New variable.
3532 (dbx_debug_hooks): Use dbxout_switch_text_section as
3533 switch_text_section debug hook.
3534 (dbxout_function_end): Switch to current_function_section
3535 rather than function_section. If crtl->has_bb_partition,
3536 output just one N_FUN, depending on in_cold_section_p.
3537 (dbxout_source_line): Remember last lineno in lastlineno.
3538 (dbxout_switch_text_section): New function.
3539 (dbxout_function_decl): Adjust dbxout_block caller.
3540 (dbx_block_with_cold_children): New function.
3541 (dbxout_block): Return true if any LBRAC/RBRAC have been
3542 emitted. Use dbx_block_with_cold_children at depth == 0
3543 in second partition. Add PARENT_BLOCKNUM argument, pass
3544 it optionally adjusted to children. Output LBRAC/RBRAC
3545 around recursive call only if the block is in the current
3546 partition, if not and anything was output, emit empty
3548 * final.c (final_scan_insn): Compute cold_function_name
3549 before calling switch_text_section debug hook. Call
3550 that hook even if dwarf2out_do_frame if not emitting
3554 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3555 TREE_READONLY decls.
3557 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3559 PR rtl-optimization/82488
3560 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3562 2017-11-26 Julia Koval <julia.koval@intel.com>
3564 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3566 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3569 2017-11-26 Julia Koval <julia.koval@intel.com>
3571 * config/i386/driver-i386.c (host_detect_local_cpu):
3572 Detect skylake-avx512.
3574 2017-11-26 Julia Koval <julia.koval@intel.com>
3576 * config.gcc: Add -march=cannonlake.
3577 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3578 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3579 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3580 (PTA_CANNONLAKE): New.
3581 (processor_target_table): Add cannonlake.
3582 (ix86_option_override_internal): Ditto.
3583 (fold_builtin_cpu): Ditto.
3584 (get_builtin_code_for_version): Handle cannonlake.
3585 (M_INTEL_COREI7_CANNONLAKE): New.
3586 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3587 * doc/invoke.texi: Add -march=cannonlake.
3589 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3591 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3592 (try_init_one_plugin): Alternative implementation for MinGW.
3593 * Makefile.in (plugin_implib): New.
3594 (gengtype-lex.c): Fix broken AIX workaround.
3595 * configure: Regenerate.
3596 * doc/plugins.texi: Document support for MinGW.
3598 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3600 PR rtl-optimization/81553
3601 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3602 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3603 is a shift where C1 has different mode than the whole shift, use C1's
3604 mode for MULT rather than the shift's mode.
3607 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3608 builtins not enabled in the currently selected ISA.
3610 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3612 PR tree-optimization/71026
3613 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3614 (insert_reciprocals): Change to insert reciprocals before a division
3615 by a square and to insert the square of a reciprocal.
3616 (execute_cse_reciprocals_1): Change to consider division by a square.
3617 (register_division_in): Add importance parameter.
3619 2017-11-24 Richard Biener <rguenther@suse.de>
3621 PR tree-optimization/82402
3622 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3623 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3625 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3627 * match.pd (0-ptr): New transformation.
3629 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3632 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3633 when skipping checking.
3635 2017-11-24 Richard Biener <rguenther@suse.de>
3637 PR tree-optimization/83128
3638 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3639 (vn_reference_lookup_3): Likewise.
3641 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3644 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3645 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3648 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3649 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3650 check for the latter with gimple_call_builtin_p. Do not handle
3651 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3653 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3655 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3658 2017-11-23 Julia Koval <julia.koval@intel.com>
3660 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3661 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3662 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3663 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3664 _mm512_maskz_expandloadu_epi16): New intrinsics.
3665 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3666 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3667 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3668 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3669 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3670 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3671 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3672 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3673 _mm256_maskz_expandloadu_epi8): New intrinsics.
3674 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3675 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3676 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3677 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3678 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3679 * config/i386/sse.md (VI248_VLBW): New iterator.
3680 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3682 2017-11-23 Julia Koval <julia.koval@intel.com>
3684 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3685 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3686 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3687 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3688 _mm512_mask_compressstoreu_epi16): New.
3689 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3690 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3691 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3692 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3693 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3694 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3695 _mm256_mask_compressstoreu_epi8): New.
3696 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3697 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3698 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3699 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3700 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3701 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3702 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3703 __builtin_ia32_compresshi128_mask,
3704 __builtin_ia32_compressstoreuqi512_mask,
3705 __builtin_ia32_compressstoreuhi512_mask,
3706 __builtin_ia32_compressstoreuqi256_mask,
3707 __builtin_ia32_compressstoreuqi128_mask,
3708 __builtin_ia32_compressstoreuhi256_mask,
3709 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3710 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3712 (ix86_expand_special_args_builtin): Handle new types.
3713 (s4fma_expand): Handle new builtin array.
3714 * config/i386/immintrin.h: Include new headers.
3715 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
3716 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
3718 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
3721 fixes. Declare temp and g variables at the top in order to avoid
3722 {} in most of the cases.
3724 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
3726 * match.pd (ptr-0): New transformation.
3728 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3730 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
3731 (struct addr_mode_cost_table): New.
3732 (struct tune_params): Add field addr_mode_costs.
3733 * config/arm/arm.c (generic_addr_mode_costs): New.
3734 (arm_slowmul_tune): Initialise addr_mode_costs field.
3735 (arm_fastmul_tune): Likewise.
3736 (arm_strongarm_tune): Likewise.
3737 (arm_xscale_tune): Likewise.
3738 (arm_9e_tune): Likewise.
3739 (arm_marvell_pj4_tune): Likewise.
3740 (arm_v6t2_tune): Likewise.
3741 (arm_cortex_tune): Likewise.
3742 (arm_cortex_a8_tune): Likewise.
3743 (arm_cortex_a7_tune): Likewise.
3744 (arm_cortex_a15_tune): Likewise.
3745 (arm_cortex_a35_tune): Likewise.
3746 (arm_cortex_a53_tune): Likewise.
3747 (arm_cortex_a57_tune): Likewise.
3748 (arm_exynosm1_tune): Likewise.
3749 (arm_xgene1_tune): Likewise.
3750 (arm_cortex_a5_tune): Likewise.
3751 (arm_cortex_a9_tune): Likewise.
3752 (arm_cortex_a12_tune): Likewise.
3753 (arm_cortex_a73_tune): Likewise.
3754 (arm_v7m_tune): Likewise.
3755 (arm_cortex_m7_tune): Likewise.
3756 (arm_v6m_tune): Likewise.
3757 (arm_fa726te_tune): Likewise.
3758 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3760 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3762 * config/arm/arm.c (arm_mem_costs): New function.
3763 (arm_rtx_costs_internal): Use arm_mem_costs.
3765 2017-11-23 Mark Wielaard <mark@klomp.org>
3767 * dwarf2out.c (init_sections_and_labels): Use generation to create
3768 unique ranges_section_label and ranges_base_label. Return generation.
3769 (output_rnglists): Add generation argument. Use generation to create
3770 unique ranges labels.
3771 (dwarf2out_finish): Get generation from init_sections_and_labels
3772 and pass generation to output_rnglists.
3774 2017-11-23 Mike Stump <mikestump@comcast.net>
3775 Eric Botcazou <ebotcazou@adacore.com>
3777 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3778 * cfgloop.h (struct loop): Add unroll field.
3779 * function.h (struct function): Add has_unroll bitfield.
3780 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3781 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3782 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3783 (pass_rtl_unroll_loops::gate): Likewise.
3784 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3785 for which loop->unroll==1.
3786 (decide_unroll_constant_iterations): Use note for consistency and
3787 take loop->unroll into account. Return early if loop->unroll is set.
3788 Fix thinko in existing test.
3789 (decide_unroll_runtime_iterations): Use note for consistency and
3790 take loop->unroll into account.
3791 (decide_unroll_stupid): Likewise.
3792 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3793 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3794 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3796 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3797 (print_loop): Print loop->unroll if set.
3798 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3799 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3800 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3802 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3803 loop->unroll is set and smaller than the trip count. Otherwise bypass
3804 entirely the heuristics if loop->unroll is set. Remove dead note.
3805 Fix off-by-one bug in other note.
3806 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3807 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3809 (tree_unroll_loops_completely): Make static.
3810 (pass_complete_unroll::execute): Use correct type for variable.
3811 (pass_complete_unrolli::execute): Fix formatting.
3812 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3814 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3816 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3817 enable when TARGET_PREFER_AVX128 is set.
3819 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3821 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3823 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3824 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3826 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3828 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3829 frequencies. Fix estimation of aggregate parameters.
3831 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3833 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3836 2017-11-23 Tom de Vries <tom@codesourcery.com>
3838 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3840 2017-11-23 Tom de Vries <tom@codesourcery.com>
3842 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3845 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3848 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3849 sibcall_value_pcrel_fdpic): Use local variable instead of
3851 (calli_tbr_rel): Add missing operand 2.
3852 (call_valuei_tbr_rel): Add missing operand 3.
3854 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3857 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3858 bitpos/bitsize covering the whole destination, use store_expr only if
3859 the complex mode is the same. Otherwise, use expand_normal and if
3860 it returns CONCAT, subreg each part separately instead of trying to
3861 subreg the whole result.
3863 2017-11-23 Richard Biener <rguenther@suse.de>
3865 PR tree-optimization/23094
3866 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3867 come here from walking over backedges in the first iteration.
3868 (vn_reference_lookup_3): Skip clobbers that store the same value.
3870 2017-11-23 Richard Biener <rguenther@suse.de>
3872 PR tree-optimization/81403
3873 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3874 a block we need a leader relative to.
3875 (phi_translate_1): For nary processing require a leader from
3876 get_representative_for given we run expression simplification
3877 using match-and-simplify. Remove previous fix.
3879 2017-11-22 Jeff Law <law@redhat.com>
3881 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3882 Use new method allocate_value_range rather than accessing the
3883 vrp_value_range_pool data member directly.
3884 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3885 to use extract_range_from_stmt method to avoid need for
3886 extract_range_from_assignment method.
3887 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3888 method rather than setting values_propgated data member directly.
3889 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3890 and values propagated data members and extract_range_from_assignment
3891 method. Reorder private data members to conform to standards.
3892 Add new methods set_lattice_propagation_complete and
3893 allocate_value_range.
3895 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3897 PR rtl-optimization/83030
3898 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3899 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3900 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3901 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3902 CROSSING_JUMP_P flag is already set before setting it.
3903 * cfgrtl.c (fixup_partition_crossing): Likewise.
3904 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3907 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3909 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3910 Handle the case where both arguments are using gen_const_vec_series.
3912 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3915 * pretty-print.c (pp_format): Move quoting implementation to
3916 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3917 to pass address of "quote" local.
3918 (pp_begin_quote): New function.
3919 (pp_end_quote): New function.
3920 * pretty-print.h (printer_fn): Convert penultimate param from bool
3922 (pp_begin_quote): New decl.
3923 (pp_end_quote): New decl.
3924 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3925 param from bool to bool *.
3926 * tree-diagnostic.h (default_tree_printer): Likewise.
3928 2017-11-22 Jeff Law <law@redhat.com>
3930 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3931 Initialize vr_values.
3932 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3933 vr_values via vr_values class instance rather than delegators.
3934 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3935 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3936 (evrp_range_analyzer::push_value_range): Likewise.
3937 (evrp_range_analyzer::pop_value_range): Likewise.
3938 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3939 most delegators. Those remaining are exposed as public interfaces.
3940 Make vr_values a pointer and private.
3941 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3943 (evrp_range_analyzer::get_vr_value): New method.
3944 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3945 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3946 (class evrp_dom_walker): Attach evrp_folder class, initialize
3947 it in the ctor. Remove temporary delegators.
3948 (evrp_dom_walker::before_dom_children): Call methods in attached
3949 evrp_range_analyzer class via class instance pointer. Use
3950 free value_range_constant_singleton to remove need for
3951 op_with_constant_singleton_value delegator method. Do not
3952 create a vrp_prop class instance for every call! Narrow
3953 scope of a couple variables.
3954 (evrp_dom_walker::cleanup): Call methods in attached
3955 evrp_range_analyzer class via class instance pointer.
3956 * vr-values.h (class vr_values): Privatize many methods and
3959 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3961 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3962 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3963 * expr.c (expand_expr_real_2): Likewise.
3964 * fold-const.c (const_unop): Likewise.
3965 * optabs-tree.c (optab_for_tree_code): Likewise.
3966 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3967 * tree-inline.c (estimate_operator_cost): Likewise.
3968 * tree-pretty-print.c (dump_generic_node): Likewise.
3969 (op_code_prio): Likewise.
3970 (op_symbol_code): Likewise.
3971 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3972 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3973 * internal-fn.c (direct_internal_fn_optab): New function.
3974 (direct_internal_fn_array, direct_internal_fn_supported_p
3975 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3976 * fold-const-call.c (fold_const_reduction): New function.
3977 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3979 * tree-vect-loop.c: Include internal-fn.h.
3980 (reduction_code_for_scalar_code): Rename to...
3981 (reduction_fn_for_scalar_code): ...this and return an internal
3983 (vect_model_reduction_cost): Take an internal_fn rather than
3985 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3987 (vectorizable_reduction): Use internal functions rather than tree
3988 codes for the reduction operation. Update calls to the functions
3990 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3991 Use calls to internal functions rather than REDUC tree codes.
3992 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3994 2017-11-22 Olivier Hainque <hainque@adacore.com>
3996 * config/vxworks.c (vxworks_override_options): Pick default
3997 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3998 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3999 DWARF_GNAT_ENCODINGS_DEFAULT.
4000 * config/vxworksae.h: Likewise.
4002 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4004 PR tree-optimization/83104
4005 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4008 2017-11-22 Marek Polacek <polacek@redhat.com>
4009 H.J. Lu <hongjiu.lu@intel.com>
4010 Jason Merrill <jason@redhat.com>
4015 * calls.c (initialize_argument_information): Call
4016 warn_parameter_passing_abi target hook.
4017 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4018 argument onto stack.
4019 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4020 * common.opt: Update -fabi-version description.
4021 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4022 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4024 (ix86_is_empty_record): New function.
4025 (ix86_warn_parameter_passing_abi): New function.
4026 (TARGET_EMPTY_RECORD_P): Redefine.
4027 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4028 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4029 * doc/tm.texi: Regenerated.
4030 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4031 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4032 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4033 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4034 instead of int_size_in_bytes.
4035 * expr.c (copy_blkmode_to_reg): Likewise.
4036 * function.c (aggregate_value_p): Return 0 for empty types.
4037 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4039 (locate_and_pad_parm): Call arg size_in_bytes instead
4041 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4042 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4043 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4045 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4046 (std_gimplify_va_arg_expr): Skip empty records. Call
4047 arg_size_in_bytes instead size_in_bytes.
4048 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4049 * tree-core.h (tree_type_common): Add empty_flag.
4050 (tree_decl_common): Update comments.
4051 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4053 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4054 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4056 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4057 * tree.c (default_is_empty_type): New function.
4058 (default_is_empty_record): New function.
4059 (arg_int_size_in_bytes): New function.
4060 (arg_size_in_bytes): New function.
4061 (get_ultimate_context): New function.
4062 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4063 TRANSLATION_UNIT_WARN_EMPTY_P.
4064 (default_is_empty_record, arg_int_size_in_bytes,
4065 arg_size_in_bytes, get_ultimate_context): Declare.
4067 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4069 * config/arm/arm.c (cmse_clear_registers): New function.
4070 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4071 code by call to cmse_clear_registers.
4072 (cmse_nonsecure_entry_clear_before_return): Likewise.
4074 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4076 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4077 (vdot_s32, vdotq_s32): New.
4078 (vdot_lane_u32, vdotq_lane_u32): New.
4079 (vdot_lane_s32, vdotq_lane_s32): New.
4082 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4085 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4086 for unsigned values with fewer HWIs than the precision.
4087 (test_overflow): New function.
4088 (wide_int_cc_tests): Call it.
4090 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4092 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4093 has the same mode class as Pmode.
4095 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4097 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4098 padding_bits_to_clear_ptr.
4099 (cmse_nonsecure_entry_clear_before_return): Likewise.
4101 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4103 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4104 auto_sbitap instead of integer bitfield to control register needing
4107 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4109 PR tree-optimization/83044
4110 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4111 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4112 that need the upper bound. Subtract offset from
4113 get_addr_base_and_unit_offset only if positive and subtract it
4114 before division by eltsize rather than after it.
4117 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4118 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4119 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4123 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4124 expand_binop with *mul_widen_optab, make sure at least one of the
4125 operands doesn't have VOIDmode.
4128 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4130 PR rtl-optimization/82044
4131 PR tree-optimization/82042
4132 * dse.c (record_store): Check for overflow.
4133 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4134 add_wild_read instead of clear_rhs_from_active_local_stores on
4135 overflow and log it into dump_file.
4137 2017-11-22 Richard Biener <rguenther@suse.de>
4139 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4140 fake edges to exit when looking for a place to insert.
4141 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4143 (insert_into_preds_of_block): Commit edge insertion immediately,
4144 assert that doesn't require new BBs.
4145 (fini_pre): Release expressions.
4146 (pass_pre::execute): Shuffle things around a bit, if the fn
4147 is too large do not compute AVAIL either as this is really the
4150 2017-11-22 Richard Biener <rguenther@suse.de>
4152 PR tree-optimization/83089
4153 * tree-if-conv.c (pass_if_conversion::execute): If anything
4154 changed reset SCEV and free the number of iteration estimates.
4156 2017-11-21 Martin Sebor <msebor@redhat.com>
4158 PR tree-optimization/82945
4159 * calls.h (warn_nonstring_bound): Remove unused function.
4161 2017-11-21 Martin Sebor <msebor@redhat.com>
4163 PR tree-optimization/82945
4164 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4165 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4166 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4168 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4169 * calls.h (get_attr_nonstring_decl): Declare new function.
4170 * doc/extend.texi (attribute nonstring): Update.
4171 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4172 get_attr_nonstring_decl and handle it.
4173 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4174 detection of nul-termination.
4175 (strlen_to_stridx): Change to a pointer.
4176 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4177 (pass_strlen::execute): Same.
4179 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4181 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4182 for the new option -mprefer-vector-width=[none|128|256|512].
4183 * config/i386/i386.c (ix86_target_string): remove old style options
4184 -mprefer-avx256 and make -mprefer-avx128 as alias.
4185 (ix86_option_override_internal): Apply defaults for the
4186 -mprefer-vector-width=[128|256] option.
4187 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4188 Implement macros to work with -mprefer-vector-width=.
4189 * config/i386/i386.opt: Implemented option
4190 -mprefer-vector-width=[none|128|256|512].
4191 * doc/invoke.texi: Documentation for
4192 -mprefer-vector-width=[none|128|256|512].
4194 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4196 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4198 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4200 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4202 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4204 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4206 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4207 * expr.c (expand_expr_real_2): Likewise.
4208 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4209 fold_binary_loc): Likewise.
4210 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4211 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4212 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4213 MINUS_EXPR transformations.
4214 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4215 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4216 * tree-inline.c (estimate_operator_cost): Likewise.
4217 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4218 op_symbol_code): Likewise.
4219 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4220 * vr-values.c (extract_range_from_binary_expr): Likewise.
4221 * varasm.c (initializer_constant_valid_p_1): Likewise.
4222 * tree.def: New tree code POINTER_DIFF_EXPR.
4224 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4226 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4227 integer suffix to movbe mnemonic.
4228 (*bswaphi2_movbe): Ditto.
4229 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4231 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4234 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4235 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4236 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4238 2017-11-21 Martin Liska <mliska@suse.cz>
4240 * tree-inline.c (expand_call_inline): Remove not needed
4243 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4244 Jakub Jelinek <jakub@redhat.com>
4247 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4248 Remove static keyword from f variable.
4250 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4252 PR tree-optimization/83086
4253 * gimple-ssa-store-merging.c
4254 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4255 rather than n.base_addr.
4257 2017-11-21 Martin Liska <mliska@suse.cz>
4259 PR rtl-optimization/82044
4260 PR tree-optimization/82042
4261 * dse.c (check_mem_read_rtx): Check for overflow.
4263 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4265 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4268 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4270 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4271 and its corresponding call arg location note.
4273 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4274 Andrew Burgess <andrew.burgess@embecosm.com>
4276 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4278 (arc_return_slot_offset): Likewise.
4279 (arc_eh_return_address_location): New declaration.
4280 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4281 (MUST_SAVE_REGISTER): Add exception handler case.
4282 (MUST_SAVE_RETURN_ADDR): Likewise.
4283 (arc_frame_pointer_required): Likewise.
4284 (arc_frame_pointer_needed): New function.
4285 (arc_compute_frame_size): Changed.
4286 (arc_expand_prologue): Likewise.
4287 (arc_expand_epilogue): Likewise.
4288 (arc_initial_elimination_offset): Likewise.
4289 (arc_return_slot_offset): Delete.
4290 (arc_eh_return_address_location): New function.
4291 (arc_builtin_setjmp_frame_value): Likewise.
4292 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4293 (EH_RETURN_STACKADJ_RTX): Define.
4294 (EH_RETURN_HANDLER_RTX): Likewise.
4295 * config/arc/arc.md (eh_return): Delete.
4297 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4299 * print-tree.h (debug_vec_tree): Remove prototype.
4300 * gdbinit.in (pvt): Remove macro.
4302 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4304 PR tree-optimization/83047
4305 * gimple-ssa-store-merging.c
4306 (imm_store_chain_info::output_merged_store): If the loads with the
4307 same vuse are in different basic blocks, for load_gsi pick a load
4308 location that is dominated by the other loads.
4311 * config/i386/i386.c (ix86_memmodel_check): Start
4312 -Winvalid-memory-model diagnostics with lowercase letter.
4315 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4316 set in_cold_section_p to the partition containing loc_list->first.
4317 When seeing loc_list->last_before_switch node, update secname and
4318 perform range_across_switch second partition handling only after that.
4321 * run-rtl-passes.c: Include debug.h.
4322 (run_rtl_passes): Call debug_hooks->assembly_start.
4323 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4327 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4328 OPTAB_DIRECT in calls to expand_simple_binop.
4330 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4333 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4335 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4338 * spellcheck.h (best_match::blithely_get_best_candidate): New
4341 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4343 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4344 with lower case letter.
4346 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4348 * config/i386/i386.md (bswaphi2): New expander.
4349 (*bswaphi2_movbe): New insn pattern.
4350 (bswaphi -> rorhi pepehole2): New peephole pattern.
4352 2017-11-20 Jeff Law <law@redhat.com>
4354 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4355 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4356 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4357 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4358 gimple-ssa-evrp-analyze.h.
4360 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4361 set BB_VISITED here.
4362 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4364 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4366 * tree-predcom.c: Add general comment on Store-Store chains.
4367 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4369 (get_chain_last_ref_at): Rename into...
4370 (get_chain_last_write_at): ...this.
4371 (get_chain_last_write_before_load): New function.
4372 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4373 CT_STORE_STORE when write reference is added.
4374 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4375 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4376 (initialize_root_vars_store_elim_1): Ditto.
4377 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4378 definition is created.
4379 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4380 chain by replacing it with dominant stored value.
4382 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4384 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4386 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4388 * vr-values.c (extract_range_from_binary_expr): Use a full range
4391 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4393 * config/arm/arm.md (R4_REGNUM): Define constant.
4394 (nonsecure_call_internal): Remove r4 clobber.
4395 (nonsecure_call_value_internal): Likewise.
4396 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4397 clobber and resequence match_operands.
4398 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4399 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4400 (nonsecure_call_value_reg_thumb2): Likewise.
4402 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4404 PR tree-optimization/78821
4405 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4406 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4407 to the address of the base rather than the base itself.
4408 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4409 (find_bswap_or_nop_finalize): New function.
4410 (find_bswap_or_nop): Use it.
4411 (bswap_replace): Return a tree rather than bool, change first
4412 argument from gimple * to gimple_stmt_iterator, allow inserting
4413 into an empty sequence, allow ins_stmt to be NULL - then emit
4414 all stmts into gsi. Fix up MEM_REF address gimplification.
4415 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4416 (struct store_immediate_info): Add N and INS_STMT non-static
4418 (store_immediate_info::store_immediate_info): Initialize them
4419 from newly added ctor args.
4420 (merged_store_group::apply_stores): Formatting fixes. Sort by
4422 (stmts_may_clobber_ref_p): For stores call also
4423 refs_anti_dependent_p.
4424 (gather_bswap_load_refs): New function.
4425 (imm_store_chain_info::try_coalesce_bswap): New method.
4426 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4427 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4428 (imm_store_chain_info::output_merged_store): Fail if number of
4429 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4430 and NOP_EXPR rhs_code.
4431 (pass_store_merging::process_store): Compute n and ins_stmt, if
4432 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4433 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4435 (pass_store_merging::execute): Calculate dominators.
4437 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4438 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4439 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4440 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4441 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4442 class pass_optimize_bswap, bswap_replace,
4443 pass_optimize_bswap::execute): Moved to ...
4444 * gimple-ssa-store-merging.c: ... this file.
4445 Include optabs-tree.h.
4446 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4447 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4448 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4449 anonymous namespace, remove static keywords.
4450 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4451 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4453 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4456 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4458 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4460 * vec.h (debug_helper): New function.
4461 (DEFINE_DEBUG_VEC): New macro.
4462 * hash-set.h (debug_helper): New function.
4463 (DEFINE_DEBUG_HASH_SET): New macro.
4464 * cfg.c (debug_slim (edge)): New function.
4465 Call DEFINE_DEBUG_VEC for edges.
4466 Call DEFINE_DEBUG_HASH_SET for edges.
4467 * cfghooks.c (debug_slim (basic_block)): New function.
4468 Call DEFINE_DEBUG_VEC for basic blocks.
4469 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4470 * print-tree.c (debug_slim): New function to handle trees.
4471 Call DEFINE_DEBUG_VEC for trees.
4472 Call DEFINE_DEBUG_HASH_SET for trees.
4473 (debug (vec<tree, va_gc>) &): Remove.
4474 (debug (<vec<tree, va_gc>) *): Remove.
4475 * print-rtl.c (debug_slim): New function to handle const_rtx.
4476 Call DEFINE_DEBUG_VEC for rtx_def.
4477 Call DEFINE_DEBUG_VEC for rtx_insn.
4478 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4479 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4480 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4481 (debug (vec<rtx_insn *> *ptr): Remove.
4482 (debug_insn_vector): Remove.
4483 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4485 2017-11-20 Tom de Vries <tom@codesourcery.com>
4487 PR rtl-optimization/82020
4488 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4489 IF_THEN_ELSE condition.
4491 2017-11-19 Jeff Law <law@redhat.com>
4493 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4494 of degenerates resulting from ignoring an edge.
4496 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4499 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4501 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4504 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4505 for uninitialied counts.
4507 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4510 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4512 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4515 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4517 2017-11-19 Tom de Vries <tom@codesourcery.com>
4519 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4522 2017-11-19 Tom de Vries <tom@codesourcery.com>
4524 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4527 2017-11-19 Tom de Vries <tom@codesourcery.com>
4529 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4532 2017-11-19 Tom de Vries <tom@codesourcery.com>
4534 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4536 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4538 2017-11-19 Tom de Vries <tom@codesourcery.com>
4540 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4542 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4543 semicolon after MCORE_EXPORT_NAME call.
4545 2017-11-19 Tom de Vries <tom@codesourcery.com>
4548 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4549 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4551 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4553 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4554 (cgraph_edge::clone): Cleanup updating of profile.
4555 * ipa-cp.c (update_profiling_info): Likewise.
4556 * ipa-inline-transform.c (inline_transform): Likewise.
4557 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4558 * ipa-split.c (execute_split_functions): Do not split when function
4560 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4561 * profile-count.c (profile_count::dump): Add global0.
4562 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4564 (profile_count::to_sreal_scale): Likewise.
4565 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4566 (profile_count::combine_with_ipa_count): New function.
4567 * profile-count.h (profile_guessed_global0adjusted): New.
4568 (profile_count::adjusted_zero): New.
4569 (profile_count::global0adjusted): New.
4570 (profile_count::combine_with_ipa_count): New.
4571 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4572 correct profile of return block of split functions.
4573 (copy_cfg_body): Remove unused profile_count.
4574 (copy_body): Likewise.
4575 (expand_call_inline): Update.
4576 (tree_function_versioning): Update.
4578 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4580 * hash-set.h (hash_set::empty): New.
4581 * tree-ssa-threadbackward.h: Delete.
4582 * tree-ssa-threadbackward.c (class thread_jumps): New.
4583 Move max_threaded_paths into class.
4584 (fsm_find_thread_path): Remove arguments that are now in class.
4585 (profitable_jump_thread_path): Rename to...
4586 (thread_jumps::profitable_jump_thread_path): ...this.
4587 (convert_and_register_jump_thread_path): Rename to...
4588 (thread_jumps::convert_and_register_current_path): ...this.
4589 (check_subpath_and_update_thread_path): Rename to...
4590 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4591 (register_jump_thread_path_if_profitable): Rename to...
4592 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4593 (handle_phi): Rename to...
4594 (thread_jumps::handle_phi): ...this.
4595 (handle_assignment): Rename to...
4596 (thread_jumps::handle_assignment): ...this.
4597 (fsm_find_control_statement_thread_paths): Rename to...
4598 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4599 (find_jump_threads_backwards): Rename to...
4600 (thread_jumps::find_jump_threads_backwards): ...this.
4601 Initialize path local data.
4602 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4603 from within thread_jumps class.
4604 (pass_early_thread_jumps::execute): Same.
4606 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4608 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4610 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4612 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4613 to canonical location.
4615 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4618 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4620 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4622 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4624 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4625 rcrt1.o%s/grcrt1.o%s for -static-pie.
4627 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4629 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4630 ix86_shift_rotate_cost): Break out from ...
4631 (ix86_rtx_costs): ... here.
4632 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4635 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4637 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4638 when entry block was promoted unlikely.
4639 (estimate_bb_frequencies): Increase frequency scale.
4640 * profile-count.h (profile_count): Export precision info.
4642 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4644 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4645 disturbin profile of entry block.
4647 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4649 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4652 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4654 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4655 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4657 2017-11-17 Jeff Law <law@redhat.com>
4659 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4660 from evrp_dom_walker class. Various methods moved into new class.
4661 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4662 (evrp_range_analyzer::enter): New method.
4663 (evrp_range_analyzer::leave): New method.
4664 (evrp_dom_walker): Remove delegators no longer needed by this class.
4665 Replace vr_values data member with evrp_range_analyzer
4667 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4668 method extracted from evrp_dom_walker::before_dom_children.
4669 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4670 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4672 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4673 Add private copy constructor and move assignment operators.
4674 Privatize methods and class data where trivially possible.
4675 (evrp_dom_walker::cleanup): New function, extracted from
4676 execute_early_vrp. Simplify access to class data.
4678 * vr-values.h (get_output_for_vrp): Prototype.
4679 * vr-values.c (get_output_for_vrp): New function extracted from
4680 vrp_visit_assignment_or_call and extract_range_from_stmt.
4681 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4683 2017-11-17 Luis Machado <luis.machado@linaro.org>
4685 * config/aarch64/aarch64.c
4686 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4687 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4688 tune_params::AUTOPREFETCHER_WEAK.
4690 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4693 * config/arm/arm.c (arm_valid_target_attribute_rec):
4694 Parse "arch=" and "+<ext>".
4695 (arm_valid_target_attribute_tree): Re-init global options.
4696 (arm_option_override): Make non-static.
4697 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4698 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4699 (__ARM_FEATURE_CRC32): Support undef.
4700 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4701 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4703 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4705 * gdbinit.in (break-on-diagnostic): New command.
4707 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4709 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4710 return to indirect jump for EH return if control-flow
4711 protection is enabled. Change explicit 'false' argument in
4712 pro_epilogue_adjust_stack with a value of
4714 * config/i386/i386.md (simple_return_indirect_internal):
4715 Remove SImode restriction to support 64-bit.
4717 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4719 * combine.c (added_notes_insn): New.
4720 (try_combine): Handle added_notes_insn like added_links_insn.
4721 Rewrite return value code.
4722 (distribute_notes): Set added_notes_insn to the earliest insn we added
4725 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4727 PR rtl-optimization/82621
4728 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
4729 dest of one of those SETs is unused.
4731 2017-11-17 Richard Biener <rguenther@suse.de>
4734 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4735 * tree-pretty-print.c (dump_generic_node): Handle
4736 annot_expr_parallel_kind.
4737 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4738 * gimplify.c (gimple_boolify): Likewise.
4740 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4742 * config.gcc (extra_headers): Add cet.h for x86 targets.
4743 * config/i386/cet.h: New file.
4744 * doc/install.texi: Add --enable-cet/--disable-cet.
4746 2017-11-17 Richard Biener <rguenther@suse.de>
4748 PR tree-optimization/83017
4749 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4750 (gen_parallel_loop): Properly count iterations.
4751 (parallelize_loops): Handle loop->can_be_parallel independent
4752 of flag_loop_parallelize_all. Make static profitability test match
4754 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4755 * invoke.texi (parloops-min-per-thread): Document.
4757 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4759 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4760 upstreaming review comments.
4762 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4764 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4765 with fast unaligned access.
4766 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4768 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4770 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4771 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4773 2017-11-17 Richard Biener <rguenther@suse.de>
4775 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4776 folding of references.
4778 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4781 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4782 of replacing call to strncmp with corresponding call to strcmp when
4785 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4787 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4788 option prefer-avx256 for skylake-avx512 configuration.
4789 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4790 (get_builtin_code_for_version): Ditto.
4792 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4793 Monk Chiang <sh.chiang04@gmail.com>
4795 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4796 (FIXED_REGISTERS): Reserve more register numbers.
4797 (CALL_USED_REGISTERS): Likewise.
4798 (REG_ALLOC_ORDER): Likewise.
4799 (REG_CLASS_CONTENTS): Likewise.
4800 (REGISTER_NAMES): Likewise.
4802 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4803 Kito Cheng <kito.cheng@gmail.com>
4805 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4807 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4809 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4811 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4812 parameter list for vec_splats.
4814 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4816 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4818 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4820 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4823 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4824 with the required version of automake 1.11.6.
4826 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4828 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4830 (DU_C2_3_power9): Correct reservation combinations.
4831 (FP_DIV_power9, VEC_DIV_power9): New.
4832 (power9-alu): Split out rotate/shift...
4833 (power9-rot): ...to here, correct dispatch resource.
4834 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4836 (power9-fp): Correct latency.
4837 (power9-sdiv): Add div/sqrt resource.
4838 (power9-ddiv): Correct latency, add div/sqrt resource.
4839 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4840 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4842 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4844 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4846 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4847 switch statement mapping KF built-ins to TF built-ins if we don't
4848 have the proper ISA 3.0 assembler support.
4850 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4852 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4853 (lower_emutls_function_body): Do not compute it.
4855 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4857 * ipa-split.c (split_bb_info): Turn time to sreal.
4858 (split_point): Likewise.
4859 (dump_split_point): Likewise.
4860 (fine_split_points): Likewise.
4861 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4863 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4865 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4868 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4870 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4872 (expensive_function_p): Remove useless assert.
4873 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4875 2017-11-16 Martin Sebor <msebor@redhat.com>
4877 PR tree-optimization/82588
4878 PR tree-optimization/82583
4879 * tree-vrp.c (check_array_ref): Handle flexible array members,
4880 string literals, and inner indices.
4881 (search_for_addr_array): Add detail to diagnostics.
4883 2017-11-16 Nathan Sidwell <nathan@acm.org>
4887 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4888 (SET_DECL_ASSEMBLER_NAME): Forward to
4889 overwrite_decl_assembler_name.
4890 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4891 (overwrite_decl_assembler_name): Declare.
4892 * tree.c (overwrite_decl_assembler_name): New.
4893 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4894 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4895 (LANG_HOOKS_INITIALIZER): Add it.
4896 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4897 * langhooks.c (lhd_set_decl_assembler_name): Use
4898 SET_DECL_ASSEMBLER_NAME.
4899 (lhd_overwrite_decl_assembler_name): Default implementation.
4901 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4902 Jackson Woodruff <jackson.woodruff@arm.com>
4904 PR tree-optimization/71026
4905 * match.pd: Canonicalize constant multiplies in division.
4907 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4909 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4910 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4911 Remove OPT_fomit_frame_pointer entry.
4912 * common/config/arc/arc-common.c: Likewise.
4913 * common/config/arm/arm-common.c: Likewise.
4914 * common/config/avr/avr-common.c: Likewise.
4915 * common/config/c6x/c6x-common.c: Likewise.
4916 * common/config/cr16/cr16-common.c: Likewise.
4917 * common/config/cris/cris-common.c: Likewise.
4918 * common/config/epiphany/epiphany-common.c: Likewise.
4919 * common/config/fr30/fr30-common.c: Likewise.
4920 * common/config/frv/frv-common.c: Likewise.
4921 * common/config/ia64/ia64-common.c: Likewise.
4922 * common/config/iq2000/iq2000-common.c: Likewise.
4923 * common/config/lm32/lm32-common.c: Likewise.
4924 * common/config/m32r/m32r-common.c: Likewise.
4925 * common/config/mcore/mcore-common.c: Likewise.
4926 * common/config/microblaze/microblaze-common.c: Likewise.
4927 * common/config/mips/mips-common.c: Likewise.
4928 * common/config/mmix/mmix-common.c: Likewise.
4929 * common/config/mn10300/mn10300-common.c: Likewise.
4930 * common/config/nios2/nios2-common.c: Likewise.
4931 * common/config/pa/pa-common.c: Likewise.
4932 * common/config/pdp11/pdp11-common.c: Likewise.
4933 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4934 * common/config/riscv/riscv-common.c: Likewise.
4935 * common/config/rs6000/rs6000-common.c: Likewise.
4936 * common/config/rx/rx-common.c: Likewise.
4937 * common/config/s390/s390-common.c: Likewise.
4938 * common/config/sh/sh-common.c: Likewise.
4939 * common/config/sparc/sparc-common.c: Likewise.
4940 * common/config/tilegx/tilegx-common.c: Likewise.
4941 * common/config/tilepro/tilepro-common.c: Likewise.
4942 * common/config/v850/v850-common.c: Likewise.
4943 * common/config/visium/visium-common.c: Likewise.
4944 * common/config/xstormy16/xstormy16-common.c: Likewise.
4945 * common/config/xtensa/xtensa-common.c: Likewise.
4946 * invoke.texi (-fomit-frame-pointer): Update documentation.
4948 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4950 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4952 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4954 * cfg.c (scale_bbs_frequencies_int,
4955 cale_bbs_frequencies_gcov_type): Remove.
4956 * cfg.h (scale_bbs_frequencies_int,
4957 cale_bbs_frequencies_gcov_type): Remove.
4959 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4961 * tree-ssa-loop-manip.c
4962 (scale_dominated_blocks_in_loop): Update to profile counts.
4963 (tree_transform_and_unroll_loop): Likewise.
4965 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4967 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4968 scale_bbs_frequencies_int.
4970 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4972 * final.c (compute_alignments): Use counts rather than frequencies.
4974 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4976 * cfgloopanal.c: Include sreal.h
4977 (average_num_loop_insns): Use counts and sreal for accounting.
4979 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4981 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4984 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4985 Kito Cheng <kito.cheng@gmail.com>
4987 * config/nds32/constraints.md: Provide more constraints.
4988 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4989 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4990 support constraints usage.
4992 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4994 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4996 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4997 Kito Cheng <kito.cheng@gmail.com>
4999 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5000 * config/nds32/nds32.opt: Refine the layout.
5001 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5002 TARGET_EXT_STRING): Support new options.
5003 * config/nds32/nds32.h: Likewise.
5004 * config/nds32/nds32.md: Likewise.
5005 * config/nds32/nds32-predicates.c: Likewise.
5006 * config/nds32/constraints.md: Likewise.
5007 * common/config/nds32/nds32-common.c: Likewise.
5009 2017-11-16 Julia Koval <julia.koval@intel.com>
5012 * config/i386/gfniintrin.h: Add sse check.
5013 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5015 2017-11-16 Julia Koval <julia.koval@intel.com>
5017 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5018 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5019 (ix86_handle_option): Handle -mavx512vbmi2.
5020 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5021 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5022 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5023 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5024 (ix86_valid_target_attribute_inner_p): Ditto.
5025 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5026 * config/i386/i386.opt (mavx512vbmi2): New option.
5027 * doc/invoke.texi: Add new option.
5029 2017-11-16 Julia Koval <julia.koval@intel.com>
5031 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5032 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5033 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5034 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5035 _mm512_gf2p8mul_epi8): New intrinsics.
5036 * config/i386/i386-builtin-types.def
5037 (V64QI_FTYPE_V64QI_V64QI): New type.
5038 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5039 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5040 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5041 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5042 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5043 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5045 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5047 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5048 explicitly as a stream of bytes.
5050 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5052 * config/rs6000/altivec.h (vec_xst_be): New #define.
5053 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5054 and externalize from *altivec_vperm_<mode>_internal.
5055 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5057 (XL_BE_V8HI): Likewise.
5058 (XL_BE_V4SI): Likewise.
5059 (XL_BE_V4SI): Likewise.
5060 (XL_BE_V2DI): Likewise.
5061 (XL_BE_V4SF): Likewise.
5062 (XL_BE_V2DF): Likewise.
5063 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5064 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5065 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5066 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5067 VSX_BUILTIN_VEC_XST_BE.
5068 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5069 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5071 (altivec_init_builtins): Replace conditional calls to def_builtin
5072 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5073 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5074 with unconditional calls. Remove calls to def_builtin for
5075 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5076 __builtin_vec_xst_be.
5077 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5078 to define_expand, and add alternate RTL generation for P8.
5079 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5080 vsx_ld_elemrev_v8hi.
5081 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5082 add alternate RTL generation for P8.
5083 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5084 vsx_ld_elemrev_v16qi.
5085 (vsx_st_elemrev_v8hi): Convert define_insn
5086 to define_expand, and add alternate RTL generation for P8.
5087 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5088 vsx_st_elemrev_v8hi.
5089 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5090 add alternate RTL generation for P8.
5091 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5092 vsx_st_elemrev_v16qi.
5094 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5097 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5098 TARGET_AVX512ER check.
5099 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5100 neither -mzeroupper nor -mno-zeroupper is used and
5101 TARGET_EMIT_VZEROUPPER is set.
5102 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5103 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5105 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5107 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5108 folding of vector compares.
5109 (fold_build_vec_cmp): New helper function.
5110 (fold_compare_helper): New helper function.
5111 (builtin_function_type): Add compare builtins to the list of functions
5112 having unsigned arguments. Cosmetic updates to comment indentation.
5113 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5114 the not+eq combination.
5116 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5118 PR tree-optimization/82726
5119 PR tree-optimization/70754
5120 * tree-predcom.c (order_drefs_by_pos): New function.
5121 (combine_chains): Move code setting has_max_use_after to...
5122 (try_combine_chains): ...here. New parameter. Sort combined chains
5123 according to position information.
5124 (tree_predictive_commoning_loop): Update call to above function.
5125 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5127 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5129 PR tree-optimization/82726
5131 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5133 PR tree-optimization/70754
5134 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5135 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5136 combined stmt before it if not NULL.
5137 (combine_chains): Process refs reversely and compute dominance point
5141 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5143 PR tree-optimization/79663
5144 * tree-predcom.c (combine_chains): Process refs in reverse order
5145 only for ZERO length chains, and add explaining comment.
5147 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5149 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5150 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5151 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5153 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5155 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5157 2017-11-15 Martin Liska <mliska@suse.cz>
5159 * tree-cfg.c (pass_warn_function_return::execute):
5160 Compare warn_return_type for greater than zero.
5162 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5166 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5167 to return true on Xeon and not on Xeon Phi.
5168 (ix86_check_avx256_register): Changed to ...
5169 (ix86_check_avx_upper_register): ... this. Add extra check for
5170 VALID_AVX512F_REG_OR_XI_MODE.
5171 (ix86_avx_u128_mode_needed): Changed
5172 ix86_check_avx256_register to ix86_check_avx_upper_register.
5173 (ix86_check_avx256_stores): Changed to ...
5174 (ix86_check_avx_upper_stores): ... this. Changed
5175 ix86_check_avx256_register to ix86_check_avx_upper_register.
5176 (ix86_avx_u128_mode_after): Changed
5177 avx_reg256_found to avx_upper_reg_found. Changed
5178 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5179 (ix86_avx_u128_mode_entry): Changed
5180 ix86_check_avx256_register to ix86_check_avx_upper_register.
5181 (ix86_avx_u128_mode_exit): Ditto.
5182 * config/i386/i386.h: (host_detect_local_cpu): New define.
5184 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5186 * config/arm/xgene1.md (xgene1): Split into automatons
5187 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5188 (xgene1_f_load): Adjust reservations and/or types.
5189 (xgene1_f_store): Likewise.
5190 (xgene1_load_pair): Likewise.
5191 (xgene1_store_pair): Likewise.
5192 (xgene1_fp_load1): Likewise.
5193 (xgene1_load1): Likewise.
5194 (xgene1_store1): Likewise.
5195 (xgene1_move): Likewise.
5196 (xgene1_alu): Likewise.
5197 (xgene1_simd): Likewise.
5198 (xgene1_bfm): Likewise.
5199 (xgene1_neon_load1): Likewise.
5200 (xgene1_neon_store1): Likewise.
5201 (xgene1_neon_logic): Likewise.
5202 (xgene1_neon_st1): Likewise.
5203 (xgene1_neon_ld1r): Likewise.
5204 (xgene1_alu_cond): Added.
5205 (xgene1_shift_reg): Likwise.
5206 (xgene1_bfx): Likewise.
5207 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5209 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5212 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5214 (can_widen_mult_without_libcall): New function.
5215 (expand_mul_overflow): If only checking unsigned mul overflow,
5216 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5217 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5218 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5219 (expand_DIVMOD): Formatting fix.
5220 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5221 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5222 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5224 PR tree-optimization/82977
5225 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5226 constructed temporary to strlen_to_stridx.put.
5228 2017-11-15 Martin Liska <mliska@suse.cz>
5230 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5231 * configure: Regenerate.
5233 2017-11-15 Martin Liska <mliska@suse.cz>
5236 * config/sh/sh-mem.cc: Use proper probability for
5239 2017-11-14 Jeff Law <law@redhat.com>
5241 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5242 the red zone for stack_clash_protection_final_dynamic_probe targets
5243 when the total dynamic stack size is zero bytes.
5245 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5246 blocks is post order.
5248 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5250 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5251 * final.c (rest_of_clean_state): Set it for the
5252 -fcompare-debug dump.
5253 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5254 class when TDF_COMPARE_DEBUG is set.
5256 * dwarf2out.c (gen_producer_string): Discard
5259 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5262 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5263 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5264 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5265 (__TGMATH_REAL_2_3): Remove macros.
5266 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5267 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5269 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5271 (remquo): Define using __TGMATH_REAL_3.
5273 2017-11-14 Jeff Law <law@redhat.com>
5275 * vr-values.c: New file with contents extracted from tree-vrp.c.
5276 * Makefile.in (OBJS): Add vr-values.o
5277 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5278 (set_value_range, set_and_canonicalize_value_range): Likewise.
5279 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5280 (value_range_constant_singleton, symbolic_range_p): Likewise.
5281 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5282 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5283 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5284 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5285 (find_case_label_range, find_case_label_index): Likewise.
5286 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5287 (range_int_cst_singleton_p, value_inside_range): Likewise.
5288 (get_single_symbol): Likewise.
5289 (switch_update): Move structure definition here.
5290 (to_remove_edges, to_update_switch_stmts): Provide externs.
5291 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5292 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5293 (vrp_val_is_min, set_value_range): Likewise.
5294 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5295 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5296 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5297 (range_is_nonnull, range_int_cst_p): Likewwise.
5298 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5299 (get_single_symbol, operand_less_p): Likewise
5300 (compare_values_warnv, compare_values): Likewise.
5301 (value_inside_range, value_range_constant_singleton): Likewise.
5302 (zero_nonzero_bitgs_from_vr): Likewise.
5303 (extract_range_from_binary_expr_1): Likewise.
5304 (overflow_comparison_p): Likewise.
5305 (to_remove_edges, to_update_switch_stmts): Likewise.
5306 (find_case_label-index, find_case_label_range): Likewise.
5307 (switch_update, set_value_range_to_nonnegative): Remove.
5308 (set_value_range_to_truthvalue): Likewise.
5309 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5310 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5311 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5312 (find_case_label_ranges, test_for_singularity): Likewise.
5313 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5314 (x_vr_values): Move to its remaining use site.
5316 2017-11-10 Jeff Law <law@redhat.com>
5318 * vr-values.h (VR_INITIALIZER): Move #define here.
5319 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5320 * Makefile.in (OBJS): Add tree-evrp.o
5321 * tree-vrp.h (assert_info): Move structure definition here.
5322 (set_value_range_to_varying): Prototype.
5323 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5324 (infer_value_range, register_edge_assert_for): Likewise.
5325 (stmt_interesting_for_vrp): Likewise.
5326 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5327 (set_value_range_to_varying): No longer static.
5328 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5329 (infer_value_range, register_edge_assert_for): Likewise.
5331 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5333 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5334 generate the XXBRD instruction.
5336 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5337 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5338 as being compatible if -mabi=ieeelongdouble.
5339 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5340 to setup float128 built-ins with hardware support.
5341 (BU_FLOAT128_HW_2): Likewise.
5342 (BU_FLOAT128_HW_3): Likewise.
5343 (BU_FLOAT128_HW_VSX_1): Likewise.
5344 (BU_FLOAT128_HW_VSX_2): Likewise.
5345 (scalar_extract_expq): Change float128 built-in functions to
5346 accommodate having both KFmode and TFmode functions. Use the
5347 KFmode variant as the default.
5348 (scalar_extract_sigq): Likewise.
5349 (scalar_test_neg_qp): Likewise.
5350 (scalar_insert_exp_q): Likewise.
5351 (scalar_insert_exp_qp): Likewise.
5352 (scalar_test_data_class_qp): Likewise.
5353 (sqrtf128_round_to_odd): Delete processing the round to odd
5354 built-in functions as special built-in functions, and define them
5355 as float128 built-ins. Use the KFmode variant as the default.
5356 (truncf128_round_to_odd): Likewise.
5357 (addf128_round_to_odd): Likewise.
5358 (subf128_round_to_odd): Likewise.
5359 (mulf128_round_to_odd): Likewise.
5360 (divf128_round_to_odd): Likewise.
5361 (fmaf128_round_to_odd): Likewise.
5362 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5363 support for KFmode and TFmode xststdcqp calls.
5364 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5365 point, switch the built-in handlers for the get/set float128
5366 exponent, get float128 mantissa, float128 test built-ins, and the
5367 float128 round to odd built-in functions. Eliminate creating the
5368 float128 round to odd built-in functions as special built-ins.
5369 (rs6000_init_builtins): Eliminate special creation of the float128
5370 round to odd built-in functions.
5371 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5372 function insns to support both TFmode and KFmode variants.
5373 (xsxsigqp_<mode>): Likewise.
5374 (xsiexpqpf_<mode>): Likewise.
5375 (xsiexpqp_<mode>): Likewise.
5376 (xststdcqp_<mode>): Likewise.
5377 (xststdcnegqp_<mode>): Likewise.
5378 (xststdcqp_<mode>): Likewise.
5380 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5382 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5383 unused path_in_freq_ptr parameter.
5384 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5386 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5388 * ipa-inline.c (edge_badness): Dump sreal frequency.
5389 (compute_inlined_call_time): Match natural implementaiton ...
5390 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5391 forgotten division by CGRAPH_FREQ_BASE.
5393 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5395 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5396 Solaris 11. Update comment.
5397 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5399 * config/sol2.h (STARTFILE_SPEC): Likewise.
5400 * configure: Regenerate.
5402 2017-11-14 Carl Love <cel@us.ibm.com>
5404 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5405 le_ and be_ prefixes to swap* variables. Remove
5406 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5409 2017-11-14 Jason Merrill <jason@redhat.com>
5411 Support GTY((cache)) on hash_map.
5412 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5413 (ggc_cache_remove): Override it instead of ggc_mx.
5414 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5415 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5416 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5417 (simple_cache_map_traits): Override maybe_mx.
5418 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5419 (hash_map): Friend gt_cleare_cache.
5420 (gt_cleare_cache): New.
5421 * tree.h (tree_cache_traits): New hash_map traits class.
5422 (tree_cache_map): New typedef.
5424 2017-11-14 Richard Biener <rguenther@suse.de>
5426 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5427 paramter and handling.
5428 (cleanup_control_flow_bb): Likewise.
5429 (cleanup_control_flow_pre): New helper performing a DFS walk
5430 to call cleanup_control_flow_bb in PRE order.
5431 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5432 via cleanup_control_flow_pre.
5434 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5436 * config/aarch64/aarch64-simd.md
5437 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5438 (*aarch64_simd_bsl<mode>_alt): Likewise.
5439 (aarch64_simd_bsldi_internal): New.
5440 (aarch64_simd_bsldi_alt): Likewise.
5442 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5444 * tracer.c (better_p): Do not compare frequencies.
5445 * reg-stack.c (better_edge): Likewise.
5446 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5449 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5451 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5452 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5453 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5454 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5455 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5456 * predict.c (maybe_hot_frequency_p): Inline to ...
5457 (maybe_hot_count_p): ... here; rewrite to counts.
5458 (counts_to_freqs): Rename to ...
5459 (update_max_bb_count): ... this one.
5460 (expensive_function_p): Use counts.
5461 (estimate_bb_frequencies): Update.
5462 (rebuild_frequencies): Update.
5463 * predict.h (counts_to_freqs): Rename to ...
5464 (update_max_bb_count): ... this one.
5465 * profile.c (compute_branch_probabilities): Add debug info
5466 * tree-inline.c (expand_call_inline): Update debug info.
5467 (optimize_inline_calls): Use update_max_bb_count..
5468 (tree_function_versioning): Use update_max_bb_count..
5469 * value-prof.c (gimple_value_profile_transformations):
5470 Do not use update_max_bb_count.
5472 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5474 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5475 always use frequencies.
5477 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5479 * bb-reorder.c: Remove frequencies from comments.
5480 (better_edge_p): Use profile counts.
5481 (find_traces): Dump profile counts.
5482 (rotate_loop): Use profile counts.
5483 (find_traces_1_round): Likewise.
5484 (connect_better_edge_p): Use counts instead of probabilities for
5486 (copy_bb_p): Drop early check for non-0 frequency.
5487 (sanitize_hot_paths): Update comments.
5489 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5491 * ipa-split.c (struct split_point): Add count.
5492 (consider_split): Do not compute incoming frequency; compute incoming
5493 count and store it to split_point.
5494 (split_function): Set count of the call to split part correctly.
5496 2017-11-13 Carl Love <cel@us.ibm.com>
5498 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5500 2017-11-13 Tom Tromey <tom@tromey.com>
5502 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5504 2017-11-13 Carl Love <cel@us.ibm.com>
5506 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5507 Add support for builtins:
5508 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5509 vector {un,}signed {char,int,short},
5510 vector {un,}signed {char,int,short}) arguments.
5511 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5512 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5513 Add BU_P9V_AV_2 expansions for the builtins.
5514 * config/rs6000/altivec.h (vec_first_match_index,
5515 vec_first_mismatch_index, vec_first_match_or_eos_index,
5516 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5517 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5518 new extern declaration.
5519 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5520 * config/rs6000/vsx.md (first_match_index_<mode>,
5521 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5522 first_mismatch_or_eos_index_<mode>): Add define expand.
5523 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5524 * doc/extend.texi: Update the built-in documenation file for the new
5527 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5529 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5530 into the min/max operations for _Float<N> and _Float<N>X types.
5532 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5535 * dwarf2out.c (do_eh_frame): New static variable.
5536 (dwarf2out_begin_prologue): Set it.
5537 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5539 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5541 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5543 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5545 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5548 * debug.h (dwarf2out_do_eh_frame): Declare.
5549 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5550 (dwarf2out_do_frame): Use it.
5551 (dwarf2out_do_cfi_asm): Likewise.
5552 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5553 (dwarf2out_assembly_start): Likewise.
5554 (dwarf2out_begin_prologue): Fix comment.
5555 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5556 if the target needs either debug or unwind DWARF2 info.
5557 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5558 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5559 -ftrapping-math, -ftrapv and -fwrapv.
5561 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5563 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5564 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5565 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5566 (estimate_edge_size_and_time): Likewise.
5567 (ipa_merge_fn_summary_after_inlining): Likewise.
5568 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5569 (compute_uninlined_call_time): Use sreal_frequency.
5570 (compute_inlined_call_time): Likewise.
5571 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5572 * profile-count.c: Include sreal.h.
5573 (profile_count::to_sreal_scale): New.
5574 * profile-count.h: Forward declare sreal.
5575 (profile_count::to_sreal_scale): Declare.
5577 2017-11-13 Nathan Sidwell <nathan@acm.org>
5579 * diagnostic.c (maybe_line_and_column): New.
5580 (diagnostic_get_location_text): Use it.
5581 (diagnostic_report_current_module): Likewise.
5582 (test_diagnostic_get_location_text): Add tests.
5584 2017-11-13 Luis Machado <luis.machado@linaro.org>
5586 * doc/md.texi (Specifying processor pipeline description): Fix
5587 incorrect latency for the div instruction example.
5589 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5591 PR tree-optimization/78821
5592 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5593 that bit_not_p is the same.
5594 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5595 (split_group): Count precisely bit_not_p bits in each statement.
5596 (invert_op): New function.
5597 (imm_store_chain_info::output_merged_store): Use invert_op to
5598 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5599 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5601 2017-11-13 Martin Liska <mliska@suse.cz>
5603 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5604 (struct source_info): Likewise.
5605 (add_branch_counts): Likewise.
5606 (add_line_counts): Likewise.
5607 (function_summary): Likewise.
5608 (output_intermediate_line): Likewise.
5609 (generate_results): Likewise.
5611 2017-11-13 Martin Liska <mliska@suse.cz>
5613 * gcov.c (struct block_info): Remove typedef for block_t.
5614 (struct line_info): Likewise.
5615 (line_info::has_block): Likewise.
5616 (EXIT_BLOCK): Likewise.
5617 (unblock): Likewise.
5618 (circuit): Likewise.
5619 (get_cycles_count): Likewise.
5620 (process_file): Likewise.
5621 (read_graph_file): Likewise.
5622 (solve_flow_graph): Likewise.
5623 (find_exception_blocks): Likewise.
5624 (add_line_counts): Likewise.
5625 (accumulate_line_info): Likewise.
5626 (output_line_details): Likewise.
5628 2017-11-13 Martin Liska <mliska@suse.cz>
5630 * gcov.c (struct arc_info): Remove typedef for arc_t.
5631 (struct line_info): Likewise.
5632 (add_branch_counts): Likewise.
5633 (output_branch_count): Likewise.
5634 (function_info::~function_info): Likewise.
5635 (circuit): Likewise.
5636 (output_intermediate_line): Likewise.
5637 (read_graph_file): Likewise.
5638 (solve_flow_graph): Likewise.
5639 (find_exception_blocks): Likewise.
5640 (add_line_counts): Likewise.
5641 (accumulate_line_info): Likewise.
5642 (output_line_details): Likewise.
5643 (output_function_details): Likewise.
5645 2017-11-13 Martin Liska <mliska@suse.cz>
5647 * gcov.c (struct function_info): Remove typedef for function_t.
5648 (struct source_info): Likewise.
5649 (source_info::get_functions_at_location): Likewise.
5650 (solve_flow_graph): Likewise.
5651 (find_exception_blocks): Likewise.
5652 (add_line_counts): Likewise.
5653 (output_intermediate_file): Likewise.
5654 (process_file): Likewise.
5655 (generate_results): Likewise.
5656 (release_structures): Likewise.
5657 (read_graph_file): Likewise.
5658 (read_count_file): Likewise.
5659 (accumulate_line_counts): Likewise.
5660 (output_lines): Likewise.
5662 2017-11-13 Martin Liska <mliska@suse.cz>
5664 * gcov.c (function_info::function_info): Remove num_counts
5665 and add vector<gcov_type>.
5666 (function_info::~function_info): Use the vector.
5667 (process_file): Likewise.
5668 (read_graph_file): Likewise.
5669 (read_count_file): Likewise.
5670 (solve_flow_graph): Likewise.
5672 2017-11-13 Martin Liska <mliska@suse.cz>
5674 * gcov.c (function_info::is_artificial): New function.
5675 (process_file): Erase all artificial early.
5676 (generate_results): Skip as all artificial are already
5679 2017-11-13 Martin Liska <mliska@suse.cz>
5681 * gcov.c (read_graph_file): Store to global vector of functions.
5682 (read_count_file): Iterate the vector.
5683 (process_file): Likewise.
5684 (generate_results): Likewise.
5685 (release_structures): Likewise.
5687 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5689 PR tree-optimization/82954
5690 * gimple-ssa-store-merging.c
5691 (imm_store_chain_info::coalesce_immediate_stores): If
5692 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5694 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5696 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5697 Upddate call to ENDIAN_LANE_N.
5698 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5699 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5700 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5701 and use aarch64_endian_lane_rtx.
5702 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5704 2017-11-12 Tom de Vries <tom@codesourcery.com>
5706 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5708 2017-11-12 Tom de Vries <tom@codesourcery.com>
5710 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5712 2017-11-12 Tom de Vries <tom@codesourcery.com>
5714 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
5716 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5717 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5718 * defaults.h (ASM_OUTPUT_LABELREF): Same.
5720 2017-11-11 Martin Sebor <msebor@redhat.com>
5723 * doc/extend.texi (attribute nonstring): Remove spurious argument.
5726 * prefic.c (translate_name): Replace strncpy with memcpy to
5727 avoid -Wstringop-truncation.
5729 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5731 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
5733 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5735 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5737 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5739 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5742 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5744 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5745 enable generating XXBRH if the value is in a vector register.
5746 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5747 value is in a vector register.
5748 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5749 register to register bswap64's instead of doing the GPR sequence
5750 used on previous machines.
5751 (bswapdi2_xxbrd): New insn.
5752 (bswapdi2_reg): Disallow on ISA 3.0.
5753 (register to register bswap64 splitter): Do not split the insn on
5754 ISA 3.0 systems that use XXBRD.
5756 2017-11-10 Martin Sebor <msebor@redhat.com>
5759 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5761 (find_subframework_file): Same.
5763 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5765 * auto-profile.c (afdo_indirect_call): Drop frequency.
5766 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5767 (cgraph_node::create_edge): Drop frequency argument.
5768 (cgraph_node::create_indirect_edge): Drop frequency argument.
5769 (cgraph_edge::make_speculative): Drop frequency arguments.
5770 (cgraph_edge::resolve_speculation): Do not update frequencies
5771 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5772 (cgraph_node::dump): Check consistency in IPA mode.
5773 (cgraph_edge::maybe_hot_p): Use IPA counter.
5774 (cgraph_edge::verify_count_and_frequency): Rename to ...
5775 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5776 (cgraph_node::verify_node): Update.
5777 * cgraph.h (struct cgraph_edge): Drop frequency.
5778 (cgraph_edge::frequency): New function.
5779 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5781 (cgraph_edge::rebuild_edges): Likewise.
5782 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5783 (duplicate_thunk_for_node): Do not pass frequency.
5784 (cgraph_node::create_clone): Scale only counts.
5785 (cgraph_node::create_virtual_clone): Do not pass frequency.
5786 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5787 (cgraph_node::create_version_clone): Do not pass frequency.
5788 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5789 (cgraph_node::expand_thunk): Do not pass frequency.
5790 (cgraph_node::create_wrapper): Do not pass frequency.
5791 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5793 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5794 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5795 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5796 (gather_caller_stats): Use frequency function.
5797 (ipcp_cloning_candidate_p): Use frequency function.
5798 (ipcp_propagate_stage): Use frequency function.
5799 (get_info_about_necessary_edges): Use frequency function.
5800 (update_profiling_info): Update only IPA profile.
5801 (update_specialized_profile): Use frequency functoin.
5802 (perhaps_add_new_callers): Update only IPA profile.
5803 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5804 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5805 (dump_ipa_call_summary): Use frequency function.
5806 (estimate_edge_size_and_time): Use frequency function.
5807 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5808 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5809 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5810 (update_noncloned_counts): ... ths one; scale counts only.
5811 (clone_inlined_nodes): Do not scale frequency.
5812 (inline_call): Do not pass frequency.
5813 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5814 (compute_inlined_call_time): Use IPA profile.
5815 (want_inline_small_function_p): Use IPA profile.
5816 (want_inline_self_recursive_call_p): Use IPA profile.
5817 (edge_badness): Use IPA profile.
5818 (lookup_recursive_calls): Use IPA profile.
5819 (recursive_inlining): Do not pass frequency.
5820 (resolve_noninline_speculation): Do not update frequency.
5821 (inline_small_functions): Collect max of IPA profile.
5822 (dump_overall_stats): Dump IPA porfile.
5823 (dump_inline_stats): Dump IPA porfile.
5824 (ipa_inline): Collect IPA stats.
5825 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5826 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5827 (ipa_propagate_frequency): Use frequency function.
5828 (ipa_profile): Cleanup.
5829 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5830 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5831 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5832 (input_node): Do not stream frequency.
5833 (input_edge): Do not stream frequency.
5834 (merge_profile_summaries): Scale only IPA profiles.
5835 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5836 * predict.c (drop_profile): Do not recompute frequency.
5837 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5838 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5839 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5840 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5841 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5842 * tree-inline.c (copy_bb): Do not scale frequency.
5843 (expand_call_inline): Do not scale frequency.
5844 (tree_function_versioning): Do not scale frequency.
5845 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5847 2017-11-10 Julia Koval <julia.koval@intel.com>
5849 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5850 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5851 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5852 (_mm256_maskz_gf2p8affine_epi64_epi8)
5853 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5854 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5855 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5856 (__builtin_ia32_vgf2p8affineqb_v32qi)
5857 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5858 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5860 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5864 (arm_option_override): Refactor.
5865 (arm_option_reconfigure_globals): New.
5866 (arm_options_perform_arch_sanity_checks): New.
5867 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5869 (arm_options_perform_arch_sanity_checks): Likewise
5871 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5873 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5874 (power9-qpmul): New.
5875 * rs6000/rs6000.md ("type" attr): Add qmul.
5876 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5877 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5878 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5880 2017-11-10 Martin Sebor <msebor@redhat.com>
5883 * builtins.c (compute_objsize): Handle arrays that
5884 compute_builtin_object_size likes to fail for. Make extern.
5885 * builtins.h (compute_objsize): Declare.
5886 (check_strncpy_sizes): New function.
5887 (expand_builtin_strncpy): Call check_strncpy_sizes.
5888 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5889 -Wstringop-truncation.
5890 (gimple_fold_builtin_strncat): Same.
5891 * gimple.c (gimple_build_call_from_tree): Set call location.
5892 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5893 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5894 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5895 (handle_builtin_strlen): Use strlen_to_stridx.
5896 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5898 Use strlen_to_stridx.
5899 (pass_strlen::execute): Release strlen_to_stridx.
5900 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5901 (-Wstringop-truncation): Document new option.
5903 2017-11-10 Martin Liska <mliska@suse.cz>
5905 PR gcov-profile/82702
5906 * gcov.c (main): Handle intermediate files in a different
5908 (get_gcov_intermediate_filename): New function.
5909 (output_gcov_file): Remove support of intermediate files.
5910 (generate_results): Allocate intermediate file.
5911 (release_structures): Clean-up properly fn_end.
5912 (output_intermediate_file): Start iterating with line 1.
5914 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5916 PR tree-optimization/82929
5917 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5918 ops_swapped_p non-static data member.
5919 (store_immediate_info::store_immediate_info): Clear it.
5920 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5921 ops set ops_swapped_p.
5922 (count_multiple_uses): Handle ops_swapped_p.
5924 2017-11-10 Martin Liska <mliska@suse.cz>
5926 * coverage.c (coverage_init): Stream information about
5927 support of has_unexecuted_blocks.
5928 * doc/gcov.texi: Document that.
5929 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5930 * gcov.c (read_graph_file): Likewise.
5931 (output_line_beginning): Fix a small issue with
5934 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5936 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5937 reference of trivial component.
5939 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5942 * gimple-ssa-store-merging.c
5943 (pass_store_merging::terminate_all_aliasing_chains): For
5944 gimple_store_p stmts also call refs_output_dependent_p.
5946 PR rtl-optimization/82913
5947 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5950 2017-11-09 Jeff Law <law@redhat.com>
5952 * vr-values.h: New file with vr_values class.
5953 * tree-vrp.c: Include vr-values.h
5954 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5955 data objects into the vr_values class.
5956 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5957 (get_value_range): Make it a member function within vr_values class.
5958 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5959 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5960 (op_with_constant_singleton_value_range): Likewise.
5961 (extract_range_for_var_from_comparison_expr): Likewise.
5962 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5963 (extract_range_from_binary_expr): Likewise.
5964 (extract_range_from_unary_expr): Likewise.
5965 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5966 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5967 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5968 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5969 (compare_name_with_value, compare_names): Likewise.
5970 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5971 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5972 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5973 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5974 (extract_range_from_phi_node): Likewise.
5975 (simplify_truth_ops_using_ranges): Likewise.
5976 (simplify_div_or_mod_using_ranges): Likewise.
5977 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5978 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5979 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5980 (simplify_float_conversion_using_ranges): Likewise.
5981 (simplify_internal_call_using_ranges): Likewise.
5982 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5983 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5984 poitner via x_vr_values for calls into gimple folder.
5985 (vrp_initialize_lattice): Make this the vr_values ctor.
5986 (vrp_free_lattice): Make this the vr_values dtor.
5987 (set_vr_value): New function.
5988 (class vrp_prop): Add vr_values data member. Add various member
5989 functions as well as member functions that delegate to vr_values.
5990 (check_array_ref): Make a member function within vrp_prop class.
5991 (search_for_addr_array, vrp_initialize): Likewise.
5992 (vrp_finalize): Likewise. Revamp to avoid direct access to
5993 vr_value, values_propagated, etc.
5994 (check_array_bounds): Extract vrp_prop class instance pointer from
5995 walk info structure. Use it to call member functions.
5996 (check_all_array_refs): Make a member function within vrp_prop class.
5997 Smuggle class instance pointer via walk info structure.
5998 (x_vr_values): New local static.
5999 (vrp_valueize): Use x_vr_values to get class instance.
6000 (vr_valueize_1): Likewise.
6001 (class vrp_folder): Add vr_values data member. Add various member
6002 functions as well as member functions that delegate to vr_values.
6003 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6004 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6005 class instance from vr_values. Use it to call member functions.
6006 (vrp_dom_walker): Add vr_values data member.
6007 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6008 instance via x_vr_values.
6009 (identify_jump_threads): Accept vr_values as argument. Store
6010 it into the walker structure.
6011 (evrp_dom_walker): Add vr_values class data member. Add various
6013 (evrp_dom_walker::try_find_new_range): Use vr_values data
6014 member to access the memory allocator.
6015 (evrp_dom_walker::before_dom_children): Store vr_values class
6016 instance into the vrp_folder class.
6017 (evrp_dom_walker::push_value_range): Rework to avoid direct
6018 access to num_vr_values and vr_value.
6019 (evrp_dom_walker::pop_value_range): Likewise.
6020 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6021 Use vr_values to get to dump_all_value_ranges member function.
6022 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6023 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6024 Pass vr_values class instance down to identify_jump_threads.
6025 Remove call to vrp_free_lattice.
6026 (debug_all_value_ranges): Remove.
6028 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6029 (vrp_folder): Likewise.
6031 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6032 Get it from the existing bitmap instead.
6033 (vrp_intersect_ranges_1): Likewise.
6035 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6037 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6039 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6040 set bit_not_p to it.
6041 (imm_store_chain_info::coalesce_immediate_stores): Break group
6042 if bit_not_p is different.
6043 (count_multiple_uses, split_group,
6044 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6045 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6046 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6047 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6048 to store_immediate_info ctor.
6050 2017-11-09 Jim Wilson <jimw@sifive.com>
6052 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6053 (scan_prog_file): Likewise.
6055 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6057 * bb-reorder.c (max_entry_frequency): Remove.
6058 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6059 connect_traces, push_to_next_round_p): Remove prototypes.
6060 (find_traces_1_round): Use counts only.
6061 (push_to_next_round_p): Likewise.
6062 (find_traces): Likewise.
6063 (rotate_loop): Likewise.
6064 (find_traces_1_round): Likewise.
6065 (connect_traces): Likewise.
6066 (edge_order): Likewise.
6068 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6070 * config/arm/arm.c (output_return_instruction): Add comments to
6071 indicate requirement for cmse_nonsecure_entry return to account
6072 for the size of clearing instruction output here.
6073 (thumb_exit): Likewise.
6074 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6075 return in hardfloat mode.
6077 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6079 * config/rs6000/rs6000.c (machine_function): Add a bool,
6080 "toc_is_wrapped_separately".
6081 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6082 if it wasn't explicitly set or unset, we are optimizing for speed, and
6083 doing separate shrink-wrapping.
6084 (rs6000_get_separate_components): Enable the TOC component if
6085 saving the TOC register in the prologue.
6086 (rs6000_components_for_bb): Handle the TOC component.
6087 (rs6000_emit_prologue_components): Store the TOC register where needed.
6088 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6089 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6091 2017-11-09 Martin Jambor <mjambor@suse.cz>
6093 * ipa-param-manipulation.c: New file.
6094 * ipa-param-manipulation.h: Likewise.
6095 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6096 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6097 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6098 (ipa_parm_adjustment): Likewise.
6099 (ipa_parm_adjustment_vec): Likewise.
6100 (ipa_get_vector_of_formal_parms): Moved declaration to
6101 ipa-param-manipulation.h.
6102 (ipa_get_vector_of_formal_parm_types): Likewise.
6103 (ipa_modify_formal_parameters): Likewise.
6104 (ipa_modify_call_arguments): Likewise.
6105 (ipa_combine_adjustments): Likewise.
6106 (ipa_dump_param_adjustments): Likewise.
6107 (ipa_modify_expr): Likewise.
6108 (ipa_get_adjustment_candidate): Likewise.
6109 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6110 ipa-param-manipulation.c.
6111 (ipa_get_vector_of_formal_parm_types): Likewise.
6112 (ipa_modify_formal_parameters): Likewise.
6113 (ipa_modify_call_arguments): Likewise.
6114 (ipa_modify_expr): Likewise.
6115 (get_ssa_base_param): Likewise.
6116 (ipa_get_adjustment_candidate): Likewise.
6117 (index_in_adjustments_multiple_times_p): Likewise.
6118 (ipa_combine_adjustments): Likewise.
6119 (ipa_dump_param_adjustments): Likewise.
6120 * tree-sra.c: Also include ipa-param-manipulation.h
6121 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6124 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6125 Alan Hayward <alan.hayward@arm.com>
6126 David Sherwood <david.sherwood@arm.com>
6128 * doc/sourcebuild.texi (vect_masked_store): Document.
6130 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6131 Alan Hayward <alan.hayward@arm.com>
6132 David Sherwood <david.sherwood@arm.com>
6134 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6136 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6137 Alan Hayward <alan.hayward@arm.com>
6138 David Sherwood <david.sherwood@arm.com>
6140 * doc/sourcebuild.texi (vect_variable_length): Document.
6142 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6143 Alan Hayward <alan.hayward@arm.com>
6144 David Sherwood <david.sherwood@arm.com>
6146 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6148 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6149 Alan Hayward <alan.hayward@arm.com>
6150 David Sherwood <david.sherwood@arm.com>
6152 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
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 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6159 previously undocumented selectors.
6160 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6162 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6164 * doc/rtl.texi (const_vector): Say that elements can be
6165 const_wide_ints too.
6166 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6167 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6168 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6169 * optabs.c (expand_vector_broadcast): Likewise.
6171 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6172 Alan Hayward <alan.hayward@arm.com>
6173 David Sherwood <david.sherwood@arm.com>
6175 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6176 scaled index even if the unscaled address was invalid.
6177 Don't increase the complexity of using a scale in that case.
6179 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6180 Alan Hayward <alan.hayward@arm.com>
6181 David Sherwood <david.sherwood@arm.com>
6183 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6184 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6185 UNITS_PER_WORD bytes.
6186 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6187 based on REGMODE_NATURAL_SIZE of the inner mode.
6188 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6189 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6190 * expmed.c (lowpart_bit_field_p): Divide the value up into
6191 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6192 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6193 whether something is likely to occupy more than one register.
6195 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6198 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6199 profile_count::adjust_for_ipa_scaling.
6200 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6201 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6203 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6205 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6207 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6208 (split_group): Add total_orig and total_new arguments, estimate the
6209 number of statements related to the store group without store merging
6210 and with store merging.
6211 (imm_store_chain_info::output_merged_store): Adjust split_group
6212 callers, punt if estimated number of statements with store merging
6213 is not smaller than estimated number of statements without it.
6215 (handled_load): Remove has_single_use checks.
6216 (pass_store_merging::process_store): Likewise.
6218 2017-11-09 Richard Biener <rguenther@suse.de>
6220 PR tree-optimization/82902
6221 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6223 2017-11-09 Martin Liska <mliska@suse.cz>
6226 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6229 2017-11-09 Martin Liska <mliska@suse.cz>
6231 PR tree-optimization/82669
6232 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6234 2017-11-09 Martin Liska <mliska@suse.cz>
6236 PR gcov-profile/48463
6237 * coverage.c (coverage_begin_function): Output also end locus
6238 of a function and information whether the function is
6240 * gcov-dump.c (tag_function): Parse and print the information.
6241 * gcov.c (INCLUDE_MAP): Add include.
6242 (INCLUDE_SET): Likewise.
6243 (struct line_info): Move earlier in the source file because
6244 of vector<line_info> in function_info structure.
6245 (line_info::line_info): Likewise.
6246 (line_info::has_block): Likewise.
6247 (struct source_info): Add new member index.
6248 (source_info::get_functions_at_location): New function.
6249 (function_info::group_line_p): New function.
6250 (output_intermediate_line): New function.
6251 (output_intermediate_file): Use the mentioned function.
6252 (struct function_start): New.
6253 (struct function_start_pair_hash): Likewise.
6254 (process_file): Add code that identifies group functions.
6255 Assign lines either to global or function scope.
6256 (generate_results): Skip artificial functions.
6257 (find_source): Assign index for each source file.
6258 (read_graph_file): Read new flag artificial and end_line.
6259 (add_line_counts): Assign it either to global of function scope.
6260 (accumulate_line_counts): Isolate core of the function to
6261 accumulate_line_info and call it for both function and global
6263 (accumulate_line_info): New function.
6264 (output_line_beginning): Fix GNU coding style.
6265 (print_source_line): New function.
6266 (output_line_details): Likewise.
6267 (output_function_details): Likewise.
6268 (output_lines): Iterate both source (global) scope and function
6270 (struct function_line_start_cmp): New class.
6271 * doc/gcov.texi: Reflect changes in documentation.
6273 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6276 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6277 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6278 and similarly for not instead of neg.
6280 2017-11-08 Andi Kleen <ak@linux.intel.com>
6282 * config/i386/i386.opt: Add -mforce-indirect-call.
6283 * config/i386/predicates.md: Check for flag_force_indirect_call.
6284 * doc/invoke.texi: Document -mforce-indirect-call
6286 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6288 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6290 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6291 config/riscv/riscv.c (predict.h): New include.
6292 (riscv_slow_unaligned_access_p): No longer static.
6293 (riscv_block_move_straight): Add require.
6294 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6296 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6299 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6300 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6301 nonimmediate_operand predicate for operand 1 instead of
6304 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6306 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6308 * config/aarch64/constraints.md (Uml): New constraint.
6309 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6312 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6314 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6315 of two vec_duplicates into a vec_concat.
6317 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6319 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6320 Simplify vec_merge of vec_duplicate and vec_concat.
6321 * config/aarch64/constraints.md (Utq): New constraint.
6322 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6325 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6327 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6328 Simplify vec_merge of vec_duplicate and const_vector.
6329 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6331 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6332 mode iterator. Update predicate on operand 1 to
6333 handle non-const_vec constants. Delete constraints.
6334 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6336 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6338 PR tree-optimization/78821
6339 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6341 (store_operand_info::store_operand_info): Initialize it to false.
6342 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6343 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6344 store in the group, and if chain_info is non-NULL, to ignore altogether
6346 (compatible_load_p): Fail if bit_not_p does not match.
6347 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6348 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6349 (pass_store_merging::process_store): Adjust
6350 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6351 call terminate_all_aliasing_chains newly when adding a store into
6352 a chain with non-NULL chain_info.
6354 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6356 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6358 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6360 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6362 (aarch64_layout_frame): Initialise emit_frame_chain.
6363 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6364 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6366 2017-11-08 Martin Liska <mliska@suse.cz>
6368 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6369 of gimple_call_internal_p.
6371 2017-11-07 Tom de Vries <tom@codesourcery.com>
6373 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6375 2017-11-07 Tom de Vries <tom@codesourcery.com>
6377 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6380 2017-11-08 Martin Liska <mliska@suse.cz>
6383 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6385 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6387 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6389 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6392 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6393 (*cmp<mode>_ccz_1): New insn with $k alternative.
6396 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6397 RTX_COMM_COMPARE as commutative as well.
6398 (ix86_binary_operator_ok): Formatting fix.
6399 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6400 *<code><mode>3<mask_name><round_saeonly_name>,
6401 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6402 *<s>mul<mode>3_highpart<mask_name>,
6403 *vec_widen_umult_even_v16si<mask_name>,
6404 *vec_widen_umult_even_v8si<mask_name>,
6405 *vec_widen_umult_even_v4si<mask_name>,
6406 *vec_widen_smult_even_v16si<mask_name>,
6407 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6408 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6409 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6410 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6411 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6412 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6413 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6414 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6415 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6416 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6417 ix86_binary_operator_ok. Formatting fixes.
6418 (*<plusminus_insn><mode>3<mask_name><round_name>,
6419 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6422 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6424 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6425 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6426 eq and ne if TARGET_ISEL.
6427 (cmp): New code_iterator.
6428 (UNS, UNSU_, UNSIK): New code_attrs.
6429 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6430 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6432 ("eq<mode>3"): ... this.
6433 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6435 ("ne<mode>3"): ... this.
6437 2017-11-07 Julia Koval <julia.koval@intel.com>
6440 * common/config/i386/i386-common.c
6441 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6442 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6443 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6444 * config/i386/i386.opt: Ditto.
6445 * config/i386/i386.c (ix86_target_string): Ditto.
6446 (ix86_option_override_internal): Ditto.
6447 (ix86_init_mpx_builtins): Move MPX to args2.
6448 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6449 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6450 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6451 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6452 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6453 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6454 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6456 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6459 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6460 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6461 (zero-extendsidi peephole2): Remove peephole.
6463 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6466 * stor-layout.c: Include attribs.h.
6467 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6468 explicit lookup of "aligned" attribute.
6470 2017-11-07 Andrew Waterman <andrew@sifive.com>
6472 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6473 (riscv_expand_block_move): Likewise.
6474 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6476 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6477 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6478 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6479 (riscv_adjust_block_mem): Likewise.
6480 (riscv_block_move_loop): Likewise.
6481 (riscv_expand_block_move): Likewise.
6482 * config/riscv/riscv.md (movmemsi): New pattern.
6484 2017-11-07 Michael Clark <michaeljclark@mac.com>
6486 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6487 (MUSL_DYNAMIC_LINKER): Likewise.
6489 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6491 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6492 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6495 2017-11-07 Richard Biener <rguenther@suse.de>
6497 * match.pd: Fix build.
6499 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6500 Jackson Woodruff <jackson.woodruff@arm.com>
6502 PR tree-optimization/71026
6503 * match.pd: Canonicalize negate in division.
6505 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6508 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6510 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6512 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6513 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6515 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6517 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6518 non-scalar integral types.
6519 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6520 (-(A-B), -(~A)): New transformations.
6522 2017-11-07 Tom de Vries <tom@codesourcery.com>
6524 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6525 semicolon after "do {} while (0)".
6526 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6527 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6528 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6529 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6530 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6532 2017-11-07 Tom de Vries <tom@codesourcery.com>
6534 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6535 after "do {} while (0)".
6536 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6537 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6538 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6539 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6540 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6542 2017-11-07 Tom de Vries <tom@codesourcery.com>
6545 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6547 (arm_rtx_costs_internal): Add missing semicolon after
6548 HANDLE_NARROW_SHIFT_ARITH call.
6550 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6552 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6553 disable isel if it was not set explicitly.
6555 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6557 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6558 (add_type_attribute) likewise.
6560 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6562 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6563 of struct ix86_frame.
6564 (ix86_initial_elimination_offset): Likewise.
6565 (ix86_expand_split_stack_prologue): Likewise.
6567 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6569 * tree-vrp.h (enum value_range_type): Update stale comment.
6571 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6572 Alan Hayward <alan.hayward@arm.com>
6573 David Sherwood <david.sherwood@arm.com>
6575 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6576 (aarch64_expand_vec_perm_const): Take the number of units too.
6577 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6578 (aarch64_expand_vec_perm_const): Likewise.
6579 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6580 (vec_perm<mode>): Update accordingly.
6582 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6583 Alan Hayward <alan.hayward@arm.com>
6584 David Sherwood <david.sherwood@arm.com>
6586 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6587 Take the number of units too.
6588 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6589 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6590 but check for a vector mode before rather than after the call.
6591 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6592 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6593 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6594 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6595 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6596 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6597 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6598 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6599 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6600 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6601 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6602 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6603 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6604 (aarch64_sqdmull2_n<mode>): Update accordingly.
6606 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6607 Alan Hayward <alan.hayward@arm.com>
6608 David Sherwood <david.sherwood@arm.com>
6610 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6611 the number of units too.
6612 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6613 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6614 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6615 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6616 (vec_store_lanesxi<mode>): Update accordingly.
6618 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6619 Alan Hayward <alan.hayward@arm.com>
6620 David Sherwood <david.sherwood@arm.com>
6622 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6623 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6624 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6625 of units rather than the mode.
6626 * config/aarch64/iterators.md (nunits): New mode attribute.
6627 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6628 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6629 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6630 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6631 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6632 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6633 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6634 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6635 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6636 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6637 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6638 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6639 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6640 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6641 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6642 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6643 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6644 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6645 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6646 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6647 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6648 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6649 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6650 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6651 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6652 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6653 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6654 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6655 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6656 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6657 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6658 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6659 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6660 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6661 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6662 (aarch64_simd_vec_setv2di): Likewise.
6664 2017-11-06 Carl Love <cel@us.ibm.com>
6666 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6668 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6669 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6671 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6673 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6674 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6675 Add power 8 macro expansions.
6676 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6677 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6678 power 8 instructions. (VSX_XXBR): Add iterator.
6680 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6682 * config/arm/arm.md (predicable_short_it): Change default to "no",
6683 improve documentation, remove uses that are identical to the default.
6684 (enabled_for_depr_it): Rename to enabled_for_short_it.
6685 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6686 Remove default uses.
6687 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6688 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6689 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6690 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6692 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6695 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6696 float128 helper macros, which are no longer used after deleting
6697 the old 'q' built-in functions, and moving the round to odd
6698 built-in functions to being special built-in functions.
6699 (BU_FLOAT128_2): Likewise.
6700 (BU_FLOAT128_1_HW): Likewise.
6701 (BU_FLOAT128_2_HW): Likewise.
6702 (BU_FLOAT128_3_HW): Likewise.
6703 (FABSQ): Delete old 'q' built-in functions.
6704 (COPYSIGNQ): Likewise.
6705 (SQRTF128_ODD): Move round to odd built-in functions to be
6706 special built-in functions, so that we can handle
6707 -mabi=ieeelongdouble.
6708 (TRUNCF128_ODD): Likewise.
6709 (ADDF128_ODD): Likewise.
6710 (SUBF128_ODD): Likewise.
6711 (MULF128_ODD): Likewise.
6712 (DIVF128_ODD): Likewise.
6713 (FMAF128_ODD): Likewise.
6714 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
6715 built-in names to 'f128'.
6716 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
6717 old 'q' built-in functions, as the machine independent code for
6718 'f128' built-in functions handles this.
6719 (rs6000_expand_builtin): Add expansion for float128 round to odd
6720 functions, keying off on -mabi=ieeelongdouble of whether to use
6721 the KFmode or TFmode variant.
6722 (rs6000_init_builtins): Initialize the _Float128 round to odd
6724 * doc/extend.texi (PowerPC Built-in Functions): Document the old
6725 _Float128 'q' built-in functions are now mapped into the new
6726 'f128' built-in functions.
6728 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
6730 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
6731 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
6733 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6736 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6737 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6738 * toplev.c: Include "ipa-fnsummary.h".
6739 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6741 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6743 PR tree-optimization/82838
6744 * gimple-ssa-store-merging.c
6745 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6746 on a separate gimple_seq which is then appended to seq.
6748 2017-11-06 Jeff Law <law@redhat.com>
6751 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6752 (get_probe_interval): New functions.
6753 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6754 (ix86_adjust_stack_and_probe): Likewise.
6755 (output_adjust_stack_and_probe): Likewise.
6756 (ix86_emit_probe_stack_range): Likewise.
6757 (ix86_expand_prologue): Likewise.
6759 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6761 PR tree-optimization/82816
6762 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6763 if the modes of the two types are the same.
6764 (convert_plusminus_to_widen): Likewise.
6766 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6768 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6770 (usadv16qi): New define_expand.
6771 (usadv8hi): New define_expand.
6773 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6776 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6778 (inline_transform): Likewise.
6779 * predict.c (counts_to_freqs): Remove useless conditional.
6780 * profile-count.h (profile_count::apply_scale): Move sanity check.
6781 * tree-inline.c (copy_bb): Always scale.
6782 (copy_cfg_body): Likewise.
6784 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6787 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6789 (*cmp_ite0): Add enabled_for_depr_it attribute.
6790 (*cmp_ite1): Likewise.
6792 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6794 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6797 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6799 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6800 return the value by pointer.
6801 (extract_range_from_multiplicative_op_1): Update accordingly.
6802 Return as soon as an operation fails.
6804 2017-11-05 Tom de Vries <tom@codesourcery.com>
6807 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6808 (DEF_SANITIZER_BUILTIN): ... here.
6809 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6810 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6812 2017-11-05 Tom de Vries <tom@codesourcery.com>
6815 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6817 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6818 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6819 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6821 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6822 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6823 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6825 2017-11-05 Tom de Vries <tom@codesourcery.com>
6828 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6831 2017-11-04 Michael Clark <michaeljclark@mac.com>
6833 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6834 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6836 (*addsi3_extended): Likewise.
6837 (*addsi3_extended2): Likewise.
6838 (<optab>si3): Likewise.
6839 (<optab>di3): Likewise.
6840 (<optab><mode>3): Likewise.
6841 (<*optabe>si3_internal): Likewise.
6842 (zero_extendqi<SUPERQI:mode>2): Likewise.
6843 (*add<mode>hi3): Likewise.
6844 (*xor<mode>hi3): Likewise.
6845 (<optab>di3): Likewise.
6846 (*<optab>si3_extend): Likewise.
6847 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6848 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6849 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6851 2017-11-04 Andrew Waterman <andrew@sifive.com>
6853 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6854 TARGET_STRICT_ALIGN based upon -mtune argument.
6856 2017-11-04 Andrew Waterman <andrew@sifive.com>
6858 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6860 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6862 * config/i386/i386.c (choose_basereg): Use optional scratch
6863 register and add assertion.
6864 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6865 needed, and don't allocate stack.
6866 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6867 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6868 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6870 2017-11-03 Jeff Law <law@redhat.com>
6872 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6873 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6874 to probe at the start of a noreturn function.
6876 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6878 PR tree-optimization/78821
6879 * gimple-ssa-store-merging.c: Update the file comment.
6880 (MAX_STORE_ALIAS_CHECKS): Define.
6881 (struct store_operand_info): New type.
6882 (store_operand_info::store_operand_info): New constructor.
6883 (struct store_immediate_info): Add rhs_code and ops data members.
6884 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6885 and op1r arguments to the ctor, initialize corresponding data members.
6886 (struct merged_store_group): Add load_align_base and load_align
6888 (merged_store_group::merged_store_group): Initialize them.
6889 (merged_store_group::do_merge): Update them.
6890 (merged_store_group::apply_stores): Pick the constant for
6891 encode_tree_to_bitpos from one of the two operands, or skip
6892 encode_tree_to_bitpos if neither operand is a constant.
6893 (class pass_store_merging): Add process_store method decl. Remove
6894 bool argument from terminate_all_aliasing_chains method decl.
6895 (pass_store_merging::terminate_all_aliasing_chains): Remove
6896 var_offset_p argument and corresponding handling.
6897 (stmts_may_clobber_ref_p): New function.
6898 (compatible_load_p): New function.
6899 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6900 if there is overlap and rhs_code is not INTEGER_CST. For
6901 non-overlapping stores terminate group if rhs is not mergeable.
6902 (get_alias_type_for_stmts): Change first argument from
6903 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6904 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6905 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6907 (get_location_for_stmts): Change first argument from
6908 auto_vec<gimple *> & to vec<gimple *> &.
6909 (struct split_store): Remove orig_stmts data member, add orig_stores.
6910 (split_store::split_store): Create orig_stores rather than orig_stmts.
6911 (find_constituent_stmts): Renamed to ...
6912 (find_constituent_stores): ... this. Change second argument from
6913 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6914 to info structures rather than the statements.
6915 (split_group): Rename ALLOW_UNALIGNED argument to
6916 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6917 it. Adjust find_constituent_stores caller.
6918 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6919 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6920 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6921 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6922 (mem_valid_for_store_merging): New function.
6923 (handled_load): New function.
6924 (pass_store_merging::process_store): New method.
6925 (pass_store_merging::execute): Use process_store method. Adjust
6926 terminate_all_aliasing_chains caller.
6928 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6930 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6931 Return true for more constants, symbols and label references.
6932 (aarch64_valid_floating_const): Remove unused function.
6934 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6937 * config/aarch64/aarch64.c (aarch64_layout_frame):
6938 Undo forcing of LR at bottom of frame.
6940 2017-11-03 Jeff Law <law@redhat.com>
6943 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6944 for int_registers_saved.
6946 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6947 extracted from tree-ssa-dom.c.
6948 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6949 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6950 (record_equivalences_from_incoming_edge): Add additional argument
6951 to single_pred_edge_ignoring_loop_edges call.
6952 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6953 (uncprop_dom_walker::before_dom_children): Add additional argument
6954 to single_pred_edge_ignoring_loop_edges call.
6955 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6956 single_pred_edge_ignoring_loop_edges rather than open coding.
6957 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6959 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6961 * match.pd (-(-A)): Rewrite.
6963 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6965 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6966 (rs6000_emit_int_cmove): New declaration.
6967 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6968 (rs6000_emit_sISEL): Delete.
6969 (rs6000_emit_int_cmove): Make non-static.
6970 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6971 instead of rs6000_emit_sISEL.
6973 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6975 * asan.c (create_cond_insert_point): Maintain profile.
6976 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6978 * basic-block.h (struct basic_block_def): Remove frequency.
6979 (EDGE_FREQUENCY): Use to_frequency
6980 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6982 (find_traces): Update to use to_frequency.
6983 (find_traces_1_round): Likewise; use only IPA counts.
6984 (bb_to_key): Likewise.
6985 (connect_traces): Use IPA counts only.
6986 (copy_bb_p): Update to use to_frequency.
6987 (fix_up_crossing_landing_pad): Likewise.
6988 (sanitize_hot_paths): Likewise.
6989 * bt-load.c (basic_block_freq): Likewise.
6990 * cfg.c (init_flow): Set count_max to uninitialized.
6991 (check_bb_profile): Remove frequencies; check counts.
6992 (dump_bb_info): Do not dump frequencies.
6993 (update_bb_profile_for_threading): Update counts only.
6994 (scale_bbs_frequencies_int): Likewise.
6995 (MAX_SAFE_MULTIPLIER): Remove.
6996 (scale_bbs_frequencies_gcov_type): Update counts only.
6997 (scale_bbs_frequencies_profile_count): Update counts only.
6998 (scale_bbs_frequencies): Update counts only.
6999 * cfg.h (struct control_flow_graph): Add count-max.
7000 (update_bb_profile_for_threading): Update prototype.
7001 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7002 (find_many_sub_basic_blocks): Likewise.
7003 * cfgcleanup.c (try_forward_edges): Likewise.
7004 (try_crossjump_to_edge): Likewise.
7005 * cfgexpand.c (expand_gimple_cond): Likewise.
7006 (expand_gimple_tailcall): Likewise.
7007 (construct_init_block): Likewise.
7008 (construct_exit_block): Likewise.
7009 * cfghooks.c (verify_flow_info): Check consistency of counts.
7010 (dump_bb_for_graph): Do not dump frequencies.
7011 (split_block_1): Do not update frequencies.
7012 (split_edge): Do not update frequencies.
7013 (make_forwarder_block): Do not update frequencies.
7014 (duplicate_block): Do not update frequencies.
7015 (account_profile_record): Do not update frequencies.
7016 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7017 for global heuristics.
7018 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7019 (expected_loop_iterations_unbounded): Use counts only.
7020 * cfgloopmanip.c (scale_loop_profile): Simplify.
7021 (create_empty_loop_on_edge): Simplify
7023 (duplicate_loop_to_header_edge): Simplify
7024 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7025 (update_br_prob_note): Take care of removing note when profile
7027 (relink_block_chain): Do not dump frequency.
7028 (rtl_account_profile_record): Use to_frequency.
7029 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7030 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7031 (cgraph_update_edges_for_call_stmt_node): Likewise.
7032 (cgraph_edge::verify_count_and_frequency): Update.
7033 (cgraph_node::verify_node): Temporarily disable frequency verification.
7034 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7035 to_cgraph_frequency.
7036 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7037 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7039 (cgraph_node::expand_thunk): Update profile.
7040 * except.c (dw2_build_landing_pads): Do not update frequency.
7041 * final.c (compute_alignments): Use to_frequency.
7042 (dump_basic_block_info): Do not dump frequency.
7043 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7044 (dump_gimple_bb_header): Do not dump frequency.
7045 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7047 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7048 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7049 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7050 (init_before_recovery): Do not update frequency.
7051 (sched_create_recovery_edges): Do not update frequency.
7052 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7053 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7054 (ipa_cp_c_finalize): Set max_count to uninitialized.
7055 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7056 (param_change_prob): Use counts.
7057 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7059 * ipa-split.c (consider_split): Use to_frequency.
7060 (split_function): Use to_frequency.
7061 * ira-build.c (loop_compare_func): Likewise.
7062 (mark_loops_for_removal): Likewise.
7063 (mark_all_loops_for_removal): Likewise.
7064 * loop-doloop.c (doloop_modify): Do not update frequency.
7065 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7067 * lto-streamer-in.c (input_function): Update count_max.
7068 * omp-expand.c (expand_omp_taskreg): Update count_max.
7069 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7070 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7071 (maybe_hot_count_p): Use ipa counts only.
7072 (maybe_hot_bb_p): Simplify.
7073 (maybe_hot_edge_p): Simplify.
7074 (probably_never_executed): Do not take frequency argument.
7075 (probably_never_executed_bb_p): Do not pass frequency.
7076 (probably_never_executed_edge_p): Likewise.
7077 (combine_predictions_for_bb): Check that profile is nonzero.
7078 (propagate_freq): Do not set frequency.
7079 (drop_profile): Simplify.
7080 (counts_to_freqs): Simplify.
7081 (expensive_function_p): Use to_frequency.
7082 (propagate_unlikely_bbs_forward): Simplify.
7083 (determine_unlikely_bbs): Simplify.
7084 (estimate_bb_frequencies): Add hack to silence graphite issues.
7085 (compute_function_frequency): Use ipa counts.
7086 (pass_profile::execute): Update.
7087 (rebuild_frequencies): Use counts only.
7088 (force_edge_cold): Use counts only.
7089 * profile-count.c (profile_count::dump): Dump new count types.
7090 (profile_count::differs_from_p): Check compatiblity.
7091 (profile_count::to_frequency): New function.
7092 (profile_count::to_cgraph_frequency): New function.
7093 * profile-count.h (struct function): Declare.
7094 (enum profile_quality): Add profile_guessed_local and
7095 profile_guessed_global0.
7096 (class profile_proability): Decrease number of bits to 29;
7097 update from_reg_br_prob_note and to_reg_br_prob_note.
7098 (class profile_count: Update comment; decrease number of bits
7099 to 61. Check compatibility.
7100 (profile_count::compatible_p): New private member function.
7101 (profile_count::ipa_p): New member function.
7102 (profile_count::operator<): Handle global zero correctly.
7103 (profile_count::operator>): Handle global zero correctly.
7104 (profile_count::operator<=): Handle global zero correctly.
7105 (profile_count::operator>=): Handle global zero correctly.
7106 (profile_count::nonzero_p): New member function.
7107 (profile_count::force_nonzero): New member function.
7108 (profile_count::max): New member function.
7109 (profile_count::apply_scale): Handle IPA scalling.
7110 (profile_count::guessed_local): New member function.
7111 (profile_count::global0): New member function.
7112 (profile_count::ipa): New member function.
7113 (profile_count::to_frequency): Declare.
7114 (profile_count::to_cgraph_frequency): Declare.
7115 * profile.c (OVERLAP_BASE): Delete.
7116 (compute_frequency_overlap): Delete.
7117 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7118 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7119 * sched-ebb.c (rank): Use counts only.
7120 * shrink-wrap.c (handle_simple_exit): Use counts only.
7121 (try_shrink_wrapping): Use counts only.
7122 (place_prologue_for_one_component): Use counts only.
7123 * tracer.c (find_best_predecessor): Use to_frequency.
7124 (find_trace): Use to_frequency.
7125 (tail_duplicate): Use to_frequency.
7126 * trans-mem.c (expand_transaction): Do not update frequency.
7127 * tree-call-cdce.c: Do not update frequency.
7128 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7129 (gimple_merge_blocks): Likewise.
7130 (gimple_split_edge): Likewise.
7131 (gimple_duplicate_sese_region): Likewise.
7132 (gimple_duplicate_sese_tail): Likewise.
7133 (move_sese_region_to_fn): Likewise.
7134 (gimple_account_profile_record): Likewise.
7135 (insert_cond_bb): Likewise.
7136 * tree-complex.c (expand_complex_div_wide): Likewise.
7137 * tree-eh.c (lower_resx): Update profile.
7138 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7140 (initialize_cfun): Do not initialize frequencies
7141 (freqs_to_counts): Delete.
7142 (copy_cfg_body): Ignore count parameter.
7143 (copy_body): Update.
7144 (expand_call_inline): Update count_max.
7145 (optimize_inline_calls): Update count_max.
7146 (tree_function_versioning): Update count_max.
7147 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7148 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7150 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7151 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7152 (try_peel_loop): Likewise.
7153 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7155 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7156 (tree_transform_and_unroll_loop): Do not use frequencies
7157 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7158 Use reliable prediction only.
7159 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7160 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7161 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7162 probability scaling.
7163 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7164 not update frequency
7165 (any_remaining_duplicated_blocks): Likewise.
7166 (update_profile): Likewise.
7167 (estimated_freqs_path): Delete.
7168 (freqs_to_counts_path): Delete.
7169 (clear_counts_path): Delete.
7170 (ssa_fix_duplicate_block_edges): Likewise.
7171 (duplicate_thread_path): Likewise.
7172 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7173 * tree-tailcall.c (decrease_profile): Do not update frequency.
7174 (eliminate_tail_call): Likewise.
7175 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7176 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7177 (optimize_mask_stores): Likewise.
7178 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7179 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7180 (ubsan_expand_ptr_ifn): Update profile.
7181 * value-prof.c (gimple_ic): Simplify.
7182 * value-prof.h (gimple_ic): Update prototype.
7183 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7184 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7186 (want_inline_self_recursive_call_p): Likewise.
7187 (resolve_noninline_speculation): Only cummulate defined counts.
7188 (inline_small_functions): Use nonzero_p.
7189 (ipa_inline): Do not access freed node.
7191 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7193 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7194 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7196 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7198 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7199 non-legitimate address.
7201 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7203 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7204 (*lt0_<mode>di, *lt0_<mode>si): New.
7206 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7208 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7209 TARGET_PAIRED_FLOAT.
7211 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7212 Jim Wilson <jim.wilson@linaro.org>
7214 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7215 * config/aarch64/aarch64-tune.md: Regenerated.
7216 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7217 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7219 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7221 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7222 (arc_expand_prologue): Restore blink for millicode.
7223 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7225 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7228 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7229 gen_vec_duplicate after forcing the scalar into a register.
7231 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7233 * combine (try_combine): Print the insns input to try_combine to the
7236 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7239 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7240 Remove second argument from aarch64_process_target_attr call.
7241 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7243 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7245 (aarch64_handle_attr_arch): Remove second argument.
7246 (aarch64_handle_attr_cpu): Ditto.
7247 (aarch64_handle_attr_tune): Ditto.
7248 (aarch64_handle_attr_isa_flags): Ditto.
7249 (aarch64_process_one_target_attr): Ditto.
7250 (aarch64_process_target_attr): Ditto.
7251 (aarch64_option_valid_attribute_p): Remove second argument.
7252 on aarch64_process_target_attr call.
7254 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7256 * diagnostic.c: Include "selftest-diagnostic.h".
7257 (selftest::assert_location_text): New function.
7258 (selftest::test_diagnostic_get_location_text): New function.
7259 (selftest::diagnostic_c_tests): Call it.
7261 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7263 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7264 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7265 (class selftest::test_diagnostic_context): Move to...
7266 * selftest-diagnostic.c: New file.
7267 * selftest-diagnostic.h: New file.
7269 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7271 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7272 offset range for FT32B.
7273 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7274 * config/ft32/ft32.md: Add TARGET_NOPM.
7275 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7276 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7278 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7280 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7282 2017-11-02 Jeff Law <law@redhat.com>
7284 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7285 virtual keyword on FINAL OVERRIDE members.
7287 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7288 virtuals together. Add virtual destructor.
7289 (substitute_and_fold_engine): Similarly.
7291 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7293 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7295 2017-11-02 Richard Biener <rguenther@suse.de>
7297 PR tree-optimization/82795
7298 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7300 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7302 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7304 (gcc_GAS_CHECK_FEATURE): Remove.
7305 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7306 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7307 * configure: Regenerate.
7309 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7311 * config/arc/arc.c (hwloop_optimize): Account for empty
7314 2017-11-02 Richard Biener <rguenther@suse.de>
7317 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7318 Truncate ARRAY_REF index and element size.
7320 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7322 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7324 2017-11-01 Jeff Law <law@redhat.com>
7326 * tree-ssa-ccp.c (ccp_folder): New class derived from
7327 substitute_and_fold_engine.
7328 (ccp_folder::get_value): New member function.
7329 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7330 (ccp_fold_stmt): Remove prototype.
7331 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7332 * tree-ssa-copy.c (copy_folder): New class derived from
7333 substitute_and_fold_engine.
7334 (copy_folder::get_value): Renamed from get_value.
7335 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7336 * tree-vrp.c (vrp_folder): New class derived from
7337 substitute_and_fold_engine.
7338 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7339 (vrp_folder::get_value): New member function.
7340 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7341 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7342 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7343 provide a class interface to folder/substitute routines.
7344 (ssa_prop_fold_stmt_fn): Remove typedef.
7345 (ssa_prop_get_value_fn): Likewise.
7346 (subsitute_and_fold): Remove prototype.
7347 (replace_uses_in): Likewise.
7348 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7349 Renamed from replace_uses_in. Call the virtual member function
7350 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7351 (substitute_and_fold_dom_walker): Remove initialization of
7352 data member entries for calbacks. Add substitute_and_fold_engine
7353 member and initialize it.
7354 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7355 member functions for get_value, replace_phi_args_in c
7356 replace_uses_in, and fold_stmt calls.
7357 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7358 substitute_and_fold. Remove assert. Update ctor call.
7360 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7361 (ssa_prop_visit_phi_fn): Likewise.
7362 (class ssa_propagation_engine): New class to provide an interface
7364 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7366 (ssa_prop_visit_phi): Likewise.
7367 (ssa_propagation_engine::simulate_stmt): Moved into class.
7368 Call visit_phi/visit_stmt from the class rather than via
7369 file scoped static variables.
7370 (ssa_propagation_engine::simulate_block): Moved into class.
7371 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7372 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7373 set file scoped statics for the visit_stmt/visit_phi callbacks.
7374 * tree-complex.c (complex_propagate): New class derived from
7375 ssa_propagation_engine.
7376 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7377 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7378 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7380 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7381 ssa_propagation_engine.
7382 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7383 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7384 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7385 * tree-ssa-copy.c (copy_prop): New class derived from
7386 ssa_propagation_engine.
7387 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7388 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7389 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7390 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7391 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7392 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7393 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7395 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7397 PR rtl-optimization/82778
7398 PR rtl-optimization/82597
7399 * compare-elim.c (struct comparison): Add in_a_setter field.
7400 (find_comparison_dom_walker::before_dom_children): Remove killed
7401 bitmap and df_simulate_find_defs call, instead walk the defs.
7402 Compute last_setter and initialize in_a_setter. Merge definitions
7403 with first initialization for a few variables.
7404 (try_validate_parallel): Use insn_invalid_p instead of
7405 recog_memoized. Return insn rather than just the pattern.
7406 (try_merge_compare): Fix up comment. Don't uselessly test if
7407 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7409 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7413 Alan Hayward <alan.hayward@arm.com>
7414 David Sherwood <david.sherwood@arm.com>
7416 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7417 aarch64_hard_regno_nregs to get the number of registers
7420 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7421 Alan Hayward <alan.hayward@arm.com>
7422 David Sherwood <david.sherwood@arm.com>
7424 * config/aarch64/constraints.md (Upl): Rename to...
7426 * config/aarch64/aarch64.md
7427 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7430 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7431 Alan Hayward <alan.hayward@arm.com>
7432 David Sherwood <david.sherwood@arm.com>
7434 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7435 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7438 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7439 Alan Hayward <alan.hayward@arm.com>
7440 David Sherwood <david.sherwood@arm.com>
7442 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7443 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7444 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7446 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7447 of a single element.
7448 * config/aarch64/iterators.md: Add a comment above the permute
7449 unspecs to say that they are generated directly by
7450 aarch64_expand_vec_perm_const.
7451 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7453 2017-11-01 Nathan Sidwell <nathan@acm.org>
7455 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7457 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7459 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7460 and medany code models, and describe what they do.
7462 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7464 Revert accidental duplicate:
7466 * combine.c (can_change_dest_mode): Reject changes in
7467 REGMODE_NATURAL_SIZE.
7469 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7471 PR rtl-optimization/64682
7472 PR rtl-optimization/69567
7473 PR rtl-optimization/69737
7474 PR rtl-optimization/82683
7475 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7476 register mentioned in the note, drop the note, unless it came from I3,
7477 in which case it should go to I3 again.
7479 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7481 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7482 and return false if not.
7483 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7485 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7487 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7488 range is known to be empty.
7490 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7491 Alan Hayward <alan.hayward@arm.com>
7492 David Sherwood <david.sherwood@arm.com>
7494 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7495 and CONST_VECTOR_NUNITS instead of computing the number of units from
7496 the byte sizes of the vector and element.
7497 (simplify_binary_operation_1): Likewise.
7498 (simplify_const_binary_operation): Likewise.
7499 (simplify_ternary_operation): Likewise.
7501 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7502 Alan Hayward <alan.hayward@arm.com>
7503 David Sherwood <david.sherwood@arm.com>
7505 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7506 (int_mem_offset): ...this new function.
7507 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7508 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7509 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7512 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7513 Alan Hayward <alan.hayward@arm.com>
7514 David Sherwood <david.sherwood@arm.com>
7516 * lower-subreg.c (interesting_mode_p): New function.
7517 (compute_costs, find_decomposable_subregs, decompose_register)
7518 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7519 (resolve_clobber, dump_choices): Use it.
7521 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7522 Alan Hayward <alan.hayward@arm.com>
7523 David Sherwood <david.sherwood@arm.com>
7525 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7527 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7528 Alan Hayward <alan.hayward@arm.com>
7529 David Sherwood <david.sherwood@arm.com>
7531 * alias.c (find_base_value, find_base_term): Only process integer
7532 truncations. Check the precision rather than the size.
7534 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7535 Alan Hayward <alan.hayward@arm.com>
7536 David Sherwood <david.sherwood@arm.com>
7538 * machmode.h (is_narrower_int_mode): New function
7539 * optabs.c (expand_float, expand_fix): Use it.
7540 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7542 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7543 Alan Hayward <alan.hayward@arm.com>
7544 David Sherwood <david.sherwood@arm.com>
7546 * rtl.h (narrower_subreg_mode): New function.
7547 * ira-color.c (update_costs_from_allocno): Use it.
7549 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7550 Alan Hayward <alan.hayward@arm.com>
7551 David Sherwood <david.sherwood@arm.com>
7553 * optabs-query.h (convert_optab_p): New function, split out from...
7554 (convert_optab_handler): ...here.
7555 (widening_optab_handler): Delete.
7556 (find_widening_optab_handler): Remove permit_non_widening parameter.
7557 (find_widening_optab_handler_and_mode): Likewise. Provide an
7558 override that operates on mode class wrappers.
7559 * optabs-query.c (widening_optab_handler): Delete.
7560 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7561 parameter. Assert that the two modes are the same class and that
7562 the "from" mode is narrower than the "to" mode. Use
7563 convert_optab_handler instead of widening_optab_handler.
7564 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7565 instead of widening_optab_handler.
7566 * expr.c (expand_expr_real_2): Update calls to
7567 find_widening_optab_handler.
7568 * optabs.c (expand_widen_pattern_expr): Likewise.
7569 (expand_binop_directly): Take the insn_code as a parameter.
7570 (expand_binop): Only call find_widening_optab_handler for
7571 conversion optabs; use optab_handler otherwise. Update calls
7572 to find_widening_optab_handler and expand_binop_directly.
7573 Use convert_optab_handler instead of widening_optab_handler.
7574 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7575 find_widening_optab_handler and use scalar_mode rather than
7577 (convert_plusminus_to_widen): Likewise.
7579 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7580 Alan Hayward <alan.hayward@arm.com>
7581 David Sherwood <david.sherwood@arm.com>
7583 * machmode.h (fixed_size_mode): New class.
7584 * rtl.h (get_pool_mode): Return fixed_size_mode.
7585 * gengtype.c (main): Add fixed_size_mode.
7586 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7587 (get_raw_arg_mode): Likewise.
7588 * doc/tm.texi: Regenerate.
7589 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7590 * targhooks.c (default_get_reg_raw_mode): Likewise.
7591 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7592 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7593 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7594 (msp430_get_raw_result_mode): Likewise.
7595 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7596 * dbxout.c (dbxout_parms): Require fixed-size modes.
7597 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7598 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7599 * omp-low.c (lower_oacc_reductions): Likewise.
7600 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7601 (simplify_subreg): Update accordingly.
7602 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7603 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7604 that aren't fixed-size.
7605 (get_pool_mode): Return a fixed_size_mode.
7606 (output_constant_pool_2): Take a fixed_size_mode.
7608 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7609 Alan Hayward <alan.hayward@arm.com>
7610 David Sherwood <david.sherwood@arm.com>
7612 * doc/rtl.texi (vec_series): Document.
7613 (const): Say that the operand can be a vec_series.
7614 * rtl.def (VEC_SERIES): New rtx code.
7615 * rtl.h (const_vec_series_p_1): Declare.
7616 (const_vec_series_p): New function.
7617 * emit-rtl.h (gen_const_vec_series): Declare.
7618 (gen_vec_series): Likewise.
7619 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7620 (gen_vec_series): Likewise.
7621 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7622 * simplify-rtx.c (simplify_unary_operation): Handle negations
7624 (simplify_binary_operation_series): New function.
7625 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7626 (test_vector_ops_series): New function.
7627 (test_vector_ops): Call it.
7628 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7629 gen_const_vec_series.
7630 (altivec_lvsr): Likewise.
7631 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7633 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7634 Alan Hayward <alan.hayward@arm.com>
7635 David Sherwood <david.sherwood@arm.com>
7637 * doc/rtl.texi (const): Update description of address constants.
7638 Say that vector constants are allowed too.
7639 * common.md (E, F): Use CONSTANT_P instead of checking for
7641 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7642 checking for CONST_VECTOR.
7643 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7645 * expr.c (expand_expr_real_2): Check for vector modes instead
7646 of checking for CONST_VECTOR.
7647 * rtl.h (const_vec_p): New function.
7648 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7649 (unwrap_const_vec_duplicate): Handle them here too.
7651 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7652 David Malcolm <dmalcolm@redhat.com>
7653 Alan Hayward <alan.hayward@arm.com>
7654 David Sherwood <david.sherwood@arm.com>
7656 * rtl.h (vec_duplicate_p): New function.
7657 * selftest-rtl.c (assert_rtx_eq_at): New function.
7658 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7659 (assert_rtx_eq_at): Declare.
7660 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7661 * selftest-run-tests.c (selftest::run_tests): Call it.
7662 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7663 (simplify_unary_operation_1): Recursively handle vector duplicates.
7664 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7666 (simplify_subreg): Handle subregs of vector duplicates.
7667 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7668 (selftest::simplify_rtx_c_tests): New functions.
7670 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7671 Alan Hayward <alan.hayward@arm.com>
7672 David Sherwood <david.sherwood@arm.com>
7674 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7675 (gen_vec_duplicate): Likewise.
7676 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7678 (gen_const_vector): ...here.
7679 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7680 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7681 whose elements are all equal.
7682 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7683 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7684 (simplify_relational_operation): Likewise.
7685 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7687 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7688 (aarch64_expand_vector_init): Likewise.
7689 * config/arm/arm.c (neon_vdup_constant): Likewise.
7690 (neon_expand_vector_init): Likewise.
7691 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7692 (arm_block_set_unaligned_vect): Likewise.
7693 (arm_block_set_aligned_vect): Likewise.
7694 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7695 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7696 (expand_vec_perm_even_odd_pack): Likewise.
7697 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7698 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7699 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7700 gen_const_vec_duplicate.
7701 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7702 * config/mips/mips.c (mips_gen_const_int_vector): Use
7703 gen_const_vec_duplicate.
7704 (mips_expand_vector_init): Use CONST0_RTX.
7705 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7706 (define_split): Use gen_const_vec_duplicate.
7707 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7708 (define_split): Use gen_const_vec_duplicate.
7709 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7710 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7711 * config/spu/spu.c (spu_const): Likewise.
7713 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7714 Alan Hayward <alan.hayward@arm.com>
7715 David Sherwood <david.sherwood@arm.com>
7717 * combine.c (can_change_dest_mode): Reject changes in
7718 REGMODE_NATURAL_SIZE.
7720 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
7722 * configure.ac (--enable-libssp): New.
7723 (gcc_cv_libc_provides_ssp): Check for explicit setting before
7724 trying to determine target-specific default. Adjust indentation.
7725 * configure: Regenerated.
7726 * doc/install.texi (Configuration): Expand --disable-libssp
7729 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
7731 config/i386/i386.c (ix86_expand_epilogue): Correct stack
7734 2017-10-31 Martin Jambor <mjambor@suse.cz>
7737 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7739 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7741 * auto-profile.c (autofdo_source_profile::read): Use
7742 UNKNOWN_LOCATION rather than 0.
7743 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7744 (warning_at): ...this overload.
7745 (warning_at_rich_loc_n): Rename to...
7746 (warning_n): ...this overload.
7747 (error_at_rich_loc): Rename to...
7748 (error_at): ...this overload.
7749 (pedwarn_at_rich_loc): Rename to...
7750 (pedwarn): ...this overload.
7751 (permerror_at_rich_loc): Rename to...
7752 (permerror): ...this overload.
7753 (inform_at_rich_loc): Rename to...
7754 (inform): ...this overload.
7755 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7756 (diagnostic_n_impl_richloc): Rename to...
7757 (diagnostic_n_impl): ...this rich_location *-based decl.
7758 (inform_at_rich_loc): Rename to...
7759 (inform): ...this, and add an assertion.
7760 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7761 (warning_at_rich_loc): Rename to...
7762 (warning_at): ...this, and add an assertion.
7763 (warning_at_rich_loc_n): Rename to...
7764 (warning_n): ...this, and add an assertion.
7765 (warning_n): Update location_t-based implementation for removal of
7766 location_t-based diagnostic_n_impl.
7767 (pedwarn_at_rich_loc): Rename to...
7768 (pedwarn): ...this, and add an assertion.
7769 (permerror_at_rich_loc): Rename to...
7770 (permerror): ...this, and add an assertion.
7771 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7772 (error_at_rich_loc): Rename to...
7773 (error_at): ...this, and add an assertion.
7774 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7775 (driver::do_spec_on_infiles): Likewise.
7776 * substring-locations.c (format_warning_va): Update for renaming
7777 of inform_at_rich_loc.
7779 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7781 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7782 _Float<N>X built-in functions so that the variant without the
7783 "__builtin_" prefix is only enabled for the GNU C and Objective C
7784 languages when they are in non-strict ANSI/ISO mode.
7785 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7786 * target.def (floatn_builtin_p): Add a target hook to control
7787 whether _Float<N> and _Float<N>X built-in functions without the
7788 "__builtin_" prefix are enabled, and return true for C and
7789 Objective C in the default hook. Include langhooks.h in
7791 * targhooks.h (default_floatn_builtin_p): Likewise.
7792 * targhooks.c (default_floatn_builtin_p): Likewise.
7793 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7794 floatn_builtin_p target hook.
7795 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7797 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7798 Eric Botcazou <ebotcazou@adacore.com>
7800 PR rtl-optimization/81803
7801 * lra-constraints.c (curr_insn_transform): Also reload the whole
7802 register for a strict subreg no wider than a word if this is for
7803 a WORD_REGISTER_OPERATIONS target.
7805 2017-10-31 Jason Merrill <jason@redhat.com>
7807 * gdbinit.in: Skip over inlines from timevar.h.
7809 2017-10-31 Martin Liska <mliska@suse.cz>
7811 * doc/gcov.texi: Document new option.
7812 * gcov.c (print_usage): Likewise print it.
7813 (process_args): Support the argument.
7814 (format_count): New function.
7815 (format_gcov): Use the function.
7817 2017-10-31 Martin Liska <mliska@suse.cz>
7819 * gcov.c (struct name_map): do not use typedef.
7820 Define operator== and operator<.
7821 (name_search): Remove.
7822 (name_sort): Remove.
7823 (main): Do not allocate names.
7824 (process_file): Add vertical space.
7825 (generate_results): Use std::find.
7826 (release_structures): Do not release memory.
7827 (find_source): Use std::find.
7829 2017-10-31 Martin Liska <mliska@suse.cz>
7831 * gcov.c (struct line_info): Remove it's typedef.
7832 (line_info::line_info): Add proper ctor.
7833 (line_info::has_block): Do not use a typedef.
7834 (struct source_info): Do not use typedef.
7835 (circuit): Likewise.
7836 (get_cycles_count): Likewise.
7837 (output_intermediate_file): Iterate via vector iterator.
7838 (add_line_counts): Use std::vector methods.
7839 (accumulate_line_counts): Likewise.
7840 (output_lines): Likewise.
7842 2017-10-31 Martin Liska <mliska@suse.cz>
7844 * gcov.c (struct source_info): Remove typedef.
7845 (source_info::source_info): Add proper ctor.
7846 (accumulate_line_counts): Use struct, not it's typedef.
7847 (output_gcov_file): Likewise.
7848 (output_lines): Likewise.
7849 (main): Do not allocate an array.
7850 (output_intermediate_file): Use size of vector container.
7851 (process_file): Resize the vector.
7852 (generate_results): Do not preallocate, use newly added vector
7854 (release_structures): Do not release sources.
7855 (find_source): Use vector methods.
7856 (add_line_counts): Do not use typedef.
7858 2017-10-31 Martin Liska <mliska@suse.cz>
7860 * doc/gcov.texi: Document that.
7861 * gcov.c (add_line_counts): Mark lines with a non-executed
7863 (output_line_beginning): Handle such lines.
7864 (output_lines): Pass new argument.
7865 (output_intermediate_file): Print it in intermediate format.
7867 2017-10-31 Martin Liska <mliska@suse.cz>
7869 * color-macros.h: New file.
7870 * diagnostic-color.c: Factor out color related to macros to
7872 * doc/gcov.texi: Document -k option.
7873 * gcov.c (INCLUDE_STRING): Include string.h.
7874 (print_usage): Add -k option.
7875 (process_args): Parse it.
7876 (pad_count_string): New function.
7877 (output_line_beginning): Likewise.
7878 (DEFAULT_LINE_START): New macro.
7879 (output_lines): Support color output.
7881 2017-10-31 Martin Liska <mliska@suse.cz>
7883 PR gcov-profile/82633
7884 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7885 their interaction with GCOV infrastructure.
7886 * configure.ac: Add -fkeep-{inline,static}-functions to
7888 * configure: Regenerate.
7890 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7893 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7895 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7898 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7899 into a register if it does not fit into an immediate offset field.
7901 2017-10-31 Olivier Hainque <hainque@adacore.com>
7903 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7905 2017-10-31 Julia Koval <julia.koval@intel.com>
7907 * config.gcc: Add gfniintrin.h.
7908 * config/i386/gfniintrin.h: New.
7909 * config/i386/i386-builtin-types.def
7910 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7911 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7912 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7913 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7914 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7915 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7916 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7917 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7918 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7919 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7920 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7921 * config/i386/immintrin.h: Include gfniintrin.h.
7922 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7924 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7926 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7928 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7930 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7931 (arm_ashldi3_1bit): Remove pattern.
7932 (ashrdi3): Remove shift by 1 expansion.
7933 (arm_ashrdi3_1bit): Remove pattern.
7934 (lshrdi3): Remove shift by 1 expansion.
7935 (arm_lshrdi3_1bit): Remove pattern.
7936 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7937 cost of ashldi3 by 1.
7938 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7939 (<shift>di3_neon): Likewise.
7941 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7943 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7944 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7945 and (*aarch64_simd_mov<VQ:mode>).
7946 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7947 pattern alternative.
7948 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7949 attributes to match pattern alternative.
7951 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7953 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7954 * config/rs6000/emmintrin.h: New file.
7955 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7957 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7959 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7960 * (movdi_vfp_cortexa8): Remove pattern.
7962 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7964 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7966 (Specific, alpha*-dec-osf5.1): Remove.
7967 (Specific, mips-sgi-irix5): Remove.
7968 (Specific, mips-sgi-irix6): Remove.
7970 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7973 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7974 arguments to clear_bit_region_be.
7976 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7978 * gimplify.c: Include memmodel.h.
7980 2017-10-30 Martin Jambor <mjambor@suse.cz>
7982 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7983 condition whether loop should be executed at all.
7985 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7987 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7988 gimple folding of vec_madd() intrinsics.
7989 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7990 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7991 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7993 2017-10-30 Richard Biener <rguenther@suse.de>
7995 PR tree-optimization/82762
7997 2017-10-23 Richard Biener <rguenther@suse.de>
7999 PR tree-optimization/82129
8001 2017-08-01 Richard Biener <rguenther@suse.de>
8003 PR tree-optimization/81181
8004 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8005 (compute_antic): ... end of iteration here.
8007 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8009 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8010 -std=iso9899:2017 and -std=gnu17.
8011 * doc/standards.texi (C Language): Document C17 support.
8012 * doc/cpp.texi (Overview): Mention -std=c17.
8013 (Standard Predefined Macros): Document C11 and C17 values of
8014 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8015 * doc/extend.texi (Inline): Do not list individual options for
8016 standards newer than C99.
8017 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8019 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8022 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8024 * asan.c (asan_finish_file): Align asan globals array by shadow
8027 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8030 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8031 (struct store_immediate_info): Add bitregion_start and bitregion_end
8033 (store_immediate_info::store_immediate_info): Add brs and bre
8034 arguments and initialize bitregion_{start,end} from those.
8035 (struct merged_store_group): Add bitregion_start, bitregion_end,
8036 align_base and mask fields. Drop unnecessary struct keyword from
8037 struct store_immediate_info. Add do_merge method.
8038 (clear_bit_region_be): Use memset instead of loop storing zeros.
8039 (merged_store_group::do_merge): New method.
8040 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8041 stores as long as the surrounding bitregions have no gaps.
8042 (merged_store_group::merge_overlapping): Use do_merge.
8043 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8044 is byte aligned, rather than requiring that start and width are
8045 byte aligned. Drop unnecessary struct keyword from
8046 struct store_immediate_info. Allocate and populate also mask array.
8047 Make start of the arrays relative to bitregion_start rather than
8048 start and size them according to bitregion_{end,start} difference.
8049 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8050 struct store_immediate_info.
8051 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8052 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8053 struct keyword from struct store_immediate_info.
8054 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8055 between stores as long as the surrounding bitregions have no gaps.
8057 (struct split_store): Add orig non-static data member.
8058 (split_store::split_store): Initialize orig to false.
8059 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8060 if there is exactly a single original stmt. Change stmts argument
8061 to pointer from reference, if NULL, don't push anything to it. Add
8062 first argument, use it to optimize skipping over orig stmts that
8063 are known to be before bitpos already. Simplify.
8064 (split_group): Return unsigned int count how many stores are or
8065 would be needed rather than a bool. Add allow_unaligned argument.
8066 Change split_stores argument from reference to pointer, if NULL,
8067 only do a dry run computing how many stores would be produced.
8068 Rewritten algorithm to use both alignment and misalign if
8069 !allow_unaligned and handle bitfield stores with gaps.
8070 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8071 from bitregion_start instead of start. Compute allow_unaligned
8072 here, if true, do 2 split_group dry runs to compute which one
8073 produces fewer stores and prefer aligned if equal. Punt if
8074 new count is bigger or equal than original before emitting any
8075 statements, rather than during that. Remove no longer needed
8076 new_ssa_names tracking. Replace num_stmts with
8077 split_stores.length (). Use 32-bit stack allocated entries
8078 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8079 if possible. Handle bitfields with gaps.
8080 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8081 Compute bitregion_{start,end} for the stores and construct
8082 store_immediate_info with that. Formatting fixes.
8084 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8087 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8088 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8090 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8092 * gimplify.c: Include tm_p.h.
8094 * common.opt (gcoff): Re-add as ignored option.
8095 (gcoff1, gcoff2, gcoff3): Likewise.
8097 * Makefile.in (OBJS): Delete sdbout.o.
8098 (GTFILES): Delete $(srcdir)/sdbout.c.
8099 * debug.h: Delete sdb_debug_hooks.
8100 * final.c: Delete sdbout.h include.
8101 (final_scan_insn): Delete SDB_DEBUG check.
8102 (rest_of_clean_state): Likewise.
8103 * output.h: Delete sdb_begin_function_line.
8106 * toplev.c: Delete sdbout.h include.
8107 (process_options): Delete SDB_DEBUG check.
8108 * tree-core.h (tree_type_common): Delete pointer field of
8110 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8111 TYPE_SYMTAB_POINTER.
8112 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8113 (TYPE_SYMTAB_IS_POINTER): Delete.
8114 (TYPE_SYMTAB_IS_DIE): Renumber.
8115 * xcoffout.c: Refer to former sdbout.c file.
8116 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8118 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8119 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8120 (Debugging Options): Delete -gcoff.
8121 (-gstabs): Delete SDB reference.
8123 (-gcoff@var{level}): Delete.
8124 * doc/passes.texi (Debugging information output): Delete SDB and
8125 sdbout.c references.
8126 * doc/tm.texi: Regenerate.
8127 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8128 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8130 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8131 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8132 (SDB_DEBUGGING_INFO): Delete.
8133 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8134 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8135 * target.def (output_source_filename): Delete COFF reference.
8137 * common.opt (gcoff): Delete.
8138 (gxcoff+): Update Negative chain.
8139 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8141 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8142 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8143 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8145 (expand_function_start): Change sdb reference to past tense.
8146 (expand_function_end): Change sdb reference to past tense.
8147 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8148 * opts.c (debug_type_names): Delete coff entry.
8149 (common_handle_option): Delete OPT_gcoff case.
8150 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8152 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8153 * config/cris/cris.h: Delete SDB reference in comment.
8154 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8155 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8156 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8157 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8159 (ix86_expand_prologue): Likewise.
8160 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8161 * config/ia64/ia64.h: Likewise.
8162 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8163 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8165 * config/mmix/mmix.h: Likewise.
8166 * config/nds32/nds32.c: Likewise.
8167 * config/stormy/storym16.h: Likewise.
8168 * config/visium/visium.h: Likewise.
8169 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8171 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8173 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8174 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8175 FRAME_POINTER_REGNUM point at high end of local var area.
8177 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8179 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8180 Move comment around. Do not reset best_edge for a copiable
8181 destination if the copy would cause a partition change.
8182 (better_edge_p): Remove redundant check.
8184 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8186 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8188 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8190 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8191 for math functions that have _Float<N> and _Float<N>X variants.
8192 (mathfn_built_in_2): Add support for math functions that have
8193 _Float<N> and _Float<N>X variants.
8194 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8195 (expand_builtin_mathfn_ternary): Add support for fma with
8196 _Float<N> and _Float<N>X variants.
8197 (expand_builtin): Likewise.
8198 (fold_builtin_3): Likewise.
8199 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8200 create math function _Float<N> and _Float<N>X variants as external
8202 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8203 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8204 the __builtin_ prefix and if not strict ansi, without the prefix.
8205 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8206 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8207 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8208 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8209 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8210 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8211 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8212 function signatures for fma _Float<N> and _Float<N>X variants.
8213 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8214 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8215 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8216 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8217 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8218 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8219 * gencfn-macros.c (print_case_cfn): Add support for math functions
8220 that have _Float<N> and _Float<N>X variants.
8221 (print_define_operator_list): Likewise.
8222 (fltfn_suffixes): Likewise.
8224 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8225 for math functions that have _Float<N> and _Float<N>X variants.
8226 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8227 and _Float<N>X variants.
8228 (COPYSIGN): Likewise.
8231 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8232 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8234 (integer_valued_read_call_p): Likewise.
8235 * fold-const-call.c (fold_const_call_ss): Likewise.
8236 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8237 _Float<N> and _Float<N>X variants.
8238 (fold_const_call_ssss): Add support for fma _Float<N> and
8239 _Float<N>X variants.
8240 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8241 support for copysign and fma _Float<N> and _Float<N>X variants.
8242 (backprop::process_builtin_call_use): Likewise.
8243 * tree-call-cdce.c (can_test_argument_range); Add support for
8244 sqrt _Float<N> and _Float<N>X variants.
8245 (edom_only_function): Likewise.
8246 (get_no_error_domain): Likewise.
8247 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8248 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8249 copysign _Float<N> and _Float<N>X variants.
8250 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8251 handled by machine independent code.
8252 (FMAF128): Likewise.
8253 * doc/cpp.texi (Common Predefined Macros): Document defining
8254 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8255 fma _Float<N> and _Float<N>X variants.
8257 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8260 * config/i386/i386-modes.def (CCFPU): Remove definition.
8261 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8262 (ix86_cc_modes_compatible): Ditto.
8263 (ix86_expand_carry_flag_compare): Ditto.
8264 (ix86_expand_int_movcc): Ditto.
8265 (ix86_expand_int_addcc): Ditto.
8266 (ix86_reverse_condition): Ditto.
8267 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8268 Return true/false for unordered/ordered fp comparisons.
8269 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8270 (ix86_prepare_fp_compare_args): Update for rename.
8271 (ix86_expand_fp_compare): Update for rename. Generate unordered
8272 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8273 (ix86_expand_sse_compare_and_jump): Ditto.
8274 * config/i386/predicates.md (fcmov_comparison_operator):
8275 Remove CCFPU mode handling.
8276 (ix86_comparison_operator): Ditto.
8277 (ix86_carry_flag_operator): Ditto.
8278 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8279 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8280 (*cmpu<mode>_cc_i387): Ditto.
8281 (FPCMP): Remove mode iterator.
8282 (unord): Remove mode attribute.
8283 (unord_subst): New define_subst transformation
8284 (unord): New define_subst attribute.
8286 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8287 (*cmpi<unord>xf_i387): Ditto.
8288 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8289 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8290 using unord_subst transformation.
8291 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8292 (round_saeonly): Also handle CCFP mode.
8293 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8294 Remove UNSPEC_SAHF unspec handling.
8296 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8298 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8300 2017-10-27 Jeff Law <law@redhat.com>
8302 * gimple-ssa-sprintf.c: Include domwalk.h.
8303 (class sprintf_dom_walker): New class, derived from dom_walker.
8304 (sprintf_dom_walker::before_dom_children): New function.
8305 (struct call_info): Moved into sprintf_dom_walker class
8306 (compute_formath_length, handle_gimple_call): Likewise.
8307 (sprintf_length::execute): Call the dominator walker rather
8308 than walking the statements.
8310 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8311 gimple statement locations.
8312 (check_array_bounds): Corresponding changes. Get the statement's
8313 location directly from wi->stmt.
8315 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8318 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8320 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8322 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8323 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8325 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8328 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8329 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8330 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8331 maybe_get_pool_constant.
8332 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8335 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8337 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8338 2.26 caveat. Update gas and gld versions.
8339 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8342 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8344 * cgraph.h (set_malloc_flag): Declare.
8345 * cgraph.c (set_malloc_flag_1): New function.
8346 (set_malloc_flag): Likewise.
8347 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8348 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8350 (read_ipa_call_summary): Add support for reading is_return_callee.
8351 (write_ipa_call_summary): Stream is_return_callee.
8352 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8353 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8354 ipa-prop.h, ipa-fnsummary.h.
8355 (pure_const_names): Change to static.
8356 (malloc_state_e): Define.
8357 (malloc_state_names): Define.
8358 (funct_state_d): Add field malloc_state.
8359 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8360 (check_retval_uses): New function.
8361 (malloc_candidate_p): Likewise.
8362 (analyze_function): Add support for malloc attribute.
8363 (pure_const_write_summary): Stream malloc_state.
8364 (pure_const_read_summary): Add support for reading malloc_state.
8365 (dump_malloc_lattice): New function.
8366 (propagate_malloc): New function.
8367 (warn_function_malloc): New function.
8368 (ipa_pure_const::execute): Call propagate_malloc and
8369 ipa_free_fn_summary.
8370 (pass_local_pure_const::execute): Add support for malloc attribute.
8371 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8372 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8374 2017-10-27 Martin Liska <mliska@suse.cz>
8376 PR gcov-profile/82457
8377 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8378 for fork-like functions to be properly instrumented.
8380 2017-10-27 Richard Biener <rguenther@suse.de>
8383 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8384 info when we redirected EH.
8386 2017-10-26 Michael Collison <michael.collison@arm.com>
8388 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8390 (<optab>_trunchf<GPI:mode>2: New pattern.
8391 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8392 * config/aarch64/iterators.md (wv): New mode attribute.
8393 (vf, VF): New mode attributes.
8394 (vgp, VGP): New mode attributes.
8395 (s): Update attribute with SImode and DImode prefixes.
8397 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8399 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8400 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8401 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8402 (nios2_option_overide): Initialize it. Don't allow R0-relative
8403 addressing with PIC.
8404 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8405 (nios2_symbolic_constant_p): Likewise.
8406 (nios2_legitimate_address_p): Likewise.
8407 (nios2_r0rel_section_name_p): New.
8408 (nios2_symbol_ref_in_r0rel_data_p): New.
8409 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8410 (r0rel_constant_p): New.
8411 (nios2_print_operand_address): Handle r0rel_constant_p.
8412 (nios2_cdx_narrow_form_p): Likewise.
8413 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8414 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8415 (Nios II Options): Document -mr0rel-sec.
8417 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8419 * config/nios2/nios2.c: Include xregex.h.
8420 (nios2_gprel_sec_regex): New.
8421 (nios2_option_overide): Initialize it. Don't allow GP-relative
8422 addressing with PIC.
8423 (nios2_small_section_name_p): Check for regex match.
8424 * config/nios2/nios2.opt (mgprel-sec=): New option.
8425 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8426 (Nios II Options): Document -mgprel-sec.
8428 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8430 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8432 2017-10-26 Tom de Vries <tom@codesourcery.com>
8434 PR tree-optimization/82707
8435 * gimple.c (gimple_copy): Fix unsharing of
8436 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8438 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8440 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8441 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8442 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8443 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8444 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8445 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8446 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8447 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8448 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8449 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8450 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8451 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8452 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8453 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8454 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8455 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8456 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8458 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8460 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8462 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8463 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8464 warning to rs6000.c. Remove the Undocumented flag, since it has
8466 (-mabi=ibmlongdouble): Likewise.
8467 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8468 already set, set the default format for long double.
8469 (rs6000_debug_reg_global): Print whether long double is IBM or
8471 (rs6000_option_override_internal): Rework setting long double
8472 format. Only warn if the user is changing the long double default
8473 and they did not use -Wno-psabi.
8474 * doc/invoke.texi (PowerPC options): Update the documentation for
8475 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8477 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8478 Alan Hayward <alan.hayward@arm.com>
8479 David Sherwood <david.sherwood@arm.com>
8481 * rtl.h (wider_subreg_mode): New function.
8482 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8483 rather than an unsigned int *.
8484 * ira-color.c (regno_max_ref_width): Replace with...
8485 (regno_max_ref_mode): ...this new variable.
8486 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8487 Use wider_subreg_mode.
8488 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8489 rather than an unsigned int *.
8490 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8491 (process_alt_operands): Likewise.
8492 (invariant_p): Likewise.
8493 * lra-spills.c (assign_mem_slot): Likewise.
8494 (add_pseudo_to_slot): Likewise.
8495 * lra.c (collect_non_operand_hard_regs): Likewise.
8496 (add_regs_to_insn_regno_info): Likewise.
8497 * reload1.c (regno_max_ref_width): Replace with...
8498 (regno_max_ref_mode): ...this new variable.
8499 (reload): Update accordingly. Update call to
8500 ira_sort_regnos_for_alter_reg.
8501 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8502 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8503 (scan_paradoxical_subregs): Likewise.
8505 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8507 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8508 (aarch64_frame): Add emit_frame_chain boolean.
8509 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8510 Move eh_return case to aarch64_layout_frame.
8511 (aarch64_layout_frame): Initialize emit_frame_chain.
8512 (aarch64_expand_prologue): Use emit_frame_chain.
8514 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8516 * config/aarch64/aarch64.c (aarch64_layout_frame):
8517 Ensure LR is always stored at the bottom of the callee-saves.
8518 Remove rarely used frame layout which saves callee-saves at top of
8519 frame, so the store of LR can be used as a valid probe in all cases.
8521 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8523 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8524 Improve unaligned TImode/TFmode base/offset split.
8526 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8527 Alan Hayward <alan.hayward@arm.com>
8528 David Sherwood <david.sherwood@arm.com>
8530 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8531 * combine.c (find_single_use_1): Likewise.
8532 (expand_field_assignment): Likewise.
8533 (move_deaths): Likewise.
8534 * lra-constraints.c (simplify_operand_subreg): Likewise.
8535 (curr_insn_transform): Likewise.
8536 * lra.c (collect_non_operand_hard_regs): Likewise.
8537 (add_regs_to_insn_regno_info): Likewise.
8538 * rtlanal.c (reg_referenced_p): Likewise.
8539 (covers_regno_no_parallel_p): Likewise.
8541 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8543 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8544 Don't print any bits outside the precision of the value.
8545 * wide-int.cc (test_printing): Add some new tests.
8547 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8549 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8550 supports -xbrace_comment option.
8551 * configure: Regenerate.
8552 * config.in: Regenerate.
8553 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8554 (ASM_CPU_SPEC): Use it.
8556 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8558 * target.def (static_rtx_alignment): New hook.
8559 * targhooks.h (default_static_rtx_alignment): Declare.
8560 * targhooks.c (default_static_rtx_alignment): New function.
8561 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8562 * doc/tm.texi: Regenerate.
8563 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8564 instead of targetm.constant_alignment. Remove call to
8566 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8567 (cris_preferred_mininum_alignment): New function, split out from...
8568 (cris_constant_alignment): ...here.
8569 (cris_static_rtx_alignment): New function.
8570 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8572 (ix86_constant_alignment): ...here.
8573 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8574 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8575 (mmix_static_rtx_alignment): New function.
8576 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8577 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8579 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8582 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8583 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8585 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8587 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8588 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8589 mark as a sign-extending load.
8590 (local_pic_load_u): Define.
8592 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8595 * fold-const.c (operand_equal_for_comparison_p): Also return true
8596 if ARG0 is a simple variant of ARG1 with narrower precision.
8597 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8599 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8601 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8603 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8604 scatter_static, scatter_per_elt.
8605 * x86-tune-costs.h: Add new cost entries.
8607 2017-10-25 Richard Biener <rguenther@suse.de>
8609 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8610 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8611 class pass_fre): Move to ...
8612 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8613 class pass_fre): ... here and adjust for statistics.
8615 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8618 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8619 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8620 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8621 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8624 2017-10-25 Richard Biener <rguenther@suse.de>
8626 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8627 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8628 eliminate_push_avail, eliminate_insert): Move inside...
8629 (class eliminate_dom_walker): ... this class in preparation
8631 (fini_eliminate): Remove by merging with ...
8632 (eliminate): ... this function. Adjust for class changes.
8633 (pass_pre::execute): Remove fini_eliminate call.
8634 (pass_fre::execute): Likewise.
8636 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8639 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8640 (VPERMI2, VPERMI2I): New mode iterators.
8641 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8642 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8644 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8645 mode iterator. Remove 3 old define_insn patterns.
8646 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8647 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8648 VPERMI2 mode iterator, remove the other two expanders.
8649 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8650 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8651 instructions, remove the other two patterns.
8652 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8653 mode iterator, remove the other two patterns.
8654 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8655 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8656 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8657 and adjust argument order accordingly.
8658 (ix86_expand_vec_perm): Adjust caller.
8659 (expand_vec_perm_1): Likewise.
8660 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8661 (expand_vec_perm_vpermt2_vpshub2): ... this.
8662 (ix86_expand_vec_perm_const_1): Adjust caller.
8663 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8666 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8667 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8668 (vec_shl_<mode>): Remove unused expander.
8669 (avx512bw_<shift_insn><mode>3): New define_insn.
8670 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8671 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8673 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8676 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8679 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8681 PR rtl-optimization/82396
8682 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8683 (autopref_multipass_init): Simplify initialization.
8684 (autopref_rank_data): Simplify sort order.
8685 * gcc/sched-int.h (autopref_multipass_data_): Remove
8686 multi_mem_insn_p, min_offset and max_offset.
8688 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8691 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8692 Check special value of flag_omit_frame_pointer.
8693 (aarch64_can_eliminate): Likewise.
8694 (aarch64_override_options_after_change_1): Simplify handling of
8695 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8697 2017-10-24 Richard Biener <rguenther@suse.de>
8699 PR tree-optimization/82697
8700 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8701 zero for conditional load and unconditional store.
8703 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8705 * doc/install.texi: Document bootstrap-cet.
8707 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8710 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8711 ENDBR instruction at function entrance if function is only
8714 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8717 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
8718 patterns to better describe from which operation the CF is computed.
8719 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
8720 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
8721 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
8722 is 0, use _0 suffixed expanders instead of emitting a comparison
8725 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
8727 * config/i386/i386.md(*movsf_internal, *movdf_internal):
8728 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
8730 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
8733 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
8734 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8735 * loop-iv.c (iv_get_reaching_def): Likewise.
8736 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8737 variable is promoted and the partition contains undefined values.
8739 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8741 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8743 (nios2_address_cost): Define.
8744 (nios2_legitimize_address): Recognize (exp + constant) directly.
8745 (TARGET_ADDRESS_COST): Define.
8747 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8749 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8750 (nios2_symbolic_memory_operand_p): Declare.
8751 (nios2_split_large_constant): Declare.
8752 (nios2_split_symbolic_memory_operand): Declare.
8753 * config/nios2/nios2.c: Adjust includes.
8754 (nios2_symbolic_constant_allowed): New.
8755 (nios2_symbolic_constant_p): New.
8756 (nios2_plus_symbolic_constant_p): New.
8757 (nios2_valid_addr_expr_p): Recognize addresses involving
8759 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8760 (nios2_symbolic_memory_operand_p): New.
8761 (nios2_large_constant_p): New.
8762 (nios2_split_large_constant): New.
8763 (nios2_split_plus_large_constant): New.
8764 (nios2_split_symbolic_memory_operand): New.
8765 (nios2_legitimize_address): Code refactoring. Handle addresses
8766 involving symbolic constants.
8767 (nios2_emit_move_sequence): Likewise.
8768 (nios2_print_operand): Improve error output.
8769 (nios2_print_operand_address): Handle LO_SUM.
8770 (nios2_cdx_narrow_form_p): Likewise.
8771 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8772 operands involving symbolic constants.
8773 (movhi_internal, movsi_internal): Likewise.
8774 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8775 (extendhisi2, extendqi<mode>2): Likewise.
8777 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8779 * tree-pass.h (PROP_rtl_split_insns): Define.
8780 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8782 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8784 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8786 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8789 * target.def (const_not_ok_for_debug_p): Default to
8790 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8791 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8792 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8793 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8794 which targetm.const_not_ok_for_debug_p returned true.
8795 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8797 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8799 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8800 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8801 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8803 (ix86_const_not_ok_for_debug_p): New function.
8804 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8805 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8807 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8810 * system.h: Conditionally include "unique-ptr.h" if
8811 INCLUDE_UNIQUE_PTR is defined.
8812 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8813 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8815 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8817 * config/rl78/rl78.md: New define_expand "subdi3".
8819 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8822 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8823 DF_REF_INSN if DF_REF_INSN_INFO is false.
8825 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8827 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8828 xmm_move instead of sse_move.
8829 (sse_store_index): New function.
8830 (ix86_register_move_cost): Be more sensible about mismatch stall;
8831 model AVX moves correctly; make difference between sse->integer and
8833 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8834 moves; make difference between SSE and AVX.
8835 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8836 and zmm_move. Increase size of sse load and store tables;
8837 add unaligned load and store tables; add ssemmx_to_integer.
8838 * x86-tune-costs.h: Update all entries according to real
8839 move latencies from Agner Fog's manual and chip documentation.
8841 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8844 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8845 * config/i386/constraints.md (Wf): New constraint.
8846 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8847 (cmp<dwi>_doubleword): Removed.
8848 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8849 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8850 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8851 expand with cmp<dwi>_doubleword. For LTU and GEU use
8852 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8854 * common.opt (gcolumn-info): Enable by default.
8855 * doc/invoke.texi (gcolumn-info): Document new default.
8857 2017-10-23 Richard Biener <rguenther@suse.de>
8859 PR tree-optimization/82672
8860 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8861 Fold the stmt if we propagated into it.
8863 2017-10-23 Richard Biener <rguenther@suse.de>
8865 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8866 (bitmap_remove_expr_from_set): ... this. All callers call this
8867 for non-constant values.
8868 (bitmap_set_subtract): Rename to...
8869 (bitmap_set_subtract_expressions): ... this. Adjust and
8871 (bitmap_set_contains_value): Remove superfluous check.
8872 (bitmap_set_replace_value): Inline into single caller ...
8873 (bitmap_value_replace_in_set): ... here and simplify.
8874 (dependent_clean): Merge into ...
8875 (clean): ... this using an overload. Adjust.
8876 (prune_clobbered_mems): Adjust.
8877 (compute_antic_aux): Likewise.
8878 (compute_partial_antic_aux): Likewise.
8880 2017-10-23 Richard Biener <rguenther@suse.de>
8882 PR tree-optimization/82129
8884 2017-08-01 Richard Biener <rguenther@suse.de>
8886 PR tree-optimization/81181
8887 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8888 (compute_antic): ... end of iteration here.
8890 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8892 * target.def (starting_frame_offset): New hook.
8893 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8894 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8895 * doc/tm.texi.in: Regenerate.
8896 * hooks.h (hook_hwi_void_0): Declare.
8897 * hooks.c (hook_hwi_void_0): New function.
8898 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8899 STARTING_FRAME_OFFSET.
8900 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8901 * reload1.c (reload): Likewise.
8902 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8903 instead of STARTING_FRAME_OFFSET.
8904 * function.c (try_fit_stack_local): Likewise.
8905 (assign_stack_local_1): Likewise
8906 (instantiate_virtual_regs): Likewise.
8907 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8908 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8909 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8910 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8911 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8912 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8913 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8914 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8915 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8916 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8917 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8918 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8919 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8920 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8921 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8922 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8923 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8924 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8925 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8926 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8927 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8928 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8929 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8930 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8931 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8932 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8933 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8934 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8935 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8936 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8937 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8938 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8939 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8940 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8941 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8942 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8943 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8944 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8945 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8946 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8947 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8948 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8949 return a HOST_WIDE_INT.
8950 (avr_builtin_setjmp_frame_value): Use it instead of
8951 STARTING_FRAME_OFFSET.
8952 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8953 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8954 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8956 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8957 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8958 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8959 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8960 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8961 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8962 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8963 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8964 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8965 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8966 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8967 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8969 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8970 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8971 * config/mips/mips.c (mips_compute_frame_info): Refer to
8972 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8973 (mips_starting_frame_offset): New function.
8974 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8975 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8976 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8977 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8978 and return a HOST_WIDE_INT.
8979 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8980 (mmix_initial_elimination_offset): Refer to
8981 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8982 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8983 * config/pa/pa.c (pa_starting_frame_offset): New function.
8984 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8985 (pa_expand_prologue): Likewise.
8986 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8987 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8988 !FRAME_GROWS_DOWNWARD handling to...
8989 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8990 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8991 !FRAME_GROWS_DOWNWARD handling to...
8992 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8993 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8994 !FRAME_GROWS_DOWNWARD handling to...
8995 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8996 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8998 (rs6000_starting_frame_offset): New function.
8999 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9000 !FRAME_GROWS_DOWNWARD handling to...
9001 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9002 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9003 !FRAME_GROWS_DOWNWARD handling to...
9004 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9005 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9006 !FRAME_GROWS_DOWNWARD handling to...
9007 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9008 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9009 (rs6000_starting_frame_offset): New function.
9010 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9011 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9012 * config/vax/vax.c (vax_starting_frame_offset): New function.
9013 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9014 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9015 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9016 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9017 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9018 * system.h (STARTING_FRAME_OFFSET): Poison.
9020 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9022 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9023 SCALAR_TYPE_MODE instead of TYPE_MODE.
9025 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9026 Alan Hayward <alan.hayward@arm.com>
9027 David Sherwood <david.sherwood@arm.com>
9029 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9031 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9032 Alan Hayward <alan.hayward@arm.com>
9033 David Sherwood <david.sherwood@arm.com>
9035 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9037 2017-10-23 Richard Biener <rguenther@suse.de>
9039 PR tree-optimization/82129
9040 * tree-ssa-pre.c (bitmap_set_and): Remove.
9041 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9042 canonicalizing expressions in the set to those with lowest
9043 ID rather than taking that from the first edge.
9045 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9047 * combine.c (rtx_equal_for_field_assignment_p): Use
9048 byte_lowpart_offset.
9050 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9051 Alan Hayward <alan.hayward@arm.com>
9052 David Sherwood <david.sherwood@arm.com>
9054 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9055 to a SUBREG_PROMOTED_VAR.
9057 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9058 Alan Hayward <alan.hayward@arm.com>
9059 David Sherwood <david.sherwood@arm.com>
9061 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9062 (expand_debug_source_expr): Likewise.
9063 * combine.c (combine_simplify_rtx): Likewise.
9064 * cse.c (fold_rtx): Likewise.
9065 * optabs.c (expand_float): Likewise.
9066 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9067 (simplify_binary_operation_1): Likewise.
9069 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9070 Alan Hayward <alan.hayward@arm.com>
9071 David Sherwood <david.sherwood@arm.com>
9073 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9074 (record_promoted_value): Likewise.
9075 * expr.c (expand_expr_real_2): Likewise.
9076 * ree.c (update_reg_equal_equiv_notes): Likewise.
9077 (combine_set_extension): Likewise.
9078 * rtlanal.c (low_bitmask_len): Likewise.
9079 * simplify-rtx.c (neg_const_int): Likewise.
9080 (simplify_binary_operation_1): Likewise.
9082 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9083 Alan Hayward <alan.hayward@arm.com>
9084 David Sherwood <david.sherwood@arm.com>
9086 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9087 * regcprop.c (maybe_mode_change): Likewise.
9088 * reload1.c (alter_reg): Likewise.
9090 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9092 * inchash.h (inchash::hash::add_wide_int): New function.
9093 * lto-streamer-out.c (hash_tree): Use it.
9095 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9097 * inchash.h (inchash::hash::add_wide_int): Rename to...
9098 (inchash::hash::add_hwi): ...this.
9099 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9100 (polymorphic_call_target_hasher::hash): Likewise.
9101 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9102 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9103 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9104 * lto-streamer-out.c (hash_tree): Likewise.
9105 * optc-save-gen.awk: Likewise.
9106 * tree.c (add_expr): Likewise.
9108 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9111 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9112 for ordered inequality comparisons even with TARGET_IEEE_FP.
9114 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9117 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9118 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9119 Expand with cmp<dwi>_doubleword.
9121 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9123 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9124 List CET intrinsics.
9125 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9126 specific to -fcf-protection option.
9128 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9130 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9131 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9132 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9133 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9134 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9135 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9136 (extra_objs): Add cet.o for Linux/x86 targets.
9137 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9138 * config/i386/cet.c: New file.
9139 * config/i386/cetintrin.h: Likewise.
9140 * config/i386/t-cet: Likewise.
9141 * config/i386/cpuid.h (bit_SHSTK): New.
9142 (bit_IBT): Likewise.
9143 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9144 pass IBT and SHSTK bits.
9145 * config/i386/i386-builtin-types.def
9146 (VOID_FTYPE_UNSIGNED_PVOID): New.
9147 (VOID_FTYPE_UINT64_PVOID): Likewise.
9148 * config/i386/i386-builtin.def: Add CET intrinsics.
9149 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9150 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9151 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9152 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9154 * config/i386/i386.c (rest_of_insert_endbranch): New.
9155 (pass_data_insert_endbranch): Likewise.
9156 (pass_insert_endbranch): Likewise.
9157 (make_pass_insert_endbranch): Likewise.
9158 (ix86_notrack_prefixed_insn_p): Likewise.
9159 (ix86_target_string): Add -mibt, -mshstk flags.
9160 (ix86_option_override_internal): Add flag_cf_protection
9162 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9163 (ix86_print_operand): Add 'notrack' prefix output.
9164 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9165 (ix86_expand_builtin): Expand CET intrinsics.
9166 (x86_output_mi_thunk): Add 'endbranch' instruction.
9167 * config/i386/i386.h (TARGET_IBT): New.
9168 (TARGET_IBT_P): Likewise.
9169 (TARGET_SHSTK): Likewise.
9170 (TARGET_SHSTK_P): Likewise.
9171 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9172 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9173 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9174 (builtin_setjmp_setup): New pattern.
9175 (builtin_longjmp): Likewise.
9176 (rdssp<mode>): Likewise.
9177 (incssp<mode>): Likewise.
9178 (saveprevssp): Likewise.
9179 (rstorssp): Likewise.
9180 (wrss<mode>): Likewise.
9181 (wruss<mode>): Likewise.
9182 (setssbsy): Likewise.
9183 (clrssbsy): Likewise.
9184 (nop_endbr): Likewise.
9185 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9187 * config/i386/immintrin.h: Include <cetintrin.h>.
9188 * config/i386/linux-common.h
9189 (file_end_indicate_exec_stack_and_cet): New prototype.
9190 (TARGET_ASM_FILE_END): New.
9192 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9194 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9195 latencies instead of having separate table; make difference between
9196 integer and float costs.
9197 * i386.h (processor_costs): Remove scalar_stmt_cost,
9198 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9199 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9201 * x86-tune-costs.h: Remove entries which has been removed in
9202 procesor_costs from all tables; make cond_taken_branch_cost
9203 and cond_not_taken_branch_cost COST_N_INSNS based.
9205 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9207 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9209 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9211 * config/i386/i386.md (isa): Remove fma_avx512f.
9212 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9213 <avx512>_fmadd_<mode>_mask3<round_name>,
9214 <avx512>_fmsub_<mode>_mask<round_name>,
9215 <avx512>_fmsub_<mode>_mask3<round_name>,
9216 <avx512>_fnmadd_<mode>_mask<round_name>,
9217 <avx512>_fnmadd_<mode>_mask3<round_name>,
9218 <avx512>_fnmsub_<mode>_mask<round_name>,
9219 <avx512>_fnmsub_<mode>_mask3<round_name>,
9220 <avx512>_fmaddsub_<mode>_mask<round_name>,
9221 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9222 <avx512>_fmsubadd_<mode>_mask<round_name>,
9223 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9224 (*vec_widen_umult_even_v16si<mask_name>,
9225 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9226 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9228 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9230 * extend.texi: Add 'nocf_check' documentation.
9231 * gimple.texi: Add second parameter to
9232 gimple_build_call_from_tree.
9233 * invoke.texi: Add -fcf-protection documentation.
9234 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9236 2017-10-20 Richard Biener <rguenther@suse.de>
9238 PR tree-optimization/82473
9239 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9240 the largest input type.
9242 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9244 * c-attribs.c (handle_nocf_check_attribute): New function.
9245 (c_common_attribute_table): Add 'nocf_check' handling.
9246 * gimple-parser.c: Add second argument NULL to
9247 gimple_build_call_from_tree.
9248 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9249 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9251 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9252 * common.opt: Add fcf-protection flag.
9253 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9254 * flag-types.h: Add enum cf_protection_level.
9255 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9256 Add 'nocf_check' attribute propagation to gimple call.
9257 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9258 (gimple_build_call_from_tree): Update prototype.
9259 (gimple_call_nocf_check_p): New function.
9260 (gimple_call_set_nocf_check): Likewise.
9261 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9262 * ipa-icf.c: Add nocf_check attribute in statement hash.
9263 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9264 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9265 * toplev.c (process_options): Add flag_cf_protection handling.
9267 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9269 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9271 2017-10-20 Richard Biener <rguenther@suse.de>
9273 PR tree-optimization/82603
9274 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9275 remove false predicated stores.
9277 2017-10-20 Richard Biener <rguenther@suse.de>
9279 * graphite-isl-ast-to-gimple.c
9280 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9281 Remove return value and simplify, dump copied stmt after lhs
9283 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9284 Reduce dump verbosity.
9285 (gsi_insert_earliest): Likewise.
9286 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9287 * graphite.c (print_global_statistics): Adjust dumping.
9288 (print_graphite_scop_statistics): Likewise.
9289 (print_graphite_statistics): Do not dump loops here.
9290 (graphite_transform_loops): But here.
9292 2017-10-20 Nicolas Roche <roche@adacore.com>
9294 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9295 * configure: Regenerate.
9297 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9300 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9301 functions when optimizing replace GIMPLE_RETURN stmts with
9302 calls to __builtin_unreachable ().
9305 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9306 for -fsanitize=thread link of executables.
9307 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9308 link of executables.
9311 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9313 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9314 define_insns for logical vector shifts to use VI248_AVX512BW
9315 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9316 condition, useless isa and prefix attributes. Change the first
9317 2 of these define_insns to ...
9318 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9319 define_insn for avx512vl.
9320 (<shift_insn><mode>3): ... and this, new define_insn without
9321 masking for non-avx512vl.
9324 * config/i386/sse.md (*andnot<mode>3,
9325 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9326 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9327 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9328 not applied use empty suffix even for TARGET_AVX512VL.
9329 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9330 is applied, supply evex,evex or evex,evex,evex instead of just
9333 2017-10-20 Julia Koval <julia.koval@intel.com>
9335 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9336 (OPTION_MASK_ISA_GFNI_UNSET): New.
9337 (ix86_handle_option): Handle OPT_mgfni.
9338 * config/i386/cpuid.h (bit_GFNI): New.
9339 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9340 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9341 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9342 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9343 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9344 * config/i386/i386.opt: Add mgfni.
9346 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9348 * config/msp430/msp430.c (msp430_option_override): Disable
9349 -fdelete-null-pointer-checks.
9350 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9352 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9354 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9355 of x87 and SSE instructions.
9357 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9359 * asan.c (create_cond_insert_point): Do not update edge count.
9360 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9361 (afdo_propagate_circuit): Likewise.
9362 (afdo_calculate_branch_prob): Likewise.
9363 (afdo_annotate_cfg): Likewise.
9364 * basic-block.h (struct edge_def): Remove count.
9365 (edge_def::count): New accessor.
9366 * bb-reorder.c (rotate_loop): Update.
9367 (find_traces_1_round): Update.
9368 (connect_traces): Update.
9369 (sanitize_hot_paths): Update.
9370 * cfg.c (unchecked_make_edge): Update.
9371 (make_single_succ_edge): Update.
9372 (check_bb_profile): Update.
9373 (dump_edge_info): Update.
9374 (update_bb_profile_for_threading): Update.
9375 (scale_bbs_frequencies_int): Update.
9376 (scale_bbs_frequencies_gcov_type): Update.
9377 (scale_bbs_frequencies_profile_count): Update.
9378 (scale_bbs_frequencies): Update.
9379 * cfganal.c (connect_infinite_loops_to_exit): Update.
9380 * cfgbuild.c (compute_outgoing_frequencies): Update.
9381 (find_many_sub_basic_blocks): Update.
9382 * cfgcleanup.c (try_forward_edges): Update.
9383 (try_crossjump_to_edge): Update
9384 * cfgexpand.c (expand_gimple_cond): Update
9385 (expand_gimple_tailcall): Update
9386 (construct_exit_block): Update
9387 * cfghooks.c (verify_flow_info): Update
9388 (redirect_edge_succ_nodup): Update
9389 (split_edge): Update
9390 (make_forwarder_block): Update
9391 (duplicate_block): Update
9392 (account_profile_record): Update
9393 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9394 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9395 * cfgloopmanip.c (scale_loop_profile): Update.
9397 (lv_adjust_loop_entry_edge): Update.
9398 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9399 (force_nonfallthru_and_redirect): Update.
9400 (purge_dead_edges): Update.
9401 (rtl_flow_call_edges_add): Update.
9402 * cgraphunit.c (init_lowered_empty_function): Update.
9403 (cgraph_node::expand_thunk): Update.
9404 * gimple-pretty-print.c (dump_probability): Update.
9405 (dump_edge_probability): Update.
9406 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9407 * haifa-sched.c (sched_create_recovery_edges): Update.
9408 * hsa-gen.c (convert_switch_statements): Update.
9409 * ifcvt.c (dead_or_predicable): Update.
9410 * ipa-inline-transform.c (inline_transform): Update.
9411 * ipa-split.c (split_function): Update.
9412 * ipa-utils.c (ipa_merge_profiles): Update.
9413 * loop-doloop.c (add_test): Update.
9414 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9415 * lto-streamer-in.c (input_cfg): Update.
9416 (input_function): Update.
9417 * lto-streamer-out.c (output_cfg): Update.
9418 * modulo-sched.c (sms_schedule): Update.
9419 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9420 * predict.c (maybe_hot_edge_p): Update.
9421 (unlikely_executed_edge_p): Update.
9422 (probably_never_executed_edge_p): Update.
9423 (dump_prediction): Update.
9424 (drop_profile): Update.
9425 (propagate_unlikely_bbs_forward): Update.
9426 (determine_unlikely_bbs): Update.
9427 (force_edge_cold): Update.
9428 * profile.c (compute_branch_probabilities): Update.
9429 * reg-stack.c (better_edge): Update.
9430 * shrink-wrap.c (handle_simple_exit): Update.
9431 * tracer.c (better_p): Update.
9432 * trans-mem.c (expand_transaction): Update.
9433 (split_bb_make_tm_edge): Update.
9434 * tree-call-cdce.c: Update.
9435 * tree-cfg.c (gimple_find_sub_bbs): Update.
9436 (gimple_split_edge): Update.
9437 (gimple_duplicate_sese_region): Update.
9438 (gimple_duplicate_sese_tail): Update.
9439 (gimple_flow_call_edges_add): Update.
9440 (insert_cond_bb): Update.
9441 (execute_fixup_cfg): Update.
9442 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9443 * tree-complex.c (expand_complex_div_wide): Update.
9444 * tree-eh.c (lower_resx): Update.
9445 (unsplit_eh): Update.
9446 (cleanup_empty_eh_move_lp): Update.
9447 * tree-inline.c (copy_edges_for_bb): Update.
9448 (freqs_to_counts): Update.
9449 (copy_cfg_body): Update.
9450 * tree-ssa-dce.c (remove_dead_stmt): Update.
9451 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9452 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9453 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9454 (unloop_loops): Update.
9455 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9456 * tree-ssa-loop-split.c (connect_loops): Update.
9457 (split_loop): Update.
9458 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9459 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9460 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9461 * tree-ssa-reassoc.c (branch_fixup): Update.
9462 * tree-ssa-tail-merge.c (replace_block_by): Update.
9463 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9464 (compute_path_counts): Update.
9465 (update_profile): Update.
9466 (recompute_probabilities): Update.
9467 (update_joiner_offpath_counts): Update.
9468 (estimated_freqs_path): Update.
9469 (freqs_to_counts_path): Update.
9470 (clear_counts_path): Update.
9471 (ssa_fix_duplicate_block_edges): Update.
9472 (duplicate_thread_path): Update.
9473 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9474 (case_bit_test_cmp): Update.
9475 (collect_switch_conv_info): Update.
9476 (gen_inbound_check): Update.
9477 (do_jump_if_equal): Update.
9478 (emit_cmp_and_jump_insns): Update.
9479 * tree-tailcall.c (decrease_profile): Update.
9480 (eliminate_tail_call): Update.
9481 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9482 (vect_do_peeling): Update.
9483 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9484 * ubsan.c (ubsan_expand_null_ifn): Update.
9485 (ubsan_expand_ptr_ifn): Update.
9486 * value-prof.c (gimple_divmod_fixed_value): Update.
9487 (gimple_mod_pow2): Update.
9488 (gimple_mod_subtract): Update.
9489 (gimple_ic): Update.
9490 (gimple_stringop_fixed_value): Update.
9492 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9495 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9497 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9499 PR rtl-optimization/82395
9500 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9501 based on non_spilled_static_chain_regno_p.
9503 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9505 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9506 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9507 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9509 2017-10-19 Martin Sebor <msebor@redhat.com>
9511 PR tree-optimization/82596
9512 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9514 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9516 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9517 (get_mem_refs_of_builtin_call): Likewise.
9518 * builtins.c (expand_builtin_apply): Adjust call to
9519 allocate_dynamic_stack_space.
9520 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9521 the third argument to allocate_dynamic_stack_space, otherwise -1.
9522 (expand_builtin): Deal with all alloca variants.
9523 (is_inexpensive_builtin): Likewise.
9524 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9525 * calls.c (special_function_p): Deal with all alloca variants.
9526 (initialize_argument_information): Adjust call to
9527 allocate_dynamic_stack_space.
9528 (expand_call): Likewise.
9529 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9530 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9531 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9532 use it for the stack usage computation.
9533 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9534 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9535 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9536 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9537 (in_loop_p): Remove first argument and useless check.
9538 (pass_walloca::execute): Remove useless test and adjust call to above.
9539 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9540 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9541 (gimplify_call_expr): Deal with all alloca variants.
9542 * hsa-gen.c (gen_hsa_alloca): Likewise.
9543 (gen_hsa_insns_for_call): Likewise.
9544 * ipa-pure-const.c (special_builtin_state): Likewise.
9545 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9546 * tree-object-size.c (alloc_object_size): Likewise.
9547 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9548 (call_may_clobber_ref_p_1): Likewise.
9549 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9550 (ccp_fold_stmt): Likewise.
9551 (optimize_stack_restore): Likewise.
9552 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9553 (mark_all_reaching_defs_necessary_1): Likewise.
9554 (propagate_necessity): Likewise.
9555 (eliminate_unnecessary_stmts): Likewise.
9556 * tree.c (build_common_builtin_nodes): Build
9557 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9558 (build_alloca_call_expr): New function.
9559 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9560 (CASE_BUILT_IN_ALLOCA): Likewise.
9561 (build_alloca_call_expr): Declare.
9562 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9564 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9567 * dwarf2out.c (new_die_raw): New static inline function.
9568 (new_die): Use it to create the DIE.
9569 (add_AT_external_die_ref): Likewise.
9570 (clone_die): Likewise.
9571 (clone_as_declaration): Likewise.
9572 (dwarf2out_vms_debug_main_pointer): Likewise.
9573 (base_type_die): Likewise. Remove early return for corner cases.
9574 Do not call add_pubtype on the DIE here.
9575 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9576 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9577 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9578 native order exists for base types, attach the DIE manually and call
9579 add_pubtype on it. Do not equate a reverse order DIE to the type.
9581 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9583 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9584 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9586 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9588 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9590 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9591 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9592 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9593 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9594 * opts.c (sanitizer_opts): Add builtin.
9595 * ubsan.c (instrument_builtin): New function.
9596 (pass_ubsan::execute): Call it.
9597 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9598 * doc/invoke.texi: Document -fsanitize=builtin.
9600 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9601 builtins, store max (log2 (align), 0) into uchar field instead of
9602 align into uptr field.
9603 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9604 store uchar 0 field instead of uptr 0 field.
9605 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9606 instead of passing one address of struct with 2 locations pass
9607 two addresses of structs with 1 location each.
9608 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9609 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9610 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9611 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9612 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9613 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9614 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9615 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9617 2017-10-19 Martin Liska <mliska@suse.cz>
9620 * file-find.c (remove_prefix): Remove.
9621 * file-find.h (remove_prefix): Likewise.
9622 * gcc-ar.c: Remove smartness of lookup.
9624 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9626 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9627 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9628 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9630 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9633 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9634 (setcc + and to xor + setcc): New peephole2.
9636 2017-10-19 Tom de Vries <tom@codesourcery.com>
9638 * doc/sourcebuild.texi (Test Directives, Variants of
9639 dg-require-support): Add dg-require-stack-size.
9641 2017-10-19 Martin Liska <mliska@suse.cz>
9644 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9645 that have a large alignment.
9646 (gimplify_target_expr): Likewise.
9648 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9650 PR rtl-optimization/82602
9651 * ira.c (rtx_moveable_p): Return false for volatile asm.
9653 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9656 * config/i386/i386-modes.def (CCGZ): New CC mode.
9657 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9658 * config/i386/predicates.md (ix86_comparison_operator):
9660 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9661 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9662 with double-word subtraction.
9663 (put_condition_code): Handle CCGZmode.
9665 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9667 * wide-int.cc (debug (const wide_int &)): New.
9668 (debug (const wide_int *)): New.
9669 (debug (const widest_int &)): New.
9670 (debug (const widest_int *)): New.
9672 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9675 * lra-constraints.c (curr_insn_transform): Use non-input operand
9676 instead of output one for matched reload.
9678 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9680 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9681 (tree-ssa-loop-ivopts.h): New header file.
9682 (struct builtin_info): New fields.
9683 (classify_builtin_1): Compute and record base and offset parts for
9684 memset builtin partition by calling strip_offset.
9685 (offset_cmp, fuse_memset_builtins): New functions.
9686 (finalize_partitions): Fuse adjacent memset partitions by calling
9688 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9689 Expose the interface.
9690 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9692 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9694 PR tree-optimization/82574
9695 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9696 that data reference must be executed exactly once per iteration
9697 against the outermost loop in nest.
9698 (classify_partition): Update call to above function.
9700 2017-10-18 Richard Biener <rguenther@suse.de>
9702 PR tree-optimization/82591
9703 * graphite.c (graphite_transform_loops): Move code gen message
9705 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9706 Here. Handle scop_to_isl_ast failing.
9707 (scop_to_isl_ast): Limit the number of ISL operations.
9709 2017-10-18 Richard Biener <rguenther@suse.de>
9711 * graphite-isl-ast-to-gimple.c
9712 (translate_isl_ast_to_gimple::set_rename): Simplify.
9713 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
9714 (graphite_copy_stmts_from_block): ... here.
9715 (copy_bb_and_scalar_dependences): Simplify.
9716 (add_parameters_to_ivs_params): Canonicalize.
9717 (generate_entry_out_of_ssa_copies): Simplify.
9718 * graphite-sese-to-poly.c (extract_affine_name): Simplify
9719 by passing in ISL dimension.
9720 (parameter_index_in_region_1): Rename to ...
9721 (parameter_index_in_region): ... this.
9722 (extract_affine): Adjust assert, pass down parameter index.
9723 (add_param_constraints): Use range-info when available.
9724 (build_scop_context): Adjust.
9725 * sese.c (new_sese_info): Adjust.
9726 (free_sese_info): Likewise.
9727 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
9728 Remove unused typedefs.
9729 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
9731 2017-10-18 Martin Liska <mliska@suse.cz>
9733 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9735 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9737 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9738 (s390_sched_init): Do not reset s390_sched_state if we entered the
9739 current basic block via a fallthru edge and all others are unlikely.
9741 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9743 * config/s390/s390.c (NUM_SIDES): New variable.
9744 (LONGRUNNING_THRESHOLD): New variable.
9745 (LATENCY_FACTOR): New variable.
9746 (s390_sched_score): Decrease score for long-running instructions on
9748 (s390_sched_variable_issue): Perform bookkeeping for long-running
9751 2017-10-18 Richard Biener <rguenther@suse.de>
9753 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9754 Simplify with removal of the parameter rename map.
9755 (set_rename): Likewise.
9756 (should_copy_to_new_region): Likewise.
9757 (graphite_copy_stmts_from_block): Likewise.
9758 (copy_bb_and_scalar_dependences): Remove initialization of
9759 unused copied_bb_map.
9761 (copy_internal_parameters): Likewise.
9762 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9763 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9764 Use INTEGRAL_TYPE_P.
9765 (parameter_index_in_region_1): Rename to ...
9766 (assign_parameter_index_in_region): ... this. Assert we have
9767 a parameter we handle.
9768 (scan_tree_for_params): Adjust.
9769 * sese.h (parameter_rename_map_t): Remove.
9770 (struct sese_info_t): Remove unused parameter_rename_map and
9771 copied_bb_map members.
9772 * sese.c (new_sese_info): Adjust.
9773 (free_sese_info): Likewise.
9775 2017-10-18 Martin Liska <mliska@suse.cz>
9778 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9779 on an abnormal edge.
9781 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9783 * doc/invoke.texi (ffunction-sections and fdata-sections):
9786 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9788 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9789 the use statement can throw internally.
9791 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9793 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9794 any RTX present on the RHS of a SET.
9795 * compare-elim.c (try_eliminate_compare): Restore comment.
9797 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9799 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9800 may be also called on tcc_type nodes.
9801 * langhooks.c (lhd_tree_size): Likewise.
9803 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9805 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9806 format_warning_at_substring.
9807 (maybe_warn): Convert source_range * param to a location_t. Pass
9808 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9809 (format_directive): Remove code to extract source_ranges and
9810 source_range * in favor of just a location_t.
9811 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9813 * substring-locations.c (format_warning_va): Convert
9814 source_range * param to a location_t.
9815 (format_warning_at_substring): Likewise.
9816 * substring-locations.h (format_warning_va): Likewise.
9817 (format_warning_at_substring): Likewise.
9819 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9821 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9823 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9824 and scatter/gather ops.
9826 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9827 vec_gather_load and vec_scatter_store.
9828 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9829 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9830 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9831 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9832 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9833 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9835 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9837 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9838 Detect FTST instruction and handle its register pops. Only pop
9839 second operand if can_pop_second_op is true.
9840 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9841 set can_pop_second_op to false in the compare_for_stack_reg call.
9843 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9844 output_fp_compare for stack register operands.
9845 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9846 instructions here. Do not emit stack register pops here. Assert
9847 that FCOMPP pops next to top stack register. Rewrite function.
9849 2017-10-17 Nathan Sidwell <nathan@acm.org>
9852 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9853 use DECL_ASSEMBLER_NAME_RAW.
9856 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9859 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9860 Wilco Dijkstra <wilco.dijkstra@arm.com>
9862 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9863 converstion to Pmode from the buf_addr.
9865 2017-10-17 Richard Biener <rguenther@suse.de>
9867 * graphite-dependences.c (scop_get_reads_and_writes): Change
9868 output parameters to references.
9870 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9872 PR 71026/tree-optimization
9873 * fold-const.c (distribute_real_division): Removed.
9874 (fold_binary_loc): Remove calls to distribute_real_divison.
9876 2017-10-17 Richard Biener <rguenther@suse.de>
9878 * graphite-scop-detection.c
9879 (scop_detection::stmt_has_simple_data_refs_p): Always use
9880 the full nest as region.
9881 (try_generate_gimple_bb): Likewise.
9882 * sese.c (scalar_evolution_in_region): Simplify now that
9883 SCEV can handle instantiation in regions.
9884 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9885 in the non-loop part of a function if requested.
9887 2017-10-17 Richard Biener <rguenther@suse.de>
9889 PR tree-optimization/82563
9890 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9892 (graphite_regenerate_ast_isl): Call it.
9893 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9895 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9897 PR tree-optimization/82549
9898 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9899 Formatting fixes. Instead of calling make_bit_field_ref with negative
9902 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9904 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9905 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9906 _mm_maskz_reduce_ss): New.
9907 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9908 __builtin_ia32_reducess_mask): Ditto..
9909 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9910 * config/i386/sse.md (reduces<mode>): Renamed to ...
9911 (reduces<mode><mask_scalar_name>): ... this.
9912 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9914 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9915 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9917 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9919 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9920 * selftest-run-tests.c (selftest::run_tests): Call
9921 selftest::unique_ptr_tests_cc_tests.
9922 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9923 * unique-ptr-tests.cc: New file.
9925 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9928 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9930 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9931 (make_hard_regno_born, make_hard_regno_dead): Update
9932 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9934 2017-10-16 Jeff Law <law@redhat.com>
9936 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9938 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9940 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9942 2017-10-16 Olivier Hainque <hainque@adacore.com>
9944 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9945 with_cpu if we were configured for an e500v2 target cpu name.
9947 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9949 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9950 * doc/invoke.texi: Document +nodsp as a valid extension for
9953 2017-10-16 Martin Liska <mliska@suse.cz>
9955 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9956 (test_set_range): Likewise.
9957 (test_range_functions): Rename to ...
9958 (test_bit_in_range): ... this.
9959 (sbitmap_c_tests): Add new test.
9961 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9963 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9965 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9966 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9968 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9970 * config/aarch64/aarch64-builtins.c
9971 (aarch64_types_quadopu_lane_qualifiers): New.
9972 (TYPES_QUADOPU_LANE): New.
9973 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9974 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9975 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9976 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9977 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9978 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9979 (Vdottype, DOTPROD): New.
9980 (sur): Add SDOT and UDOT.
9982 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9984 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9985 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9986 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9988 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9989 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9990 Enable TARGET_DOTPROD.
9991 (cortex-a75.cortex-a55): Likewise.
9992 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9994 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9996 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9997 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9999 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10001 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10002 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10003 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10004 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10005 * config/arm/types.md (neon_dot, neon_dot_q): New.
10006 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10008 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10010 * config/arm/arm.h (TARGET_DOTPROD): New.
10011 * config/arm/arm.c (arm_arch_dotprod): New.
10012 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10013 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10014 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10015 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10016 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10017 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10018 * doc/invoke.texi (armv8.2-a): Document dotprod
10020 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10022 * i386.c (ix86_vec_cost): New function.
10023 (ix86_rtx_costs): Handle vector operations better.
10024 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10025 * x86-tune-costs.h: Add new costs to all tables.
10027 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10029 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10031 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10032 divsd, sqrtss and sqrtsd
10033 * x86-tune-costs.h: Add new entries to all costs.
10034 (znver1_cost): Fix to match real instruction latencies.
10036 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10037 Michael Collison <michael.collison@arm.com>
10039 * compare-elim.c: Include emit-rtl.h.
10040 (can_merge_compare_into_arith): New function.
10041 (try_validate_parallel): Likewise.
10042 (try_merge_compare): Likewise.
10043 (try_eliminate_compare): Call the above when no previous clobber
10045 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10048 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10050 PR middle-end/62263
10051 PR middle-end/82498
10052 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10053 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10055 PR middle-end/62263
10056 PR middle-end/82498
10057 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10058 to be any operand_equal_p operands. For & (B - 1) require
10059 B to be power of 2. Recognize
10060 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10062 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10065 * optabs.c (expand_memory_blockage): Fix call of
10066 targetm.have_memory_blockage.
10068 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10071 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10072 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10073 objects to extra_objs instead of overwriting it.
10075 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10077 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10078 Use any_fp_register_operand as operand[3] predicate. Simplify
10079 equality test for operands[2] and operands[4] memory location.
10080 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10081 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10082 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10083 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10084 any_fp_register_operand as operand[1] predicate. Simplify
10085 equality test for operands[0] and operands[3] memory location.
10086 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10087 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10088 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10090 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10092 * target-insns.def: Add memory_blockage.
10093 * optabs.c (expand_memory_blockage): New function.
10094 (expand_asm_memory_barrier): Rename ...
10095 (expand_asm_memory_blockage): ... to this.
10096 (expand_mem_thread_fence): Call expand_memory_blockage
10097 instead of expand_asm_memory_barrier.
10098 (expand_mem_singnal_fence): Ditto.
10099 (expand_atomic_load): Ditto.
10100 (expand_atomic_store): Ditto.
10101 * doc/md.texi (Standard Pattern Names For Generation):
10102 Document memory_blockage instruction pattern.
10104 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10106 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10107 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10108 * config/rl78/rl78.md: New define_expand "adddi3".
10110 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10112 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10115 2017-10-13 Jeff Law <law@redhat.com>
10117 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10119 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10122 * internal-fn.c (expand_mul_overflow): If both operands have
10123 the same highpart of -1 or 0 and the topmost bit of lowpart
10124 is different, overflow is if res <= 0 rather than res < 0.
10126 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10128 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10129 TARGET_P9_VECTOR code for unaligned_load case.
10131 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10133 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10135 2017-10-13 Nathan Sidwell <nathan@acm.org>
10137 * tree-core.h (tree_contains_struct): Make bool.
10138 * tree.c (tree_contains_struct): Likewise.
10139 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10140 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10141 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10142 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10143 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10145 2017-10-13 Richard Biener <rguenther@suse.de>
10147 * graphite-isl-ast-to-gimple.c
10148 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10149 parameters and dominance check.
10150 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10151 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10152 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10153 Do not update SSA form here or do intermediate IL verification.
10154 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10155 (graphite_initialize): Remove check on the number of loops in
10156 the function and inline into graphite_transform_loops.
10157 (graphite_finalize): Inline into graphite_transform_loops.
10158 (graphite_transform_loops): Perform SSA update and IL verification
10160 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10162 2017-10-13 Richard Biener <rguenther@suse.de>
10164 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10165 graphite_expression_type_precision): Avoid global constructor
10167 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10168 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10169 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10170 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10171 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10172 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10174 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10177 * config/i386/i386.h (ix86_red_zone_size): New.
10178 * config/i386/i386.md (push peephole2s): Replace
10179 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10181 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10182 Alan Hayward <alan.hayward@arm.com>
10183 David Sherwood <david.sherwood@arm.com>
10185 * combine.c (can_change_dest_mode): Reject changes in
10186 REGMODE_NATURAL_SIZE.
10188 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10189 Alan Hayward <alan.hayward@arm.com>
10190 David Sherwood <david.sherwood@arm.com>
10192 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10193 (expand_debug_source_expr): Likewise.
10194 * combine.c (combine_simplify_rtx): Likewise.
10195 * cse.c (fold_rtx): Likewise.
10196 * fwprop.c (canonicalize_address): Likewise.
10197 * targhooks.c (default_shift_truncation_mask): Likewise.
10199 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10200 Alan Hayward <alan.hayward@arm.com>
10201 David Sherwood <david.sherwood@arm.com>
10203 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10204 (widened_mode): Likewise.
10205 (expand_unop): Likewise.
10206 * ree.c (transform_ifelse): Likewise.
10207 (merge_def_and_ext): Likewise.
10208 (combine_reaching_defs): Likewise.
10209 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10211 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10212 Alan Hayward <alan.hayward@arm.com>
10213 David Sherwood <david.sherwood@arm.com>
10215 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10216 * combine.c (gen_lowpart_for_combine): Likewise.
10217 * dwarf2out.c (rtl_for_decl_location): Likewise.
10218 * final.c (alter_subreg): Likewise.
10219 * rtlhooks.c (gen_lowpart_general): Likewise.
10220 (gen_lowpart_if_possible): Likewise.
10222 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10223 Alan Hayward <alan.hayward@arm.com>
10224 David Sherwood <david.sherwood@arm.com>
10226 * calls.c (expand_call): Use subreg_lowpart_offset.
10227 * cse.c (cse_insn): Likewise.
10228 * regcprop.c (copy_value): Likewise.
10229 (copyprop_hardreg_forward_1): Likewise.
10231 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10234 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10235 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10236 =Q constraints to +Q and into insn condition add check
10237 that operands[0] and operands[1] are equal.
10238 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10239 =Q constraints to +Q and into insn condition add check
10240 that operands[0] is equal to either operands[1] or operands[2].
10243 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10244 instead of handling MINUS_EXPR twice (once for each argument),
10245 canonicalize operand order and handle just once, use rtype where
10246 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10249 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10250 any values of __C while still being pattern recognizable as a simple
10251 rotate instruction.
10253 2017-10-13 Richard Biener <rguenther@suse.de>
10255 PR tree-optimization/82451
10257 2017-10-02 Richard Biener <rguenther@suse.de>
10259 PR tree-optimization/82355
10260 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10261 a mapping for the enclosing loop but avoid generating one for
10262 the loop tree root.
10263 (copy_bb_and_scalar_dependences): Remove premature codegen
10264 error on PHIs in blocks duplicated into multiple places.
10265 * graphite-scop-detection.c
10266 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10267 in the region use it as loop and nest to analyze the DR in.
10268 (try_generate_gimple_bb): Likewise.
10269 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10270 (add_loop_constraints): For blocks in a loop not in the region
10271 create a dimension with a single iteration.
10272 * sese.h (gbb_loop_at_index): Remove assert.
10274 * cfgloop.c (loop_preheader_edge): For the loop tree root
10275 return the single successor of the entry block.
10276 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10277 Reset the SCEV hashtable and niters.
10278 * graphite-scop-detection.c
10279 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10280 assert that we only have POLYNOMIAL_CHREC that vary in loops
10281 contained in the region.
10282 (scop_detection::graphite_can_represent_expr): Adjust.
10283 (scop_detection::stmt_has_simple_data_refs_p): For loops
10284 not in the region set loop to NULL. The nest is now the
10285 entry edge to the region.
10286 (try_generate_gimple_bb): Likewise.
10287 * sese.c (scalar_evolution_in_region): Adjust for
10288 instantiate_scev change.
10289 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10290 Make nest parameter the edge into the region.
10291 (create_data_ref): Likewise.
10292 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10293 entry edge into a region and adjust instantiate_scev calls.
10294 (create_data_ref): Likewise.
10295 (graphite_find_data_references_in_stmt): Likewise.
10296 (find_data_references_in_stmt): Pass the loop preheader edge
10297 from the nest argument.
10298 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10299 parameter the edge into the region.
10300 (instantiate_parameters): Use the loop preheader edge as entry.
10301 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10303 (get_instantiated_value_entry): Make instantiate_below parameter
10304 the edge into the region.
10305 (instantiate_scev_name): Likewise. Adjust dominance checks,
10306 when we cannot use loop-based instantiation instantiate by
10307 walking use-def chains.
10308 (instantiate_scev_poly): Adjust.
10309 (instantiate_scev_binary): Likewise.
10310 (instantiate_scev_convert): Likewise.
10311 (instantiate_scev_not): Likewise.
10312 (instantiate_array_ref): Remove.
10313 (instantiate_scev_3): Likewise.
10314 (instantiate_scev_2): Likewise.
10315 (instantiate_scev_1): Likewise.
10316 (instantiate_scev_r): Do not blindly handle N-operand trees.
10317 Do not instantiate array-refs. Handle all constants and invariants.
10318 (instantiate_scev): Make instantiate_below parameter
10319 the edge into the region.
10320 (resolve_mixers): Use the loop preheader edge for the region
10321 parameter to instantiate_scev_r.
10322 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10324 2017-10-13 Richard Biener <rguenther@suse.de>
10326 PR tree-optimization/82525
10327 * graphite-isl-ast-to-gimple.c
10328 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10330 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10331 Fail code generation when we cannot represent the isl integer.
10332 (binary_op_to_tree): Elide modulo operations that are no-ops
10333 in the type we code generate. Remove now superfluous code
10336 2017-10-13 Richard Biener <rguenther@suse.de>
10338 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10339 (scop_detection::harmful_loop_in_region): Remove premature
10340 IV type restriction.
10341 (scop_detection::graphite_can_represent_scev): We can handle
10342 pointer IVs just fine.
10344 2017-10-13 Alan Modra <amodra@gmail.com>
10346 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10347 "Clobbers and Scratch Registers". Add paragraph on
10348 alternative to clobbers for scratch registers and OpenBLAS
10351 2017-10-13 Alan Modra <amodra@gmail.com>
10353 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10354 example of a memory input for a string of known length. Move
10355 commentary out of table. Add a number of new examples
10356 covering array memory inputs.
10358 2017-10-12 Martin Liska <mliska@suse.cz>
10360 PR tree-optimization/82493
10361 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10362 (test_range_functions): New function.
10363 (sbitmap_c_tests): Likewise.
10364 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10365 * selftest.h (sbitmap_c_tests): New function.
10367 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10369 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10371 * config/rs6000/amo.h: Fix spacing issue.
10373 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10376 * config/i386/i386.md (*ashl<mode>3_mask_1,
10377 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10378 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10381 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10383 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10384 (profile_probability): Set max_probability
10385 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10387 * profile-count.c (profile_probability::differs_from_p): Do not
10388 rely on max_probaiblity == 10000
10390 2017-10-12 Jeff Law <law@redhat.com>
10392 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10395 2017-10-12 Martin Sebor <msebor@redhat.com>
10399 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10400 (handle_alias_pairs): Call it.
10401 * common.opt (-Wattribute-alias): New option.
10402 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10403 * doc/invoke.texi (-Wattribute-alias): Document.
10405 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10408 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10410 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10412 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10413 (make_hard_regno_born, make_hard_regno_dead): Update
10414 bb_killed_pseudos and bb_gen_pseudos.
10416 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10418 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10420 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10422 * config/alpha/alpha.c (alpha_split_conditional_move):
10423 Use std::swap instead of manually swapping.
10424 (alpha_stdarg_optimize_hook): Ditto.
10425 (alpha_canonicalize_comparison): Ditto.
10427 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10429 * tree-loop-distribution.c (struct builtin_info): New struct.
10430 (struct partition): Refactor fields into struct builtin_info.
10431 (partition_free): Free struct builtin_info.
10432 (build_size_arg_loc, build_addr_arg_loc): Delete.
10433 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10434 information from struct builtin_info.
10435 (find_single_drs): New function refactored from classify_partition.
10436 Also moved builtin validity checks to this function.
10437 (compute_access_range, alloc_builtin): New functions.
10438 (classify_builtin_st, classify_builtin_ldst): New functions.
10439 (classify_partition): Refactor code into functions find_single_drs,
10440 classify_builtin_st and classify_builtin_ldst.
10441 (distribute_loop): Don't do runtime alias check when distributing
10443 (find_seed_stmts_for_distribution): New function.
10444 (pass_loop_distribution::execute): Refactor code finding seed
10445 stmts into above function. Support distribution for the innermost
10446 two-level loop nest. Adjust dump information.
10448 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10450 * tree-loop-distribution.c: Adjust the general comment.
10451 (NUM_PARTITION_THRESHOLD): New macro.
10452 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10453 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10454 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10455 in call to build_partition_graph.
10456 (finalize_partitions): New parameter. Make loop distribution more
10457 conservative by fusing more partitions.
10458 (distribute_loop): Don't do runtime alias check in case of loop nest
10460 (find_seed_stmts_for_distribution): New function.
10461 (prepare_perfect_loop_nest): New function.
10462 (pass_loop_distribution::execute): Refactor code finding seed stmts
10463 and loop nest into above functions. Support loop nest distribution.
10464 Adjust dump information accordingly.
10466 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10468 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10469 and set PTYPE_SEQUENTIAL for merged partition.
10471 2017-10-12 Richard Biener <rguenther@suse.de>
10473 PR tree-optimization/69728
10475 2017-09-19 Richard Biener <rguenther@suse.de>
10477 PR tree-optimization/69728
10478 * graphite-sese-to-poly.c (schedule_error): New global.
10479 (add_loop_schedule): Handle empty domain by failing the
10481 (build_original_schedule): Handle schedule_error.
10483 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10484 domain by returning an unchanged schedule.
10486 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10488 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10489 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10491 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10493 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10496 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10499 * expr.c (store_field): Don't optimize away bitsize == 0 store
10500 from CALL_EXPR with addressable return type.
10502 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10504 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10505 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10506 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10507 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10508 TARGET_ISEL instead of TARGET_ISEL<sel>.
10510 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10512 * config/rs6000/rs6000.c
10513 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10515 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10517 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10519 (reg_or_cint_operand): Fix comment.
10520 (reg_or_zero_operand): New predicate.
10521 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10522 * config/rs6000/rs6000.c (output_isel): Delete.
10523 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10524 instead of reg_or_cint_operand. Output instruction directly (not via
10526 (isel_unsigned_<mode>): Ditto.
10527 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10528 gpc_reg_operand. Add an instruction alternative for this. Output
10529 instruction directly.
10530 (*isel_reversed_unsigned_<mode>): Ditto.
10532 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10534 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10535 (TARGET_CANONICALIZE_COMPARISON): Define.
10537 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10540 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10541 Check whether the dest is REG before adding REG_EQUIV note.
10543 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10546 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10548 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10549 (make_hard_regno_born, make_hard_regno_dead): Update
10550 bb_killed_pseudos and bb_gen_pseudos.
10552 2017-10-11 Nathan Sidwell <nathan@acm.org>
10554 * incpath.h (enum incpath_kind): Name enum, prefix values.
10555 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10556 * incpath.c (heads, tails): Use INC_MAX.
10557 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10558 (merge_include_chains, split_quote_chain,
10559 register_include_chains): Update incpath_kind names.
10560 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10561 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10563 (add_framework_path, darwin_register_objc_includes): Likewise.
10564 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10566 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10568 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10569 Do not use float_operator operator predicate.
10570 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10571 * config/i386/predicates.md (float_operator): Remove predicate.
10573 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10575 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10576 (*jccxf_i387): Ditto.
10577 (*jcc<mode>_i387): Ditto.
10578 (*jccu<mode>_i387): Ditto.
10579 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10580 (*jcc_*_i387 splitters): Remove.
10581 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10582 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10583 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10586 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10588 * profile-count.h (slow_safe_scale_64bit): New function.
10589 (safe_scale_64bit): New inline.
10590 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10591 * profile-count.c: Include wide-int.h
10592 (slow_safe_scale_64bit): New.
10594 2017-10-11 Nathan Sidwell <nathan@acm.org>
10596 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10597 HAS_DECL_ASSEMBLER_NAME_P.
10598 * gimple-expr.c (gimple_decl_printable_name: Check
10599 HAS_DECL_ASSEMBLER_NAME_P too.
10600 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10601 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10602 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10603 * passes.c (rest_of_decl_compilation): Check
10604 HAS_DECL_ASSEMBLER_NAME_P too.
10605 * recog.c (verify_changes): Likewise.
10606 * tree-pretty-print.c (dump_decl_name): Likewise.
10607 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10609 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10610 (SET_DECL_ASSEMBLER_NAME): Use it.
10611 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10612 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10613 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10615 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10617 * config.gcc (i386, x86_64): Add extra objects.
10618 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10619 (ix86_min_insn_size): Declare.
10620 (ix86_issue_rate): Declare.
10621 (ix86_adjust_cost): Declare.
10622 (ia32_multipass_dfa_lookahead): Declare.
10623 (ix86_macro_fusion_p): Declare.
10624 (ix86_macro_fusion_pair_p): Declare.
10625 (ix86_bd_has_dispatch): Declare.
10626 (ix86_bd_do_dispatch): Declare.
10627 (ix86_core2i7_init_hooks): Declare.
10628 (ix86_atom_sched_reorder): Declare.
10629 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10630 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10631 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10632 (rip_relative_addr_p): Rename to ...
10633 (ix86_rip_relative_addr_p): ... this one; export.
10634 (memory_address_length): Update.
10635 (ix86_issue_rate): Move to x86-tune-sched.c.
10636 (ix86_flags_dependent): Move to x86-tune-sched.c.
10637 (ix86_agi_dependent): Move to x86-tune-sched.c.
10638 (exact_dependency_1): Move to x86-tune-sched.c.
10639 (exact_store_load_dependency): Move to x86-tune-sched.c.
10640 (ix86_adjust_cost): Move to x86-tune-sched.c.
10641 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10642 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10643 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10644 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10645 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10646 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10647 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10648 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10649 (min_insn_size): Rename to ...
10650 (ix86_min_insn_size): ... this one; export.
10651 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10652 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10653 (core2i7_first_cycle_multipass_backtrack): Move to
10654 x86-tune-sched-core.c.
10655 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10656 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10657 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10658 (ix86_avoid_jump_mispredicts): Update.
10659 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10660 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10661 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10662 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10663 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10664 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10665 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10666 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10667 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10668 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10669 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10670 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10671 (BIG): Move to ix86-tune-sched-bd.c.
10672 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10673 (enum insn_path): Move to ix86-tune-sched-bd.c.
10674 (get_mem_group): Move to ix86-tune-sched-bd.c.
10675 (is_cmp): Move to ix86-tune-sched-bd.c.
10676 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10677 (is_branch): Move to ix86-tune-sched-bd.c.
10678 (is_prefetch): Move to ix86-tune-sched-bd.c.
10679 (init_window): Move to ix86-tune-sched-bd.c.
10680 (allocate_window): Move to ix86-tune-sched-bd.c.
10681 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10682 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10683 (process_end_window): Move to ix86-tune-sched-bd.c.
10684 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10685 (find_constant): Move to ix86-tune-sched-bd.c.
10686 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10687 (has_immediate): Move to ix86-tune-sched-bd.c.
10688 (get_insn_path): Move to ix86-tune-sched-bd.c.
10689 (get_insn_group): Move to ix86-tune-sched-bd.c.
10690 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10691 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10692 (add_insn_window): Move to ix86-tune-sched-bd.c.
10693 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10694 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10695 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10696 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10697 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10698 (do_dispatch): Move to ix86-tune-sched-bd.c.
10699 (has_dispatch): Move to ix86-tune-sched-bd.c.
10700 * config/i386/t-i386: Add new object files.
10701 * config/i386/x86-tune-costs.h: New file.
10702 * config/i386/x86-tune-sched-atom.c: New file.
10703 * config/i386/x86-tune-sched-bd.c: New file.
10704 * config/i386/x86-tune-sched-core.c: New file.
10705 * config/i386/x86-tune-sched.c: New file.
10707 2017-10-11 Liu Hao <lh_mouse@126.com>
10709 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10710 the generic version below instead.
10711 (should_colorize): Recognize Windows consoles as terminals
10713 * pretty-print.c [__MINGW32__] (write_all): New function.
10714 [__MINGW32__] (find_esc_head): Likewise.
10715 [__MINGW32__] (find_esc_terminator): Likewise.
10716 [__MINGW32__] (eat_esc_sequence): Likewise.
10717 [__MINGW32__] (mingw_ansi_fputs): New function that handles
10719 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
10722 2017-10-11 Richard Biener <rguenther@suse.de>
10724 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
10725 Properly call analyze_scalar_evolution with the loop of the stmt.
10727 2017-10-11 Richard Biener <rguenther@suse.de>
10729 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
10730 * tree-core.h (tree_base): Add chrec_var union member.
10731 * tree.h (CHREC_VAR): Remove.
10732 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
10733 * tree-chrec.h (build_polynomial_chrec): Adjust.
10734 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10735 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10737 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10739 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10740 * match.pd: ... here.
10741 ((T) X == (T) Y): Relax condition.
10743 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10745 PR tree-optimization/82472
10746 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10748 (break_alias_scc_partitions): Update postorder number.
10750 2017-10-11 Martin Liska <mliska@suse.cz>
10753 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10755 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10757 2017-10-11 Martin Liska <mliska@suse.cz>
10759 * print-rtl.c (print_insn): Move declaration of idbuf
10760 to same scope as name.
10762 2017-10-11 Martin Liska <mliska@suse.cz>
10767 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10768 volatile arguments.
10770 2017-10-11 Martin Liska <mliska@suse.cz>
10773 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10774 volatile arguments.
10776 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10778 * config.gcc (default_gnu_indirect_function): Default to yes for
10779 arm*-*-linux* with glibc.
10781 2017-10-11 Richard Biener <rguenther@suse.de>
10783 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10784 default-defs and types we do not want to analyze.
10785 (interpret_loop_phi): Replace unreachable code with an assert.
10786 (compute_scalar_evolution_in_loop): Remove and inline ...
10787 (analyze_scalar_evolution_1): ... here, replacing condition with
10788 what makes the intent clearer. Remove handling of cases
10789 get_scalar_evolution now handles.
10791 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10793 PR rtl-optimization/81434
10794 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10795 comment for main loop. In sched_group_found if, also add checks for
10796 pass and min_cost_group.
10798 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10800 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10801 (rs6000_insn_cost): New function.
10802 * config/rs6000/rs6000.md (cost): New attribute.
10804 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10805 H.J. Lu <hongjiu.lu@intel.com>
10809 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10810 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10811 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10812 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10813 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10814 OPTION_MASK_ISA_MMX and some other bit set require both
10815 mmx and the ISAs without the mmx bit.
10816 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10817 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10818 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10819 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10820 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10821 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10822 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10823 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10824 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10825 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10826 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10827 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10828 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10829 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10830 Add OPTION_MASK_ISA_MMX.
10832 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10834 * config.gcc (armv7*-*-freebsd*): New target.
10835 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10837 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10839 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10840 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10843 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10845 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10846 a template parameter.
10847 (WIDE_INT_REF_FOR): Update accordingly.
10848 * tree.h (wi::int_traits <const_tree>): Delete.
10849 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10850 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10851 (wi::tree_to_wide_ref): New typedef.
10852 (wi::to_wide): New function.
10853 * calls.c (get_size_range): Use wi::to_wide when operating on
10854 trees as wide_ints.
10855 * cgraph.c (cgraph_node::create_thunk): Likewise.
10856 * config/i386/i386.c (ix86_data_alignment): Likewise.
10857 (ix86_local_alignment): Likewise.
10858 * dbxout.c (stabstr_O): Likewise.
10859 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10860 * expr.c (const_vector_from_tree): Likewise.
10861 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10862 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10863 (fold_negate_expr_1, int_const_binop_1, const_binop)
10864 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10865 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10866 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10867 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10868 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10869 (fold_not_const, round_up_loc): Likewise.
10870 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10871 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10872 (alloca_call_type): Likewise.
10873 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10874 * godump.c (go_output_typedef): Likewise.
10875 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10876 * internal-fn.c (get_min_precision): Likewise.
10877 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10878 * ipa-polymorphic-call.c
10879 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10880 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10881 (ipa_modify_call_arguments): Likewise.
10882 * match.pd: Likewise.
10883 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10884 * print-tree.c (print_node_brief, print_node): Likewise.
10885 * stmt.c (expand_case): Likewise.
10886 * stor-layout.c (layout_type): Likewise.
10887 * tree-affine.c (tree_to_aff_combination): Likewise.
10888 * tree-cfg.c (group_case_labels_stmt): Likewise.
10889 * tree-data-ref.c (dr_analyze_indices): Likewise.
10890 (prune_runtime_alias_test_list): Likewise.
10891 * tree-dump.c (dequeue_and_dump): Likewise.
10892 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10893 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10894 * tree-pretty-print.c (dump_generic_node): Likewise.
10895 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10896 (simple_iv_with_niters): Likewise.
10897 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10898 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10899 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10900 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10901 (refine_value_range_using_guard, number_of_iterations_ne_max)
10902 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10903 (get_cst_init_from_scev, record_nonwrapping_iv)
10904 (scev_var_range_cant_overflow): Likewise.
10905 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10906 * tree-ssa-pre.c (compute_avail): Likewise.
10907 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10908 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10909 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10910 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10911 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10912 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10913 (dump_case_nodes, try_switch_expansion): Likewise.
10914 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10915 (vect_do_peeling): Likewise.
10916 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10917 * tree-vect-stmts.c (vectorizable_load): Likewise.
10918 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10919 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10920 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10921 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10922 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10923 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10924 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10925 (evrp_dom_walker::before_dom_children): Likewise.
10926 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10927 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10928 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10929 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10930 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10931 (get_range_pos_neg): Likewise.
10932 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10933 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10934 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10935 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10936 * config/avr/avr.c (avr_fold_builtin): Likewise.
10937 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10938 * config/msp430/msp430.c (msp430_attr): Likewise.
10939 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10940 * config/powerpcspe/powerpcspe-c.c
10941 (altivec_resolve_overloaded_builtin): Likewise.
10942 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10943 (rs6000_expand_ternop_builtin): Likewise.
10944 * config/rs6000/rs6000-c.c
10945 (altivec_resolve_overloaded_builtin): Likewise.
10946 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10947 (rs6000_expand_ternop_builtin): Likewise.
10948 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10950 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10952 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10953 when copying loop nest with only one inner loop.
10955 2017-10-10 Richard Biener <rguenther@suse.de>
10957 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10958 blocks if SCEV is active.
10959 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10961 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10962 (scev_initialize): Assert we are not yet initialized.
10964 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10966 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10967 inner loop's exit stmt by making it always exit the loop, otherwise
10968 we would generate an infinite empty loop.
10970 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10972 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10973 renaming variables in new preheader if it's deleted.
10975 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10977 * tree-loop-distribution.c (struct partition): Remove unused field
10978 loops of the structure.
10979 (partition_alloc, partition_free): Ditto.
10980 (build_rdg_partition_for_vertex): Ditto.
10982 2017-10-09 Jeff Law <law@redhat.com>
10984 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10985 return type to match prototype and documentation.
10987 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10989 * config/rs6000/rs6000.c (processor_costs): Move to ...
10990 * config/rs6000/rs6000.h: ... here.
10991 (rs6000_cost): Declare.
10993 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10995 * except.c (setjmp_fn): New global variable.
10996 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10997 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10998 if DONT_USE_BUILTIN_SETJMP is defined.
11000 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11002 * target.def (insn_cost): New hook.
11003 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11004 * doc/tm.texi: Regenerate.
11005 * rtlanal.c (insn_cost): Use the new hook.
11007 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11009 * combine.c (combine_validate_cost): Compute the new insn_cost,
11010 not just pattern_cost.
11011 (try_combine): Adjust comment.
11013 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11015 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11017 * combine.c (uid_insn_cost): Adjust comment.
11018 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11020 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11021 * dse.c (find_shift_sequence): Ditto.
11022 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11023 (bb_valid_for_noce_process_p): Use pattern_cost.
11024 * rtl.h (insn_rtx_cost): Delete.
11025 (pattern_cost): New prototype.
11026 (insn_cost): New prototype.
11027 * rtlanal.c (insn_rtx_cost): Rename to...
11028 (pattern_cost): ... this.
11031 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11033 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11034 (*jcc<mode>_0_r_i387): Ditto.
11035 (*jccxf_r_i387): Ditto.
11036 (*jcc<mode>_r_i387): Ditto.
11037 (*jccu<mode>_r_i387): Ditto.
11038 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11039 (*jcc): Rename from *jcc_1.
11041 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11043 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11044 deferred rescans after the lvx/stvx recombination pre-pass.
11046 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11048 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11049 memory operation instruction support.
11050 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11051 (rs6000-ibm-aix[789]*): Likewise.
11052 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11053 Document new functions.
11055 2017-10-09 Richard Biener <rguenther@suse.de>
11057 PR tree-optimization/82397
11058 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11059 equality only for semantically equal trees.
11061 2017-10-09 Richard Biener <rguenther@suse.de>
11063 PR tree-optimization/82449
11064 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11065 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11066 allow constant addresses.
11067 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11070 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11072 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11075 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11078 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11081 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11084 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11086 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11089 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11090 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11092 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11094 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11095 (WI_BINARY_PREDICATE_RESULT): Likewise.
11096 (wi::binary_traits::operator_result): New type.
11097 (wi::binary_traits::predicate_result): Likewise.
11098 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11099 (generic_wide_int::operator==, generic_wide_int::operator!=)
11100 (generic_wide_int::operator&, generic_wide_int::and_not)
11101 (generic_wide_int::operator|, generic_wide_int::or_not)
11102 (generic_wide_int::operator^, generic_wide_int::operator+
11103 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11105 (operator~, unary operator-, operator==, operator!=, operator&)
11106 (operator|, operator^, operator+, binary operator-, operator*): New
11108 * expr.c (get_inner_reference): Use wi::bit_and_not.
11109 * fold-const.c (fold_binary_loc): Likewise.
11110 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11111 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11112 (bit_value_binop): Likewise.
11113 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11114 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11115 (extract_range_from_binary_expr_1): Likewise.
11116 (masked_increment): Likewise.
11117 (simplify_bit_ops_using_ranges): Likewise.
11119 2017-10-09 Martin Jambor <mjambor@suse.cz>
11122 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11123 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11124 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11126 (hsa_fixup_mov_insn_type): New function.
11127 (hsa_op_with_type::get_in_type): Use it.
11128 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11129 immediates in an assert.
11130 (hsa_op_with_type::extend_int_to_32bit): New method.
11131 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11132 types. Convert to dest type if necessary.
11133 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11134 (reg_for_gimple_ssa): Pass false as min32int to
11135 hsa_type_for_scalar_tree_type.
11136 (gen_hsa_addr): Fixup type when creating addresable temporary.
11137 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11138 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11139 necessary. Call hsa_fixup_mov_insn_type.
11140 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11141 extend operands and convert to dest type if necessary.
11142 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11143 to dest type if necessary.
11144 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11145 if conversion nt necessary and size matches.
11146 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11147 to dest type if necessary.
11148 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11149 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11151 (gen_hsa_clrsb): Likewise.
11152 (gen_hsa_ffs): Likewise.
11153 (gen_hsa_divmod): Extend operands and convert to dest type if
11155 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11157 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11159 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11160 Remove empty default arguments. Use a brace block as output
11162 (conditional return): Ditto.
11164 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11165 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11167 (group_ending_nop): Ditto.
11168 (doloop_end): Ditto.
11169 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11170 (splitters for those): Ditto.
11172 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11174 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11175 a conditional jump (and the compare for it) so that pc_rtx is the
11177 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11178 for the deleted and renamed ctr<mode>_internal[234] patterns.
11179 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11180 Delete second conditional return pattern.
11181 (ctr<mode>_internal2): Delete this second bdnz pattern.
11182 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11183 (ctr<mode>_internal4): Delete this second bdz pattern.
11185 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11187 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11188 (always_initialized_rtx_for_ssa_name_p): New predicate.
11189 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11190 (finish_out_of_ssa): Free new field of SA.
11191 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11192 * tree-ssa-coalesce.c: Include tree-ssa.h.
11193 (get_parm_default_def_partitions): Remove extern keyword.
11194 (get_undefined_value_partitions): New function.
11195 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11196 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11197 uninitialized bits.
11198 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11200 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11202 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11204 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11206 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11207 for targets that preffer 128bit.
11209 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11211 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11213 2017-10-08 Olivier Hainque <hainque@adacore.com>
11215 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11216 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11218 (thumb_set_return_address): Likewise.
11220 2017-10-08 Olivier Hainque <hainque@adacore.com>
11222 * common/config/arm/arm-common.c (arm_except_unwind_info):
11223 Handle DWARF2_UNWIND_INFO.
11225 2017-10-07 Michael Collison <michael.collison@arm.com>
11227 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11230 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11232 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11233 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11234 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11235 defined, force the creation of a new block for a dispatch label.
11237 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11239 * invoke.texi (Wsuggest-attribute=cold): Document.
11240 * common.opt (Wsuggest-attribute=cold): New
11241 * ipa-pure-const.c (warn_function_cold): New function.
11242 * predict.c (compute_function_frequency): Use it.
11243 * predict.h (warn_function_cold): Declare.
11245 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11247 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11250 2017-10-06 Martin Liska <mliska@suse.cz>
11252 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11253 keyword for member functions.
11254 (struct sanopt_tree_couple): New struct.
11255 (struct sanopt_tree_couple_hash): New function.
11256 (struct sanopt_ctx): Add new hash_map.
11257 (has_dominating_ubsan_ptr_check): New function.
11258 (record_ubsan_ptr_check_stmt): Likewise.
11259 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11260 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11261 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11263 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11266 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11267 aarch64_simd_valid_immediate on CONST_VECTORs.
11268 (aarch64_reg_or_bic_imm): Likewise.
11270 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11272 PR rtl-optimization/82396
11273 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11275 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11277 * graphite-dependences.c (scop_get_reads): Move code to...
11278 (scop_get_must_writes): Move code to...
11279 (scop_get_may_writes): Move code to...
11280 (scop_get_reads_and_writes): ... here.
11281 (scop_get_dependences): Call scop_get_reads_and_writes.
11283 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11285 PR tree-optimization/82434
11286 * fold-const.h (can_native_encode_type_p,
11287 can_native_encode_string_p): Remove.
11288 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11289 don't encode anything, just return what would be otherwise returned.
11290 (native_encode_fixed, native_encode_complex, native_encode_vector):
11292 (native_encode_string): Likewise. Inline by hand
11293 can_native_encode_string_p.
11294 (can_native_encode_type_p): Remove.
11295 (can_native_encode_string_p): Remove.
11296 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11297 STRING_CSTs using can_native_encode_string_p, test all
11298 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11299 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11300 argument from native_encode_expr.
11301 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11302 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11305 2017-10-06 Richard Biener <rguenther@suse.de>
11307 PR tree-optimization/82397
11308 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11309 operand_equal_p but rely on data_ref_compare_tree for detecting
11311 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11312 to match up with dr_group_sort_cmp.
11314 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11317 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11319 * config/s390/s390-builtin-types.def: Regenerate.
11321 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11324 * config/s390/s390-builtin-types.def: Regenerate.
11325 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11326 Change flag from B_VXE to B_VX.
11327 (s390_vec_min_dbl): Remove B_VXE flag.
11329 2017-10-06 Richard Biener <rguenther@suse.de>
11331 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11332 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11333 translate_isl_ast_to_gimple::is_valid_rename,
11334 translate_isl_ast_to_gimple::get_rename,
11335 translate_isl_ast_to_gimple::get_def_bb_for_const,
11336 translate_isl_ast_to_gimple::get_new_name,
11337 translate_isl_ast_to_gimple::collect_all_ssa_names,
11338 translate_isl_ast_to_gimple::copy_loop_phi_args,
11339 translate_isl_ast_to_gimple::collect_all_ssa_names,
11340 translate_isl_ast_to_gimple::copy_loop_phi_args,
11341 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11342 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11343 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11344 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11345 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11346 translate_isl_ast_to_gimple::copy_cond_phi_args,
11347 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11348 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11349 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11350 translate_isl_ast_to_gimple::rename_uses,
11351 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11352 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11353 (set_rename_for_each_def): Likewise.
11354 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11355 here. Handle rewriting SCEV analyzable uses here.
11356 (copy_bb_and_scalar_dependences): Generate code for PHI
11358 (graphite_regenerate_ast_isl): Adjust.
11359 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11360 (add_write, add_read): New functions.
11361 (build_cross_bb_scalars_def): Use it and simplify.
11362 (build_cross_bb_scalars_use): Likewise.
11363 (graphite_find_cross_bb_scalar_vars): Inline into...
11364 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11365 simulating out-of-SSA. Compute liveout and add dependencies.
11366 (build_scops): Force an empty entry block.
11367 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11369 (sese_build_liveouts): Declare.
11370 (sese_trivially_empty_bb_p): Likewise.
11371 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11372 compute liveout and debug_liveout.
11373 (sese_bad_liveouts_use): Remove.
11374 (sese_reset_debug_liveouts_bb): Likewise.
11375 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11376 (sese_build_liveouts): Build liveout and debug_liveout and store
11378 (new_sese_info): Adjust.
11379 (free_sese_info): Likewise.
11380 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11381 do not build liveout here.
11382 (move_sese_in_condition): Adjust region entry.
11383 (scev_analyzable_p): Match up with chrec_apply requirements.
11384 (sese_trivially_empty_bb_p): New.
11385 * tree-into-ssa.c (get_reaching_def): Properly support generating
11386 default-defs for incremental rewrite of anonymous names.
11388 2017-10-06 Richard Biener <rguenther@suse.de>
11390 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11391 precision do not perform modulo reduction.
11393 2017-10-06 Richard Biener <rguenther@suse.de>
11395 PR tree-optimization/82436
11396 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11397 conservatively choose the vectorization factor when checking
11398 whether we can perform the required load permutation.
11399 (vect_transform_slp_perm_load): Assert when we may not fail.
11401 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11403 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11404 message for incompatible -msdata=* and -mcall-* options.
11406 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11408 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11409 rate for post-reload scheduling.
11411 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11413 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11415 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11417 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11418 to improve monte carlo in scimark.
11420 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11422 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11423 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11424 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11425 pentium4_cost, nocona_cost): Set reassociation width to 1.
11426 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11427 width to 2 for fp operations and 1 otherwise.
11428 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11430 (atom_cost): Set reassociation width to 2.
11431 (slm_cost, generic_cost): Set fp reassociation width
11432 to 2 and 1 otherwise.
11433 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11434 (core_cost): Set fp reassociation width to 4 and vector to 2.
11435 (ix86_reassociation_width): Rewrite using cost table; special case
11436 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11437 and TARGET_AVX128_OPTIMAL.
11438 * config/i386/i386.h (processor_costs): Add
11439 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11440 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11441 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11442 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11443 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11444 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11446 2017-10-05 Nathan Sidwell <nathan@acm.org>
11448 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11450 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11452 * config/arm/arm.c (arm_test_fpu_data): New.
11453 (arm_run_selftests): Call arm_test_fpu_data.
11455 2017-10-04 Nathan Sidwell <nathan@acm.org>
11457 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11459 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11461 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11463 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11464 check type for aarch64_simd_valid_immediate.
11465 (aarch64_output_simd_mov_immediate): Update prototype.
11466 (aarch64_simd_valid_immediate): Update prototype.
11467 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11468 support for ORR-immediate.
11469 (and<mode>3): modified pattern to add support for BIC-immediate.
11470 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11471 now checks for valid immediate for BIC and ORR based on new enum
11473 (aarch64_output_simd_mov_immediate): Function now used to output
11474 BIC/ORR imm as well based on new enum argument.
11475 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11477 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11478 (aarch64_reg_or_bic_imm): Likewise.
11480 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11482 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11483 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11485 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11489 PR rtl-optimization/82396
11490 * haifa-sched.c (autopref_multipass_init): Simplify
11492 (autopref_rank_data): Simplify sort order.
11493 * sched-int.h (autopref_multipass_data_): Remove
11494 multi_mem_insn_p, min_offset and max_offset.
11496 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11498 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11500 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11502 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11503 vect_doubleint_cvt.
11505 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11507 * doc/sourcebuild.texi: Document vect_long_mult.
11509 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11511 PR tree-optimization/82413
11512 * fold-const.c (build_range_check): Use widest_int when comparing
11513 the maximum ETYPE value with HIGH.
11515 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11517 PR rtl-optimization/82396
11518 * haifa-sched.c (autopref_multipass_init): Simplify
11520 (autopref_rank_data): Simplify sort order.
11521 * sched-int.h (autopref_multipass_data_): Remove
11522 multi_mem_insn_p, min_offset and max_offset.
11524 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11526 PR tree-optimization/82381
11527 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11528 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11531 PR tree-optimization/82374
11532 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11533 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11534 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11535 current_function_decl to the new decl.
11537 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11539 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11540 helper macro for IEEE float128 hardware built-in functions.
11541 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11543 (TRUNCF128_ODD): Likewise.
11544 (ADDF128_ODD): Likewise.
11545 (SUBF128_ODD): Likewise.
11546 (MULF128_ODD): Likewise.
11547 (DIVF128_ODD): Likewise.
11548 (FMAF128_ODD): Likewise.
11549 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11550 UNSPEC_TRUNC_ROUND_TO_ODD.
11551 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11552 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11553 floating point round to odd instructions.
11554 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11555 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11556 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11557 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11558 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11559 (trunc<mode>sf2_hw): Change the truncate with round to odd
11560 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11561 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11562 to odd hardware instructions.
11563 (sub<mode>3_odd): Likewise.
11564 (mul<mode>3_odd): Likewise.
11565 (div<mode>3_odd): Likewise.
11566 (sqrt<mode>2_odd): Likewise.
11567 (fma<mode>4_odd): Likewise.
11568 (fms<mode>4_odd): Likewise.
11569 (nfma<mode>4_odd): Likewise.
11570 (nfms<mode>4_odd): Likewise.
11571 (trunc<mode>df2_odd): Change the truncate with round to odd
11572 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11574 * doc/extend.texi (PowerPC built-in functions): Update documentation
11575 for existing IEEE float128-bit built-in functions. Add built-in
11576 functions that generate the IEEE 128-bit floating point round to
11579 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11581 PR rtl-optimization/77729
11582 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11583 to (X&(C1&~C2))|C2 transformations.
11585 2017-10-03 Martin Jambor <mjambor@suse.cz>
11587 PR tree-optimization/82363
11588 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11589 mismatch, mark lacc written regardless of racc.
11591 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11593 PR tree-optimization/82381
11594 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11595 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11596 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11597 fallthrough into reassoc_stmt_dominates_stmt_p.
11600 * combine.c (combine_instructions): Don't combine in unreachable
11603 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11606 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11607 function to not use the have_cpu variable. Do not set cpu_index,
11608 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11609 or the default cpu.
11610 (rs6000_valid_attribute_p): Remove duplicate initializations of
11611 old_optimize and func_optimize.
11612 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11613 (rs6000_activate_target_options): Make global.
11614 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11617 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11619 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11620 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11621 Set *poffset to 0 and *psize and *pmax_size to -1 if
11622 *poffset + *psize overflows in HOST_WIDE_INT.
11624 PR tree-optimization/82387
11625 PR tree-optimization/82388
11626 PR tree-optimization/82389
11627 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11628 instead of live_bytes non-NULL.
11630 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11633 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11636 2017-10-02 Richard Biener <rguenther@suse.de>
11638 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11639 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11640 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11642 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11644 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11645 requested precision matches the type's.
11646 * calls.c (alloc_max_size): Calculate the new candidate size as
11647 a widest_int and use wi::to_widest when comparing it with the
11648 current candidate size.
11649 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11650 zero rather than integer_zero_node.
11651 * match.pd: Check for a no-op conversion before using wi::add
11652 rather than after. Use tree_to_uhwi when summing small shift
11653 counts into an unsigned int.
11655 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11660 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11661 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11662 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11663 POINTER_AND_FP_REGS.
11665 2017-10-02 Richard Biener <rguenther@suse.de>
11667 PR tree-optimization/82355
11668 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11669 a mapping for the enclosing loop but avoid generating one for
11670 the loop tree root.
11671 (copy_bb_and_scalar_dependences): Remove premature codegen
11672 error on PHIs in blocks duplicated into multiple places.
11673 * graphite-scop-detection.c
11674 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11675 in the region use it as loop and nest to analyze the DR in.
11676 (try_generate_gimple_bb): Likewise.
11677 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11678 (add_loop_constraints): For blocks in a loop not in the region
11679 create a dimension with a single iteration.
11680 * sese.h (gbb_loop_at_index): Remove assert.
11682 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11684 * omp-expand.c (adjust_context_scope): New function.
11685 (expand_parallel_call): Call adjust_context_scope.
11687 2017-10-01 Jeff Law <law@redhat.com>
11689 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11690 dom_opt_dom_walker class with direct access to private members.
11691 Add comments. Call test_for_singularity.
11692 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11693 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11694 m_dummy_cond anymore.
11695 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11697 (pass_dominator:execute): Build the dummy_cond here and pass it
11698 to the dom_opt_dom_walker ctor.
11699 (test_for_singularity): New function.
11701 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11702 Maya Rashish <coypu@sdf.org>
11704 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11705 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11706 (alpha*-*-netbsd*) Use nbsd_tm_file.
11707 (arm*-*-netbsdelf*) Likewise.
11708 (i[34567]86-*-netbsdelf*) Likewise.
11709 (x86_64-*-netbsd*) Likewise.
11710 (mips*-*-netbsd*) Likewise.
11711 (powerpc-*-netbsd*) Likewise.
11712 (sh*-*-netbsd*) Likewise.
11713 (sparc-*-netbsdelf*) Likewise.
11714 (sparc64-*-netbsd*) Likewise.
11715 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
11717 (vax-*-netbsdelf*) Likewise.
11718 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
11719 (UINT_FAST8_TYPE) Likewise.
11720 (INT_FAST16_TYPE) Check CHAR_FAST16.
11721 (UINT_FAST16_TYPE) Likewise.
11723 2017-09-30 Jakub Jelinek <jakub@redhat.com>
11726 * config/i386/i386.md
11727 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
11728 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
11729 *divmodsi4_zext_2): New define_insn_and_split.
11730 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
11731 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
11732 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
11733 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
11734 New define_insn_and_split.
11735 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11736 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11737 operands[1] having DImode when mode is SImode.
11739 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11740 always SImode for DIV and MOD in REG_EQUAL notes.
11742 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11744 PR middle-end/82319
11745 * match.pd: Fix handling of NaNs in pattern.
11747 2017-09-29 Jeff Law <law@redhat.com>
11749 * sbitmap.c (bitmap_bit_in_range_p): New function.
11750 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11751 * tree-ssa-dse.c (live_bytes_read): New function.
11752 (dse_classify_store): Ignore reads of dead bytes.
11754 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11755 typos and whitespace errors.
11756 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11757 * config/s390/s390.c (s390_emit_prologue): Likewise.
11759 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11762 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11763 with a symbol for LRA.
11765 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11767 PR rtl-optimization/82338
11768 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11770 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11772 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11773 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11774 (qsort_chk): Declare.
11775 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11776 (qsort_chk): New function.
11778 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11780 PR tree-optimization/82337
11781 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11782 phi definition if the PHI result appears in an abnormal PHI.
11783 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11784 basis appears in an abnormal PHI.
11786 2017-09-29 Richard Biener <rguenther@suse.de>
11788 * graphite-isl-ast-to-gimple.c
11789 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11790 (binary_op_to_tree): Use it.
11791 (get_rename_from_scev): Likewise.
11792 (copy_loop_phi_nodes): Likewise.
11793 (copy_bb_and_scalar_dependences): Likewise.
11794 (translate_pending_phi_nodes): Likewise.
11796 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11799 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11800 for movabsq $(i32 << shift), r64.
11802 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11804 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11805 index when encoding %esp as %rsp to avoid 0x67 prefix.
11807 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11809 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11810 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11812 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11814 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11815 Extensions with more than 16 double VFP registers.
11816 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11817 to_clear_mask and all code related to it. Replace the remaining
11818 entry by a sbitmap and adapt code accordingly.
11820 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11822 * brig-builtins.def: Change pure attributes to const.
11824 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11826 * config.gcc (default_gnu_indirect_function): Default to yes for
11827 sparc*-*-linux* with glibc.
11829 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11831 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11832 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11833 when creating .init_array and .fini_array sections with priority
11836 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11839 * config/aarch64/aarch64.c
11840 (aarch64_builtin_support_vector_misalignment): Always return false
11841 when misalignment is unknown.
11843 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11845 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11846 this function to return false if the definition used by the swap
11847 instruction is artificial, or if the memory address from which the
11848 constant value is loaded is not represented by a base address held
11849 in a register or if the base address register is a frame or stack
11850 pointer. Additionally, return false if the base address of the
11851 loaded constant is a SYMBOL_REF but is not considered to be a
11853 (replace_swapped_load_constant): New function.
11854 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11855 loaded constant vector with a load of a swapped constant vector.
11857 2017-09-27 Carl Love <cel@us.ibm.com>
11859 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11860 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11861 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11864 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11866 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11867 first, always call autopref_rank_data otherwise.
11869 2017-09-27 Richard Biener <rguenther@suse.de>
11871 * graphite-scop-detection.c (find_scop_parameters): Move
11872 loop bound handling ...
11873 (gather_bbs::before_dom_children): ... here, avoiding the need
11874 to build scop_info->loop_nest.
11875 (record_loop_in_sese): Remove.
11876 * sese.h (sese_info_t::loop_nest): Remove.
11877 * sese.c (new_sese_info): Do not allocate loop_nest.
11878 (free_sese_info): Do not free loop_nest.
11880 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11883 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11884 lhs from calls if the lhs has addressable type.
11886 2017-09-27 Richard Biener <rguenther@suse.de>
11888 * graphite.h (scop::max_alias_set): New member.
11889 * graphite-scop-detection.c: Remove references to non-existing
11890 --param in comments.
11891 (build_alias_sets): Record the maximum alias set used for drs.
11892 (build_scops): Support zero as unlimited for
11893 --param graphite-max-arrays-per-scop.
11894 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11895 and inline into ...
11896 (build_poly_sr_1): ... here. Compute alias set based on the
11897 maximum alias set used for drs rather than
11898 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11900 2017-09-27 Richard Biener <rguenther@suse.de>
11902 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11903 --param loop-block-tile-size=0 to disable tiling.
11905 2017-09-27 Richard Biener <rguenther@suse.de>
11907 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11908 (graphite-max-nb-scop-params): Document special value zero.
11909 * domwalk.h (dom_walker::STOP): New symbolical constant.
11910 (dom_walker::dom_walker): Add optional parameter for bb to
11912 (dom_walker::~dom_walker): Declare.
11913 (dom_walker::before_dom_children): Document STOP return value.
11914 (dom_walker::m_user_bb_to_rpo): New member.
11915 (dom_walker::m_bb_to_rpo): Likewise.
11916 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11917 mapping here if not provided by the user.
11918 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11919 provided by the user.
11920 (dom_walker::STOP): Define.
11921 (dom_walker::walk): Do not compute bb to RPO mapping here.
11922 Support STOP return value from before_dom_children to stop
11924 * graphite-optimize-isl.c (optimize_isl): If the schedule
11925 is the same still generate code if -fgraphite-identity
11926 or -floop-parallelize-all are given.
11927 * graphite-scop-detection.c: Include cfganal.h.
11928 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11930 (gather_bbs::before_dom_children): Return STOP for BBs
11932 (build_scops): Compute bb to RPO mapping and pass it to
11933 the domwalk. Treat --param graphite-max-nb-scop-params=0
11934 as not limiting the number of params.
11935 * graphite.c (graphite_initialize): Remove limit on the
11936 number of basic-blocks in a function.
11937 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11938 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11939 default value of 10.
11941 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11943 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11944 Adjust code to eliminate needing to do the shift right 32-bits
11945 operation after XSCVDPSPN.
11947 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11949 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11950 ((X / Y) != 0 -> X >= Y): Likewise.
11952 2017-09-26 Carl Love <cel@us.ibm.com>
11954 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11955 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11956 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11957 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11958 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11959 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11960 definitions and overloading.
11961 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11962 statement for P9V_BUILTIN_XST_LEN_R.
11963 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11964 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11965 define_expand and define_insn for the instructions and builtins.
11966 * doc/extend.texi: Update the built-in documentation file for the new
11967 built-in functions.
11968 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11969 define_insn for the instructions
11971 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11974 * gcc/config/netbsd-protos.h: New file.
11975 * gcc/config/netbsd.c: New file.
11976 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11977 * gcc/config/t-netbsd: New file.
11978 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11979 (tmake_file) Add t-netbsd.
11980 (extra_objs) Add netbsd.o.
11982 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11986 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11988 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11990 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11991 sign extension from a vector register to a GPR by doing a 32-bit
11992 direct move and then an EXTSW.
11993 (extendsi<mode>2 splitter): Likewise.
11994 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11995 right or vector extract after doing XSCVDPSPN. Use
11996 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11998 (movdi_from_sf_zero_ext): Likewise.
11999 (reload_gpr_from_vsxsf): Likewise.
12000 (p8_mfvsrd_4_disf): Delete, no longer used.
12001 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12002 then needing to move the SFmode to a GPR to use the XSCVDPSP
12003 instruction instead of FRSP and XSCVDPSPN.
12004 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12005 it is adjacent to the other XSCVSPDP insns.
12006 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12007 SFmode to be in traditional Altivec registers.
12008 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12009 (vsx_xscvspdpn): Likewise.
12010 (vsx_xscvspdpn_scalar): Likewise.
12012 2017-09-26 Martin Jambor <mjambor@suse.cz>
12014 * tree-sra.c (compare_access_positions): Put integral types first,
12015 stabilize sorting of integral types, remove conditions putting
12016 non-full-precision integers last.
12017 (sort_and_splice_var_accesses): Disable scalarization if a
12018 non-integert would be represented by a non-full-precision integer.
12020 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12022 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12023 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12024 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12025 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12026 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12027 conditionals inside the function instead of around it. Call
12028 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12029 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12031 2017-09-26 Richard Biener <rguenther@suse.de>
12033 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12035 (scop_detection::build_scop_breadth): ... this. Removed.
12036 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12037 (scop_detection::harmful_stmt_in_bb): Likewise.
12038 (scop_detection::graphite_can_represent_stmt): Likewise.
12039 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12040 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12041 (scop_detection::can_represent_loop_1): ... this. Removed.
12042 (scop_detection::harmful_loop_in_region): Simplify after inlining
12043 the above and remove more quadraticness.
12044 (build_scops): Adjust.
12045 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12048 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12051 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12052 REGNO (base) == SP_REG if base is a REG.
12054 PR middle-end/35691
12055 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12056 if it is different SSA_NAME.
12057 (optimize_range_tests_cmp_bitwise): New function.
12058 (optimize_range_tests): Call it.
12060 2017-09-26 Richard Biener <rguenther@suse.de>
12062 PR tree-optimization/82321
12063 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12064 for the def being inside the loop.
12066 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12068 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12070 * config/s390/s390-builtins.def: Fix constraint on op4.
12072 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12074 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12075 independent expanders.
12076 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12077 ("vec_ordered", "vec_unordered"): New expanders.
12079 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12081 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12084 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12086 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12087 vec_unpacks_lo_v16qi.
12088 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12090 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12092 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12093 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12094 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12096 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12098 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12100 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12102 ("*vec_slb<mode>"): New insn pattern.
12103 ("vec_shr_<mode>"): New expander.
12104 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12105 and force the shift count operand to V16QImode.
12106 ("vec_srb<mode>"): Set shift count mode to V16QI.
12108 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12110 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12111 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12112 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12114 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12117 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12119 2017-09-26 Richard Biener <rguenther@suse.de>
12121 PR tree-optimization/82320
12122 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12125 2017-09-25 Jeff Law <law@redhat.com>
12127 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12128 prototype for new argument.
12129 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12130 mostly extracted from rs6000_emit_allocate_stack.
12131 (rs6000_emit_probe_stack_range_stack_clash): New function.
12132 (rs6000_emit_allocate_stack): Call
12133 rs6000_emit_probe_stack_range_stack_clash as needed.
12134 (rs6000_emit_probe_stack_range): Add additional argument
12135 to call to gen_probe_stack_range{si,di}.
12136 (output_probe_stack_range): New.
12137 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12138 (output_probe_stack_range_stack_clash): New.
12139 (rs6000_emit_prologue): Emit notes into dump file as requested.
12140 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12141 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12142 Add additional operand and pass it to output_probe_stack_range.
12144 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12146 PR tree-optimization/82163
12147 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12148 (checking_verify_loop_closed_ssa): New parameter.
12149 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12150 (check_loop_closed_ssa_stmt): Delete.
12151 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12152 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12153 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12156 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12158 * brig-builtins.def: Treat HSAIL barrier builtins as
12159 setjmp/longjump style functions.
12161 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12163 * target.def (constant_alignment): New hook.
12164 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12165 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12166 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12167 * doc/tm.texi: Regenerate.
12168 * targhooks.h (default_constant_alignment): Declare.
12169 (constant_alignment_word_strings): Likewise.
12170 * targhooks.c (default_constant_alignment): New function.
12171 (constant_alignment_word_strings): Likewise.
12172 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12173 instead of CONSTANT_ALIGNMENT.
12174 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12175 (force_const_mem): Likewise.
12176 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12177 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12178 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12179 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12180 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12182 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12183 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12184 constant_alignment_word_strings.
12185 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12186 (CONSTANT_ALIGNMENT): Likewise.
12187 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12188 (arm_constant_alignment): New function.
12189 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12190 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12191 constant_alignment_word_strings.
12192 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12193 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12194 constant_alignment_word_strings.
12195 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12196 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12197 (cris_constant_alignment): New function.
12198 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12199 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12200 (epiphany_constant_alignment): New function.
12201 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12202 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12203 constant_alignment_word_strings.
12204 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12205 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12206 constant_alignment_word_strings.
12207 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12208 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12209 constant_alignment_word_strings.
12210 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12211 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12212 * config/i386/i386.c (ix86_constant_alignment): Make static.
12213 Use the same interface as the target hook.
12214 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12215 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12216 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12217 constant_alignment_word_strings.
12218 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12219 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12220 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12221 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12222 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12223 constant_alignment_word_strings.
12224 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12225 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12226 constant_alignment_word_strings.
12227 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12228 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12229 constant_alignment_word_strings.
12230 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12231 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12233 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12234 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12235 * config/mips/mips.c (mips_constant_alignment): New function.
12236 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12237 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12238 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12239 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12240 (mmix_constant_alignment): Make static. Use the same interface
12241 as the target hook.
12242 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12243 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12244 constant_alignment_word_strings.
12245 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12246 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12247 constant_alignment_word_strings.
12248 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12249 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12250 constant_alignment_word_strings.
12251 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12252 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12253 (rs6000_constant_alignment): New function.
12254 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12255 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12256 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12257 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12258 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12259 (rs6000_constant_alignment): New function.
12260 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12261 * config/s390/s390.c (s390_constant_alignment): New function.
12262 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12263 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12264 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12265 constant_alignment_word_strings.
12266 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12267 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12268 (sparc_constant_alignment): New function.
12269 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12270 * config/spu/spu.c (spu_constant_alignment): New function.
12271 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12272 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12273 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12274 constant_alignment_word_strings.
12275 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12276 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12277 constant_alignment_word_strings.
12278 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12279 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12280 constant_alignment_word_strings.
12281 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12282 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12283 (visium_constant_alignment): New function.
12284 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12285 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12286 (xtensa_constant_alignment): New function.
12287 * system.h (CONSTANT_ALIGNMENT): Poison.
12289 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12291 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12292 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12293 (rs6000_builtin_valid_without_lhs): New helper function.
12294 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12295 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12297 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12299 * target.h (vec_perm_indices): Use unsigned short rather than
12301 (auto_vec_perm_indices): Likewise.
12302 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12303 Use unsigned int rather than unsigned char.
12304 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12306 2017-09-25 Richard Biener <rguenther@suse.de>
12308 * cfgloop.h (sort_sibling_loops): Declare.
12309 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12310 (sort_sibling_loops): New function sorting the sibling loop list
12312 * graphite.c (graphite_transform_loops): Sort sibling loops.
12314 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12316 * target.def (vec_perm_const_ok): Change sel parameter to
12318 * optabs-query.c (can_vec_perm_p): Update accordingly.
12319 * doc/tm.texi: Regenerate.
12320 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12321 auto_vec_perm_indices and remove separate nelt field.
12322 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12323 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12324 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12325 (aarch64_expand_vec_perm_const): Update accordingly.
12326 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12327 to vec_perm_indices.
12328 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12329 auto_vec_perm_indices and remove separate nelt field.
12330 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12331 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12332 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12334 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12335 to vec_perm_indices.
12336 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12337 sel to vec_perm_indices.
12338 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12339 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12340 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12342 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12345 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12348 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12349 on the FUNCTION_DECL function context if it has a DIE that is a
12352 2017-09-25 Richard Biener <rguenther@suse.de>
12354 PR tree-optimization/82285
12355 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12358 2017-09-25 Tom de Vries <tom@codesourcery.com>
12362 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12365 2017-09-25 Richard Biener <rguenther@suse.de>
12367 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12368 ISL errors other than isl_error_quota happen. Dump if the
12369 schedule is the same.
12370 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12371 errors instead of aborting inside ISL.
12373 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12376 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12377 of libgcc_eh for m64.
12378 * config/i386/darwin64.h: Likewise.
12380 2017-09-25 Richard Biener <rguenther@suse.de>
12382 PR middle-end/82144
12383 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12384 attribute for incomplete types nor twice for complete ones.
12386 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12389 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12390 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12393 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12396 * config/i386/i386.opt (mprefer-avx256): Use
12397 ix86_target_flags variable.
12398 * config/i386/i386.c (ix86_target_string): Move
12399 -mprefer-avx256 to flag2_opts.
12401 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12403 PR middle-end/35691
12404 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12405 and x != -1 | y != -1 into (x & y) != -1.
12407 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12409 * config.gcc: Add new case statement to set
12410 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12411 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12412 s390x-*-linux* case statements. Added aarch64 to the list of
12413 supported architectures.
12415 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12417 PR tree-optimization/82289
12418 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12419 STMT_VINFO_RELEVANT_P.
12421 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12422 Alan Hayward <alan.hayward@arm.com>
12423 David Sherwood <david.sherwood@arm.com>
12425 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12426 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12427 (extract_range_from_binary_expr_1): Don't call
12428 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12430 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12431 Alan Hayward <alan.hayward@arm.com>
12432 David Sherwood <david.sherwood@arm.com>
12434 * target.def (preferred_vector_alignment): New hook.
12435 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12437 * doc/tm.texi: Regenerate.
12438 * targhooks.h (default_preferred_vector_alignment): Declare.
12439 * targhooks.c (default_preferred_vector_alignment): New function.
12440 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12442 (DR_TARGET_ALIGNMENT): New macro.
12443 (aligned_access_p): Update commentary.
12444 (vect_known_alignment_in_bytes): New function.
12445 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12447 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12448 Calculate the misalignment based on the target alignment rather than
12450 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12451 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12452 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12453 the target alignment, rather than masking the element misalignment
12454 with the number of elements in a vector. Also use the target
12455 alignment when calculating the maximum number of peels.
12456 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12457 instead of TYPE_ALIGN_UNIT.
12458 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12459 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12460 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12461 (vect_create_data_ref_ptr): Likewise.
12462 (vect_setup_realignment): Realign by ANDing with
12463 -DR_TARGET_MISALIGNMENT.
12464 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12465 the number of peels based on DR_TARGET_ALIGNMENT.
12466 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12467 with the guaranteed alignment boundary when deciding whether
12469 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12470 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12471 (ensure_base_align): Remove stmt_info parameter. Get the
12472 target base alignment from DR_TARGET_ALIGNMENT.
12473 (vectorizable_store): Update call accordingly. Interpret
12474 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12476 (vectorizable_load): Likewise.
12478 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12479 Alan Hayward <alan.hayward@arm.com>
12480 David Sherwood <david.sherwood@arm.com>
12482 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12483 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12484 (vect_enhance_data_refs_alignment): Likewise.
12486 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12488 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12489 error. Only quit immediately if parsing is complete.
12490 (BEGIN): Initialize fatal_err and parse_done.
12491 (begin fpu, end fpu): Check number of arguments.
12492 (begin arch, end arch): Likewise.
12493 (begin cpu, end cpu): Likewise.
12494 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12495 (optalias): Likewise.
12497 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12499 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12500 * config/arm/arm-isa.h: Delete. Move definitions to ...
12501 * arm-cpus.in: ... here. Use new feature and fgroup values.
12502 * config/arm/arm.c (arm_option_override): Use lower case for feature
12504 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12505 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12506 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12508 (print_isa_bits_for): New function.
12509 (gen_isa): New function.
12510 (gen_comm_data): Use print_isa_bits_for.
12511 (define feature): New keyword.
12512 (define fgroup): New keyword.
12513 * config/arm/t-arm (TM_H): Remove.
12514 (GTM_H): Add arm-isa.h.
12515 (arm-isa.h): Add rule to generate file.
12516 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12517 case for feature bit names.
12519 2017-09-22 Richard Biener <rguenther@suse.de>
12521 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12523 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12524 print of no dependency loops ...
12525 * graphite.c (graphite_transform_loops): ... here.
12526 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12528 (same_close_phi_node, remove_duplicate_close_phi,
12529 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12530 (canonicalize_loop_closed_ssa): ... here and simplify.
12531 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12532 (optimize_isl): Use dump_printf_loc to tell when we stopped
12533 optimizing because of an ISL timeout.
12535 2017-09-22 Richard Biener <rguenther@suse.de>
12537 PR tree-optimization/82291
12538 * tree-if-conv.c (predicate_mem_writes): Make sure to
12539 remove writes in blocks predicated with false.
12541 2017-09-22 Richard Biener <rguenther@suse.de>
12543 * sese.c: Include cfganal.h.
12544 (if_region_set_false_region): Remove.
12545 (create_if_region_on_edge): Likewise.
12546 (move_sese_in_condition): Re-implement without destroying
12549 2017-09-22 Richard Biener <rguenther@suse.de>
12551 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12552 Verify both BBs contain loop PHI nodes before dispatching to
12553 copy_loop_phi_args.
12554 (graphite_regenerate_ast_isl): Do not recompute dominators,
12555 do not verify three times. Restructure for clarity.
12556 * graphite-scop-detection.c (same_close_phi_node,
12557 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12558 defined_in_loop_p, canonicalize_loop_closed_ssa,
12559 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12560 checking and SSA rewrite, move to ...
12561 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12562 (graphite_initialize): Do not pass in ctx, do not reset the
12563 SCEV cache, compute only dominators.
12564 (graphite_transform_loops): Allocate ISL ctx after
12565 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12566 Maintain post-dominators only around build_scops.
12567 * sese.c (if_region_set_false_region): Make static. Free
12568 and recompute dominators.
12569 (move_sese_in_condition): Assert we don't get called with
12570 post-dominators computed.
12571 * sese.h (if_region_set_false_region): Remove.
12573 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12575 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12576 mode attribute for TARGET_AVX512VL.
12578 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12580 * config/i386/i386.opt (mprefer-avx256): New option.
12581 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12583 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12584 for TARGET_PREFER_AVX256.
12585 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12587 2017-09-21 Jeff Law <law@redhat.com>
12589 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12590 Fix dump output if the only stack space is for pushed registers.
12592 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12594 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12597 2017-09-21 Martin Sebor <msebor@redhat.com>
12600 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12601 code (in C++) or code that triggers warnings.
12603 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12605 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12607 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12609 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12610 * sched-rgn.c: Ditto.
12611 * sel-sched-ir.c: Ditto.
12613 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12615 * toplev.h (set_random_seed): Adjust return type.
12616 * toplev.c (init_local_tick): Move eager initialization of random_seed
12617 to get_random_seed. Adjust comment.
12618 (init_random_seed): Inline to get_random_seed, delete.
12619 (get_random_seed): Initialize random_seed lazily.
12620 (set_random_seed): Do not return previous value.
12621 (print_switch_value): Do not call get_random_seed.
12623 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12625 * cgraph.c (delete_function_version): New, broken out from...
12626 (cgraph_node::delete_function_version): ...here. Rename to
12627 cgraph_node::delete_function_version_by_decl. Update all uses.
12628 (cgraph_node::remove): Call delete_function_version.
12630 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12633 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12634 VAR_DECLs to which addressable non-volatile parameters are mapped
12635 and for id->retvar after the return value assignment. Clear
12636 id->retval and id->retbnd after inlining.
12638 2017-09-21 Richard Biener <rguenther@suse.de>
12640 PR tree-optimization/82276
12641 PR tree-optimization/82244
12642 * tree-vrp.c (build_assert_expr_for): Set
12643 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12645 (remove_range_assertions): Revert earlier change.
12647 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12650 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12652 2017-09-21 Richard Biener <rguenther@suse.de>
12654 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12655 Restore valid IL after code generation errors.
12656 * graphite.c (graphite_transform_loops): Diagnose code
12657 generation issues as MSG_MISSED_OPTIMIZATION and continue
12658 with processing SCOPs.
12660 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12661 Alan Hayward <alan.hayward@arm.com>
12662 David Sherwood <david.sherwood@arm.com>
12664 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12665 rather than choosing between plus_constant and gen_rtx_<CODE>.
12666 * expr.c (emit_push_insn): Likewise.
12667 (expand_expr_real_2): Likewise.
12669 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12670 Alan Hayward <alan.hayward@arm.com>
12671 David Sherwood <david.sherwood@arm.com>
12673 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12675 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12676 Alan Hayward <alan.hayward@arm.com>
12677 David Sherwood <david.sherwood@arm.com>
12679 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12680 calling tree_to_uhwi.
12682 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12683 Alan Hayward <alan.hayward@arm.com>
12684 David Sherwood <david.sherwood@arm.com>
12686 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12687 INTEGER_CST rather than a negative test for ADDR_EXPR.
12689 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12690 Alan Hayward <alan.hayward@arm.com>
12691 David Sherwood <david.sherwood@arm.com>
12693 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12694 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12696 2017-09-21 Richard Biener <rguenther@suse.de>
12698 PR tree-optimization/71351
12699 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12700 graphite_create_new_loop_guard): Remove, fold remaining parts
12702 (translate_isl_ast_node_for): ... here and simplify.
12704 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12707 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12708 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12709 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12710 alternative always use QI mode, for -Os imov (=R,R) alternative
12711 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
12714 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12715 Jeff Law <law@redhat.com>
12717 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
12718 (allocate_stack_space): New function, partially extracted from
12719 s390_emit_prologue.
12720 (s390_emit_prologue): Track offset to most recent stack probe.
12721 Code to allocate space moved into allocate_stack_space.
12722 Dump actions when no stack is allocated.
12723 (s390_prologue_plus_offset): New function.
12724 (s390_emit_stack_probe): Likewise.
12726 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
12728 * common.opt (Wa, Wl, Wp, g, gz=): Add
12730 (gno-column-info): Remove.
12731 (gcolumn-info): Drop RejectNegative.
12732 (gno-): New prefix.
12733 (gno-record-gcc-switches): Remove.
12734 (grecord-gcc-switches): Drop RejectNegative.
12735 (gno-split-dwarf): Remove.
12736 (gsplit-dwarf): Drop RejectNegative.
12737 (gno-strict-dwarf): Remove.
12738 (gstrict-dwarf): Drop RejectNegative.
12739 * config/darwin.opt (gfull, gused): Add RejectNegative.
12740 * dwarf2out.c (gen_producer_string): Drop
12741 gno-record-gcc-switches handler.
12742 * optc-gen.awk: Add g to prefixes with negative forms.
12743 * opts-common.c (remapping_prefix_p): New.
12744 (find_opt): Check it.
12745 (generate_canonical_option): Test g prefix.
12746 (option_map): Add -gno- mapping.
12747 (add_misspelling_candidates): Check remapping_prefix_p.
12749 2017-09-20 Jeff Law <law@redhat.com>
12751 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12752 thinko in stack clash protection support.
12754 * explow.c (compute_stack_clash_protection_loop_data): Use
12755 CONST_INT_P instead of explicit test. Verify object is a
12756 CONST_INT_P before looking at INTVAL.
12757 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12758 instead of explicit test.
12760 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12763 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12764 address instead of to r1 and r11.
12766 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12768 * config.gcc: Support "knm".
12769 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12770 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12772 * config/i386/i386.c (m_KNM): Define.
12773 (processor_target_table): Add "knm".
12775 (ix86_option_override_internal): Add "knm".
12776 (ix86_issue_rate): Add PROCESSOR_KNM.
12777 (ix86_adjust_cost): Ditto.
12778 (ia32_multipass_dfa_lookahead): Ditto.
12779 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12780 (fold_builtin_cpu): Add M_INTEL_KNM.
12781 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12782 (processor_type): Add PROCESSOR_KNM.
12783 * config/i386/x86-tune.def: Add m_KNM.
12784 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12786 2017-09-20 Richard Biener <rguenther@suse.de>
12788 PR tree-optimization/80213
12789 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12790 are allowed in empty BBs as well.
12791 (canonicalize_loop_closed_ssa): Also look for other complex
12793 (scop_detection::get_sese): Include the loop-closed PHI block
12795 (scop_detection::merge_sese): Remove code adding extra blocks.
12796 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12797 (build_scops): Assert the final returned scop is invalid.
12799 2017-09-20 Richard Biener <rguenther@suse.de>
12801 PR tree-optimization/82264
12802 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12804 (vn_phi_lookup): Likewise.
12805 (vn_phi_insert): Likewise.
12807 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12809 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12810 that fits into uhwi or shwi, add DW_AT_const_value regardless
12811 of early_dwarf without going through RTL, using add_AT_unsigned
12814 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12815 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12816 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12817 (DEBUG_MACRO_SECTION): Likewise.
12818 (DEBUG_LTO_MACRO_SECTION): Likewise.
12819 (DEBUG_STR_DWO_SECTION): Likewise.
12820 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12821 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12822 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12823 (DEBUG_LTO_LINE_STR_SECTION): Define.
12824 (init_sections_and_labels): Initialize debug_line_str_section
12825 variable. Initialize debug_loc_section for -gdwarf-5 to
12826 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12828 2017-09-20 Richard Biener <rguenther@suse.de>
12830 * graphite-sese-to-poly.c (extract_affine): Properly handle
12831 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12833 2017-09-20 Richard Biener <rguenther@suse.de>
12835 PR tree-optimization/81373
12836 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12837 Force SESE live-out defs to be handled even if they are
12840 2017-09-19 Jeff Law <law@redhat.com>
12842 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12843 nothing for stack adjustments with REG_STACK_CHECK.
12844 * sched-deps.c (parse_add_or_inc): Reject insns with
12845 REG_STACK_CHECK from dependency breaking.
12846 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12847 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12848 * reg-notes.def (STACK_CHECK): New note.
12850 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12851 (ix86_expand_prologue): Dump stack clash info as needed.
12852 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12854 * function.c (dump_stack_clash_frame_info): New function.
12855 * function.h (dump_stack_clash_frame_info): Prototype.
12856 (enum stack_clash_probes): New enum.
12858 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12859 flag_stack_clash_protection.
12860 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12861 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12862 (arm_frame_pointer_required): Likewise.
12863 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12864 (ia64_expand_prologue): Likewise.
12865 * config/mips/mips.c (mips_expand_prologue): Likewise.
12866 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12867 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12868 (sparc_flat_expand_prologue): Likewise.
12869 * config/spu/spu.c (spu_expand_prologue): Likewise.
12871 * explow.c: Include "params.h".
12872 (anti_adjust_stack_and_probe_stack_clash): New function.
12873 (get_stack_check_protect): Likewise.
12874 (compute_stack_clash_protection_loop_data): Likewise.
12875 (emit_stack_clash_protection_loop_start): Likewise.
12876 (emit_stack_clash_protection_loop_end): Likewise.
12877 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12878 Use anti_adjust_stack_and_probe_stack_clash.
12879 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12880 (emit_stack_clash_protection_loop_start): Likewise.
12881 (emit_stack_clash_protection_loop_end): Likewise.
12882 * rtl.h (get_stack_check_protect): Prototype.
12883 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12884 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12885 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12887 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12889 * doc/tm.texi: Rebuilt.
12890 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12891 get_stack_check_protect.
12892 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12893 * config/arm/arm.c (arm_expand_prologue): Likewise.
12894 (arm_frame_pointer_required): Likewise.
12895 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12896 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12897 * config/mips/mips.c (mips_expand_prologue): Likewise.
12898 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12899 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12900 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12901 (sparc_flat_expand_prologue): Likewise.
12903 * common.opt (-fstack-clash-protection): New option.
12904 * flag-types.h (enum stack_check_type): Note difference between
12905 -fstack-check= and -fstack-clash-protection.
12906 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12907 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12908 * toplev.c (process_options): Issue warnings/errors for cases
12909 not handled with -fstack-clash-protection.
12910 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12911 (-fstack-check): Note additional problem with -fstack-check=generic.
12912 Note that -fstack-check is primarily for Ada and refer users
12913 to -fstack-clash-protection for stack-clash-protection.
12914 Document new params for stack clash protection.
12916 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12918 * config/i386/i386.c (ix86_split_long_move): Do not handle
12919 address used for LEA in a special way.
12921 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12923 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12925 2017-09-19 Martin Sebor <msebor@redhat.com>
12928 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12929 of incompatible types.
12931 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12934 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12935 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12936 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12938 2017-09-19 Richard Biener <rguenther@suse.de>
12940 PR tree-optimization/82244
12941 * tree-vrp.c (remove_range_assertions): Do not propagate
12942 a constant to abnormals but replace the assert with a copy.
12944 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12946 PR rtl-optimization/57878
12947 PR rtl-optimization/68988
12948 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12949 avoidance test involving non_reload_pseudos. Move frequency test
12950 below the general fragmentation avoidance test.
12952 2017-09-19 Richard Biener <rguenther@suse.de>
12954 PR tree-optimization/69728
12955 * graphite-sese-to-poly.c (schedule_error): New global.
12956 (add_loop_schedule): Handle empty domain by failing the
12958 (build_original_schedule): Handle schedule_error.
12960 2017-09-19 Richard Biener <rguenther@suse.de>
12962 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12963 Do not iterate to sibling loops but only to siblings of inner
12966 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12969 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12970 that the registers are different.
12972 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12974 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12975 to processor_model and "amdfam17h" to arch_names_table.
12976 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12978 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12981 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12983 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12984 Alan Hayward <alan.hayward@arm.com>
12985 David Sherwood <david.sherwood@arm.com>
12987 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12989 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12991 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12992 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12993 vect_schedule_slp_instance.
12994 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12995 Update call to vect_slp_analyze_node_operations. Simplify return
12997 (vect_slp_analyze_bb_1): Update call accordingly.
12998 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12999 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13000 (vect_schedule_slp): Update call accordingly.
13002 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13003 Alan Hayward <alan.hayward@arm.com>
13004 David Sherwood <david.sherwood@arm.com>
13006 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13007 with types that aren't in fact scalar.
13009 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13011 * tree-vect-slp.c (vect_record_max_nunits): New function,
13013 (vect_build_slp_tree_1): ...here.
13014 (vect_build_slp_tree_2): Call it for phis too.
13016 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13018 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13019 to vect_get_vec_def_for_operand when getting the mask operand.
13021 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13022 Alan Hayward <alan.hayward@arm.com>
13023 David Sherwood <david.sherwood@arm.com>
13025 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13028 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13029 Alan Hayward <alan.hayward@arm.com>
13030 David Sherwood <david.sherwood@arm.com>
13032 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13033 calculation for vector booleans.
13035 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13036 Alan Hayward <alan.hayward@arm.com>
13037 David Sherwood <david.sherwood@arm.com>
13039 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13041 (vect_transform_stmt): ...here.
13042 (vect_analyze_stmt): Use it instead of calling
13043 vectorizable_live_operation directly.
13045 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13047 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13048 non-SIMT targets in acc vector loops.
13050 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13052 * configure.ac: Add arc and check if assembler supports gdwarf2.
13053 * configure: Regenerate.
13055 2017-09-18 Richard Biener <rguenther@suse.de>
13057 PR tree-optimization/82220
13058 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13059 epilogue niters from the min_profitable_iters compute.
13061 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13064 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13065 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13066 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13068 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13071 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13072 switching to a new text section.
13074 2017-09-18 Richard Biener <rguenther@suse.de>
13076 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13078 (build_alias_set): Reject aliases with no access function.
13080 2017-09-18 Richard Biener <rguenther@suse.de>
13082 PR tree-optimization/79622
13083 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13085 (build_cross_bb_scalars_use): Likewise.
13087 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13089 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13091 2017-09-18 Alan Modra <amodra@gmail.com>
13094 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13095 stack_pointer_rtx for count 0. Update comments. Break up
13096 large rtl expression.
13098 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13100 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13101 Increase to 20 bytes.
13102 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13103 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13104 or avx version of the stub.
13106 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13109 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13110 compute the minimum stack alignment. Also update preferred stack
13111 boundary for leaf functions.
13113 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13115 PR tree-optimization/82228
13116 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13119 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13121 * common/config/nds32/nds32-common.c
13122 (nds32_option_optimization_table): Refine formatting.
13123 (nds32_option_optimization_table): Use -fsched-pressure and
13124 -fomit-frame-pointer for specific optimization level.
13126 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13128 * config/nds32/nds32.c: Refine formatting and comments.
13129 * config/nds32/nds32.h: Likewise.
13130 * config/nds32/nds32.md: Likewise.
13131 * config/nds32/nds32-cost.c: Likewise.
13132 * config/nds32/nds32-isr.c: Likewise.
13133 * config/nds32/nds32-md-auxiliary.c: Likewise.
13134 * config/nds32/nds32-multiple.md: Likewise.
13135 * config/nds32/nds32-predicates.c: Likewise.
13137 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13138 Jakub Jelinek <jakub@redhat.com>
13140 Add support for -std=c++2a.
13141 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13143 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13145 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13148 * doc/extend.texi (Common Function Attributes): Add
13149 references to ARM, AArch64, and S/390 specific attributes.
13150 (Function Specific Option Pragmas): Add AArch64 and S/390
13151 to list of back ends that support the target pragma.
13153 2017-09-15 Nathan Sidwell <nathan@acm.org>
13155 * doc/standards.texi: Fix C++17 description. Update URLs for
13158 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13160 * common.opt (Wcast-align=strict): New warning option.
13161 * doc/invoke.texi: Document -Wcast-align=strict.
13163 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13165 * cgraph.h (cgraph_thunk_info): Add comments.
13166 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13167 assert for VIRTUAL_* arguments stricter.
13169 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13171 PR tree-optimization/71026
13172 * match.pd: Move RDIV patterns from fold-const.c
13173 * fold-const.c (distribute_real_division): Removed.
13174 (fold_binary_loc): Remove calls to distribute_real_divison.
13176 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13178 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13179 c++1z and gnu++1z as deprecated. Change other references to
13180 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13181 Change -Wc++1z-compat to -Wc++17-compat.
13182 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13183 * dwarf2out.c (highest_c_language): Handle C++17.
13184 (gen_compile_unit_die): Likewise.
13186 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13188 PR rtl-optimization/82192
13189 * combine.c (make_extraction): Don't look through non-paradoxical
13190 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13193 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13194 Alan Hayward <alan.hayward@arm.com>
13195 David Sherwood <david.sherwood@arm.com>
13197 * target.def (function_arg_offset): New hook.
13198 * targhooks.h (default_function_arg_offset): Declare.
13199 * targhooks.c (default_function_arg_offset): New function.
13200 * function.c (locate_and_pad_parm): Use
13201 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13202 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13203 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13204 * doc/tm.texi: Regenerate.
13205 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13206 * config/spu/spu.c (spu_function_arg_offset): New function.
13207 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13208 * system.h (FUNCTION_ARG_OFFSET): Poison.
13210 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13211 Alan Hayard <alan.hayward@arm.com>
13212 David Sherwood <david.sherwood@arm.com>
13214 * target.def (truly_noop_truncation): New hook.
13215 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13216 than TRULY_NOOP_TRUNCATION.
13217 * hooks.h (hook_bool_uint_uint_true): Declare.
13218 * hooks.c (hook_bool_uint_uint_true): New function.
13219 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13220 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13221 * doc/tm.texi: Regenerate.
13222 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13223 rather than TRULY_NOOP_TRUNCATION in comments.
13224 (simplify_comparison): Likewise.
13225 (record_truncated_value): Likewise.
13226 * expmed.c (extract_bit_field_1): Likewise.
13227 (extract_split_bit_field): Likewise.
13228 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13229 instead of TRULY_NOOP_TRUNCATION.
13230 * function.c (assign_parm_setup_block): Likewise.
13231 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13232 * rtlhooks.c: Include target.h.
13233 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13234 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13235 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13236 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13237 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13238 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13239 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13240 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13241 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13242 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13243 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13244 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13245 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13246 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13247 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13248 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13249 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13250 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13251 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13252 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13253 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13254 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13255 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13256 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13257 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13258 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13259 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13260 rather than TRULY_NOOP_TRUNCATION in comments.
13261 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13262 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13263 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13264 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13265 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13266 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13267 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13268 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13269 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13270 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13271 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13272 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13273 rather than TRULY_NOOP_TRUNCATION in comments.
13274 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13275 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13276 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13277 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13278 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13279 TRULY_NOOP_TRUNCATION condition.
13280 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13281 (TRULY_NOOP_TRUNCATION): Delete.
13282 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13283 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13284 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13285 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13286 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13287 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13288 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13289 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13290 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13291 rather than TRULY_NOOP_TRUNCATION in comments.
13292 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13293 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13294 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13295 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13296 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13297 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13299 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13302 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13303 (*cmp_ior): Likewise.
13304 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13305 (*ior_scc_scc_cmp): Likewise.
13306 (*and_scc_scc): Likewise.
13307 (*and_scc_scc_cmp): Likewise.
13309 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13310 Alan Hayard <alan.hayward@arm.com>
13311 David Sherwood <david.sherwood@arm.com>
13313 * target.def (can_change_mode_class): New hook.
13314 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13315 (hard_regno_nregs): Likewise.
13316 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13317 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13318 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13319 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13320 (LOAD_EXTEND_OP): Update accordingly.
13321 * doc/tm.texi: Regenerate.
13322 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13323 CANNOT_CHANGE_MODE_CLASS.
13324 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13325 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13326 * combine.c (simplify_set): Update accordingly.
13327 * emit-rtl.c (validate_subreg): Likewise.
13328 * recog.c (general_operand): Likewise.
13329 * regcprop.c (mode_change_ok): Likewise.
13330 * reload1.c (choose_reload_regs): Likewise.
13331 (inherit_piecemeal_p): Likewise.
13332 * rtlanal.c (simplify_subreg_regno): Likewise.
13333 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13334 instead of CANNOT_CHANGE_MODE_CLASS.
13335 (reload_cse_simplify_operands): Likewise.
13336 * reload.c (push_reload): Use targetm.can_change_mode_class
13337 instead of CANNOT_CHANGE_MODE_CLASS.
13338 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13339 REG_CANNOT_CHANGE_MODE_P.
13340 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13341 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13342 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13343 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13344 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13345 (arm_can_change_mode_class): New function.
13346 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13347 than CANNOT_CHANGE_MODE_CLASS in comments.
13348 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13349 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13350 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13351 (ix86_can_change_mode_class): ...this new function, inverting the
13352 sense of the return value.
13353 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13354 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13355 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13356 (ia64_can_change_mode_class): New function.
13357 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13358 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13359 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13360 (m32c_can_change_mode_class): ...this new function, inverting the
13361 sense of the return value.
13362 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13363 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13364 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13365 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13366 (mips_can_change_mode_class): ...this new function, inverting the
13367 sense of the return value.
13368 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13369 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13370 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13371 (msp430_can_change_mode_class): New function.
13372 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13373 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13374 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13375 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13376 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13377 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13378 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13379 (pa_cannot_change_mode_class): Replace with...
13380 (pa_can_change_mode_class): ...this new function, inverting the
13381 sense of the return value.
13382 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13383 than CANNOT_CHANGE_MODE_CLASS in comments.
13384 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13385 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13386 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13387 (pdp11_cannot_change_mode_class): Replace with...
13388 (pdp11_can_change_mode_class): ...this new function, inverting the
13389 sense of the return value.
13390 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13391 * config/powerpcspe/powerpcspe-protos.h
13392 (rs6000_cannot_change_mode_class_ptr): Delete.
13393 * config/powerpcspe/powerpcspe.c
13394 (rs6000_cannot_change_mode_class_ptr): Delete.
13395 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13396 (rs6000_option_override_internal): Assign to
13397 targetm.can_change_mode_class instead of
13398 rs6000_cannot_change_mode_class_ptr.
13399 (rs6000_cannot_change_mode_class): Replace with...
13400 (rs6000_can_change_mode_class): ...this new function, inverting the
13401 sense of the return value.
13402 (rs6000_debug_cannot_change_mode_class): Replace with...
13403 (rs6000_debug_can_change_mode_class): ...this new function.
13404 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13405 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13406 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13407 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13408 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13410 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13411 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13412 (rs6000_option_override_internal): Assign to
13413 targetm.can_change_mode_class instead of
13414 rs6000_cannot_change_mode_class_ptr.
13415 (rs6000_cannot_change_mode_class): Replace with...
13416 (rs6000_can_change_mode_class): ...this new function, inverting the
13417 sense of the return value.
13418 (rs6000_debug_cannot_change_mode_class): Replace with...
13419 (rs6000_debug_can_change_mode_class): ...this new function.
13420 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13421 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13422 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13423 (s390_can_change_mode_class): ...this new function, inverting the
13424 sense of the return value.
13425 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13426 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13427 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13428 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13429 (sh_cannot_change_mode_class): Replace with...
13430 (sh_can_change_mode_class): ...this new function, inverting the
13431 sense of the return value.
13432 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13433 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13434 (sparc_can_change_mode_class): New function.
13435 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13436 * config/spu/spu.c (spu_can_change_mode_class): New function.
13437 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13438 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13439 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13440 (visium_can_change_mode_class): New function.
13441 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13443 2017-09-15 Richard Biener <rguenther@suse.de>
13445 PR tree-optimization/82217
13446 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13447 but not undefined case.
13449 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13452 * postreload.c (reload_cse_simplify_operands): Skip
13453 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13455 2017-09-15 Richard Biener <rguenther@suse.de>
13457 PR tree-optimization/68823
13458 * graphite-scop-detection.c (build_alias_set): If we have a
13459 possible dependence check whether we can handle them by just
13460 looking at the DRs DR_ACCESS_FNs.
13461 (build_scops): If build_alias_set fails, fail the SCOP.
13463 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13465 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13466 to support float128 built-in functions that require the ISA 3.0
13468 (BU_FLOAT128_3_HW): Likewise.
13469 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13470 built-in functions.
13471 (FMAF128): Likewise.
13473 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13474 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13475 floating point instructions.
13476 (rs6000_invalid_builtin): Likewise.
13477 (rs6000_builtin_mask_names): Likewise.
13478 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13479 (RS6000_BTM_FLOAT128_HW): Likewise.
13480 (RS6000_BTM_COMMON): Likewise.
13481 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13483 * doc/extend.texi (RS/6000 built-in functions): Document the
13484 IEEE 128-bit floating point square root and fused multiply-add
13485 built-in functions.
13487 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13489 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13490 reg (r2) isn't in the set of registers defined in the prologue.
13492 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13493 Alan Hayward <alan.hayward@arm.com>
13494 David Sherwood <david.sherwood@arm.com>
13496 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13497 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13498 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13499 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13500 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13502 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13503 max_vectorization_factor.
13504 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13506 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13507 Alan Hayward <alan.hayward@arm.com>
13508 David Sherwood <david.sherwood@arm.com>
13510 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13511 (vect_worthwhile_without_simd_p): Declare.
13512 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13513 (vectorizable_reduction): Use it.
13514 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13515 (vectorizable_operation): Likewise.
13517 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13518 Alan Hayward <alan.hayward@arm.com>
13519 David Sherwood <david.sherwood@arm.com>
13521 * tree-vectorizer.h (vect_get_num_copies): New function.
13522 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13523 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13524 (vectorizable_induction): Likewise.
13525 (vectorizable_live_operation): Likewise.
13526 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13527 (vectorizable_bswap): Likewise.
13528 (vectorizable_call): Likewise.
13529 (vectorizable_conversion): Likewise.
13530 (vectorizable_assignment): Likewise.
13531 (vectorizable_shift): Likewise.
13532 (vectorizable_operation): Likewise.
13533 (vectorizable_store): Likewise.
13534 (vectorizable_load): Likewise.
13535 (vectorizable_condition): Likewise.
13536 (vectorizable_comparison): Likewise.
13537 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13539 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13540 Alan Hayward <alan.hayward@arm.com>
13541 David Sherwood <david.sherwood@arm.com>
13543 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13544 of vect_init_vector.
13546 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13547 Alan Hayward <alan.hayward@arm.com>
13548 David Sherwood <david.sherwood@arm.com>
13550 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13551 an inline wrapper that provides a location.
13552 (gimple_build_vector): Likewise.
13553 * gimple-fold.c (gimple_build_vector_from_val): New function.
13554 (gimple_build_vector): Likewise.
13555 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13556 functions to build the initial value. Always return a gimple value.
13557 (get_initial_defs_for_reduction): Likewise. Only compute
13559 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13560 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13561 (vectorizable_induction): Use gimple_build_vector rather than
13564 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13565 Alan Hayward <alan.hayward@arm.com>
13566 David Sherwood <david.sherwood@arm.com>
13568 * target.h (vec_perm_indices): New typedef.
13569 (auto_vec_perm_indices): Likewise.
13570 * optabs-query.h: Include target.h
13571 (can_vec_perm_p): Take a vec_perm_indices *.
13572 * optabs-query.c (can_vec_perm_p): Likewise.
13573 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13574 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13575 * tree-vect-generic.c (lower_vec_perm): Likewise.
13576 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13577 (vect_grouped_load_supported): Likewise.
13578 (vect_shift_permute_load_chain): Likewise.
13579 (vect_permute_store_chain): Use auto_vec_perm_indices.
13580 (vect_permute_load_chain): Likewise.
13581 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13582 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13583 Update uses of can_vec_perm_p.
13584 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13585 mode with a number of elements. Take a vec_perm_indices *.
13586 (vect_create_epilog_for_reduction): Update accordingly.
13587 Use auto_vec_perm_indices.
13588 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13589 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13590 (vect_transform_slp_perm_load): Likewise.
13591 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13592 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13593 (vect_gen_perm_mask_checked): Likewise.
13594 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13595 (vect_gen_perm_mask_checked): Likewise.
13596 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13597 (vectorizable_store): Likewise.
13598 (vectorizable_load): Likewise.
13599 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13600 (vectorizable_bswap): Likewise.
13602 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13603 Alan Hayward <alan.hayward@arm.com>
13604 David Sherwood <david.sherwood@arm.com>
13606 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13607 * tree.c (build_vector): Likewise.
13608 (build_vector_from_ctor): Update accordingly.
13609 (build_vector_from_val): Likewise.
13610 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13611 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13612 * tree-vect-generic.c (add_rshift): Likewise.
13613 (expand_vector_divmod): Likewise.
13614 (optimize_vector_constructor): Likewise.
13615 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13616 (vect_transform_slp_perm_load): Likewise.
13617 (vect_schedule_slp_instance): Likewise.
13618 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13619 (vectorizable_call): Likewise.
13620 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13621 * expmed.c (make_tree): Likewise.
13622 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13623 a vector passed to build_vector.
13624 (fold_convert_const): Likewise.
13625 (exact_inverse): Likewise.
13626 (fold_ternary_loc): Likewise.
13627 (fold_relational_const): Likewise.
13628 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13629 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13630 (const_unop): Likewise. Store the reduction accumulator in a
13631 variable rather than an array.
13632 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13633 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13634 the new vector, rather than constructing it after the input arrays.
13635 (native_interpret_vector): Use auto_vec<tree> when building
13636 a vector passed to build_vector. Add elements in order.
13637 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13638 auto_vec<tree> when building a vector passed to build_vector.
13639 (vect_create_epilog_for_reduction): Likewise.
13640 (vectorizable_induction): Likewise.
13641 (get_initial_def_for_reduction): Likewise. Fix indentation of
13643 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13645 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13646 passed to build_vector.
13648 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13649 Alan Hayward <alan.hayward@arm.com>
13650 David Sherwood <david.sherwood@arm.com>
13652 * tree-core.h (tree_base::u): Add an "nelts" field.
13653 (tree_vector): Use VECTOR_CST_NELTS as the length.
13654 * tree.c (tree_size): Likewise.
13655 (make_vector): Initialize VECTOR_CST_NELTS.
13656 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13657 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13658 TYPE_VECTOR_SUBPARTS.
13659 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13660 as the number of units, setting it from VECTOR_CST_NELTS.
13661 (const_vector_from_tree): Likewise.
13662 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13663 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13664 (fold_negate_expr_1): Likewise.
13665 (fold_convert_const): Likewise.
13666 (const_binop): Likewise. Differentiate the number of output and
13668 (const_unop): Likewise.
13669 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13670 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13671 in cases that did the opposite.
13673 2017-09-14 Richard Biener <rguenther@suse.de>
13675 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13678 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13680 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13682 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13685 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13686 if and where to split a bb, except for splitting before debug insn
13687 sequences followed by non-label real insn. Delete debug insns
13688 in between basic blocks.
13690 * combine.c (make_compound_operation_int): Formatting fixes.
13692 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13693 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13694 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13695 * config/sol2.h (LINK_EH_SPEC): Likewise.
13696 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13697 * config/s390/linux.h (LINK_SPEC): Likewise.
13698 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13699 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13700 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13701 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13702 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13703 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13704 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13705 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13707 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13709 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13711 (ENDFILE_LINUX_SPEC): Likewise.
13712 (LINK_EH_SPEC): Likewise.
13713 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
13714 (LINK_OS_LINUX_SPEC32): Likewise.
13715 (LINK_OS_LINUX_SPEC64): Likewise.
13716 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
13717 (LINK_OS_LINUX_SPEC): Likewise.
13719 2017-09-13 Martin Liska <mliska@suse.cz>
13721 PR middle-end/82154
13722 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
13723 CASE_HIGH is NULL_TREE.
13725 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13726 Alan Hayward <alan.hayward@arm.com>
13727 David Sherwood <david.sherwood@arm.com>
13729 * target.def (secondary_memory_needed): New hook.
13730 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13731 instead of SECONDARY_MEMORY_NEEDED.
13732 (secondary_memory_needed_mode): Likewise.
13733 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
13734 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13735 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13736 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13737 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13738 * doc/tm.texi: Regenerate.
13739 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13740 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13741 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13742 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13743 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13744 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13745 mode argument first and change the reg_class arguments to reg_class_t.
13746 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13747 Make static. Update the call to inline_secondary_memory_needed.
13748 (ix86_register_move_cost): Update the call to
13749 inline_secondary_memory_needed.
13750 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13751 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13753 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13754 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13756 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13757 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13758 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13759 and match hook interface. Add comment from mips.h.
13760 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13761 * config/mmix/mmix.md (truncdfsf2): Refer to
13762 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13764 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13765 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13766 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13767 (pa_secondary_memory_needed): New function.
13768 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13769 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13770 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13771 (pdp11_secondary_memory_needed): Make static and match hook interface.
13772 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13773 * config/powerpcspe/powerpcspe-protos.h
13774 (rs6000_secondary_memory_needed_ptr): Delete.
13775 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13777 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13778 (rs6000_option_override_internal): Assign to
13779 targetm.secondary_memory_needed rather than
13780 rs6000_secondary_memory_needed_ptr.
13781 (rs6000_secondary_memory_needed): Match hook interface.
13782 (rs6000_debug_secondary_memory_needed): Likewise.
13783 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13784 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13785 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13786 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13787 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13788 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13790 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13791 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13792 (rs6000_option_override_internal): Assign to
13793 targetm.secondary_memory_needed rather than
13794 rs6000_secondary_memory_needed_ptr.
13795 (rs6000_secondary_memory_needed): Match hook interface.
13796 (rs6000_debug_secondary_memory_needed): Likewise.
13797 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13798 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13799 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13800 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13801 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13802 (sparc_secondary_memory_needed): New function.
13803 * lra-constraints.c (check_and_process_move): Refer to
13804 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13806 (curr_insn_transform): Likewise.
13807 (process_alt_operands): Use targetm.secondary_memory_needed
13808 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13809 (check_secondary_memory_needed_p): Likewise.
13810 (choose_split_class): Likewise.
13811 * reload.c: Unconditionally include code that was previously
13812 conditional on SECONDARY_MEMORY_NEEDED.
13813 (push_secondary_reload): Use targetm.secondary_memory_needed
13814 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13815 (push_reload): Likewise.
13816 * reload1.c: Unconditionally include code that was previously
13817 conditional on SECONDARY_MEMORY_NEEDED.
13818 (choose_reload_regs): Use targetm.secondary_memory_needed
13819 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13820 (gen_reload): Likewise.
13821 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13823 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13824 Alan Hayward <alan.hayward@arm.com>
13825 David Sherwood <david.sherwood@arm.com>
13827 * target.def (secondary_memory_needed_mode): New hook:
13828 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13829 * targhooks.h (default_secondary_memory_needed_mode): New function.
13830 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13831 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13832 * doc/tm.texi: Regenerate.
13833 * lra-constraints.c (check_and_process_move): Use
13834 targetm.secondary_memory_needed_mode instead of
13835 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13836 (curr_insn_transform): Likewise.
13837 * reload.c (get_secondary_mem): Likewise.
13838 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13839 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13841 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13842 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13843 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13844 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13845 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13847 * config/powerpcspe/powerpcspe-protos.h
13848 (rs6000_secondary_memory_needed_mode): Delete.
13849 * config/powerpcspe/powerpcspe.c
13850 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13851 (rs6000_secondary_memory_needed_mode): Make static.
13852 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13853 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13855 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13857 (rs6000_secondary_memory_needed_mode): Make static.
13858 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13859 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13860 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13861 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13862 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13864 (sparc_secondary_memory_needed_mode): New function.
13865 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13867 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13869 * config/aarch64/constraints.md (Umq): New constraint.
13870 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13872 (mov<mode>): Update condition.
13874 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13876 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13877 when bitposition is the same.
13879 2017-09-13 Richard Biener <rguenther@suse.de>
13881 * dwarf2out.c (output_die_symbol): Remove.
13882 (output_die): Do not output a DIEs symbol.
13884 2017-09-13 Richard Biener <rguenther@suse.de>
13886 PR middle-end/82128
13887 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13888 default-def to avoid breaking iterator update with the weird
13889 interaction with cgraph_update_edges_for_call_stmt_node.
13891 2017-09-13 Richard Biener <rguenther@suse.de>
13893 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13894 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13895 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13896 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13898 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13900 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13901 Disable pc relative literal load irrespective of
13902 TARGET_FIX_ERR_A53_84341 for default.
13904 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13906 * config/sparc/sparc.c (output_return): Output the source location of
13907 the insn in the delay slot, if any.
13908 (output_sibcall): Likewise.
13910 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13913 * common.opt (-static-pie): New alias.
13914 (shared): Negate static-pie.
13915 (-no-pie): Update help text.
13917 (static-pie): New option.
13918 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13919 -static-pie support.
13920 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13921 (LINK_EH_SPEC): Likewise.
13922 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13923 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13924 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13925 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13926 (init_gcc_specs): Likewise.
13927 (init_spec): Likewise.
13928 (display_help): Update help message for -pie.
13929 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13932 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13934 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13935 (movdi_aarch64): Likewise.
13936 (movti_aarch64): Likewise.
13938 2017-09-12 Simon Wright <simon@pushface.org>
13941 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13942 calculation of the minor version, always output as 0.
13944 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13948 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13949 on it early, rather than manual conversion late. For
13950 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13951 instead of performing manual conversion.
13953 2017-09-12 Carl Love <cel@us.ibm.com>
13955 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13956 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13958 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13959 VMULOSW): Add definitions.
13960 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13961 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13962 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13963 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13964 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13966 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13968 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13970 (movti_aarch64): Likewise.
13971 (movdf_aarch64): Likewise.
13972 (movtf_aarch64): Likewise.
13973 (load_pairdi): Likewise.
13974 (store_pairdi): Likewise.
13975 (load_pairdf): Likewise.
13976 (store_pairdf): Likewise.
13977 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13978 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13979 (ldr_got_small_<mode>): Likewise.
13980 (ldr_got_small_28k_<mode>): Likewise.
13981 (ldr_got_tiny): Likewise.
13982 * config/aarch64/iterators.md (ldst_sz): New.
13983 (ldpstp_sz): Likewise.
13984 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13986 (thunderx_load): Split load_8 to load_16.
13987 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13989 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13990 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13991 (xgene1_store_pair): Split store_8 to store_16.
13992 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13993 (falkor_st_0_st_sd): Split store_8 to store_16.
13995 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13997 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13998 and store1/2/3/4 to store_4/8/12/16.
13999 * config/aarch64/aarch64.md: Update for rename.
14000 * config/arm/arm.md: Likewise.: Likewise.
14001 * config/arm/arm.c: Likewise.
14002 * config/arm/thumb1.md: Likewise.
14003 * config/arm/thumb2.md: Likewise.
14004 * config/arm/vfp.md: Likewise.
14005 * config/arm/arm-generic.md: Likewise.
14006 * config/arm/arm1020e.md: Likewise.
14007 * config/arm/arm1026ejs.md: Likewise.
14008 * config/arm/arm1136jfs.md: Likewise.
14009 * config/arm/arm926ejs.md: Likewise.
14010 * config/arm/cortex-a15.md: Likewise.
14011 * config/arm/cortex-a17.md: Likewise.
14012 * config/arm/cortex-a5.md: Likewise.
14013 * config/arm/cortex-a53.md: Likewise.
14014 * config/arm/cortex-a57.md: Likewise.
14015 * config/arm/cortex-a7.md: Likewise.
14016 * config/arm/cortex-a8.md: Likewise.
14017 * config/arm/cortex-a9.md: Likewise.
14018 * config/arm/cortex-m4.md: Likewise.
14019 * config/arm/cortex-m7.md: Likewise.
14020 * config/arm/cortex-r4.md: Likewise.
14021 * config/arm/exynos-m1.md: Likewise.
14022 * config/arm/fa526.md: Likewise.
14023 * config/arm/fa606te.md: Likewise.
14024 * config/arm/fa626te.md: Likewise.
14025 * config/arm/fa726te.md: Likewise.
14026 * config/arm/fmp626.md: Likewise.
14027 * config/arm/iwmmxt.md: Likewise.
14028 * config/arm/ldmstm.md: Likewise.
14029 * config/arm/marvell-pj4.md: Likewise.
14030 * config/arm/xgene1.md: Likewise.
14031 * config/aarch64/thunderx.md: Likewise.
14032 * config/aarch64/thunderx2t99.md: Likewise.
14033 * config/aarch64/falkor.md: Likewise.
14035 2017-09-12 Martin Liska <mliska@suse.cz>
14037 * attribs.c (private_lookup_attribute): New function.
14038 * attribs.h (private_lookup_attribute): Declared here.
14039 (lookup_attribute): Called from this place.
14041 2017-09-12 Richard Biener <rguenther@suse.de>
14043 PR tree-optimization/82157
14044 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14045 stmts with side-effects.
14047 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14048 Alan Hayward <alan.hayward@arm.com>
14049 David Sherwood <david.sherwood@arm.com>
14051 * target.def (hard_regno_nregs): New hook.
14052 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14053 * targhooks.h (default_hard_regno_nregs): Declare.
14054 * targhooks.c (default_hard_regno_nregs): New function.
14055 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14056 (TARGET_HARD_REGNO_NREGS): ...this hook.
14057 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14058 (CLASS_MAX_NREGS): Likewise.
14059 * doc/tm.texi: Regenerate.
14060 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14061 instead of HARD_REGNO_NREGS.
14062 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14063 HARD_REGNO_NREGS in the comment.
14064 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14065 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14066 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14067 Return an unsigned int.
14068 (TARGET_HARD_REGNO_NREGS): Redefine.
14069 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14070 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14071 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14072 (arc_hard_regno_nregs): New function.
14073 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14074 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14075 (arm_hard_regno_nregs): New function.
14076 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14077 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14078 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14079 (TARGET_HARD_REGNO_NREGS): Redefine.
14080 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14081 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14082 (HARD_REGNO_NREGS): Delete.
14083 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14084 (cr16_hard_regno_nregs): New function.
14085 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14086 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14087 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14088 (cris_hard_regno_nregs): New function.
14089 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14090 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14091 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14092 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14093 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14094 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14095 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14096 (frv_hard_regno_nregs): Make static. Take and return an
14098 (frv_class_max_nregs): Remove outdated copy of documentation.
14099 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14100 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14101 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14102 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14103 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14104 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14105 (TARGET_HARD_REGNO_NREGS): Redefine.
14106 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14107 (CLASS_MAX_NREGS): Update comment.
14108 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14109 (ia64_hard_regno_nregs): New function.
14110 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14111 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14112 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14113 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14114 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14116 (m32c_hard_regno_nregs): Likewise. Make static.
14117 (TARGET_HARD_REGNO_NREGS): Redefine.
14118 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14119 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14120 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14121 (m68k_hard_regno_nregs): New function.
14122 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14123 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14124 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14125 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14126 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14127 Take and return an unsigned int.
14128 (TARGET_HARD_REGNO_NREGS): Redefine.
14129 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14130 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14131 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14132 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14133 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14134 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14135 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14136 (msp430_hard_regno_nregs): Make static. Take and return an
14138 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14139 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14140 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14141 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14142 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14143 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14144 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14145 (TARGET_HARD_REGNO_NREGS): Redefine.
14146 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14147 (PA_HARD_REGNO_NREGS): ...this.
14148 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14149 (PA_HARD_REGNO_NREGS): ...this.
14150 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14151 (pa_hard_regno_nregs): New function.
14152 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14153 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14154 (pdp11_hard_regno_nregs): New function.
14155 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14156 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14157 (rs6000_hard_regno_nregs_hook): New function.
14158 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14159 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14160 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14161 Take and return an unsigned int. Move earlier in file.
14162 (TARGET_HARD_REGNO_NREGS): Redefine.
14163 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14164 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14165 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14166 (rl78_hard_regno_nregs): Make static. Take and return an
14168 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14169 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14170 (rs6000_hard_regno_nregs_hook): New function.
14171 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14172 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14173 (TARGET_HARD_REGNO_NREGS): Redefine.
14174 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14175 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14176 instead of HARD_REGNO_NREGS.
14177 (s390_hard_regno_nregs): New function.
14178 (s390_hard_regno_mode_ok): Add comment from s390.h.
14179 (TARGET_HARD_REGNO_NREGS): Redefine.
14180 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14181 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14182 (sh_hard_regno_nregs): New function.
14183 (sh_pass_in_reg_p): Use it.
14184 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14185 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14186 (sparc_hard_regno_nregs): New function.
14187 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14188 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14189 (spu_function_arg_advance): Use it, supplying a valid register number.
14190 (TARGET_HARD_REGNO_NREGS): Redefine.
14191 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14192 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14193 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14194 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14195 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14196 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14197 (CLASS_MAX_NREGS): Remove copy of old documentation.
14198 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14199 (visium_hard_regno_nregs): New function.
14200 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14201 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14202 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14203 xtensa_hard_regno_nregs): New function.
14204 * system.h (HARD_REGNO_NREGS): Poison.
14206 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14208 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14209 hard_regno_nregs instead of HARD_REGNO_NREGS.
14210 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14211 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14212 (c6x_expand_epilogue): Likewise.
14213 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14214 (frv_read_iacc_argument): Likewise.
14215 * config/sh/sh.c: Include regs.h.
14216 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14217 (regs_used): Likewise.
14218 (output_stack_adjust): Likewise.
14219 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14220 * expmed.c: Include regs.h.
14221 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14222 * ree.c: Include regs.h.
14223 (combine_reaching_defs): Use hard_regno_nregs instead of
14225 (add_removable_extension): Likewise.
14227 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14229 * regs.h (hard_regno_nregs): Turn into a function.
14230 (end_hard_regno): Update accordingly.
14231 * caller-save.c (setup_save_areas): Likewise.
14232 (save_call_clobbered_regs): Likewise.
14233 (replace_reg_with_saved_mem): Likewise.
14234 (insert_restore): Likewise.
14235 (insert_save): Likewise.
14236 * combine.c (can_change_dest_mode): Likewise.
14237 (move_deaths): Likewise.
14238 (distribute_notes): Likewise.
14239 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14240 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14241 (rs6000_split_multireg_move): Likewise.
14242 (rs6000_register_move_cost): Likewise.
14243 (rs6000_memory_move_cost): Likewise.
14244 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14245 (rs6000_split_multireg_move): Likewise.
14246 (rs6000_register_move_cost): Likewise.
14247 (rs6000_memory_move_cost): Likewise.
14248 * cselib.c (cselib_reset_table): Likewise.
14249 (cselib_lookup_1): Likewise.
14250 * emit-rtl.c (set_mode_and_regno): Likewise.
14251 * function.c (aggregate_value_p): Likewise.
14252 * ira-color.c (setup_profitable_hard_regs): Likewise.
14253 (check_hard_reg_p): Likewise.
14254 (calculate_saved_nregs): Likewise.
14255 (assign_hard_reg): Likewise.
14256 (improve_allocation): Likewise.
14257 (calculate_spill_cost): Likewise.
14258 * ira-emit.c (modify_move_list): Likewise.
14259 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14260 (ira_hard_reg_in_set_p): Likewise.
14261 * ira.c (setup_reg_mode_hard_regset): Likewise.
14262 (clarify_prohibited_class_mode_regs): Likewise.
14263 (check_allocation): Likewise.
14264 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14265 (lra_setup_reg_renumber): Likewise.
14266 (setup_try_hard_regno_pseudos): Likewise.
14267 (spill_for): Likewise.
14268 (assign_hard_regno): Likewise.
14269 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14270 * lra-constraints.c (in_class_p): Likewise.
14271 (lra_constraint_offset): Likewise.
14272 (simplify_operand_subreg): Likewise.
14273 (lra_constraints): Likewise.
14274 (split_reg): Likewise.
14275 (split_if_necessary): Likewise.
14276 (invariant_p): Likewise.
14277 (inherit_in_ebb): Likewise.
14278 * lra-lives.c (process_bb_lives): Likewise.
14279 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14280 (get_hard_regs): Likewise.
14281 (do_remat): Likewise.
14282 * lra-spills.c (assign_spill_hard_regs): Likewise.
14283 * mode-switching.c (create_pre_exit): Likewise.
14284 * postreload.c (reload_combine_recognize_pattern): Likewise.
14285 * recog.c (peep2_find_free_register): Likewise.
14286 * regcprop.c (kill_value_regno): Likewise.
14287 (set_value_regno): Likewise.
14288 (copy_value): Likewise.
14289 (maybe_mode_change): Likewise.
14290 (find_oldest_value_reg): Likewise.
14291 (copyprop_hardreg_forward_1): Likewise.
14292 * regrename.c (check_new_reg_p): Likewise.
14293 (regrename_do_replace): Likewise.
14294 * reload.c (push_reload): Likewise.
14295 (combine_reloads): Likewise.
14296 (find_dummy_reload): Likewise.
14297 (operands_match_p): Likewise.
14298 (find_reloads): Likewise.
14299 (find_equiv_reg): Likewise.
14300 (reload_adjust_reg_for_mode): Likewise.
14301 * reload1.c (count_pseudo): Likewise.
14302 (count_spilled_pseudo): Likewise.
14303 (find_reg): Likewise.
14304 (clear_reload_reg_in_use): Likewise.
14305 (free_for_value_p): Likewise.
14306 (allocate_reload_reg): Likewise.
14307 (choose_reload_regs): Likewise.
14308 (reload_adjust_reg_for_temp): Likewise.
14309 (emit_reload_insns): Likewise.
14310 (delete_output_reload): Likewise.
14311 * rtlanal.c (subreg_get_info): Likewise.
14312 * sched-deps.c (sched_analyze_reg): Likewise.
14313 * sel-sched.c (init_regs_for_mode): Likewise.
14314 (mark_unavailable_hard_regs): Likewise.
14315 (choose_best_reg_1): Likewise.
14316 (verify_target_availability): Likewise.
14317 * valtrack.c (dead_debug_insert_temp): Likewise.
14318 * var-tracking.c (track_loc_p): Likewise.
14319 (emit_note_insn_var_location): Likewise.
14320 * varasm.c (make_decl_rtl): Likewise.
14321 * reginfo.c (choose_hard_reg_mode): Likewise.
14322 (init_reg_modes_target): Refer directly to
14323 this_target_regs->x_hard_regno_nregs.
14325 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14327 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14328 instead of hard_regno_nregs.
14330 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14332 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14333 end_hard_regno instead of hard_regno_nregs.
14334 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14335 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14336 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14337 * ira-color.c (improve_allocation): Likewise.
14338 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14339 * lra-lives.c (mark_regno_live): Likewise.
14340 (mark_regno_dead): Likewise.
14341 * lra-remat.c (operand_to_remat): Likewise.
14342 * lra.c (collect_non_operand_hard_regs): Likewise.
14343 * postreload.c (reload_combine_note_store): Likewise.
14344 (move2add_valid_value_p): Likewise.
14345 * reload.c (regno_clobbered_p): Likewise.
14347 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14349 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14351 * config/v850/v850.c (v850_reorg): Likewise.
14352 * reload.c (refers_to_regno_for_reload_p): Likewise.
14353 (find_equiv_reg): Likewise.
14354 * reload1.c (reload_reg_reaches_end_p): Likewise.
14356 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14358 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14360 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14361 * config/arm/arm.c (output_move_neon): Likewise.
14362 (arm_attr_length_move_neon): Likewise.
14363 (neon_split_vcombine): Likewise.
14364 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14365 (c6x_mark_reg_written): Likewise.
14366 (c6x_dwarf_register_span): Likewise.
14367 * config/i386/i386.c (ix86_save_reg): Likewise.
14368 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14369 (rws_access_reg): Likewise.
14370 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14371 * mode-switching.c (create_pre_exit): Likewise.
14372 * ree.c (combine_reaching_defs): Likewise.
14373 (add_removable_extension): Likewise.
14374 * regcprop.c (find_oldest_value_reg): Likewise.
14375 (copyprop_hardreg_forward_1): Likewise.
14376 * reload.c (reload_inner_reg_of_subreg): Likewise.
14377 (push_reload): Likewise.
14378 (combine_reloads): Likewise.
14379 (find_dummy_reload): Likewise.
14380 (reload_adjust_reg_for_mode): Likewise.
14381 * reload1.c (find_reload_regs): Likewise.
14382 (forget_old_reloads_1): Likewise.
14383 (reload_reg_free_for_value_p): Likewise.
14384 (reload_adjust_reg_for_temp): Likewise.
14385 (emit_reload_insns): Likewise.
14386 (delete_output_reload): Likewise.
14387 * sel-sched.c (choose_best_reg_1): Likewise.
14388 (choose_best_pseudo_reg): Likewise.
14390 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14391 Alan Hayward <alan.hayward@arm.com>
14392 David Sherwood <david.sherwood@arm.com>
14394 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14395 * target.def (slow_unaligned_access): New hook.
14396 * targhooks.h (default_slow_unaligned_access): Declare.
14397 * targhooks.c (default_slow_unaligned_access): New function.
14398 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14399 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14400 * doc/tm.texi: Regenerate.
14401 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14402 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14403 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14405 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14406 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14408 (rs6000_slow_unaligned_access): New function.
14409 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14410 (expand_block_compare): Likewise.
14411 (expand_strn_compare): Likewise.
14412 (rs6000_rtx_costs): Likewise.
14413 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14414 (riscv_slow_unaligned_access): Likewise.
14415 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14416 (riscv_slow_unaligned_access_p): ...this and make static.
14417 (riscv_option_override): Update accordingly.
14418 (riscv_slow_unaligned_access): New function.
14419 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14420 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14421 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14422 (rs6000_slow_unaligned_access): New function.
14423 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14424 (rs6000_rtx_costs): Likewise.
14425 * config/rs6000/rs6000-string.c (expand_block_compare)
14426 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14427 of SLOW_UNALIGNED_ACCESS.
14428 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14429 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14430 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14431 of SLOW_UNALIGNED_ACCESS.
14432 * expmed.c (simple_mem_bitfield_p): Likewise.
14433 * expr.c (alignment_for_piecewise_move): Likewise.
14434 (emit_group_load_1): Likewise.
14435 (emit_group_store): Likewise.
14436 (copy_blkmode_from_reg): Likewise.
14437 (emit_push_insn): Likewise.
14438 (expand_assignment): Likewise.
14439 (store_field): Likewise.
14440 (expand_expr_real_1): Likewise.
14441 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14442 * lra-constraints.c (simplify_operand_subreg): Likewise.
14443 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14444 * gimple-ssa-store-merging.c: Likewise in block comment at start
14446 * tree-ssa-strlen.c: Include target.h.
14447 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14448 of SLOW_UNALIGNED_ACCESS.
14449 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14451 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14453 PR rtl-optimization/82185
14454 * expmed.c (emit_store_flag_int): Only test tem if it has been
14457 2017-09-12 Richard Biener <rguenther@suse.de>
14459 PR middle-end/82149
14460 * match.pd ((FTYPE) N CMP CST): Fix typo.
14462 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14464 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14466 (mips_near_type_p): Add 'short_call' attribute as a synonym
14468 * doc/extend.texi (short_call): Document new function attribute.
14470 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14473 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14474 assertion check that in the condition.
14475 (get_atomic_generic_size): Likewise. Before testing if parameter
14476 has pointer type, if it has array type, call for C++
14477 default_conversion to perform array-to-pointer conversion.
14479 2017-09-12 Richard Biener <rguenther@suse.de>
14481 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14482 for operations we cannot scalarize.
14484 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14486 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14487 vectors heap vectors. Clean up comments.
14488 Make visited_bbs a reference.
14489 (profitable_jump_thread_path): Make GC
14490 vectors heap vectors. Clean up comments.
14492 (convert_and_register_jump_thread_path): Make GC vectors heap
14494 (check_subpath_and_update_thread_path): Same. Clean up comments.
14495 Make visited_bbs a reference.
14496 (handle_phi): Abstract common code to to
14497 register_jump_thread_path_if_profitable.
14498 Rename VAR_BB to DEF_BB.
14500 Make GC vectors heap vectors.
14501 Make visited_bbs a reference.
14502 (handle_assignment): Same.
14503 (register_jump_thread_path_if_profitable): New.
14504 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14506 Make GC vectors heap vectors. Clean up comments.
14507 Make visited_bbs a reference.
14508 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14509 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14512 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14515 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14516 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14518 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14520 Revert r251800 and r251799.
14522 2017-09-11 Martin Jambor <mjambor@suse.cz>
14525 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14526 arguments in advance.
14527 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14528 use it to find predecessor edges.
14529 (naive_outof_ssa): Collect vector of predecessors.
14531 2017-09-08 Jason Merrill <jason@redhat.com>
14533 PR c++/70029 - ICE with ref-qualifier and -flto
14534 * langhooks.h (struct lang_hooks_for_types): Add
14535 copy_lang_qualifiers.
14536 * attribs.c (build_type_attribute_qual_variant): Use it.
14537 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14539 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14540 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14542 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14545 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14546 (*mulsi3_sp64): New instruction.
14547 (mulsi3): New expander.
14549 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14551 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14553 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14555 * sancov.c: Include memmodel.h.
14557 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14560 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14563 2017-09-07 Carl Love <cel@us.ibm.com>
14565 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14566 the sldi instruction.
14568 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14570 * sancov.c: Include tm_p.h.
14572 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14575 * output.h (switch_to_other_text_partition): New declaration.
14576 * varasm.c (switch_to_other_text_partition): New function.
14577 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14578 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14579 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14580 to the other text partition before emitting LCL label and switch back
14581 after emitting the word after it.
14583 2017-09-07 Richard Biener <rguenther@suse.de>
14585 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14586 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14587 critical edges here, after loop init.
14588 (pass_data_pre): Remove PROP_no_crit_edges flags.
14589 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14590 for valueization of call args to avoid leaking VN_TOP.
14591 (visit_use): Assert we do not visit default defs.
14592 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14593 Use error_mark_node to more easily detect leaking VN_TOP.
14594 All default-defs are varying, not VN_TOP. Mark them visited.
14595 (run_scc_vn): Make code match comment.
14597 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14599 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14600 OPTION_MASK_FLOAT128_KEYWORD.
14601 (POWERPC_MASKS): Likewise.
14602 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14603 support for the -mfloat128-type option, and make -mfloat128
14604 default on PowerPC Linux systems. Define or undefine
14605 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14606 Define __float128 to be __ieee128 if IEEE 128-bit support is
14607 enabled, or undefine it.
14608 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14609 Delete defining __FLOAT128_TYPE__.
14610 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14611 -mfloat128-type option and make -mfloat128 default on PowerPC
14613 (TARGET_FLOAT128_TYPE): Likewise.
14614 (-mfloat128-type): Likewise.
14615 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14616 Delete the -mfloat128-type option and make -mfloat128 default on
14617 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14618 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14619 128-bit floating point is enabled. Change tests from using
14620 -mfloat128-type to -mfloat128.
14621 (rs6000_mangle_type): Use the correct mangling for the __float128
14622 type even if normal long double is restricted to 64-bits.
14623 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14625 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14626 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14627 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14629 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14630 documentation for -mfloat128.
14632 2017-09-06 Olivier Hainque <hainque@adacore.com>
14634 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14636 2017-09-06 Wish Wu <wishwu007@gmail.com>
14637 Jakub Jelinek <jakub@redhat.com>
14639 * asan.c (initialize_sanitizer_builtins): Add
14640 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14641 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14642 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14643 BT_FN_VOID_UINT64_PTR variables.
14644 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14645 (BT_FN_VOID_UINT16_UINT16): Likewise.
14646 (BT_FN_VOID_UINT32_UINT32): Likewise.
14647 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14648 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14649 (BT_FN_VOID_UINT64_PTR): Likewise.
14650 * common.opt (flag_sanitize_coverage): New variable.
14651 (fsanitize-coverage=trace-pc): Remove.
14652 (fsanitize-coverage=): Add.
14653 * flag-types.h (enum sanitize_coverage_code): New enum.
14654 * fold-const.c (fold_range_test): Disable non-short-circuit
14655 optimization if flag_sanitize_coverage.
14656 (fold_truth_andor): Likewise.
14657 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14658 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14659 (coverage_sanitizer_opts): New array.
14660 (get_closest_sanitizer_option): Add OPTS argument, handle also
14661 OPT_fsanitize_coverage_.
14662 (parse_sanitizer_options): Adjusted to also handle
14663 OPT_fsanitize_coverage_.
14664 (common_handle_option): Add OPT_fsanitize_coverage_.
14665 * sancov.c (instrument_comparison, instrument_switch): New function.
14666 (sancov_pass): Add trace-cmp support.
14667 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14668 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14669 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14670 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14671 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14672 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14673 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14674 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14675 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14676 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14678 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14680 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14681 error. Only quit immediately if parsing is complete.
14682 (BEGIN): Initialize fatal_err and parse_done.
14683 (begin fpu, end fpu): Check number of arguments.
14684 (begin arch, end arch): Likewise.
14685 (begin cpu, end cpu): Likewise.
14686 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14687 (optalias): Likewise.
14689 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14691 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14692 * config/arm/arm-isa.h: Delete. Move definitions to ...
14693 * arm-cpus.in: ... here. Use new feature and fgroup values.
14694 * config/arm/arm.c (arm_option_override): Use lower case for feature
14696 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14697 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14698 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14700 (print_isa_bits_for): New function.
14701 (gen_isa): New function.
14702 (gen_comm_data): Use print_isa_bits_for.
14703 (define feature): New keyword.
14704 (define fgroup): New keyword.
14705 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14706 (arm-isa.h): Add rule to generate file.
14707 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14708 case for feature bit names.
14710 2017-09-06 Richard Biener <rguenther@suse.de>
14712 * tree-ssa-pre.c (NECESSARY): Remove.
14713 (create_expression_by_pieces): Do not touch pass-local flags.
14714 (insert_into_preds_of_block): Likewise.
14715 (do_pre_regular_insertion): Likewise.
14716 (eliminate_insert): Likewise.
14717 (eliminate_dom_walker::before_dom_children): Likewise.
14718 (fini_eliminate): Do not look at inserted_exprs.
14719 (mark_operand_necessary): Remove.
14720 (remove_dead_inserted_code): Replace with simple work-list
14721 algorithm based on inserted_exprs and SSA uses.
14722 (pass_pre::execute): Re-order fini_eliminate and
14723 remove_dead_inserted_code.
14725 2017-09-06 Olivier Hainque <hainque@adacore.com>
14727 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
14728 for VxWorks 7. Adjust surrounding comments.
14730 2017-09-06 Richard Biener <rguenther@suse.de>
14732 * gimple-ssa-strength-reduction.c
14733 (find_candidates_dom_walker::before_dom_children): Also allow
14736 2017-09-06 Richard Biener <rguenther@suse.de>
14738 PR tree-optimization/82108
14739 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14740 for gap in the non-permutation SLP case.
14742 2017-09-06 Martin Jambor <mjambor@suse.cz>
14744 PR tree-optimization/82078
14745 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14746 add_access_to_work_queue...
14747 (build_accesses_from_assign): ...here.
14748 (propagate_all_subaccesses): Make sure racc is the group
14749 representative, if there is one.
14751 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14753 PR middle-end/82095
14754 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14757 2017-09-06 Richard Biener <rguenther@suse.de>
14759 * gimple-ssa-strength-reduction.c
14760 (find_candidates_dom_walker::before_doom_children): Use a
14761 type and not a mode check.
14763 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14766 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14768 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14770 2017-09-05 Jeff Law <law@redhat.com>
14772 PR tree-optimization/64910
14773 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14774 cases where we have 3 or more operands.
14776 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14778 PR middle-end/81768
14779 * omp-low.c (lower_omp_for): Recompute tree invariant if
14780 gimple_omp_for_initial/final is ADDR_EXPR.
14782 PR middle-end/81768
14783 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14784 into gimple val before gimplification fo the COND_EXPR.
14786 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14788 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14789 REGION_COPY argument.
14790 (thread_through_all_blocks): Remove unused argument to
14791 duplicate_thread_path.
14793 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14794 Alan Hayward <alan.hayward@arm.com>
14795 David Sherwood <david.sherwood@arm.com>
14797 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14798 Take a scalar_mode rather than a machine_mode.
14799 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14800 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14801 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14802 (aarch64_gen_adjusted_ldpstp): Likewise.
14803 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14805 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14806 Alan Hayward <alan.hayward@arm.com>
14807 David Sherwood <david.sherwood@arm.com>
14809 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14810 Take a scalar_int_mode instead of a machine_mode.
14811 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14812 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14813 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14814 (aarch64_simd_attr_length_rglist): Delete.
14815 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14816 a scalar_int_mode instead of a machine_mode.
14817 (aarch64_add_offset): Likewise.
14818 (aarch64_internal_mov_immediate): Likewise
14819 (aarch64_add_constant_internal): Likewise.
14820 (aarch64_add_constant): Likewise.
14821 (aarch64_movw_imm): Likewise.
14822 (aarch64_rtx_arith_op_extract_p): Likewise.
14823 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14824 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14825 Remove assert that the mode isn't a vector.
14826 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14827 (aarch64_expand_mov_immediate): Update calls after above changes.
14828 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14829 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14830 (aarch64_move_imm): Likewise.
14831 (aarch64_can_const_movi_rtx_p): Likewise.
14832 (aarch64_strip_extend): Likewise.
14833 (aarch64_extr_rtx_p): Likewise.
14834 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14835 a CONST_INT when the mode parameter is VOIDmode.
14836 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14838 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14840 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14841 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14842 (bitwise_type_for_mode): Update accordingly.
14844 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14846 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14847 * stor-layout.c (mode_for_size_tree): Likewise.
14848 (mode_for_array): Update accordingly.
14849 (layout_decl): Likewise.
14850 (compute_record_mode): Likewise. Only set the mode once.
14852 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14854 * target.def (get_mask_mode): Change return type to opt_mode.
14856 * doc/tm.texi: Regenerate.
14857 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14858 * targhooks.c (default_get_mask_mode): Likewise.
14859 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14860 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14861 targetm.get_mask_mode.
14862 * tree.c (build_truth_vector_type): Likewise.
14864 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14866 * machmode.h (mode_for_vector): Return an opt_mode.
14867 * stor-layout.c (mode_for_vector): Likewise.
14868 (mode_for_int_vector): Update accordingly.
14869 (layout_type): Likewise.
14870 * config/i386/i386.c (emit_memmov): Likewise.
14871 (ix86_expand_set_or_movmem): Likewise.
14872 (ix86_expand_vector_init): Likewise.
14873 (ix86_get_mask_mode): Likewise.
14874 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14876 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14877 * expmed.c (extract_bit_field_1): Likewise.
14878 * expr.c (expand_expr_real_2): Likewise.
14879 * optabs-query.c (can_vec_perm_p): Likewise.
14880 (can_vec_mask_load_store_p): Likewise.
14881 * optabs.c (expand_vec_perm): Likewise.
14882 * targhooks.c (default_get_mask_mode): Likewise.
14883 * tree-vect-stmts.c (vectorizable_store): Likewise.
14884 (vectorizable_load): Likewise.
14885 (get_vectype_for_scalar_type_and_size): Likewise.
14887 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14889 * machmode.h (mode_for_int_vector): New function.
14890 * stor-layout.c (mode_for_int_vector): Likewise.
14891 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14892 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14893 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14894 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14895 (s390_expand_vcond): Likewise.
14897 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14899 * machmode.h (opt_machine_mode): New type.
14900 (opt_mode<T>): Allow construction from anything that can be
14902 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14903 (mode_for_size): Return an opt_machine_mode.
14904 * stor-layout.c (mode_for_size): Likewise.
14905 (mode_for_size_tree): Update call accordingly.
14906 (bitwise_mode_for_mode): Likewise.
14907 (make_fract_type): Likewise.
14908 (make_accum_type): Likewise.
14909 * caller-save.c (replace_reg_with_saved_mem): Update call
14911 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14912 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14913 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14914 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14915 * expmed.c (extract_bit_field_1): Likewise.
14916 * reload.c (get_secondary_mem): Likewise.
14917 * varasm.c (assemble_integer): Likewise.
14918 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14921 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14923 * machmode.h (decimal_float_mode_for_size): New function.
14924 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14925 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14926 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14927 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14928 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14929 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14931 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14933 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14934 (get_builtin_sync_mode): Likewise.
14935 (expand_ifn_atomic_compare_exchange): Likewise.
14936 (expand_builtin_atomic_clear): Likewise.
14937 (expand_builtin_atomic_test_and_set): Likewise.
14938 (fold_builtin_atomic_always_lock_free): Likewise.
14939 * calls.c (compute_argument_addresses): Likewise.
14940 (emit_library_call_value_1): Likewise.
14941 (store_one_arg): Likewise.
14942 * combine.c (combine_instructions): Likewise.
14943 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14944 * config/arm/arm.c (arm_function_value): Likewise.
14945 (aapcs_allocate_return_reg): Likewise.
14946 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14947 * config/i386/i386.c (construct_container): Likewise.
14948 (ix86_gimplify_va_arg): Likewise.
14949 (ix86_expand_sse_cmp): Likewise.
14950 (emit_memmov): Likewise.
14951 (emit_memset): Likewise.
14952 (expand_small_movmem_or_setmem): Likewise.
14953 (ix86_expand_pextr): Likewise.
14954 (ix86_expand_pinsr): Likewise.
14955 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14956 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14958 * config/mips/mips.c (mips_function_value_1) Likewise.
14959 (mips_block_move_straight): Likewise.
14960 (mips_expand_ins_as_unaligned_store): Likewise.
14961 * config/powerpcspe/powerpcspe.c
14962 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14963 (rs6000_darwin64_record_arg_flush): Likewise.
14964 * config/rs6000/rs6000.c
14965 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14966 (rs6000_darwin64_record_arg_flush): Likewise.
14967 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14968 (sparc_function_value_1): Likewise.
14969 * config/spu/spu.c (adjust_operand): Likewise.
14970 (spu_emit_branch_or_set): Likewise.
14971 (arith_immediate_p): Likewise.
14972 * emit-rtl.c (gen_lowpart_common): Likewise.
14973 * expr.c (expand_expr_real_1): Likewise.
14974 * function.c (assign_parm_setup_block): Likewise.
14975 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14976 * reload1.c (alter_reg): Likewise.
14977 * stor-layout.c (mode_for_vector): Likewise.
14978 (layout_type): Likewise.
14980 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14982 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14983 (spu_convert_move): Likewise.
14984 * lower-subreg.c (resolve_simple_move): Likewise.
14986 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14989 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14990 define_insn to a define_expand.
14991 (altivec_vsum2sws_direct): New define_insn.
14992 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14994 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14996 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14999 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15004 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15005 realign stack if stack alignment needed is less than incoming
15008 2017-09-05 Marek Polacek <polacek@redhat.com>
15011 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15014 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15016 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15018 2017-09-05 Richard Biener <rguenther@suse.de>
15020 PR tree-optimization/82084
15021 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15023 2017-09-05 Richard Biener <rguenther@suse.de>
15025 PR tree-optimization/82102
15026 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15028 2017-09-05 Martin Liska <mliska@suse.cz>
15030 PR tree-optimization/82032
15031 * tree-cfg.c (generate_range_test): New function.
15032 * tree-cfg.h (generate_range_test): Declared here.
15033 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15034 (cleanup_control_expr_graph): Use it.
15035 * tree-switch-conversion.c (try_switch_expansion): Remove
15037 (emit_case_nodes): Use generate_range_test.
15039 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15042 * config/i386/i386.md (*<btsc><mode>_mask): Add
15043 TARGET_USE_BT to insn constraint.
15044 (*btr<mode>_mask): Ditto.
15046 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15048 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15049 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15051 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15054 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15055 TARGET_NEON and TARGET_IWMMXT.
15056 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15057 TARGET_NEON and TARGET_IWMMXT.
15058 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15060 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15062 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15063 (ix86_rewrite_tls_address): Ditto.
15064 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15065 (ix86_rewrite_tls_address_1): Ditto.
15066 (ix86_rewrite_tls_address): Ditto.
15067 * config/i386/predicates.md (tls_address_pattern): New predicate.
15068 * config/i386/i386.md (TLS address splitter): New splitter.
15070 2017-09-04 Richard Biener <rguenther@suse.de>
15072 PR tree-optimization/82084
15073 * fold-const.h (can_native_encode_string_p): Declare.
15074 * fold-const.c (can_native_encode_string_p): Factor out from ...
15075 (native_encode_string): ... here.
15076 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15077 vectorizing stores from constants we later cannot handle.
15079 2017-09-04 Marek Polacek <polacek@redhat.com>
15082 * doc/invoke.texi: Update -Wtautological-compare documentation.
15084 2017-09-04 Jeff Law <law@redhat.com>
15086 PR tree-optimization/64910
15087 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15088 swap the first and last operand if the last is a constant.
15090 2017-09-04 Marek Polacek <polacek@redhat.com>
15093 * convert.c (do_narrow): When sanitizing signed integer overflows,
15094 bail out for signed types.
15095 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15097 2017-09-04 Richard Biener <rguenther@suse.de>
15099 PR tree-optimization/82060
15100 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15101 Move devirtualization after stmt folding and before EH/AB/noreturn
15102 cleanup to get the stmt refs canonicalized. Use a bool instead
15103 of gimple_modified_p since that doesn't work for NOPs. Schedule
15104 NOPs generated by folding for removal.
15106 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15107 Alan Hayward <alan.hayward@arm.com>
15108 David Sherwood <david.sherwood@arm.com>
15110 * coretypes.h (pad_direction): New enum.
15111 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15112 (FUNCTION_ARG_PADDING): Likewise.
15113 * target.def (function_arg_padding): New hook.
15114 * targhooks.h (default_function_arg_padding): Declare.
15115 * targhooks.c (default_function_arg_padding): New function.
15116 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15117 (TARGET_FUNCTION_ARG_PADDING): ...this.
15118 * doc/tm.texi: Regenerate.
15119 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15120 instead of direction.
15121 (compute_argument_addresses): Likewise.
15122 (load_register_parameters): Likewise.
15123 (emit_library_call_value_1): Likewise.
15124 (store_one_arg): Use targetm.calls.function_arg_padding instead
15125 of FUNCTION_ARG_PADDING.
15126 (must_pass_in_stack_var_size_or_pad): Likewise.
15127 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15128 (emit_group_store): Likewise.
15129 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15130 instead of FUNCTION_ARG_PADDING.
15131 (emit_push_insn): Likewise, and propagate enum change throughout
15133 * function.h (direction): Delete.
15134 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15136 * function.c (assign_parm_find_stack_rtl): Likewise.
15137 (assign_parm_setup_block_p): Likewise.
15138 (assign_parm_setup_block): Likewise.
15139 (gimplify_parameters): Likewise.
15140 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15141 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15143 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15144 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15145 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15146 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15147 (aarch64_function_arg_padding): ...this new function.
15148 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15149 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15150 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15151 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15152 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15153 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15154 (arm_pad_arg_upward): Replace with...
15155 (arm_function_arg_padding): ...this new function.
15156 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15158 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15159 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15160 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15161 (ia64_hpux_function_arg_padding): Replace with...
15162 (ia64_function_arg_padding): ...this new function. Use pad_direction
15163 instead of direction. Check for TARGET_HPUX.
15164 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15165 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15166 (iq2000_function_arg_padding): New function.
15167 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15168 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15169 (mips_function_arg_padding): ...this new function.
15170 (mips_pad_reg_upward): Update accordingly.
15171 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15172 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15173 targetm.calls.function_arg_padding.
15174 (FUNCTION_ARG_PADDING): Delete.
15175 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15176 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15177 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15178 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15179 (nios2_block_reg_padding): Return pad_direction instead of direction.
15180 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15181 instead of direction.
15182 (nios2_function_arg_padding): Likewise. Make static.
15183 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15184 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15185 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15186 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15187 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15188 (pa_function_arg_padding): Make static. Return pad_direction instead
15190 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15191 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15192 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15193 instead of direction. Use targetm.calls.function_arg_padding.
15194 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15195 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15196 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15197 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15198 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15200 (function_arg_padding): Rename to...
15201 (rs6000_function_arg_padding): ...this. Make static. Return
15202 pad_direction instead of direction.
15203 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15204 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15205 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15206 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15207 instead of direction. Use targetm.calls.function_arg_padding.
15208 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15209 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15210 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15211 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15212 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15213 (function_arg_padding): Rename to...
15214 (rs6000_function_arg_padding): ...this. Make static. Return
15215 pad_direction instead of direction.
15216 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15217 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15218 * config/s390/s390.c (s390_function_arg_padding): New function.
15219 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15220 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15221 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15222 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15223 (function_arg_padding): Rename to...
15224 (sparc_function_arg_padding): ...this. Make static. Return
15225 pad_direction instead of direction.
15226 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15227 * config/spu/spu.c (spu_function_arg_padding): New function.
15228 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15229 * system.h (FUNCTION_ARG_PADDING): Poison.
15231 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15232 Alan Hayward <alan.hayward@arm.com>
15233 David Sherwood <david.sherwood@arm.com>
15235 * target.def (modes_tieable_p): New hook.
15236 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15237 (TARGET_MODES_TIEABLE_P): ...this.
15238 * doc/tm.texi.in: Regenerate.
15239 * hooks.h (hook_bool_mode_mode_true): Declare.
15240 * hooks.c (hook_bool_mode_mode_true): New function.
15241 * combine.c (subst): Use targetm.modes_tieable_p instead of
15243 * dse.c (find_shift_sequence): Likewise.
15244 * expmed.c (extract_low_bits): Likewise.
15245 * lower-subreg.c: Include target.h.
15246 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15248 * rtlanal.c (rtx_cost): Likewise.
15249 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15250 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15251 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15252 (TARGET_MODES_TIEABLE_P): Redefine.
15253 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15254 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15255 (TARGET_MODES_TIEABLE_P): Redefine.
15256 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15257 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15258 (arc_modes_tieable_p): New function.
15259 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15260 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15261 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15262 (arm_modes_tieable_p): Make static.
15263 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15264 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15265 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15266 (TARGET_MODES_TIEABLE_P): Redefine.
15267 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15268 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15269 (TARGET_MODES_TIEABLE_P): Redefine.
15270 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15271 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15272 (cr16_modes_tieable_p): New function.
15273 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15274 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15275 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15276 (TRULY_NOOP_TRUNCATION): Update comment.
15277 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15278 (TRULY_NOOP_TRUNCATION): Update comment.
15279 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15280 (frv_modes_tieable_p): New function.
15281 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15282 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15283 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15284 (TARGET_MODES_TIEABLE_P): Redefine.
15285 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15286 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15287 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15288 (TARGET_MODES_TIEABLE_P): Redefine.
15289 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15290 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15291 (ia64_modes_tieable_p): New function.
15292 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15293 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15294 (iq2000_modes_tieable_p): New function.
15295 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15296 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15297 (lm32_modes_tieable_p): New function.
15298 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15299 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15300 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15301 (TARGET_MODES_TIEABLE_P): Redefine.
15302 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15303 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15304 (m32r_modes_tieable_p): New function.
15305 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15306 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15307 (m68k_modes_tieable_p): New function.
15308 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15309 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15310 (mcore_modes_tieable_p): New function.
15311 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15312 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15314 (TARGET_MODES_TIEABLE_P): Redefine.
15315 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15316 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15317 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15318 (TARGET_MODES_TIEABLE_P): Redefine.
15319 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15320 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15321 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15322 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15323 (mn10300_modes_tieable_p): ...this and make static.
15324 (TARGET_MODES_TIEABLE_P): Redefine.
15325 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15326 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15327 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15328 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15329 (msp430_modes_tieable_p): Make static.
15330 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15331 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15332 (TARGET_MODES_TIEABLE_P): Redefine.
15333 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15334 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15335 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15336 (TARGET_MODES_TIEABLE_P): Redefine.
15337 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15338 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15339 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15340 (TARGET_MODES_TIEABLE_P): Redefine.
15341 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15342 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15343 (pdp11_modes_tieable_p): New function.
15344 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15345 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15346 (rs6000_modes_tieable_p): New function.
15347 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15348 * config/powerpcspe/powerpcspe.md: Update comment.
15349 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15350 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15351 (TARGET_MODES_TIEABLE_P): Redefine.
15352 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15353 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15354 (rl78_modes_tieable_p): New function.
15355 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15356 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15357 (rs6000_modes_tieable_p): New function.
15358 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15359 * config/rs6000/rs6000.md: Update comment.
15360 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15361 * config/rx/rx.c (rx_modes_tieable_p): New function.
15362 (TARGET_MODES_TIEABLE_P): Redefine.
15363 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15364 * config/s390/s390.c (s390_modes_tieable_p): New function.
15365 (TARGET_MODES_TIEABLE_P): Redefine.
15366 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15367 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15368 (sh_modes_tieable_p): New function.
15369 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15370 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15371 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15372 (sparc_modes_tieable_p): Make static.
15373 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15374 * config/spu/spu.c (spu_modes_tieable_p): New function.
15375 (TARGET_MODES_TIEABLE_P): Redefine.
15376 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15377 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15378 (TARGET_MODES_TIEABLE_P): Redefine.
15379 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15380 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15381 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15382 * config/v850/v850.c (v850_modes_tieable_p): New function.
15383 (TARGET_MODES_TIEABLE_P): Redefine.
15384 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15385 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15386 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15387 (visium_modes_tieable_p): New function.
15388 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15389 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15390 (xtensa_modes_tieable_p): New function.
15391 * system.h (MODES_TIEABLE_P): Poison.
15393 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15394 Alan Hayward <alan.hayward@arm.com>
15395 David Sherwood <david.sherwood@arm.com>
15397 * target.def (hard_regno_mode_ok): New hook.
15398 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15399 (TARGET_HARD_REGNO_MODE_OK): ...this.
15400 * doc/tm.texi.in: Regenerate.
15401 * hooks.h (hook_bool_uint_mode_true): Declare.
15402 * hooks.c (hook_bool_uint_mode_true): New function.
15403 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15404 HARD_REGNO_MODE_OK.
15405 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15406 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15407 instead of HARD_REGNO_MODE_OK.
15408 * caller-save.c: Include target.h.
15409 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15410 HARD_REGNO_MODE_OK.
15411 * combine.c (can_combine_p): Likewise.
15412 (combinable_i3pat): Likewise.
15413 (can_change_dest_mode): Likewise.
15414 * expr.c (init_expr_target): Likewise.
15415 (convert_move): Likewise.
15416 (convert_modes): Likewise.
15417 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15418 (setup_prohibited_mode_move_regs): Likewise.
15419 * ira.h (target_ira): Likewise.
15420 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15421 * lra-constraints.c (process_alt_operands): Likewise.
15422 (split_reg): Likewise.
15423 * recog.c (peep2_find_free_register): Likewise.
15424 * ree.c (combine_reaching_defs): Likewise.
15425 * regcprop.c (maybe_mode_change): Likewise.
15426 * reginfo.c (init_reg_sets_1): Likewise.
15427 (choose_hard_reg_mode): Likewise.
15428 (simplifiable_subregs): Likewise.
15429 * regrename.c (check_new_reg_p): Likewise.
15430 * reload.c (find_valid_class): Likewise.
15431 (find_valid_class_1): Likewise.
15432 (reload_inner_reg_of_subreg): Likewise.
15433 (push_reload): Likewise.
15434 (combine_reloads): Likewise.
15435 (find_dummy_reload): Likewise.
15436 (find_reloads): Likewise.
15437 * reload1.c (find_reg): Likewise.
15438 (set_reload_reg): Likewise.
15439 (allocate_reload_reg): Likewise.
15440 (choose_reload_regs): Likewise.
15441 (reload_adjust_reg_for_temp): Likewise.
15442 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15443 (simplify_subreg_regno): Likewise.
15444 * sel-sched.c (init_regs_for_mode): Likewise.
15445 * varasm.c (make_decl_rtl): Likewise.
15446 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15447 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15448 HARD_REGNO_MODE_OK.
15449 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15450 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15451 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15452 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15453 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15454 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15455 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15456 (arc_mode_class): Delete.
15457 (HARD_REGNO_MODE_OK): Delete.
15458 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15459 (arc_hard_regno_mode_ok): Rename old array to...
15460 (arc_hard_regno_mode_ok_modes): ...this.
15461 (arc_conditional_register_usage): Update accordingly.
15462 (arc_mode_class): Make static.
15463 (arc_hard_regno_mode_ok): New function.
15464 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15465 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15466 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15467 (arm_hard_regno_mode_ok): Make static.
15468 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15469 HARD_REGNO_MODE_OK.
15470 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15471 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15472 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15474 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15475 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15476 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15477 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15478 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15480 * config/bfin/predicates.md (valid_reg_operand): Use
15481 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15482 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15483 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15484 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15485 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15486 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15487 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15488 (cr16_hard_regno_mode_ok): Make static and return a bool.
15489 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15490 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15491 (cris_hard_regno_mode_ok): New function.
15492 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15493 (epiphany_mode_class): Delete.
15494 (HARD_REGNO_MODE_OK): Delete.
15495 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15496 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15497 (hard_regno_mode_ok): Rename to...
15498 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15499 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15500 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15501 HARD_REGNO_MODE_OK.
15502 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15503 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15504 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15505 (frv_hard_regno_mode_ok): Make static and return a bool.
15506 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15507 HARD_REGNO_MODE_OK.
15508 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15509 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15510 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15511 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15513 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15514 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15515 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15516 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15518 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15519 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15520 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15521 (ia64_hard_regno_mode_ok): New function.
15522 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15523 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15524 (iq2000_hard_regno_mode_ok): New function.
15525 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15526 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15527 (lm32_hard_regno_mode_ok): New function.
15528 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15529 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15530 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15531 instead of HARD_REGNO_MODE_OK.
15532 (m32c_hard_regno_ok): Rename to...
15533 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15534 (m32c_cannot_change_mode_class): Update accordingly.
15535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15536 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15537 (m32r_mode_class): Delete.
15538 (HARD_REGNO_MODE_OK): Delete.
15539 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15540 (m32r_hard_regno_mode_ok): Rename to...
15541 (m32r_hard_regno_modes): ...this.
15542 (m32r_mode_class): Make static.
15543 (m32r_hard_regno_mode_ok): New function.
15544 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15545 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15546 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15547 (m68k_hard_regno_mode_ok): Make static.
15548 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15549 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15550 (mcore_hard_regno_mode_ok): New function.
15551 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15552 (HARD_REGNO_MODE_OK): Delete.
15553 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15555 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15556 (microblaze_hard_regno_mode_ok): New function.
15557 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15558 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15559 (mips_hard_regno_mode_ok): Delete.
15560 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15561 (mips_hard_regno_mode_ok_p): ...this and make static.
15562 (mips_hard_regno_mode_ok_p): Rename to...
15563 (mips_hard_regno_mode_ok_uncached): ...this.
15564 (mips_hard_regno_mode_ok): New function.
15565 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15566 of HARD_REGNO_MODE_OK.
15567 (mips_option_override): Update after above name changes.
15568 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15569 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15570 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15571 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15572 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15574 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15575 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15576 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15577 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15578 (msp430_hard_regno_mode_ok): Make static and return a bool.
15579 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15580 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15581 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15583 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15584 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15585 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15586 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15587 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15588 (PA_HARD_REGNO_MODE_OK): ...this
15589 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15590 (PA_HARD_REGNO_MODE_OK): ...this.
15591 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15592 (pa_hard_regno_mode_ok): New function.
15593 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15594 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15595 (pdp11_hard_regno_mode_ok): New function.
15596 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15597 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15599 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15601 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15602 (rs6000_hard_regno_mode_ok): Rename to...
15603 (rs6000_hard_regno_mode_ok_uncached): ...this.
15604 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15605 (rs6000_hard_regno_mode_ok): New function.
15606 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15607 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15608 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15609 (riscv_hard_regno_mode_ok): ...this and make static.
15610 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15611 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15612 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15613 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15614 (rl78_hard_regno_mode_ok): Make static and return bool.
15615 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15616 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15618 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15620 (rs6000_hard_regno_mode_ok): Rename to...
15621 (rs6000_hard_regno_mode_ok_uncached): ...this.
15622 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15623 (rs6000_hard_regno_mode_ok): New function.
15624 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15625 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15626 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15627 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15628 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15629 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15630 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15631 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15632 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15633 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15634 (sh_hard_regno_mode_ok): Make static.
15635 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15636 instead of HARD_REGNO_MODE_OK.
15637 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15638 (sparc_mode_class): Delete.
15639 (HARD_REGNO_MODE_OK): Delete.
15640 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15641 (hard_regno_mode_classes): Make static.
15642 (sparc_mode_class): Likewise.
15643 (sparc_hard_regno_mode_ok): New function.
15644 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15645 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15646 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15648 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15649 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15650 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15651 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15652 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15653 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15654 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15655 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15656 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15657 (visium_hard_regno_mode_ok): New function.
15658 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15659 instead of HARD_REGNO_MODE_OK.
15660 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15661 (HARD_REGNO_MODE_OK): Delete.
15662 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15663 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15664 (xtensa_option_override): Update accordingly.
15665 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15666 (xtensa_hard_regno_mode_ok): New function.
15667 * system.h (HARD_REGNO_MODE_OK): Poison.
15669 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15670 Alan Hayward <alan.hayward@arm.com>
15671 David Sherwood <david.sherwood@arm.com>
15673 * target.def (hard_regno_call_part_clobbered): New hook.
15674 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15675 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15676 * doc/tm.texi: Regenerate.
15677 * hooks.h (hook_bool_uint_mode_false): Declare.
15678 * hooks.c (hook_bool_uint_mode_false): New function.
15679 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15680 * cselib.c (cselib_process_insn): Use
15681 targetm.hard_regno_call_part_clobbered instead of
15682 HARD_REGNO_CALL_PART_CLOBBERED.
15683 * ira-conflicts.c (ira_build_conflicts): Likewise.
15684 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15685 * lra-constraints.c (need_for_call_save_p): Likewise.
15686 * lra-lives.c: Include target.h.
15687 (check_pseudos_live_through_calls): Use
15688 targetm.hard_regno_call_part_clobbered instead of
15689 HARD_REGNO_CALL_PART_CLOBBERED.
15690 * regcprop.c: Include target.h.
15691 (copyprop_hardreg_forward_1): Use
15692 targetm.hard_regno_call_part_clobbered instead of
15693 HARD_REGNO_CALL_PART_CLOBBERED.
15694 * reginfo.c (choose_hard_reg_mode): Likewise.
15695 * regrename.c (check_new_reg_p): Likewise.
15696 * reload.c (find_equiv_reg): Likewise.
15697 * reload1.c (emit_reload_insns): Likewise.
15698 * sched-deps.c (deps_analyze_insn): Likewise.
15699 * sel-sched.c (init_regs_for_mode): Likewise.
15700 (mark_unavailable_hard_regs): Likewise.
15701 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15702 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15703 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15705 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15706 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15707 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15709 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15711 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15712 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15713 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
15715 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15716 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15717 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
15719 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15720 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
15722 * config/powerpcspe/powerpcspe.c
15723 (rs6000_hard_regno_call_part_clobbered): New function.
15724 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15725 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15726 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
15728 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15729 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15730 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
15732 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15733 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15734 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15736 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15737 Alan Hayward <alan.hayward@arm.com>
15738 David Sherwood <david.sherwood@arm.com>
15740 * rtl.h (subreg_memory_offset): Declare.
15741 * emit-rtl.c (subreg_memory_offset): New function.
15742 * expmed.c (store_bit_field_1): Use it.
15743 * expr.c (undefined_operand_subword_p): Likewise.
15744 * simplify-rtx.c (simplify_subreg): Likewise.
15746 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15748 PR rtl-optimization/57448
15751 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15752 using atomic_load pattern.
15753 (expand_atomic_store): Likewise.
15755 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15758 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15759 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15760 handling. Use replace_call_with_value with NULL instead of
15761 gsi_replace, unlink_stmt_vdef and release_defs.
15763 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15766 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15768 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15771 * rtl.h (emit_library_call_value_1): Declare.
15772 (emit_library_call): Replace declaration with a series of overloads.
15773 Remove the parameter count argument.
15774 (emit_library_call_value): Likewise.
15775 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15776 with an "rtx_mode_t *".
15777 (emit_library_call_value): Delete.
15778 (emit_library_call): Likewise.
15779 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15780 (asan_emit_allocas_unpoison): Likewise.
15781 * builtins.c (expand_builtin_powi): Likewise.
15782 (expand_asan_emit_allocas_unpoison): Likewise.
15783 * cfgexpand.c (expand_main_function): Likewise.
15784 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15785 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15786 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15787 * config/arm/arm.c (arm_trampoline_init): Likewise.
15788 (arm_call_tls_get_addr): Likewise.
15789 (arm_expand_divmod_libfunc): Likewise.
15790 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15791 (smulsi3_highpart): Likewise.
15792 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15793 (c6x_expand_compare): Likewise.
15794 (c6x_expand_movmem): Likewise.
15795 * config/frv/frv.c (frv_trampoline_init): Likewise.
15796 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15797 (ix86_expand_divmod_libfunc): Likewise.
15798 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15799 (ia64_expand_compare): Likewise.
15800 (ia64_profile_hook): Likewise.
15801 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15802 (nonlocal_goto): Likewise.
15803 (restore_stack_nonlocal): Likewise.
15804 * config/m32r/m32r.c (block_move_call): Likewise.
15805 (m32r_trampoline_init): Likewise.
15806 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15807 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15808 (m68k_call_m68k_read_tp): Likewise.
15809 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15810 (microblaze_expand_divide): Likewise.
15811 * config/mips/mips.h (mips_args): Likewise.
15812 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15813 (MIPS_ICACHE_SYNC): Likewise.
15814 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15815 (nios2_trampoline_init): Likewise.
15816 * config/pa/pa.c (hppa_tls_call): Likewise.
15817 (pa_trampoline_init): Likewise.
15818 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15819 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15820 (expand_strn_compare): Likewise.
15821 (rs6000_generate_compare): Likewise.
15822 (rs6000_expand_float128_convert): Likewise.
15823 (output_profile_hook): Likewise.
15824 (rs6000_trampoline_init): Likewise.
15825 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15826 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15827 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15828 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15829 (rs6000_generate_compare): Likewise.
15830 (rs6000_expand_float128_convert): Likewise.
15831 (output_profile_hook): Likewise.
15832 (rs6000_trampoline_init): Likewise.
15833 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15834 * config/sh/sh.c (sh_trampoline_init): Likewise.
15835 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15836 (sparc_emit_float_lib_cmp): Likewise.
15837 (sparc32_initialize_trampoline): Likewise.
15838 (sparc64_initialize_trampoline): Likewise.
15839 (sparc_profile_hook): Likewise.
15840 * config/spu/spu.c (ea_load_store): Likewise.
15841 * config/spu/spu.md (floatunssidf2): Likewise.
15842 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15843 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15844 * config/visium/visium.c (expand_block_move_4): Likewise.
15845 (expand_block_move_2): Likewise.
15846 (expand_block_move_1): Likewise.
15847 (expand_block_set_4): Likewise.
15848 (expand_block_set_2): Likewise.
15849 (expand_block_set_1): Likewise.
15850 (visium_trampoline_init): Likewise.
15851 (visium_profile_hook): Likewise.
15852 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15853 (xtensa_setup_frame_addresses): Likewise.
15854 (xtensa_trampoline_init): Likewise.
15855 * except.c (sjlj_emit_function_enter): Likewise.
15856 (sjlj_emit_function_exit): Likewise.
15857 * explow.c (allocate_dynamic_stack_space): Likewise.
15858 (probe_stack_range): Likewise.
15859 * expr.c (convert_mode_scalar): Likewise.
15860 * optabs.c (expand_binop): Likewise.
15861 (expand_twoval_binop_libfunc): Likewise.
15862 (expand_unop): Likewise.
15863 (prepare_cmp_insn): Likewise.
15864 (prepare_float_lib_cmp): Likewise.
15865 (expand_float): Likewise.
15866 (expand_fix): Likewise.
15867 (expand_fixed_convert): Likewise.
15868 (maybe_emit_sync_lock_test_and_set): Likewise.
15869 (expand_atomic_compare_and_swap): Likewise.
15870 (expand_mem_thread_fence): Likewise.
15871 (expand_atomic_fetch_op): Likewise.
15873 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15875 * doc/generic.texi (OpenACC): Adjust URL.
15876 * doc/invoke.texi (C Dialect Options): Ditto.
15878 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15880 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15881 predicate for operand 1. Add (m,<S>) constraint.
15882 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15883 Prevent memory operand 1 with register operand 2.
15885 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15887 PR rtl-optimization/82024
15888 * combine.c (try_combine): If the combination result is a PARALLEL,
15889 and we only need to retain the SET in there that would be placed
15890 at I2, check that we can place that at I3 instead, before doing so.
15892 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15895 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15897 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15898 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15901 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15902 Jeff Law <law@redhat.com>
15904 * varasm.c (bss_initializer_p): Do not put constants into .bss
15905 (categorize_decl_for_section): Handle bss_initializer_p returning
15906 false when DECL_INITIAL is NULL.
15908 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15911 * config/s390/s390.c (s390_can_inline_p): New function.
15913 2017-09-01 Jeff Law <law@redhat.com>
15915 PR tree-optimization/82052
15916 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15917 Always initialize the returned slot after a hash table miss
15918 when INSERT is true.
15920 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15922 * config/s390/s390.md (mem_signal_fence): Remove.
15923 * doc/md.texi (mem_signal_fence): Remove.
15924 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15926 * target-insns.def (mem_signal_fence): Remove.
15928 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15931 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15934 * asan.c (create_odr_indicator): Strip name encoding from assembler
15935 name before appending it after __odr_asan_.
15937 2017-09-01 Martin Liska <mliska@suse.cz>
15939 PR tree-optimization/82059
15940 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15941 frequency only when an edge is redirected.
15943 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15945 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15946 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15947 (arc_conditional_register_usage): Remove ARC600 lp_count
15949 (arc_file_start): Emit Tag_ARC_CPU_variation.
15950 (arc_can_use_doloop_p): New conditions to use ZOLs.
15951 (hwloop_fail): New function.
15952 (hwloop_optimize): Likewise.
15953 (hwloop_pattern_reg): Likewise.
15954 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15955 (arc_reorg_loops): New function, calls reorg_loops.
15956 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15957 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15959 (arc_loop_hazard): Remove function, functionality moved into
15961 (arc_hazard): Remove arc_loop_hazard call.
15962 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15963 into hwloop_optimize.
15964 (arc_label_align): Remove ZOL handling.
15965 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15966 * config/arc/arc.md (doloop_begin): Remove pattern.
15967 (doloop_begin_i): Likewise.
15968 (doloop_end_i): Likewise.
15969 (doloop_fallback): Likewise.
15970 (doloop_fallback_m): Likewise.
15971 (doloop_end): Reimplement expand.
15972 (arc_lp): New pattern for LP instruction.
15973 (loop_end): New pattern.
15974 (loop_fail): Likewise.
15975 (decrement_and_branch_until_zero): Likewise.
15976 * config/arc/arc.opt (mlpc-width): New option.
15977 * doc/invoke.texi (mlpc-width): Document option.
15979 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15981 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15982 (arc_ccfsm_advance): Fix checking for delay slots.
15983 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15985 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15987 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15989 (movhi_insn): Update store instruction constraint which are saving
15990 6-bit short immediates.
15991 (movsi_insn): Consider also short scaled load operations.
15992 (zero_extendhisi2_i): Use Usd constraint instead of T.
15993 (extendhisi2_i): Add q constraint.
15994 (arc_clzsi2): Add type and length attributes.
15995 (arc_ctzsi2): Likewise.
15996 * config/arc/constraints.md (Usc): Update constraint, the
15997 assembler can parse two relocations for a single instruction.
15999 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16001 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16002 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16004 2017-08-31 Olivier Hainque <hainque@adacore.com>
16006 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16007 match as powerpc-wrs-vxworks*.
16009 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16011 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16012 register constraint for by-element operand.
16013 (aarch64_mls_elt_merge<mode>): Likewise.
16015 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16017 * config/arc/arc.c (arc_can_follow_jump): Check for short
16020 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16022 * config.gcc: Use g.opt for arc.
16023 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16024 functionality moved to ...
16025 (legitimate_scaled_address_p): New function, ...here.
16026 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16027 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16028 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16030 (arc_override_options): Handle G option.
16031 (arc_output_pic_addr_const): Correct function definition.
16032 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16033 (arc_decl_anon_ns_mem_p): Delete.
16034 (arc_in_small_data_p): Overhaul this function to take into
16035 consideration the value given via G option.
16036 (arc_rewrite_small_data_1): Renamed and corrected old
16037 arc_rewrite_small_data function.
16038 (arc_rewrite_small_data): New function.
16039 (small_data_pattern): Don't use pic_offset_table_rtx.
16040 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16041 * config/arc/simdext.md (movmisalignv2hi): Use
16042 prepare_move_operands function.
16044 (movmisalign*): Likewise.
16045 * doc/invoke.texi (ARC options): Document -G option.
16047 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16049 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16051 * config/arc/arc.c (arc_print_operand): Output scalled address for
16052 sdata whenever is possible.
16053 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16054 load/stores are available.
16055 (compact_sda_memory_operand): Check for the alignment required by
16056 code density instructions.
16057 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16059 * config/arc/constraints.md (Usd): Update constraint.
16060 (Us0): New constraint.
16061 (Usc): Update constraint.
16063 2017-08-31 Richard Biener <rguenther@suse.de>
16065 PR middle-end/82054
16066 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16067 function only once.
16069 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16071 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16072 Resize type_signature.
16074 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16075 Alan Hayward <alan.hayward@arm.com>
16076 David Sherwood <david.sherwood@arm.com>
16078 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16079 subregs whose inner modes can be stored in GPRs.
16080 (aarch64_classify_index): Likewise.
16082 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16083 Alan Hayward <alan.hayward@arm.com>
16084 David Sherwood <david.sherwood@arm.com>
16086 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16087 (V_INT_EQUIV): ...this.
16088 (v_cmp_result): Rename to...
16089 (v_int_equiv): ...this.
16090 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16091 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16092 (copysign<mode>3): Likewise.
16093 (aarch64_simd_bsl<mode>_internal): Likewise.
16094 (aarch64_simd_bsl<mode>): Likewise.
16095 (vec_cmp<mode><mode>): Likewise.
16096 (vcond<mode><mode>): Likewise.
16097 (vcond<v_cmp_mixed><mode>): Likewise.
16098 (vcondu<mode><v_cmp_mixed>): Likewise.
16099 (aarch64_cm<optab><mode>): Likewise.
16100 (aarch64_cmtst<mode>): Likewise.
16101 (aarch64_fac<optab><mode>): Likewise.
16102 (vec_perm_const<mode>): Likewise.
16103 (vcond_mask_<mode><v_cmp_result>): Rename to...
16104 (vcond_mask_<mode><v_int_equiv>): ...this.
16105 (vec_cmp<mode><v_cmp_result>): Rename to...
16106 (vec_cmp<mode><v_int_equiv>): ...this.
16108 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16109 Alan Hayward <alan.hayward@arm.com>
16110 David Sherwood <david.sherwood@arm.com>
16112 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16114 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16115 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16116 (UNSPEC_LD4_DREG): New unspecs.
16117 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16118 (aarch64_ld2<mode>_dreg_be): Replace with...
16119 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16121 (aarch64_ld3<mode>_dreg_le)
16122 (aarch64_ld3<mode>_dreg_be): Replace with...
16123 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16125 (aarch64_ld4<mode>_dreg_le)
16126 (aarch64_ld4<mode>_dreg_be): Replace with...
16127 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16130 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16132 PR tree-optimization/81987
16133 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16134 insert an initializer in a location not dominated by the stride
16137 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16139 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16140 on the entire header of the finally block in the fallthru case.
16142 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16144 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16146 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16148 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16149 rs6000_emit_move_from_cr and call renamed function.
16150 (rs6000_emit_prologue): Call renamed functions.
16151 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16152 movesi_from_cr, remove volatile CRs.
16154 2017-08-30 Jon Beniston <jon@beniston.com>
16155 Richard Biener <rguenther@suse.de>
16157 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16158 of VECTOR_MODE_P check.
16159 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16160 element vector types.
16162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16164 * df.h (df_read_modify_subreg_p): Remove in favor of...
16165 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16166 const_rtx instead of an rtx.
16167 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16168 * df-problems.c (df_word_lr_mark_ref): Likewise.
16169 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16170 (mark_pseudo_reg_dead): Likewise.
16171 (mark_ref_dead): Likewise.
16172 * reginfo.c (init_subregs_of_mode): Likewise.
16173 * sched-deps.c (sched_analyze_1): Likewise.
16174 * df-scan.c (df_def_record_1): Likewise.
16175 (df_uses_record): Likewise.
16176 (df_read_modify_subreg_p): Remove in favor of...
16177 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16178 const_rtx instead of an rtx.
16180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16181 Alan Hayward <alan.hayward@arm.com>
16182 David Sherwood <david.sherwood@arm.com>
16184 * rtl.h (partial_subreg_p): New function.
16185 * caller-save.c (save_call_clobbered_regs): Use it.
16186 * calls.c (expand_call): Likewise.
16187 * combine.c (combinable_i3pat): Likewise.
16188 (simplify_set): Likewise.
16189 (make_extraction): Likewise.
16190 (make_compound_operation_int): Likewise.
16191 (gen_lowpart_or_truncate): Likewise.
16192 (force_to_mode): Likewise.
16193 (make_field_assignment): Likewise.
16194 (reg_truncated_to_mode): Likewise.
16195 (record_truncated_value): Likewise.
16196 (move_deaths): Likewise.
16197 * cse.c (record_jump_cond): Likewise.
16198 (cse_insn): Likewise.
16199 * cselib.c (cselib_lookup_1): Likewise.
16200 * expmed.c (extract_bit_field_using_extv): Likewise.
16201 * function.c (assign_parm_setup_reg): Likewise.
16202 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16203 * ira-build.c (create_insn_allocnos): Likewise.
16204 * lra-coalesce.c (merge_pseudos): Likewise.
16205 * lra-constraints.c (match_reload): Likewise.
16206 (simplify_operand_subreg): Likewise.
16207 (curr_insn_transform): Likewise.
16208 * lra-lives.c (process_bb_lives): Likewise.
16209 * lra.c (new_insn_reg): Likewise.
16210 (lra_substitute_pseudo): Likewise.
16211 * regcprop.c (mode_change_ok): Likewise.
16212 (maybe_mode_change): Likewise.
16213 (copyprop_hardreg_forward_1): Likewise.
16214 * reload.c (push_reload): Likewise.
16215 (find_reloads): Likewise.
16216 (find_reloads_subreg_address): Likewise.
16217 * reload1.c (alter_reg): Likewise.
16218 (eliminate_regs_1): Likewise.
16219 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16221 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16223 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16224 back to if statements, including unpack.
16226 2017-08-30 Martin Liska <mliska@suse.cz>
16228 PR inline-asm/82001
16229 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16231 (func_checker::compare_asm_inputs_outputs): ... this function.
16232 (func_checker::compare_gimple_asm): Use the function to compare
16233 also ASM constrains.
16234 * ipa-icf-gimple.h: Rename the function.
16236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16237 Alan Hayward <alan.hayward@arm.com>
16238 David Sherwood <david.sherwood@arm.com>
16240 * coretypes.h (complex_mode): New type.
16241 * gdbhooks.py (build_pretty_printer): Handle it.
16242 * machmode.h (complex_mode): New class.
16243 (complex_mode::includes_p): New function.
16244 (is_complex_int_mode): Likewise.
16245 (is_complex_float_mode): Likewise.
16246 * genmodes.c (get_mode_class): Handle complex mode classes.
16247 * function.c (expand_function_end): Use is_complex_int_mode.
16249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16250 Alan Hayward <alan.hayward@arm.com>
16251 David Sherwood <david.sherwood@arm.com>
16253 * coretypes.h (scalar_mode_pod): New typedef.
16254 * gdbhooks.py (build_pretty_printer): Handle it.
16255 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16256 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16257 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16258 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16259 as_a <scalar_mode>.
16261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16262 Alan Hayward <alan.hayward@arm.com>
16263 David Sherwood <david.sherwood@arm.com>
16265 * machmode.h (mode_for_vector): Take a scalar_mode instead
16267 * stor-layout.c (mode_for_vector): Likewise.
16268 * explow.c (promote_mode): Use as_a <scalar_mode>.
16269 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16272 Alan Hayward <alan.hayward@arm.com>
16273 David Sherwood <david.sherwood@arm.com>
16275 * target.def (preferred_simd_mode): Take a scalar_mode
16276 instead of a machine_mode.
16277 * targhooks.h (default_preferred_simd_mode): Likewise.
16278 * targhooks.c (default_preferred_simd_mode): Likewise.
16279 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16280 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16281 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16282 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16283 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16284 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16285 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16286 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16288 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16289 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16290 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16291 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16292 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16293 * doc/tm.texi: Regenerate.
16294 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16298 Alan Hayward <alan.hayward@arm.com>
16299 David Sherwood <david.sherwood@arm.com>
16301 * target.def (scalar_mode_supported_p): Take a scalar_mode
16302 instead of a machine_mode.
16303 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16304 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16305 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16306 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16307 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16308 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16309 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16310 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16311 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16312 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16313 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16314 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16315 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16316 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16318 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16319 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16320 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16321 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16322 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16324 * doc/tm.texi: Regenerate.
16326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16327 Alan Hayward <alan.hayward@arm.com>
16328 David Sherwood <david.sherwood@arm.com>
16330 * coretypes.h (opt_scalar_mode): New typedef.
16331 * gdbhooks.py (build_pretty_printers): Handle it.
16332 * machmode.h (mode_iterator::get_2xwider): Add overload for
16334 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16336 * expr.c (convert_mode_scalar): Likewise.
16337 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16338 * optabs.c (expand_float): Likewise.
16339 (expand_fix): Likewise.
16340 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16343 Alan Hayward <alan.hayward@arm.com>
16344 David Sherwood <david.sherwood@arm.com>
16346 * optabs.c (expand_float): Explicitly check for scalars before
16347 using a branching expansion.
16348 (expand_fix): Likewise.
16350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16351 Alan Hayward <alan.hayward@arm.com>
16352 David Sherwood <david.sherwood@arm.com>
16354 * expr.c (convert_mode): Split scalar handling out into...
16355 (convert_mode_scalar): ...this new function. Treat the modes
16358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16359 Alan Hayward <alan.hayward@arm.com>
16360 David Sherwood <david.sherwood@arm.com>
16362 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16364 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16367 Alan Hayward <alan.hayward@arm.com>
16368 David Sherwood <david.sherwood@arm.com>
16370 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16371 rather than a machine_mode.
16372 (fixed_from_string): Likewise.
16373 (fixed_convert): Likewise.
16374 (fixed_convert_from_int): Likewise.
16375 (fixed_convert_from_real): Likewise.
16376 (real_convert_from_fixed): Likewise.
16377 * fixed-value.c (fixed_from_double_int): Likewise.
16378 (fixed_from_string): Likewise.
16379 (fixed_convert): Likewise.
16380 (fixed_convert_from_int): Likewise.
16381 (fixed_convert_from_real): Likewise.
16382 (real_convert_from_fixed): Likewise.
16383 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16386 Alan Hayward <alan.hayward@arm.com>
16387 David Sherwood <david.sherwood@arm.com>
16389 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16390 of separate mode class checks. Do not allow vector modes here.
16391 (immed_wide_int_const): Use as_a <scalar_mode>.
16392 * explow.c (trunc_int_for_mode): Likewise.
16393 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16394 (wi::shwi): Likewise.
16395 (wi::min_value): Likewise.
16396 (wi::max_value): Likewise.
16397 * dwarf2out.c (loc_descriptor): Likewise.
16398 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16399 for CONST_WIDE_INT.
16401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16402 Alan Hayward <alan.hayward@arm.com>
16403 David Sherwood <david.sherwood@arm.com>
16405 * tree.h (SCALAR_TYPE_MODE): New macro.
16406 * expr.c (expand_expr_addr_expr_1): Use it.
16407 (expand_expr_real_2): Likewise.
16408 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16409 (fold_convert_const_fixed_from_int): Likewise.
16410 (fold_convert_const_fixed_from_real): Likewise.
16411 (native_encode_fixed): Likewise
16412 (native_encode_complex): Likewise
16413 (native_encode_vector): Likewise.
16414 (native_interpret_fixed): Likewise.
16415 (native_interpret_real): Likewise.
16416 (native_interpret_complex): Likewise.
16417 (native_interpret_vector): Likewise.
16418 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16419 (simd_clone_adjust_argument_types): Likewise.
16420 (simd_clone_init_simd_arrays): Likewise.
16421 (simd_clone_adjust): Likewise.
16422 * stor-layout.c (layout_type): Likewise.
16423 * tree.c (build_minus_one_cst): Likewise.
16424 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16425 * tree-inline.c (estimate_move_cost): Likewise.
16426 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16427 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16428 (vectorizable_reduction): Likewise.
16429 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16430 (vect_recog_mixed_size_cond_pattern): Likewise.
16431 (check_bool_pattern): Likewise.
16432 (adjust_bool_pattern): Likewise.
16433 (search_type_for_mask_1): Likewise.
16434 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16435 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16436 (vectorizable_load): Likewise.
16437 (vectorizable_store): Likewise.
16438 * ubsan.c (ubsan_encode_value): Likewise.
16439 * varasm.c (output_constant): Likewise.
16441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16442 Alan Hayward <alan.hayward@arm.com>
16443 David Sherwood <david.sherwood@arm.com>
16445 * coretypes.h (scalar_mode): New class.
16446 * machmode.h (scalar_mode): Likewise.
16447 (scalar_mode::includes_p): New function.
16448 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16449 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16450 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16451 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16452 * expmed.c (store_bit_field_1): Likewise.
16453 (extract_bit_field_1): Likewise.
16454 * expr.c (write_complex_part): Likewise.
16455 (read_complex_part): Likewise.
16456 (emit_move_complex_push): Likewise.
16457 (expand_expr_real_2): Likewise.
16458 * function.c (assign_parm_setup_reg): Likewise.
16459 (assign_parms_unsplit_complex): Likewise.
16460 * optabs.c (expand_binop): Likewise.
16461 * rtlanal.c (subreg_get_info): Likewise.
16462 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16463 * varasm.c (output_constant_pool_2): Likewise.
16465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16466 Alan Hayward <alan.hayward@arm.com>
16467 David Sherwood <david.sherwood@arm.com>
16469 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16471 (expmed_mult_highpart_optab): Likewise.
16472 (expmed_mult_highpart): Likewise.
16474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16475 Alan Hayward <alan.hayward@arm.com>
16476 David Sherwood <david.sherwood@arm.com>
16478 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16479 instead of a machine_mode.
16480 (builtin_memset_read_str): Likewise.
16481 * builtins.c (c_readstr): Likewise.
16482 (builtin_memcpy_read_str): Likewise.
16483 (builtin_strncpy_read_str): Likewise.
16484 (builtin_memset_read_str): Likewise.
16485 (builtin_memset_gen_str): Likewise.
16486 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16487 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16488 instead of a machine_mode.
16489 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16491 (make_extraction): Likewise.
16492 (try_widen_shift_mode): Take and return scalar_int_modes instead
16494 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16495 a scalar_int_mode instead of a machine_mode.
16496 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16497 (avr_addr_space_pointer_mode): Likewise.
16498 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16499 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16500 (msp430_unwind_word_mode): Likewise.
16501 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16502 (spu_addr_space_pointer_mode): Likewise.
16503 (spu_addr_space_address_mode): Likewise.
16504 (spu_libgcc_cmp_return_mode): Likewise.
16505 (spu_libgcc_shift_count_mode): Likewise.
16506 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16507 (rl78_addr_space_pointer_mode): Likewise.
16508 (fl78_unwind_word_mode): Likewise.
16509 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16511 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16512 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16513 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16514 (mips_valid_pointer_mode): Likewise.
16515 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16516 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16517 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16519 (ft32_addr_space_address_mode): Likewise.
16520 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16521 scalar_int_mode instead of a machine_mode.
16522 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16524 (m32c_addr_space_address_mode): Likewise.
16525 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16526 (rs6000_eh_return_filter_mode): Likewise.
16527 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16528 (rs6000_eh_return_filter_mode): Likewise.
16529 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16530 (s390_libgcc_shift_count_mode): Likewise.
16531 (s390_unwind_word_mode): Likewise.
16532 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16534 * target.def (mode_rep_extended): Likewise.
16535 (valid_pointer_mode): Likewise.
16536 (addr_space.valid_pointer_mode): Likewise.
16537 (eh_return_filter_mode): Return a scalar_int_mode rather than
16539 (libgcc_cmp_return_mode): Likewise.
16540 (libgcc_shift_count_mode): Likewise.
16541 (unwind_word_mode): Likewise.
16542 (addr_space.pointer_mode): Likewise.
16543 (addr_space.address_mode): Likewise.
16544 * doc/tm.texi: Regenerate.
16545 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16547 (do_jump): Use scalar_int_mode for local variables.
16548 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16549 rather than a machine_mode.
16550 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16551 (scompare_loc_descriptor_wide): Likewise.
16552 (scompare_loc_descriptor_narrow): Likewise.
16553 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16555 * except.c (sjlj_emit_dispatch_table): Likewise.
16556 (expand_builtin_eh_copy_values): Likewise.
16557 * explow.c (convert_memory_address_addr_space_1): Likewise.
16558 Take a scalar_int_mode rather than a machine_mode.
16559 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16560 than a machine_mode.
16561 (memory_address_addr_space): Use scalar_int_mode for local variables.
16562 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16563 rather than a machine_mode.
16564 * expmed.c (mask_rtx): Likewise.
16565 (init_expmed_one_conv): Likewise.
16566 (expand_mult_highpart_adjust): Likewise.
16567 (extract_high_half): Likewise.
16568 (expmed_mult_highpart_optab): Likewise.
16569 (expmed_mult_highpart): Likewise.
16570 (expand_smod_pow2): Likewise.
16571 (expand_sdiv_pow2): Likewise.
16572 (emit_store_flag_int): Likewise.
16573 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16575 (extract_low_bits): Likewise.
16576 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16578 * expr.c (pieces_addr::adjust): Likewise.
16579 (can_store_by_pieces): Likewise.
16580 (store_by_pieces): Likewise.
16581 (clear_by_pieces_1): Likewise.
16582 (expand_expr_addr_expr_1): Likewise.
16583 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16584 (expand_expr_real_1): Likewise.
16585 (try_casesi): Likewise.
16586 * final.c (shorten_branches): Likewise.
16587 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16588 type of "mode" to machine_mode.
16589 * internal-fn.c (expand_arith_overflow_result_store): Take a
16590 scalar_int_mode rather than a machine_mode.
16591 (expand_mul_overflow): Use scalar_int_mode for local variables.
16592 * loop-doloop.c (doloop_modify): Likewise.
16593 (doloop_optimize): Likewise.
16594 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16595 than a machine_mode.
16596 (expand_doubleword_shift_condmove): Likewise.
16597 (expand_doubleword_shift): Likewise.
16598 (expand_doubleword_clz): Likewise.
16599 (expand_doubleword_popcount): Likewise.
16600 (expand_doubleword_parity): Likewise.
16601 (expand_absneg_bit): Use scalar_int_mode for local variables.
16602 (prepare_float_lib_cmp): Likewise.
16603 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16604 rather than a machine_mode.
16605 (convert_memory_address_addr_space): Likewise.
16606 (get_mode_bounds): Likewise.
16607 (get_address_mode): Return a scalar_int_mode rather than a
16609 * rtlanal.c (get_address_mode): Likewise.
16610 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16611 than a machine_mode.
16612 * targhooks.c (default_mode_rep_extended): Likewise.
16613 (default_valid_pointer_mode): Likewise.
16614 (default_addr_space_valid_pointer_mode): Likewise.
16615 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16616 than a machine_mode.
16617 (default_libgcc_cmp_return_mode): Likewise.
16618 (default_libgcc_shift_count_mode): Likewise.
16619 (default_unwind_word_mode): Likewise.
16620 (default_addr_space_pointer_mode): Likewise.
16621 (default_addr_space_address_mode): Likewise.
16622 * targhooks.h (default_eh_return_filter_mode): Likewise.
16623 (default_libgcc_cmp_return_mode): Likewise.
16624 (default_libgcc_shift_count_mode): Likewise.
16625 (default_unwind_word_mode): Likewise.
16626 (default_addr_space_pointer_mode): Likewise.
16627 (default_addr_space_address_mode): Likewise.
16628 (default_mode_rep_extended): Take a scalar_int_mode rather than
16630 (default_valid_pointer_mode): Likewise.
16631 (default_addr_space_valid_pointer_mode): Likewise.
16632 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16634 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16635 rather than a machine_mode.
16636 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16637 for local variables.
16638 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16639 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16640 than a machine_mode.
16642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16643 Alan Hayward <alan.hayward@arm.com>
16644 David Sherwood <david.sherwood@arm.com>
16646 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16647 the mode argument to scalar_int_mode.
16648 (do_jump_by_parts_zero_rtx): Likewise.
16649 (do_jump_by_parts_equality_rtx): Likewise.
16650 (do_jump_by_parts_greater): Take a mode argument.
16651 (do_jump_by_parts_equality): Likewise.
16652 (do_jump_1): Update calls accordingly.
16654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16655 Alan Hayward <alan.hayward@arm.com>
16656 David Sherwood <david.sherwood@arm.com>
16658 * is-a.h (safe_dyn_cast): New function.
16659 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16660 (jump_table_for_label): Likewise.
16661 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16662 instead of an rtx_insn *.
16663 (shorten_branches): Use dyn_cast instead of LABEL_P and
16664 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16665 rtx_jump_table_data::get_data_mode.
16666 (final_scan_insn): Likewise.
16668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16669 Alan Hayward <alan.hayward@arm.com>
16670 David Sherwood <david.sherwood@arm.com>
16672 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16673 trying to combine a full-register integer set with a subreg
16676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16677 Alan Hayward <alan.hayward@arm.com>
16678 David Sherwood <david.sherwood@arm.com>
16680 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16681 that is always either address_mode or pointer_mode.
16683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16684 Alan Hayward <alan.hayward@arm.com>
16685 David Sherwood <david.sherwood@arm.com>
16687 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16688 when the two are known to be equal.
16690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16691 Alan Hayward <alan.hayward@arm.com>
16692 David Sherwood <david.sherwood@arm.com>
16694 * simplify-rtx.c (simplify_const_unary_operation): Use
16695 is_a <scalar_int_mode> instead of checking for a nonzero
16696 precision. Forcibly convert op_mode to a scalar_int_mode
16697 in that case. More clearly differentiate the operand and
16698 result modes and use the former when deciding what the value
16699 of a count-bits operation should be. Use is_int_mode instead
16700 of checking for a MODE_INT. Remove redundant check for whether
16701 this mode has a zero precision.
16703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16704 Alan Hayward <alan.hayward@arm.com>
16705 David Sherwood <david.sherwood@arm.com>
16707 * optabs.c (widen_leading): Change the type of the mode argument
16708 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16709 (widen_bswap): Likewise.
16710 (expand_parity): Likewise.
16711 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16712 (expand_ffs): Likewise.
16713 (epand_unop): Check for scalar integer modes before calling the
16716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16717 Alan Hayward <alan.hayward@arm.com>
16718 David Sherwood <david.sherwood@arm.com>
16720 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
16722 (expand_expr_real_1): Update call accordingly.
16724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16725 Alan Hayward <alan.hayward@arm.com>
16726 David Sherwood <david.sherwood@arm.com>
16728 * expmed.c (store_bit_field_using_insv): Add op0_mode and
16729 value_mode arguments. Use scalar_int_mode internally.
16730 (store_bit_field_1): Rename the new integer mode from imode
16731 to op0_mode and use it instead of GET_MODE (op0). Update calls
16732 to store_split_bit_field, store_bit_field_using_insv and
16733 store_fixed_bit_field.
16734 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16735 Use scalar_int_mode internally. Use a bit count rather than a mode
16736 when calculating the largest bit size for get_best_mode.
16737 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16738 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16739 Remove assertion that OP0 has a scalar integer mode.
16740 (store_split_bit_field): Add op0_mode and value_mode arguments.
16741 Update calls to extract_fixed_bit_field.
16742 (extract_bit_field_using_extv): Add an op0_mode argument.
16743 Use scalar_int_mode internally.
16744 (extract_bit_field_1): Rename the new integer mode from imode to
16745 op0_mode and use it instead of GET_MODE (op0). Update calls to
16746 extract_split_bit_field, extract_bit_field_using_extv and
16747 extract_fixed_bit_field.
16748 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16749 to extract_split_bit_field and extract_fixed_bit_field_1.
16750 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16751 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16752 on the target mode.
16753 (extract_split_bit_field): Add an op0_mode argument. Update call
16754 to extract_fixed_bit_field.
16756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16757 Alan Hayward <alan.hayward@arm.com>
16758 David Sherwood <david.sherwood@arm.com>
16760 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16761 * explow.c (hard_function_value): Likewise.
16762 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16763 convert_to_mode call outside the loop.
16764 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16765 for the mode iterator. Require the mode specified by max_pieces
16767 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16769 (copy_blkmode_to_reg): Likewise.
16770 (set_storage_via_setmem): Likewise.
16771 * optabs.c (prepare_cmp_insn): Likewise.
16772 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16773 * stor-layout.c (finish_bitfield_representative): Likewise.
16775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16776 Alan Hayward <alan.hayward@arm.com>
16777 David Sherwood <david.sherwood@arm.com>
16779 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16780 * expr.c (convert_move): Use them.
16781 (convert_modes): Likewise.
16782 (store_expr_with_bounds): Likewise.
16784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16785 Alan Hayward <alan.hayward@arm.com>
16786 David Sherwood <david.sherwood@arm.com>
16788 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16789 parameter for the mode of "x". Remove the "known_x", "known_mode"
16790 and "known_ret" arguments. Change the type of the mode argument
16791 to scalar_int_mode.
16792 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16793 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16794 (reg_num_sign_bit_copies_for_combine): Likewise.
16795 * rtlanal.c (nonzero_bits1): Likewise.
16796 (num_sign_bit_copies1): Likewise.
16797 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16798 (reg_num_sign_bit_copies_general): Likewise.
16799 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16800 (reg_nonzero_bits_general): Likewise.
16802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16803 Alan Hayward <alan.hayward@arm.com>
16804 David Sherwood <david.sherwood@arm.com>
16806 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16807 than in subroutines. Return 1 for non-integer modes.
16808 (cached_num_sign_bit_copies): Change the type of the mode parameter
16809 to scalar_int_mode.
16810 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16811 classes. Handle CONST_INT_P first and then check whether X also
16812 has a scalar integer mode. Check the same thing for inner registers
16813 of a SUBREG and for values that are being extended or truncated.
16815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16816 Alan Hayward <alan.hayward@arm.com>
16817 David Sherwood <david.sherwood@arm.com>
16819 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16820 in subroutines. Return the mode mask for non-integer modes.
16821 (cached_nonzero_bits): Change the type of the mode parameter
16822 to scalar_int_mode.
16823 (nonzero_bits1): Likewise. Remove early exit for other mode
16824 classes. Handle CONST_INT_P first and then check whether X
16825 also has a scalar integer mode.
16827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16828 Alan Hayward <alan.hayward@arm.com>
16829 David Sherwood <david.sherwood@arm.com>
16831 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16832 Return a scalar_int_mode and assert that the size is greater than
16834 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16835 (op_by_pieces_d::op_by_pieces_d): Likewise.
16836 (op_by_pieces_d::run): Likewise.
16837 (can_store_by_pieces): Likewise.
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 (extract_left_shift): Add a mode argument and update
16845 (make_compound_operation_int): Change the type of the mode parameter
16846 to scalar_int_mode and update the call to extract_left_shift.
16848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16849 Alan Hayward <alan.hayward@arm.com>
16850 David Sherwood <david.sherwood@arm.com>
16852 * combine.c (simplify_and_const_int): Change the type of the mode
16853 parameter to scalar_int_mode.
16854 (simplify_and_const_int_1): Likewise. Update recursive call.
16856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16857 Alan Hayward <alan.hayward@arm.com>
16858 David Sherwood <david.sherwood@arm.com>
16860 * combine.c (simplify_compare_const): Check that the mode is a
16861 scalar_int_mode (rather than VOIDmode) before testing its
16863 (simplify_comparison): Move COMPARISON_P handling out of the
16864 loop and restrict the latter part of the loop to scalar_int_modes.
16865 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16866 and when considering SUBREG_REGs. Use is_int_mode instead of
16867 checking GET_MODE_CLASS against MODE_INT.
16869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16870 Alan Hayward <alan.hayward@arm.com>
16871 David Sherwood <david.sherwood@arm.com>
16873 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16874 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16875 shift_unit_mode and for modes involved in scalar shifts.
16877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16878 Alan Hayward <alan.hayward@arm.com>
16879 David Sherwood <david.sherwood@arm.com>
16881 * combine.c (force_int_to_mode): New function, split out from...
16882 (force_to_mode): ...here. Keep xmode up-to-date and use it
16883 instead of GET_MODE (x).
16885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16886 Alan Hayward <alan.hayward@arm.com>
16887 David Sherwood <david.sherwood@arm.com>
16889 * optabs-query.h (extraction_insn::struct_mode): Change type to
16890 opt_scalar_int_mode and update comment.
16891 (extraction_insn::field_mode): Change type to scalar_int_mode.
16892 (extraction_insn::pos_mode): Likewise.
16893 * combine.c (make_extraction): Update accordingly.
16894 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16895 (get_optab_extraction_insn): Likewise.
16896 * recog.c (simplify_while_replacing): Likewise.
16897 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16898 parameter to opt_scalar_int_mode.
16900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16901 Alan Hayward <alan.hayward@arm.com>
16902 David Sherwood <david.sherwood@arm.com>
16904 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16905 to a scalar_int_mode instead of a machine_mode.
16906 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16907 (get_best_mode): Return a boolean and use a pointer argument to store
16908 the selected mode. Replace the limit mode parameter with a bit limit.
16909 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16910 for the values returned by bit_field_mode_iterator::next_mode.
16911 (store_bit_field): Update call to get_best_mode.
16912 (store_fixed_bit_field): Likewise.
16913 (extract_fixed_bit_field): Likewise.
16914 * expr.c (optimize_bitfield_assignment_op): Likewise.
16915 * fold-const.c (optimize_bit_field_compare): Likewise.
16916 (fold_truth_andor_1): Likewise.
16917 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16918 Update for new type of m_mode.
16919 (get_best_mode): As above.
16921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16922 Alan Hayward <alan.hayward@arm.com>
16923 David Sherwood <david.sherwood@arm.com>
16925 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16926 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16927 (store_bit_field): Check is_a <scalar_int_mode> before calling
16928 strict_volatile_bitfield_p.
16929 (extract_bit_field): Likewise.
16931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16932 Alan Hayward <alan.hayward@arm.com>
16933 David Sherwood <david.sherwood@arm.com>
16935 * target.def (cstore_mode): Return a scalar_int_mode.
16936 * doc/tm.texi: Regenerate.
16937 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16938 * targhooks.h (default_cstore_mode): Likewise.
16939 * targhooks.c (default_cstore_mode): Likewise, using a forced
16941 * expmed.c (emit_cstore): Expect the target of the cstore to be
16944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16945 Alan Hayward <alan.hayward@arm.com>
16946 David Sherwood <david.sherwood@arm.com>
16948 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16950 (niter_desc): Likewise mode.
16951 (iv_analyze): Add a mode parameter.
16953 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16954 and change its type to scalar_int_mode.
16955 * loop-iv.c: Update commentary at head of file.
16956 (iv_constant): Pass the mode paraeter before the rtx it describes
16957 and change its type to scalar_int_mode. Remove VOIDmode handling.
16958 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16959 (iv_extend): Likewise.
16960 (shorten_into_mode): Likewise.
16961 (iv_add): Use scalar_int_mode.
16962 (iv_mult): Likewise.
16963 (iv_shift): Likewise.
16964 (canonicalize_iv_subregs): Likewise.
16965 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16966 it describes and change its mode to scalar_int_mode. Also change
16967 the type of the returned inner_mode to scalar_int_mode.
16968 (get_biv_step): Likewise, turning outer_mode from a pointer
16969 into a direct parameter. Update call to get_biv_step_1.
16970 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16971 iv_constant and get_biv_step.
16972 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16973 and change its type to scalar_int_mode. Don't initialise iv->mode
16974 to VOIDmode and remove later checks for its still being VOIDmode.
16975 Update calls to iv_analyze_op and iv_analyze_expr. Check
16976 is_a <scalar_int_mode> when changing the mode under consideration.
16977 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16978 Update call to iv_analyze_expr.
16979 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16980 inner register is not also a scalar_int_mode. Update call to
16982 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16983 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16984 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16985 separate mode class checks. Update calls to iv_analyze. Remove
16986 fix-up of VOIDmodes after iv_analyze_biv.
16987 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16988 don't have a scalar_int_mode. Update call to biv_p.
16990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16991 Alan Hayward <alan.hayward@arm.com>
16992 David Sherwood <david.sherwood@arm.com>
16994 * cfgexpand.c (convert_debug_memory_address): Use
16995 as_a <scalar_int_mode>.
16996 * combine.c (expand_compound_operation): Likewise.
16997 (make_extraction): Likewise.
16998 (change_zero_ext): Likewise.
16999 (simplify_comparison): Likewise.
17000 * cse.c (cse_insn): Likewise.
17001 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17002 (mem_loc_descriptor): Likewise.
17003 (loc_descriptor): Likewise.
17004 * expmed.c (init_expmed_one_mode): Likewise.
17005 (synth_mult): Likewise.
17006 (emit_store_flag_1): Likewise.
17007 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17008 of a comparison with size.
17009 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17010 (reduce_to_bit_field_precision): Likewise.
17011 * function.c (expand_function_end): Likewise.
17012 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17013 * loop-doloop.c (doloop_modify): Likewise.
17014 * optabs.c (expand_binop): Likewise.
17015 (expand_unop): Likewise.
17016 (expand_copysign_absneg): Likewise.
17017 (prepare_cmp_insn): Likewise.
17018 (maybe_legitimize_operand): Likewise.
17019 * recog.c (const_scalar_int_operand): Likewise.
17020 * rtlanal.c (get_address_mode): Likewise.
17021 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17022 (simplify_cond_clz_ctz): Likewise.
17023 * tree-nested.c (get_nl_goto_field): Likewise.
17024 * tree.c (build_vector_type_for_mode): Likewise.
17025 * var-tracking.c (use_narrower_mode): Likewise.
17027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17028 Alan Hayward <alan.hayward@arm.com>
17029 David Sherwood <david.sherwood@arm.com>
17031 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17032 * builtins.c (expand_builtin_signbit): Use it.
17033 * cfgexpand.c (expand_debug_expr): Likewise.
17034 * dojump.c (do_jump): Likewise.
17035 (do_compare_and_jump): Likewise.
17036 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17037 * expmed.c (make_tree): Likewise.
17038 * expr.c (expand_expr_real_2): Likewise.
17039 (expand_expr_real_1): Likewise.
17040 (try_casesi): Likewise.
17041 * fold-const-call.c (fold_const_call_ss): Likewise.
17042 * fold-const.c (unextend): Likewise.
17043 (extract_muldiv_1): Likewise.
17044 (fold_single_bit_test): Likewise.
17045 (native_encode_int): Likewise.
17046 (native_encode_string): Likewise.
17047 (native_interpret_int): Likewise.
17048 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17049 * internal-fn.c (expand_addsub_overflow): Likewise.
17050 (expand_neg_overflow): Likewise.
17051 (expand_mul_overflow): Likewise.
17052 (expand_arith_overflow): Likewise.
17053 * match.pd: Likewise.
17054 * stor-layout.c (layout_type): Likewise.
17055 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17056 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17057 * tree-ssanames.c (get_range_info): Likewise.
17058 * tree-switch-conversion.c (array_value_type) Likewise.
17059 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17060 (vect_recog_divmod_pattern): Likewise.
17061 (vect_recog_mixed_size_cond_pattern): Likewise.
17062 * tree-vrp.c (extract_range_basic): Likewise.
17063 (simplify_float_conversion_using_ranges): Likewise.
17064 * tree.c (int_fits_type_p): Likewise.
17065 * ubsan.c (instrument_bool_enum_load): Likewise.
17066 * varasm.c (mergeable_string_section): Likewise.
17067 (narrowing_initializer_constant_valid_p): Likewise.
17068 (output_constant): Likewise.
17070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17071 Alan Hayward <alan.hayward@arm.com>
17072 David Sherwood <david.sherwood@arm.com>
17074 * machmode.h (NARROWEST_INT_MODE): New macro.
17075 * expr.c (alignment_for_piecewise_move): Use it instead of
17076 GET_CLASS_NARROWEST_MODE (MODE_INT).
17077 (push_block): Likewise.
17078 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17080 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17083 Alan Hayward <alan.hayward@arm.com>
17084 David Sherwood <david.sherwood@arm.com>
17086 * postreload.c (move2add_valid_value_p): Change the type of the
17087 mode parameter to scalar_int_mode.
17088 (move2add_use_add2_insn): Add a mode parameter and use it instead
17090 (move2add_use_add3_insn): Likewise.
17091 (reload_cse_move2add): Update accordingly.
17093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17094 Alan Hayward <alan.hayward@arm.com>
17095 David Sherwood <david.sherwood@arm.com>
17097 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17099 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17100 * optabs.c (expand_unop): Likewise.
17102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17103 Alan Hayward <alan.hayward@arm.com>
17104 David Sherwood <david.sherwood@arm.com>
17106 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17107 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17108 (popcount_loc_descriptor): Likewise.
17109 (bswap_loc_descriptor): Likewise.
17110 (rotate_loc_descriptor): Likewise.
17111 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17112 calling the functions above.
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 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17120 (try_combine): Likewise.
17121 (simplify_if_then_else): Likewise.
17122 * cse.c (cse_insn): Likewise.
17123 * dwarf2out.c (mem_loc_descriptor): Likewise.
17124 * emit-rtl.c (gen_lowpart_common): Likewise.
17125 * simplify-rtx.c (simplify_truncation): Likewise.
17126 (simplify_binary_operation_1): Likewise.
17127 (simplify_const_relational_operation): Likewise.
17128 (simplify_ternary_operation): Likewise.
17129 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17132 Alan Hayward <alan.hayward@arm.com>
17133 David Sherwood <david.sherwood@arm.com>
17135 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17136 * reload.c (push_reload): Likewise.
17137 (find_reloads): Likewise.
17139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17140 Alan Hayward <alan.hayward@arm.com>
17141 David Sherwood <david.sherwood@arm.com>
17143 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17144 (make_compound_operation_int): Likewise.
17145 (change_zero_ext): Likewise.
17146 * expr.c (convert_move): Likewise.
17147 (convert_modes): Likewise.
17148 * fwprop.c (forward_propagate_subreg): Likewise.
17149 * loop-iv.c (get_biv_step_1): Likewise.
17150 * optabs.c (widen_operand): Likewise.
17151 * postreload.c (move2add_valid_value_p): Likewise.
17152 * recog.c (simplify_while_replacing): Likewise.
17153 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17154 (simplify_binary_operation_1): Likewise. Remove redundant
17155 mode equality check.
17157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17158 Alan Hayward <alan.hayward@arm.com>
17159 David Sherwood <david.sherwood@arm.com>
17161 * combine.c (combine_simplify_rtx): Add checks for
17162 is_a <scalar_int_mode>.
17163 (simplify_if_then_else): Likewise.
17164 (make_field_assignment): Likewise.
17165 (simplify_comparison): Likewise.
17166 * ifcvt.c (noce_try_bitop): Likewise.
17167 * loop-invariant.c (canonicalize_address_mult): Likewise.
17168 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17171 Alan Hayward <alan.hayward@arm.com>
17172 David Sherwood <david.sherwood@arm.com>
17174 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17175 is_a <scalar_int_mode> instead of != BLKmode.
17177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17178 Alan Hayward <alan.hayward@arm.com>
17179 David Sherwood <david.sherwood@arm.com>
17181 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17182 instead of != VOIDmode.
17183 * combine.c (if_then_else_cond): Likewise.
17184 (change_zero_ext): Likewise.
17185 * dwarf2out.c (mem_loc_descriptor): Likewise.
17186 (loc_descriptor): Likewise.
17187 * rtlanal.c (canonicalize_condition): Likewise.
17188 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17191 Alan Hayward <alan.hayward@arm.com>
17192 David Sherwood <david.sherwood@arm.com>
17194 * simplify-rtx.c (simplify_binary_operation_1): Use
17195 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17198 Alan Hayward <alan.hayward@arm.com>
17199 David Sherwood <david.sherwood@arm.com>
17201 * wide-int.h (int_traits<unsigned char>) New class.
17202 (int_traits<unsigned short>) Likewise.
17203 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17204 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17206 * combine.c (set_nonzero_bits_and_sign_copies): Use
17207 is_a <scalar_int_mode>.
17208 (find_split_point): Likewise.
17209 (combine_simplify_rtx): Likewise.
17210 (simplify_logical): Likewise.
17211 (expand_compound_operation): Likewise.
17212 (expand_field_assignment): Likewise.
17213 (make_compound_operation): Likewise.
17214 (extended_count): Likewise.
17215 (change_zero_ext): Likewise.
17216 (simplify_comparison): Likewise.
17217 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17218 (ucompare_loc_descriptor): Likewise.
17219 (minmax_loc_descriptor): Likewise.
17220 (mem_loc_descriptor): Likewise.
17221 (loc_descriptor): Likewise.
17222 * expmed.c (init_expmed_one_mode): Likewise.
17223 * lra-constraints.c (lra_constraint_offset): Likewise.
17224 * optabs.c (prepare_libcall_arg): Likewise.
17225 * postreload.c (move2add_note_store): Likewise.
17226 * reload.c (operands_match_p): Likewise.
17227 * rtl.h (load_extend_op): Likewise.
17228 * rtlhooks.c (gen_lowpart_general): Likewise.
17229 * simplify-rtx.c (simplify_truncation): Likewise.
17230 (simplify_unary_operation_1): Likewise.
17231 (simplify_binary_operation_1): Likewise.
17232 (simplify_const_binary_operation): Likewise.
17233 (simplify_const_relational_operation): Likewise.
17234 (simplify_subreg): Likewise.
17235 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17236 * var-tracking.c (adjust_mems): Likewise.
17237 (prepare_call_arguments): Likewise.
17239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17240 Alan Hayward <alan.hayward@arm.com>
17241 David Sherwood <david.sherwood@arm.com>
17243 * machmode.h (is_int_mode): New fuction.
17244 * combine.c (find_split_point): Use it.
17245 (combine_simplify_rtx): Likewise.
17246 (simplify_if_then_else): Likewise.
17247 (simplify_set): Likewise.
17248 (simplify_shift_const_1): Likewise.
17249 (simplify_comparison): Likewise.
17250 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17251 * cse.c (notreg_cost): Likewise.
17252 (cse_insn): Likewise.
17253 * cselib.c (cselib_lookup_1): Likewise.
17254 * dojump.c (do_jump_1): Likewise.
17255 (do_compare_rtx_and_jump): Likewise.
17256 * dse.c (get_call_args): Likewise.
17257 * dwarf2out.c (rtl_for_decl_init): Likewise.
17258 (native_encode_initializer): Likewise.
17259 * expmed.c (emit_store_flag_1): Likewise.
17260 (emit_store_flag): Likewise.
17261 * expr.c (convert_modes): Likewise.
17262 (store_field): Likewise.
17263 (expand_expr_real_1): Likewise.
17264 * fold-const.c (fold_read_from_constant_string): Likewise.
17265 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17266 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17267 * optabs.c (expand_binop): Likewise.
17268 (expand_unop): Likewise.
17269 (expand_abs_nojump): Likewise.
17270 (expand_one_cmpl_abs_nojump): Likewise.
17271 * simplify-rtx.c (mode_signbit_p): Likewise.
17272 (val_signbit_p): Likewise.
17273 (val_signbit_known_set_p): Likewise.
17274 (val_signbit_known_clear_p): Likewise.
17275 (simplify_relational_operation_1): Likewise.
17276 * tree.c (vector_type_mode): Likewise.
17278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17279 Alan Hayward <alan.hayward@arm.com>
17280 David Sherwood <david.sherwood@arm.com>
17282 * machmode.h (smallest_mode_for_size): Fix formatting.
17283 (smallest_int_mode_for_size): New function.
17284 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17285 instead of smallest_mode_for_size.
17286 * combine.c (make_extraction): Likewise.
17287 * config/arc/arc.c (arc_expand_movmem): Likewise.
17288 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17289 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17290 * config/s390/s390.c (s390_expand_insv): Likewise.
17291 * config/sparc/sparc.c (assign_int_registers): Likewise.
17292 * config/spu/spu.c (spu_function_value): Likewise.
17293 (spu_function_arg): Likewise.
17294 * coverage.c (get_gcov_type): Likewise.
17295 (get_gcov_unsigned_t): Likewise.
17296 * dse.c (find_shift_sequence): Likewise.
17297 * expmed.c (store_bit_field_1): Likewise.
17298 * expr.c (convert_move): Likewise.
17299 (store_field): Likewise.
17300 * internal-fn.c (expand_arith_overflow): Likewise.
17301 * optabs-query.c (get_best_extraction_insn): Likewise.
17302 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17303 * stor-layout.c (layout_type): Likewise.
17304 (initialize_sizetypes): Likewise.
17305 * targhooks.c (default_get_mask_mode): Likewise.
17306 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17309 Alan Hayward <alan.hayward@arm.com>
17310 David Sherwood <david.sherwood@arm.com>
17312 * machmode.h (opt_mode::else_blk): New function.
17313 (int_mode_for_mode): Declare.
17314 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17315 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17317 * cfgexpand.c (expand_debug_expr): Likewise.
17318 * combine.c (gen_lowpart_or_truncate): Likewise.
17319 (gen_lowpart_for_combine): Likewise.
17320 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17321 * config/avr/avr.c (avr_to_int_mode): Likewise.
17322 (avr_out_plus_1): Likewise.
17323 (avr_out_plus): Likewise.
17324 (avr_out_round): Likewise.
17325 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17326 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17327 (s390_expand_vcond): Likewise.
17328 * config/spu/spu.c (spu_split_immediate): Likewise.
17329 (spu_expand_mov): Likewise.
17330 * dse.c (get_stored_val): Likewise.
17331 * expmed.c (store_bit_field_1): Likewise.
17332 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17334 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17335 (extract_low_bits): Likewise.
17336 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17337 handling rather than repeating the check.
17338 (emit_group_store): Likewise.
17339 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17340 * optabs.c (expand_absneg_bit): Likewise.
17341 (expand_copysign_absneg): Likewise.
17342 (expand_copysign_bit): Likewise.
17343 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17344 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17345 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17346 * var-tracking.c (prepare_call_arguments): Likewise.
17347 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17348 int_mode_for_mode instead of mode_for_size.
17349 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17352 Alan Hayward <alan.hayward@arm.com>
17353 David Sherwood <david.sherwood@arm.com>
17355 * machmode.h (int_mode_for_size): New function.
17356 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17357 instead of mode_for_size.
17358 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17360 * combine.c (expand_field_assignment): Use int_mode_for_size
17361 instead of mode_for_size.
17362 (make_extraction): Likewise.
17363 (simplify_shift_const_1): Likewise.
17364 (simplify_comparison): Likewise.
17365 * dojump.c (do_jump): Likewise.
17366 * dwarf2out.c (mem_loc_descriptor): Likewise.
17367 * emit-rtl.c (init_derived_machine_modes): Likewise.
17368 * expmed.c (flip_storage_order): Likewise.
17369 (convert_extracted_bit_field): Likewise.
17370 * expr.c (copy_blkmode_from_reg): Likewise.
17371 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17372 * internal-fn.c (expand_mul_overflow): Likewise.
17373 * lower-subreg.c (simple_move): Likewise.
17374 * optabs-libfuncs.c (init_optabs): Likewise.
17375 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17376 * tree.c (vector_type_mode): Likewise.
17377 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17378 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17379 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17380 * tree-vect-stmts.c (vectorizable_load): Likewise.
17381 (vectorizable_store): Likewise.
17383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17384 Alan Hayward <alan.hayward@arm.com>
17385 David Sherwood <david.sherwood@arm.com>
17387 * coretypes.h (pod_mode): New type.
17388 (scalar_int_mode_pod): New typedef.
17389 * machmode.h (pod_mode): New class.
17390 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17391 * genmodes.c (emit_mode_int_n): Update accordingly.
17392 * lower-subreg.h (target_lower_subreg): Change type to
17393 scalar_int_mode_pod.
17394 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17395 scalar_int_mode_pod.
17397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17398 Alan Hayward <alan.hayward@arm.com>
17399 David Sherwood <david.sherwood@arm.com>
17401 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17402 machine_mode to scalar_int_mode.
17403 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17404 (rs6000_option_override_internal): Remove cast to int.
17405 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17406 machine_mode to scalar_int_mode.
17407 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17408 (rs6000_option_override_internal): Remove cast to int.
17409 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17410 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17412 * config/s390/s390.c (s390_expand_builtin): Likewise.
17413 * coretypes.h (scalar_int_mode): New type.
17414 (opt_scalar_int_mode): New typedef.
17415 * machmode.h (scalar_int_mode): New class.
17416 (scalar_int_mode::includes_p): New function.
17417 (byte_mode): Change type to scalar_int_mode.
17418 (word_mode): Likewise.
17419 (ptr_mode): Likewise.
17420 * emit-rtl.c (byte_mode): Likewise.
17421 (word_mode): Likewise.
17422 (ptr_mode): Likewise.
17423 (init_derived_machine_modes): Update accordingly.
17424 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17425 and MODE_PARTIAL_INT.
17426 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17427 opt_scalar_int_mode.
17429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17430 Alan Hayward <alan.hayward@arm.com>
17431 David Sherwood <david.sherwood@arm.com>
17433 * target.def (libgcc_floating_mode_supported_p): Take a
17435 * doc/tm.texi: Regenerate.
17436 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17438 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17439 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17443 Alan Hayward <alan.hayward@arm.com>
17444 David Sherwood <david.sherwood@arm.com>
17446 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17447 * doc/tm.texi: Regenerate.
17448 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17449 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17450 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17451 * targhooks.h (default_floatn_mode): Likewise.
17452 * targhooks.c (default_floatn_mode): Likewise.
17453 * tree.c (build_common_tree_nodes): Update accordingly.
17455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17456 Alan Hayward <alan.hayward@arm.com>
17457 David Sherwood <david.sherwood@arm.com>
17459 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17460 (mode_iterator::iterate_p): Likewise.
17461 (mode_iterator::get_wider): Likewise.
17462 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17465 Alan Hayward <alan.hayward@arm.com>
17466 David Sherwood <david.sherwood@arm.com>
17468 * coretypes.h (opt_scalar_float_mode): New typedef.
17469 * machmode.h (float_mode_for_size): New function.
17470 * emit-rtl.c (double_mode): Delete.
17471 (init_emit_once): Use float_mode_for_size.
17472 * stor-layout.c (layout_type): Likewise.
17473 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17476 Alan Hayward <alan.hayward@arm.com>
17477 David Sherwood <david.sherwood@arm.com>
17479 * output.h (assemble_real): Take a scalar_float_mode.
17480 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17481 * config/arm/arm.md (consttable_4): Likewise.
17482 (consttable_8): Likewise.
17483 (consttable_16): Likewise.
17484 * config/mips/mips.md (consttable_float): Likewise.
17485 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17486 * varasm.c (assemble_real): Take a scalar_float_mode.
17487 (output_constant_pool_2): Update accordingly.
17488 (output_constant): Likewise.
17490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17491 Alan Hayward <alan.hayward@arm.com>
17492 David Sherwood <david.sherwood@arm.com>
17494 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17495 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17496 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17497 (native_encode_real): Likewise.
17498 (native_interpret_real): Likewise.
17499 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17500 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17503 Alan Hayward <alan.hayward@arm.com>
17504 David Sherwood <david.sherwood@arm.com>
17506 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17507 <scalar_float_mode>. Simplify.
17508 (gen_extend_conv_libfunc): Likewise.
17510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17511 Alan Hayward <alan.hayward@arm.com>
17512 David Sherwood <david.sherwood@arm.com>
17514 * coretypes.h (scalar_float_mode): New type.
17515 * machmode.h (mode_traits::from_int): Use machine_mode if
17516 USE_ENUM_MODES is defined.
17517 (is_a): New function.
17519 (dyn_cast): Likewise.
17520 (scalar_float_mode): New class.
17521 (scalar_float_mode::includes_p): New function.
17522 (is_float_mode): Likewise.
17523 * gdbhooks.py (MachineModePrinter): New class.
17524 (build_pretty_printer): Use it for scalar_float_mode.
17525 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17526 (format_helper::format_helper): Turn into a template.
17527 * genmodes.c (get_mode_class): New function.
17528 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17529 or machine_mode if none.
17530 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17531 as_a <scalar_float_mode>.
17532 * dwarf2out.c (mem_loc_descriptor): Likewise.
17533 (insert_float): Likewise.
17534 (add_const_value_attribute): Likewise.
17535 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17536 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17537 (expand_unop): Update accordingly.
17538 (expand_abs_nojump): Likewise.
17539 (expand_copysign_absneg): Take a scalar_float_mode.
17540 (expand_copysign_bit): Likewise.
17541 (expand_copysign): Update accordingly.
17543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17544 Alan Hayward <alan.hayward@arm.com>
17545 David Sherwood <david.sherwood@arm.com>
17547 * coretypes.h (opt_mode): New class.
17548 * machmode.h (opt_mode): Likewise.
17549 (opt_mode::else_void): New function.
17550 (opt_mode::require): Likewise.
17551 (opt_mode::exists): Likewise.
17552 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17553 (GET_MODE_2XWIDER_MODE): Likewise.
17554 (mode_iterator::get_wider): Update accordingly.
17555 (mode_iterator::get_2xwider): Likewise.
17556 (mode_iterator::get_known_wider): Likewise, turning into a template.
17557 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17558 forcing a wider mode to exist.
17559 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17560 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17561 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17562 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17563 * lower-subreg.c (init_lower_subreg): Likewise.
17564 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17565 on the final iteration.
17566 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17567 a wider mode exists before asking for a move pattern.
17568 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17569 forcing a wider mode to exist.
17570 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17571 returning false if no such mode exists.
17572 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17573 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17574 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17575 Avoid checking for a MODE_INT if we already know the mode is not a
17577 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17578 forcing a wider mode to exist.
17579 (expmed_mult_highpart_optab): Likewise.
17580 (expmed_mult_highpart): Likewise.
17581 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17583 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17584 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17585 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17586 * internal-fn.c (expand_mul_overflow): Update use of
17587 GET_MODE_2XWIDER_MODE.
17588 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17589 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17590 GET_MODE_WIDER_MODE.
17591 (convert_plusminus_to_widen): Likewise.
17592 * tree-switch-conversion.c (array_value_type): Likewise.
17593 * var-tracking.c (emit_note_insn_var_location): Likewise.
17594 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17595 Return false inside rather than outside the loop if no wider mode
17597 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17598 and GET_MODE_2XWIDER_MODE
17599 (can_compare_p): Use else_void.
17600 * gdbhooks.py (OptMachineModePrinter): New class.
17601 (build_pretty_printer): Use it for opt_mode.
17603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17604 Alan Hayward <alan.hayward@arm.com>
17605 David Sherwood <david.sherwood@arm.com>
17607 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17608 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17611 Alan Hayward <alan.hayward@arm.com>
17612 David Sherwood <david.sherwood@arm.com>
17614 * machmode.h (mode_traits): New structure.
17615 (get_narrowest_mode): New function.
17616 (mode_iterator::start): Likewise.
17617 (mode_iterator::iterate_p): Likewise.
17618 (mode_iterator::get_wider): Likewise.
17619 (mode_iterator::get_known_wider): Likewise.
17620 (mode_iterator::get_2xwider): Likewise.
17621 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17622 (FOR_EACH_MODE): Likewise.
17623 (FOR_EACH_MODE_FROM): Likewise.
17624 (FOR_EACH_MODE_UNTIL): Likewise.
17625 (FOR_EACH_WIDER_MODE): Likewise.
17626 (FOR_EACH_2XWIDER_MODE): Likewise.
17627 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17628 * combine.c (simplify_comparison): Likewise
17629 * config/i386/i386.c (type_natural_mode): Likewise.
17630 * cse.c (cse_insn): Likewise.
17631 * dse.c (find_shift_sequence): Likewise.
17632 * emit-rtl.c (init_derived_machine_modes): Likewise.
17633 (init_emit_once): Likewise.
17634 * explow.c (hard_function_value): Likewise.
17635 * expmed.c (extract_fixed_bit_field_1): Likewise.
17636 (extract_bit_field_1): Likewise.
17637 (expand_divmod): Likewise.
17638 (emit_store_flag_1): Likewise.
17639 * expr.c (init_expr_target): Likewise.
17640 (convert_move): Likewise.
17641 (alignment_for_piecewise_move): Likewise.
17642 (widest_int_mode_for_size): Likewise.
17643 (emit_block_move_via_movmem): Likewise.
17644 (copy_blkmode_to_reg): Likewise.
17645 (set_storage_via_setmem): Likewise.
17646 (compress_float_constant): Likewise.
17647 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17648 * optabs-query.c (get_best_extraction_insn): Likewise.
17649 * optabs.c (expand_binop): Likewise.
17650 (expand_twoval_unop): Likewise.
17651 (expand_twoval_binop): Likewise.
17652 (widen_leading): Likewise.
17653 (widen_bswap): Likewise.
17654 (expand_parity): Likewise.
17655 (expand_unop): Likewise.
17656 (prepare_cmp_insn): Likewise.
17657 (prepare_float_lib_cmp): Likewise.
17658 (expand_float): Likewise.
17659 (expand_fix): Likewise.
17660 (expand_sfix_optab): Likewise.
17661 * postreload.c (move2add_use_add2_insn): Likewise.
17662 * reg-stack.c (reg_to_stack): Likewise.
17663 * reginfo.c (choose_hard_reg_mode): Likewise.
17664 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17665 * stor-layout.c (mode_for_size): Likewise.
17666 (smallest_mode_for_size): Likewise.
17667 (mode_for_vector): Likewise.
17668 (finish_bitfield_representative): Likewise.
17669 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17670 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17671 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17672 * var-tracking.c (prepare_call_arguments): Likewise.
17674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17675 Alan Hayward <alan.hayward@arm.com>
17676 David Sherwood <david.sherwood@arm.com>
17678 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17679 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17680 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17681 * machmode.h (mode_size): Move earlier in file.
17682 (mode_precision): Likewise.
17683 (mode_inner): Likewise.
17684 (mode_nunits): Likewise.
17685 (mode_unit_size): Likewise.
17686 (unit_unit_precision): Likewise.
17687 (mode_wider): Likewise.
17688 (mode_2xwider): Likewise.
17689 (machine_mode): New class.
17690 (mode_to_bytes): New function.
17691 (mode_to_bits): Likewise.
17692 (mode_to_precision): Likewise.
17693 (mode_to_inner): Likewise.
17694 (mode_to_unit_size): Likewise.
17695 (mode_to_unit_precision): Likewise.
17696 (mode_to_nunits): Likewise.
17697 (GET_MODE_SIZE): Use mode_to_bytes.
17698 (GET_MODE_BITSIZE): Use mode_to_bits.
17699 (GET_MODE_PRECISION): Use mode_to_precision.
17700 (GET_MODE_INNER): Use mode_to_inner.
17701 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17702 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17703 (GET_MODE_NUNITS): Use mode_to_nunits.
17704 * system.h (ALWAYS_INLINE): New macro.
17705 * config/powerpcspe/powerpcspe-c.c
17706 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17707 int for arg1_mode and arg2_mode.
17709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17710 Alan Hayward <alan.hayward@arm.com>
17711 David Sherwood <david.sherwood@arm.com>
17713 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17714 Prefix mode names with E_ in case statements.
17715 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17716 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
17717 (aarch64_split_simd_move): Likewise.
17718 (aarch64_gen_storewb_pair): Likewise.
17719 (aarch64_gen_loadwb_pair): Likewise.
17720 (aarch64_gen_store_pair): Likewise.
17721 (aarch64_gen_load_pair): Likewise.
17722 (aarch64_get_condition_code_1): Likewise.
17723 (aarch64_constant_pool_reload_icode): Likewise.
17724 (get_rsqrte_type): Likewise.
17725 (get_rsqrts_type): Likewise.
17726 (get_recpe_type): Likewise.
17727 (get_recps_type): Likewise.
17728 (aarch64_gimplify_va_arg_expr): Likewise.
17729 (aarch64_simd_container_mode): Likewise.
17730 (aarch64_emit_load_exclusive): Likewise.
17731 (aarch64_emit_store_exclusive): Likewise.
17732 (aarch64_expand_compare_and_swap): Likewise.
17733 (aarch64_gen_atomic_cas): Likewise.
17734 (aarch64_emit_bic): Likewise.
17735 (aarch64_emit_atomic_swap): Likewise.
17736 (aarch64_emit_atomic_load_op): Likewise.
17737 (aarch64_evpc_trn): Likewise.
17738 (aarch64_evpc_uzp): Likewise.
17739 (aarch64_evpc_zip): Likewise.
17740 (aarch64_evpc_ext): Likewise.
17741 (aarch64_evpc_rev): Likewise.
17742 (aarch64_evpc_dup): Likewise.
17743 (aarch64_gen_ccmp_first): Likewise.
17744 (aarch64_gen_ccmp_next): Likewise.
17745 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17746 (alpha_emit_xfloating_libcall): Likewise.
17747 (emit_insxl): Likewise.
17748 (alpha_arg_type): Likewise.
17749 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17750 (arc_preferred_simd_mode): Likewise.
17751 (arc_secondary_reload): Likewise.
17752 (get_arc_condition_code): Likewise.
17753 (arc_print_operand): Likewise.
17754 (arc_legitimate_constant_p): Likewise.
17755 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17756 * config/arc/arc.md (casesi_load): Likewise.
17757 (casesi_compact_jump): Likewise.
17758 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17759 (cc_use_register): Likewise.
17760 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17761 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17762 (arm_init_iwmmxt_builtins): Likewise.
17763 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17764 (neon_expand_vector_init): Likewise.
17765 (arm_attr_length_move_neon): Likewise.
17766 (maybe_get_arm_condition_code): Likewise.
17767 (arm_emit_vector_const): Likewise.
17768 (arm_preferred_simd_mode): Likewise.
17769 (arm_output_iwmmxt_tinsr): Likewise.
17770 (thumb1_output_casesi): Likewise.
17771 (thumb2_output_casesi): Likewise.
17772 (arm_emit_load_exclusive): Likewise.
17773 (arm_emit_store_exclusive): Likewise.
17774 (arm_expand_compare_and_swap): Likewise.
17775 (arm_evpc_neon_vuzp): Likewise.
17776 (arm_evpc_neon_vzip): Likewise.
17777 (arm_evpc_neon_vrev): Likewise.
17778 (arm_evpc_neon_vtrn): Likewise.
17779 (arm_evpc_neon_vext): Likewise.
17780 (arm_validize_comparison): Likewise.
17781 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17782 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17783 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17784 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17785 (c6x_preferred_simd_mode): Likewise.
17786 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17787 (epiphany_rtx_costs): Likewise.
17788 * config/epiphany/predicates.md (proper_comparison_operator):
17790 * config/frv/frv.c (condexec_memory_operand): Likewise.
17791 (frv_emit_move): Likewise.
17792 (output_move_single): Likewise.
17793 (output_condmove_single): Likewise.
17794 (frv_hard_regno_mode_ok): Likewise.
17795 (frv_matching_accg_mode): Likewise.
17796 * config/h8300/h8300.c (split_adds_subs): Likewise.
17797 (h8300_rtx_costs): Likewise.
17798 (h8300_print_operand): Likewise.
17799 (compute_mov_length): Likewise.
17800 (output_logical_op): Likewise.
17801 (compute_logical_op_length): Likewise.
17802 (compute_logical_op_cc): Likewise.
17803 (h8300_shift_needs_scratch_p): Likewise.
17804 (output_a_shift): Likewise.
17805 (compute_a_shift_length): Likewise.
17806 (compute_a_shift_cc): Likewise.
17807 (expand_a_rotate): Likewise.
17808 (output_a_rotate): Likewise.
17809 * config/i386/i386.c (classify_argument): Likewise.
17810 (function_arg_advance_32): Likewise.
17811 (function_arg_32): Likewise.
17812 (function_arg_64): Likewise.
17813 (function_value_64): Likewise.
17814 (ix86_gimplify_va_arg): Likewise.
17815 (ix86_legitimate_constant_p): Likewise.
17816 (put_condition_code): Likewise.
17817 (split_double_mode): Likewise.
17818 (ix86_avx256_split_vector_move_misalign): Likewise.
17819 (ix86_expand_vector_logical_operator): Likewise.
17820 (ix86_split_idivmod): Likewise.
17821 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17822 (ix86_build_const_vector): Likewise.
17823 (ix86_build_signbit_mask): Likewise.
17824 (ix86_match_ccmode): Likewise.
17825 (ix86_cc_modes_compatible): Likewise.
17826 (ix86_expand_branch): Likewise.
17827 (ix86_expand_sse_cmp): Likewise.
17828 (ix86_expand_sse_movcc): Likewise.
17829 (ix86_expand_int_sse_cmp): Likewise.
17830 (ix86_expand_vec_perm_vpermi2): Likewise.
17831 (ix86_expand_vec_perm): Likewise.
17832 (ix86_expand_sse_unpack): Likewise.
17833 (ix86_expand_int_addcc): Likewise.
17834 (ix86_split_to_parts): Likewise.
17835 (ix86_vectorize_builtin_gather): Likewise.
17836 (ix86_vectorize_builtin_scatter): Likewise.
17837 (avx_vpermilp_parallel): Likewise.
17838 (inline_memory_move_cost): Likewise.
17839 (ix86_tieable_integer_mode_p): Likewise.
17840 (x86_maybe_negate_const_int): Likewise.
17841 (ix86_expand_vector_init_duplicate): Likewise.
17842 (ix86_expand_vector_init_one_nonzero): Likewise.
17843 (ix86_expand_vector_init_one_var): Likewise.
17844 (ix86_expand_vector_init_concat): Likewise.
17845 (ix86_expand_vector_init_interleave): Likewise.
17846 (ix86_expand_vector_init_general): Likewise.
17847 (ix86_expand_vector_set): Likewise.
17848 (ix86_expand_vector_extract): Likewise.
17849 (emit_reduc_half): Likewise.
17850 (ix86_emit_i387_round): Likewise.
17851 (ix86_mangle_type): Likewise.
17852 (ix86_expand_round_sse4): Likewise.
17853 (expand_vec_perm_blend): Likewise.
17854 (canonicalize_vector_int_perm): Likewise.
17855 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17856 (expand_vec_perm_1): Likewise.
17857 (expand_vec_perm_interleave3): Likewise.
17858 (expand_vec_perm_even_odd_pack): Likewise.
17859 (expand_vec_perm_even_odd_1): Likewise.
17860 (expand_vec_perm_broadcast_1): Likewise.
17861 (ix86_vectorize_vec_perm_const_ok): Likewise.
17862 (ix86_expand_vecop_qihi): Likewise.
17863 (ix86_expand_mul_widen_hilo): Likewise.
17864 (ix86_expand_sse2_abs): Likewise.
17865 (ix86_expand_pextr): Likewise.
17866 (ix86_expand_pinsr): Likewise.
17867 (ix86_preferred_simd_mode): Likewise.
17868 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17869 * config/i386/sse.md (*andnot<mode>3): Likewise.
17870 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17871 (*<code><mode>3): Likewise.
17872 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17873 (ia64_expand_atomic_op): Likewise.
17874 (ia64_arg_type): Likewise.
17875 (ia64_mode_to_int): Likewise.
17876 (ia64_scalar_mode_supported_p): Likewise.
17877 (ia64_vector_mode_supported_p): Likewise.
17878 (expand_vec_perm_broadcast): Likewise.
17879 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17880 (iq2000_function_arg_advance): Likewise.
17881 (iq2000_function_arg): Likewise.
17882 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17883 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17884 (m68k_libcall_value): Likewise.
17885 (m68k_function_value): Likewise.
17886 (sched_attr_op_type): Likewise.
17887 * config/mcore/mcore.c (mcore_output_move): Likewise.
17888 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17890 (microblaze_function_arg): Likewise.
17891 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17892 (mips_print_operand): Likewise.
17893 (mips_mode_ok_for_mov_fmt_p): Likewise.
17894 (mips_vector_mode_supported_p): Likewise.
17895 (mips_preferred_simd_mode): Likewise.
17896 (mips_expand_vpc_loongson_even_odd): Likewise.
17897 (mips_expand_vec_unpack): Likewise.
17898 (mips_expand_vi_broadcast): Likewise.
17899 (mips_expand_vector_init): Likewise.
17900 (mips_expand_vec_reduc): Likewise.
17901 (mips_expand_msa_cmp): Likewise.
17902 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17903 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17904 (cc_flags_for_mode): Likewise.
17905 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17906 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17907 (nds32_output_casesi_pc_relative): Likewise.
17908 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17909 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17910 (nvptx_gen_unpack): Likewise.
17911 (nvptx_gen_pack): Likewise.
17912 (nvptx_gen_shuffle): Likewise.
17913 (nvptx_gen_wcast): Likewise.
17914 (nvptx_preferred_simd_mode): Likewise.
17915 * config/pa/pa.c (pa_secondary_reload): Likewise.
17916 * config/pa/predicates.md (base14_operand): Likewise.
17917 * config/powerpcspe/powerpcspe-c.c
17918 (altivec_resolve_overloaded_builtin): Likewise.
17919 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17921 (rs6000_preferred_simd_mode): Likewise.
17922 (output_vec_const_move): Likewise.
17923 (rs6000_expand_vector_extract): Likewise.
17924 (rs6000_split_vec_extract_var): Likewise.
17925 (reg_offset_addressing_ok_p): Likewise.
17926 (rs6000_legitimate_offset_address_p): Likewise.
17927 (rs6000_legitimize_address): Likewise.
17928 (rs6000_emit_set_const): Likewise.
17929 (rs6000_const_vec): Likewise.
17930 (rs6000_emit_move): Likewise.
17931 (spe_build_register_parallel): Likewise.
17932 (rs6000_darwin64_record_arg_recurse): Likewise.
17933 (swap_selector_for_mode): Likewise.
17934 (spe_init_builtins): Likewise.
17935 (paired_init_builtins): Likewise.
17936 (altivec_init_builtins): Likewise.
17937 (do_load_for_compare): Likewise.
17938 (rs6000_generate_compare): Likewise.
17939 (rs6000_expand_float128_convert): Likewise.
17940 (emit_load_locked): Likewise.
17941 (emit_store_conditional): Likewise.
17942 (rs6000_output_function_epilogue): Likewise.
17943 (rs6000_handle_altivec_attribute): Likewise.
17944 (rs6000_function_value): Likewise.
17945 (emit_fusion_gpr_load): Likewise.
17946 (emit_fusion_p9_load): Likewise.
17947 (emit_fusion_p9_store): Likewise.
17948 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17949 (fusion_gpr_mem_load): Likewise.
17950 (fusion_addis_mem_combo_load): Likewise.
17951 (fusion_addis_mem_combo_store): Likewise.
17952 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17953 (fusion_gpr_mem_load): Likewise.
17954 (fusion_addis_mem_combo_load): Likewise.
17955 (fusion_addis_mem_combo_store): Likewise.
17956 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17958 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17959 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17960 (rs6000_preferred_simd_mode): Likewise.
17961 (output_vec_const_move): Likewise.
17962 (rs6000_expand_vector_extract): Likewise.
17963 (rs6000_split_vec_extract_var): Likewise.
17964 (reg_offset_addressing_ok_p): Likewise.
17965 (rs6000_legitimate_offset_address_p): Likewise.
17966 (rs6000_legitimize_address): Likewise.
17967 (rs6000_emit_set_const): Likewise.
17968 (rs6000_const_vec): Likewise.
17969 (rs6000_emit_move): Likewise.
17970 (rs6000_darwin64_record_arg_recurse): Likewise.
17971 (swap_selector_for_mode): Likewise.
17972 (paired_init_builtins): Likewise.
17973 (altivec_init_builtins): Likewise.
17974 (rs6000_expand_float128_convert): Likewise.
17975 (emit_load_locked): Likewise.
17976 (emit_store_conditional): Likewise.
17977 (rs6000_output_function_epilogue): Likewise.
17978 (rs6000_handle_altivec_attribute): Likewise.
17979 (rs6000_function_value): Likewise.
17980 (emit_fusion_gpr_load): Likewise.
17981 (emit_fusion_p9_load): Likewise.
17982 (emit_fusion_p9_store): Likewise.
17983 * config/rx/rx.c (rx_gen_move_template): Likewise.
17984 (flags_from_mode): Likewise.
17985 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17986 (s390_slb_comparison): Likewise.
17987 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17988 (s390_vector_mode_supported_p): Likewise.
17989 (s390_cc_modes_compatible): Likewise.
17990 (s390_match_ccmode_set): Likewise.
17991 (s390_canonicalize_comparison): Likewise.
17992 (s390_emit_compare_and_swap): Likewise.
17993 (s390_branch_condition_mask): Likewise.
17994 (s390_rtx_costs): Likewise.
17995 (s390_secondary_reload): Likewise.
17996 (__SECONDARY_RELOAD_CASE): Likewise.
17997 (s390_expand_cs): Likewise.
17998 (s390_preferred_simd_mode): Likewise.
17999 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18000 * config/sh/sh.c (sh_print_operand): Likewise.
18001 (dump_table): Likewise.
18002 (sh_secondary_reload): Likewise.
18003 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18004 * config/sh/sh.md (casesi_worker_1): Likewise.
18005 (casesi_worker_2): Likewise.
18006 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18007 (fcc_comparison_operator): Likewise.
18008 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18009 (emit_soft_tfmode_cvt): Likewise.
18010 (sparc_preferred_simd_mode): Likewise.
18011 (output_cbranch): Likewise.
18012 (sparc_print_operand): Likewise.
18013 (sparc_expand_vec_perm_bmask): Likewise.
18014 (vector_init_bshuffle): Likewise.
18015 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18016 (spu_vector_mode_supported_p): Likewise.
18017 (spu_expand_insv): Likewise.
18018 (spu_emit_branch_or_set): Likewise.
18019 (spu_handle_vector_attribute): Likewise.
18020 (spu_builtin_splats): Likewise.
18021 (spu_builtin_extract): Likewise.
18022 (spu_builtin_promote): Likewise.
18023 (spu_expand_sign_extend): Likewise.
18024 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18025 (tilegx_simd_int): Likewise.
18026 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18027 (tilepro_simd_int): Likewise.
18028 * config/v850/v850.c (const_double_split): Likewise.
18029 (v850_print_operand): Likewise.
18030 (ep_memory_offset): Likewise.
18031 * config/vax/vax.c (vax_rtx_costs): Likewise.
18032 (vax_output_int_move): Likewise.
18033 (vax_output_int_add): Likewise.
18034 (vax_output_int_subtract): Likewise.
18035 * config/visium/predicates.md (visium_branch_operator): Likewise.
18036 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18037 (visium_print_operand_address): Likewise.
18038 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18039 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18040 (xtensa_expand_conditional_branch): Likewise.
18041 (xtensa_copy_incoming_a7): Likewise.
18042 (xtensa_output_literal): Likewise.
18043 * dfp.c (decimal_real_maxval): Likewise.
18044 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18047 Alan Hayward <alan.hayward@arm.com>
18048 David Sherwood <david.sherwood@arm.com>
18050 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18051 (mode_nunits_inline): Likewise.
18052 (mode_inner_inline): Likewise.
18053 (mode_unit_size_inline): Likewise.
18054 (mode_unit_precision_inline): Likewise.
18055 (emit_insn_modes_h): Likewise. Also emit a #define of the
18057 (emit_mode_wider): Add an E_ prefix to mode names.
18058 (emit_mode_complex): Likewise.
18059 (emit_mode_inner): Likewise.
18060 (emit_mode_adjustments): Likewise.
18061 (emit_mode_int_n): Likewise.
18062 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18063 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18064 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18065 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18066 (CRC32_BUILTIN, ENTRY): Likewise.
18067 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18068 (aarch64_pop_regs): Likewise.
18069 (aarch64_process_components): Likewise.
18070 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18071 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18072 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18073 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18074 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18075 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18077 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18078 (pcmpistr): Likewise.
18079 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18080 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18081 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18083 * config/rl78/rl78.c (mduc_regs): Likewise.
18084 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18085 (htm_expand_builtin): Likewise.
18086 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18087 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18088 * config/xtensa/xtensa.c (print_operand): Likewise.
18089 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18090 (NUM_MODE_VECTOR_INT): Likewise.
18091 * genoutput.c (null_operand): Likewise.
18092 (output_operand_data): Likewise.
18093 * genrecog.c (print_parameter_value): Likewise.
18094 * lra.c (debug_operand_data): Likewise.
18096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18098 * dwarf2out.c (scompare_loc_descriptor_wide)
18099 (scompare_loc_descriptor_narrow): New functions, split out from...
18100 (scompare_loc_descriptor): ...here.
18101 * expmed.c (emit_store_flag_int): New function, split out from...
18102 (emit_store_flag): ...here.
18104 2017-08-30 Richard Biener <rguenther@suse.de>
18106 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18107 (dwarf2out_early_finish): Move setting of AT_pubnames from
18108 early debug output to early finish.
18110 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18112 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18113 and -mdata-region to the assembler.
18115 2017-08-30 Richard Biener <rguenther@suse.de>
18117 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18119 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18120 (dwarf2out_early_global_decl): Only generate a DIE for the
18121 abstract origin if it doesn't already exist or is a declaration DIE.
18122 (resolve_addr): Do not add the linkage name twice when
18123 generating a stub DIE for the DW_TAG_GNU_call_site target.
18125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18127 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18128 Use machine_mode rather than int for arg1_mode.
18130 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18133 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18134 that the second argument of the built-in functions to unpack
18135 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18136 switch statement instead a lot of if statements.
18137 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18138 Allow 64-bit values to be in Altivec registers as well as
18139 traditional floating point registers.
18140 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18142 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18144 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18145 MAX_REGS_PER_ADDRESS == 1.
18147 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18149 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18150 * config/i386/i386.c (ix86_option_override_internal): Simplify
18151 setting of opts->x_flag_entry.
18153 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18154 Jakub Jelinek <jakub@redhat.com>
18155 Richard Biener <rguenther@suse.de>
18157 PR tree-optimization/81503
18158 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18159 folded constant fits in the target type; reorder tests for clarity.
18161 2017-08-29 Martin Liska <mliska@suse.cz>
18163 * passes.def: Include pass_lower_switch.
18164 * stmt.c (dump_case_nodes): Remove and move to
18165 tree-switch-conversion.
18166 (case_values_threshold): Likewise.
18167 (expand_switch_as_decision_tree_p): Likewise.
18168 (emit_case_decision_tree): Likewise.
18169 (expand_case): Likewise.
18170 (balance_case_nodes): Likewise.
18171 (node_has_low_bound): Likewise.
18172 (node_has_high_bound): Likewise.
18173 (node_is_bounded): Likewise.
18174 (emit_case_nodes): Likewise.
18175 (struct simple_case_node): New struct.
18176 (add_case_node): Remove.
18177 (emit_case_dispatch_table): Use vector instead of case_list.
18178 (reset_out_edges_aux): Remove.
18179 (compute_cases_per_edge): Likewise.
18180 (expand_case): Build list of simple_case_node.
18181 (expand_sjlj_dispatch_table): Use it.
18182 * tree-switch-conversion.c (struct case_node): Moved from
18183 stmt.c and adjusted.
18184 (emit_case_nodes): Likewise.
18185 (node_has_low_bound): Likewise.
18186 (node_has_high_bound): Likewise.
18187 (node_is_bounded): Likewise.
18188 (case_values_threshold): Likewise.
18189 (reset_out_edges_aux): Likewise.
18190 (compute_cases_per_edge): Likewise.
18191 (add_case_node): Likewise.
18192 (dump_case_nodes): Likewise.
18193 (balance_case_nodes): Likewise.
18194 (expand_switch_as_decision_tree_p): Likewise.
18195 (emit_jump): Likewise.
18196 (emit_case_decision_tree): Likewise.
18197 (try_switch_expansion): Likewise.
18198 (do_jump_if_equal): Likewise.
18199 (emit_cmp_and_jump_insns): Likewise.
18200 (fix_phi_operands_for_edge): New function.
18201 (record_phi_operand_mapping): Likewise.
18202 (class pass_lower_switch): New pass.
18203 (pass_lower_switch::execute): New function.
18204 (make_pass_lower_switch): Likewise.
18205 (conditional_probability):
18206 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18207 * tree-pass.h: Add make_pass_lower_switch.
18209 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18212 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18215 2017-08-29 Richard Biener <rguenther@suse.de>
18217 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18218 we do not add a DW_AT_inline attribute twice.
18219 (gen_subprogram_die): Remove code setting DW_AT_inline on
18220 DECL_ABSTRACT_P nodes.
18222 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18224 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18225 calls to internal functions.
18226 (gimplify_modify_expr): Likewise.
18227 * tree-call-cdce.c (use_internal_fn): Likewise.
18228 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18229 (convert_to_divmod): Set the nothrow flag.
18230 * tree-if-conv.c (predicate_mem_writes): Likewise.
18231 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18232 (vectorizable_call): Likewise.
18233 (vectorizable_store): Likewise.
18234 (vectorizable_load): Likewise.
18235 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18236 (vect_recog_mask_conversion_pattern): Likewise.
18238 2017-08-29 Martin Liska <mliska@suse.cz>
18241 * gcc.c (driver_handle_option): Add new argument.
18242 * opts-common.c (handle_option): Pass
18243 target_option_override_hook.
18244 * opts-global.c (lang_handle_option): Add new option.
18245 (set_default_handlers): Add new argument.
18246 (decode_options): Likewise.
18247 * opts.c (target_handle_option): Likewise.
18248 (common_handle_option): Call target_option_override_hook.
18249 * opts.h (struct cl_option_handler_func): Add hook for
18250 target option override.
18251 (struct cl_option_handlers): Likewise.
18252 (set_default_handlers): Add new argument.
18253 (decode_options): Likewise.
18254 (common_handle_option): Likewise.
18255 (target_handle_option): Likewise.
18256 * toplev.c (toplev::main): Pass targetm.target_option.override
18259 2017-08-29 Richard Biener <rguenther@suse.de>
18260 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18262 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18263 life to the active subtree.
18265 2017-08-28 Jeff Law <law@redhat.com>
18267 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18268 derive_equivalences.
18269 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18270 Code moved into....
18271 (edge_info::derive_equivalences): New private member function
18273 * tree-ssa-dom.c (class edge_info): Changed from a struct
18274 to a class. Add ctor/dtor, methods and data members.
18275 (edge_info::edge_info): Renamed from allocate_edge_info.
18276 Initialize additional members.
18277 (edge_info::~edge_info): New.
18278 (free_dom_edge_info): Delete the edge info.
18279 (record_edge_info): Use new class & associated member functions.
18280 Tighten forms for testing for edge equivalences.
18281 (record_temporary_equivalences): Iterate over the simple
18282 equivalences rather than assuming there's only one per edge.
18283 (cprop_into_successor_phis): Iterate over the simple
18284 equivalences rather than assuming there's only one per edge.
18285 (optimize_stmt): Use operand_equal_p rather than pointer
18286 equality for mini-DSE code.
18288 2017-08-28 Nathan Sidwell <nathan@acm.org>
18290 * gcc.c (execute): Fold SIGPIPE handling into switch
18291 statement. Adjust internal error message.
18293 2017-08-28 Richard Biener <rguenther@suse.de>
18296 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18297 Do nothing for removed DIEs.
18299 2017-08-28 Richard Biener <rguenther@suse.de>
18301 PR tree-optimization/81977
18302 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18305 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18308 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18309 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18310 using targetm.gen_mem_thread_fence.
18312 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18315 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18316 predicate to register_operand. Reorder operands.
18317 (*btr<mode>): Ditto.
18318 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18319 (*btr<mode>_mask): Ditto.
18321 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18323 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18324 * config/rs6000/xmmintrin.h: New file.
18325 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18327 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18330 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18331 parameter and_insn and return it.
18332 (recombine_lvx_pattern): Insert a copy to ensure availability of
18333 the base register of the copied masking operation at the point of
18334 the instruction replacement.
18335 (recombine_stvx_pattern): Likewise.
18337 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18339 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18340 undocumented switches.
18341 (-mpower9-dform-vector): Likewise.
18342 (-mpower9-dform): Likewise.
18343 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18344 comments to delete references to -mpower9-dform* switches.
18345 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18346 Delete reference to -mpower9-dform* switches, test for
18347 -mpower9-vector instead.
18348 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18349 (OTHER_P9_VECTOR_MASKS): Likewise.
18350 (POWERPC_MASKS): Likewise.
18351 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18352 tests against -mpower9-dform* to -mpower9-vector. Delete code
18353 that checked for -mpower9-dform* consistancy with other options.
18354 Add test for -mpower9-misc to enable other power9 switches.
18355 (rs6000_init_hard_regno_mode_ok): Likewise.
18356 (rs6000_option_override_internal): Likewise.
18357 (rs6000_emit_prologue): Likewise.
18358 (rs6000_emit_epilogue): Likewise.
18359 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18360 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18361 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18363 (emit_fusion_p9_store): Likewise.
18364 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18365 resetting these macros if the assembler does not support ISA 3.0
18367 (TARGET_P9_DFORM_VECTOR): Likewise.
18368 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18369 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18371 2017-08-25 Alan Modra <amodra@gmail.com>
18373 PR rtl-optimization/81747
18374 * cse.c (cse_extended_basic_block): Don't attempt to record
18375 equivalences for degenerate conditional branches that branch
18376 to their fall-through.
18378 2017-08-24 Martin Sebor <msebor@redhat.com>
18380 PR middle-end/81908
18381 * gimple-fold.c (size_must_be_zero_p): New function.
18382 (gimple_fold_builtin_memory_op): Call it.
18384 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18386 * config/rs6000/mm_malloc.h: New file.
18388 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18390 PR tree-optimization/81913
18391 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18392 analysis when either IVs in condition can wrap.
18394 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18396 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18397 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18399 2017-08-24 Richard Biener <rguenther@suse.de>
18402 * targhooks.c (default_target_can_inline_p): Properly
18403 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18404 is present and always compare.
18405 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18406 infer -mfpmath=sse from TARGET_SSE_P.
18407 (ix86_can_inline_p): Properly use target_option_default_node when
18408 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18410 2017-08-24 Richard Biener <rguenther@suse.de>
18413 * dwarf2out.c (output_die): Handle flag_generate_offload like
18415 (output_comp_unit): Likewise.
18416 (gen_array_type_die): Likewise.
18417 (dwarf2out_early_finish): Likewise.
18418 (note_variable_value_in_expr): Likewise.
18419 (dwarf2out_finish): Likewise. Adjust assert.
18420 * cgraphunit.c (symbol_table::compile): Move setting of
18421 flag_generate_offload earlier ...
18422 (symbol_table::finalize_compilation_unit): ... here, before
18423 early debug finalization.
18425 2017-08-24 Richard Biener <rguenther@suse.de>
18427 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18428 and ipa-fnsummary.h.
18429 (ix86_can_inline_p): When ix86_fpmath flags do not match
18430 check whether the callee uses FP math at all.
18432 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18434 PR middle-end/81931
18435 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18436 instead of TYPE_PRECISION.
18438 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18439 Alan Hayward <alan.hayward@arm.com>
18440 David Sherwood <david.sherwood@arm.com>
18442 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18444 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18446 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18448 (ix86_frame::stack_realign_allocate): New field.
18449 (struct machine_frame_state): Modify comments.
18450 (machine_frame_state::sp_realigned_fp_end): New field.
18451 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18452 layout calculation.
18453 (sp_valid_at): Add assertion to assure no attempt to access invalid
18454 offset of a realigned stack.
18455 (fp_valid_at): Likewise.
18456 (choose_baseaddr): Modify comments.
18457 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18458 ix86_expand_prologue.
18459 (ix86_expand_prologue): Modify stack realignment and allocation.
18460 (ix86_expand_epilogue): Modify comments.
18461 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18462 avx2_runtime, avx512f, and avx512f_runtime.
18464 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18466 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18467 (mstackrealign): Do not init to -1.
18468 * config/i386/i386.c (ix86_option_override_internal):
18469 Check opts_set, not opts when setting default value of
18470 opts->x_ix86_force_align_arg_pointer.
18472 2017-08-23 Richard Biener <rguenther@suse.de>
18474 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18475 lang_hooks.decl_printable_name.
18476 * print-rtl-function.c (print_rtx_function): Likewise.
18477 * tree-pretty-print.c (dump_function_header): Likewise.
18479 2017-08-23 Richard Biener <rguenther@suse.de>
18482 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18483 -g0 at compile-time.
18485 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18487 PR middle-end/19706
18488 * doc/sourcebuild.texi (Other hardware attributes):
18491 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18493 PR middle-end/19706
18494 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18495 Fix single-use check.
18497 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18499 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18501 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18503 * doc/install.texi: Modify to add more details on running selected
18506 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18508 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18509 is combined with -mabi=ms.
18510 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18513 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18515 PR tree-optimization/81488
18516 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18517 and cached_basis fields.
18518 (MAX_SPREAD): New constant.
18519 (alloc_cand_and_find_basis): Initialize new fields.
18520 (clear_visited): New function.
18521 (create_phi_basis_1): Rename from create_phi_basis, set visited
18522 and cached_basis fields.
18523 (create_phi_basis): New wrapper function.
18524 (phi_add_costs_1): Rename from phi_add_costs, add spread
18525 parameter, set visited field, short-circuit when limits reached.
18526 (phi_add_costs): New wrapper function.
18527 (record_phi_increments_1): Rename from record_phi_increments, set
18529 (record_phi_increments): New wrapper function.
18530 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18531 (phi_incr_cost): New wrapper function.
18532 (all_phi_incrs_profitable_1): Rename from
18533 all_phi_incrs_profitable, set visited field.
18534 (all_phi_incrs_profitable): New wrapper function.
18536 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18537 Alan Hayward <alan.hayward@arm.com>
18538 David Sherwood <david.sherwood@arm.com>
18540 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18541 that takes the outer and inner modes.
18542 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18543 comparison as the canonical test for a paradoxical subreg.
18544 * combine.c (simplify_set): Use paradoxical_subreg_p.
18545 (make_extraction): Likewise.
18546 (force_to_mode): Likewise.
18547 (rtx_equal_for_field_assignment_p): Likewise.
18548 (gen_lowpart_for_combine): Likewise.
18549 (simplify_comparison): Likewise.
18550 * cse.c (equiv_constant): Likewise.
18551 * expmed.c (store_bit_field_1): Likewise.
18552 * final.c (alter_subreg): Likewise.
18553 * fwprop.c (propagate_rtx): Likewise.
18554 (forward_propagate_subreg): Likewise.
18555 * ira-conflicts.c (ira_build_conflicts): Likewise.
18556 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18557 * lra-constraints.c (curr_insn_transform): Likewise.
18558 (split_reg): Likewise.
18559 * lra-eliminations.c (move_plus_up): Likewise.
18560 (lra_eliminate_regs_1): Likewise.
18561 * recog.c (general_operand): Likewise.
18562 * ree.c (combine_reaching_defs): Likewise.
18563 * reload.c (push_reload): Likewise.
18564 (find_reloads): Likewise.
18565 * reload1.c (elimination_effects): Likewise.
18566 (compute_reload_subreg_offset): Likewise.
18567 (choose_reload_regs): Likewise.
18568 * rtlanal.c (subreg_lsb_1): Likewise.
18569 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18570 (simplify_subreg): Likewise.
18571 * var-tracking.c (track_loc_p): Likewise.
18572 * emit-rtl.c (byte_lowpart_offset): Likewise.
18573 (paradoxical_subreg_p): Delete out-of-line definition.
18575 2017-08-22 Jeff Law <law@redhat.com>
18577 PR tree-optimization/81741
18578 PR tree-optimization/71947
18579 * tree-ssa-dom.c: Include tree-inline.h.
18580 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18581 equivalences if one is more expensive to compute than the other.
18582 * tree-ssa-scopedtables.h (class const_or_copies): Make
18583 record_const_or_copy_raw method private.
18584 (class avail_exprs_stack): New method simplify_binary_operation.
18585 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18586 avail_exprs_stack::simplify_binary_operation as needed.
18587 (avail_exprs_stack::simplify_binary_operation): New function.
18589 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18591 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18592 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18593 (DOT_SYMBOLS): Likewise.
18594 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18595 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18596 (RS6000_ABI_NAME): Likewise.
18597 (TARGET_CMODEL): Likewise.
18598 (TOC_SECTION_ASM_OP): Likewise.
18599 (SET_CMODEL): New macro.
18600 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18602 2017-08-22 Richard Biener <rguenther@suse.de>
18604 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18605 to allow for free-lang-data replacements similar to verify_type_variant.
18607 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18608 Alan Hayward <alan.hayward@arm.com>
18609 David Sherwood <david.sherwood@arm.com>
18611 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18612 VOIDmode for the LABEL_REF.
18614 2017-08-22 Richard Biener <rguenther@suse.de>
18616 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18618 2017-08-22 Richard Biener <rguenther@suse.de>
18620 * common.opt (feliminate-dwarf2-dups): Ignore.
18621 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18622 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18623 same_die_p_wrap, compute_section_prefix,
18624 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18625 (comdat_symbol_id, comdat_symbol_number): Likewise.
18626 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18628 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18629 (output_die): Mark unreachable path unreachable.
18630 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18631 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18632 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18633 (dwarf2out_early_finish): Likewise.
18635 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18637 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18639 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18642 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18643 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18644 (avr_attribute_table) <io, io_low, address>: Initialize
18645 .decl_required with true.
18647 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18649 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18650 undocumented debugging options.
18651 (-mvsx-scalar-double): Likewise.
18652 (-mallow-df-permute): Likewise.
18653 (-mvectorize-builtins): Likewise.
18654 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18655 (rs6000_builtin_vectorized_function): Likewise.
18656 (rs6000_builtin_md_vectorized_function): Likewise.
18657 (rs6000_opt_vars): Likewise.
18659 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18662 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18663 (*btrq_imm): Rename from *btrq.
18664 (*btcq_imm): Rename from *btcq.
18665 (btsc): New code attribute.
18666 (*<btsc><mode>): New insn pattern.
18667 (*btr<mode>): Ditto.
18668 (*<btsc><mode>_mask): New insn_and_split pattern.
18669 (*btr<mode>_mask): Ditto.
18671 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18672 Alan Hayward <alan.hayward@arm.com>
18673 David Sherwood <david.sherwood@arm.com>
18675 * function.c (pad_below): Simplify padding calculation.
18677 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18678 Alan Hayward <alan.hayward@arm.com>
18679 David Sherwood <david.sherwood@arm.com>
18681 * target.def (function_prologue): Remove frame size argument.
18682 (function_epilogue): Likewise.
18683 * doc/tm.texi: Regenerate.
18684 * final.c (final_start_function): Update call to function_prologue.
18685 (final_end_function): Update call to function_epilogue.
18686 (default_function_pro_epilogue): Remove frame size argument.
18687 * output.h (default_function_pro_epilogue): Likewise.
18688 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18689 (arm_output_function_prologue): Likewise.
18690 * config/frv/frv.c (frv_function_prologue): Likewise.
18691 (frv_function_epilogue): Likewise.
18692 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18693 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18694 (ia64_output_function_epilogue): Likewise.
18695 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18696 (m32r_output_function_epilogue): Likewise.
18697 * config/microblaze/microblaze.c (microblaze_function_prologue)
18698 (microblaze_function_epilogue): Likewise.
18699 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18700 (mips_output_function_epilogue): Likewise.
18701 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18702 (mmix_target_asm_function_epilogue): Likewise.
18703 * config/msp430/msp430.c (msp430_start_function): Likewise.
18704 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18705 (nds32_asm_function_epilogue): Likewise.
18706 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18707 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18708 (pa_output_function_epilogue): Likewise.
18709 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18710 (rs6000_output_function_epilogue): Likewise.
18711 * config/rl78/rl78.c (rl78_start_function): Likewise.
18712 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
18713 (rs6000_output_function_epilogue): Likewise.
18714 * config/rx/rx.c (rx_output_function_prologue): Likewise.
18715 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
18716 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
18717 (sparc_asm_function_epilogue): Likewise.
18719 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18721 * tree.h (type_has_mode_precision_p): New function.
18722 * convert.c (convert_to_integer_1): Use it.
18723 * expr.c (expand_expr_real_2): Likewise.
18724 (expand_expr_real_1): Likewise.
18725 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
18726 * match.pd: Likewise.
18727 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
18728 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
18729 * tree-tailcall.c (process_assignment): Likewise.
18730 * tree-vect-loop.c (vectorizable_reduction): Likewise.
18731 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
18732 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
18733 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18734 (vectorizable_assignment): Likewise.
18735 (vectorizable_shift): Likewise.
18736 (vectorizable_operation): Likewise.
18737 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18739 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18741 * match.pd: Add pow (C, x) simplification.
18743 2017-08-21 Richard Biener <rguenther@suse.de>
18745 PR tree-optimization/81900
18746 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18747 for blocks with abnormal predecessors.
18748 (compute_antic): Do not set visited flag prematurely.
18750 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18753 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18755 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18757 * stor-layout.h (vector_type_mode): Move to...
18758 * tree.h (vector_type_mode): ...here.
18759 * stor-layout.c (vector_type_mode): Move to...
18760 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18762 2017-08-21 Richard Biener <rguenther@suse.de>
18764 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18765 register_external_die hooks.
18766 (debug_false_tree_charstarstar_uhwistar): Declare.
18767 (debug_nothing_tree_charstar_uhwi): Likewise.
18768 * debug.c (do_nothing_debug_hooks): Adjust.
18769 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18770 (debug_nothing_tree_charstar_uhwi): Likewise.
18771 * dbxout.c (dbx_debug_hooks): Adjust.
18772 (xcoff_debug_hooks): Likewise.
18773 * sdbout.c (sdb_debug_hooks): Likewise.
18774 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18775 * dwarf2out.c (macinfo_label_base): New global.
18776 (dwarf2out_register_external_die): New function for the
18777 register_external_die hook.
18778 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18779 (dwarf2_debug_hooks): Use them.
18780 (dwarf2_lineno_debug_hooks): Adjust.
18781 (struct die_struct): Add with_offset flag.
18782 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18783 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18784 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18785 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18786 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18787 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18788 defining section names for the early LTO debug variants.
18789 (reset_indirect_string): New helper.
18790 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18791 (print_dw_val): Add support for offsetted symbol references.
18792 (get_ultimate_context): Split out from is_cxx.
18793 (is_cxx): Use get_ultimate_context.
18794 (is_fortran): Add decl overload.
18795 (compute_comp_unit_symbol): Split out worker from
18796 compute_section_prefix.
18797 (compute_section_prefix): Call compute_comp_unit_symbol and
18798 set comdat_type_p here.
18799 (output_die): Skip DIE symbol output for the LTO added one.
18800 Handle DIE symbol references with offset.
18801 (output_comp_unit): Guard section name mangling properly.
18802 For LTO debug sections emit a symbol at the section beginning
18803 which we use to refer to its DIEs.
18804 (add_abstract_origin_attribute): For DIEs registered via
18805 dwarf2out_register_external_die directly refer to the early
18806 DIE rather than indirectly through the shadow one we created.
18807 Remove obsolete call to dwarf2out_abstract_function for
18808 non-function/block origins.
18809 (gen_array_type_die): When generating early LTO debug do
18810 not emit DW_AT_string_length.
18811 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18812 late when in LTO. As suggested place a gcc_unreachable for
18813 the DECL_ABSTRACT_P case.
18814 (gen_subprogram_die): Avoid another specification DIE
18815 for early built declarations/definitions for the late LTO case.
18816 (gen_variable_die): Add type references for late duplicated VLA dies
18818 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18819 we have the abstract instance already.
18820 (process_scope_var): Adjust decl DIE contexts in LTO which
18821 first puts them in limbo.
18822 (gen_decl_die): Do not generate type DIEs late apart from
18823 types for VLAs or for decls we do not yet have a DIE. Do not
18824 call dwarf2out_abstract_function late.
18825 (dwarf2out_early_global_decl): Make sure to create DIEs
18826 for abstract instances of a decl first.
18827 (dwarf2out_late_global_decl): Adjust comment.
18828 (output_macinfo_op): With multiple macro sections use
18829 macinfo_label_base to distinguish labels.
18830 (output_macinfo): Likewise. Update macinfo_label_base.
18831 Pass in the line info label.
18832 (note_variable_value_in_expr): When generating LTO resolve
18833 all variable values here by generating DIEs as needed.
18834 (init_sections_and_labels): Add early LTO debug flag parameter
18835 and generate different sections and names if set. Add generation
18836 counter for the labels so we can have multiple of them.
18837 (reset_dies): Helper to allow DIEs to be output multiple times.
18838 (dwarf2out_finish): When outputting DIEs to the fat part of an
18839 LTO object first reset DIEs.
18840 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18841 (modified_type_die): Check for decl_ultimate_origin being self
18843 (gen_type_die_with_usage): Likewise.
18844 (gen_typedef_die): Allow decl_ultimate_origin being self.
18845 (set_decl_abstract_flags): Remove.
18846 (set_block_abstract_flags): Likewise.
18847 (dwarf2out_abstract_function): Treat the early generated DIEs
18848 as the abstract copy and only add DW_AT_inline and
18849 DW_AT_artificial here and call set_decl_origin_self.
18850 If the DIE has an abstract origin don't do anything.
18851 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18852 if we have none yet (Go fails to build one, PR78628).
18853 (variably_modified_type_p): Prevent endless recursion for Ada
18854 cyclic pointer types.
18855 * lto-streamer-in.c: Include debug.h.
18856 (dref_queue): New global.
18857 (lto_read_tree_1): Stream in DIE references.
18858 (lto_input_tree): Register DIE references.
18859 (input_function): Stream DECL_DEBUG_ARGS.
18860 * lto-streamer-out.c: Include debug.h.
18861 (lto_write_tree_1): Output DIE references.
18862 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18863 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18864 (output_function): Stream DECL_DEBUG_ARGS.
18865 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18866 Stream DECL_ABSTRACT_ORIGIN.
18867 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18868 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18869 DECL_CONTEXT for file-scope decls.
18870 * lto-streamer.h (struct dref_entry): Declare.
18871 (dref_queue): Likewise.
18872 * cfgexpand.c (pass_expand::execute): Do not call the
18873 outlining_inline_function hook here.
18874 * lto-wrapper.c (debug_obj): New global.
18875 (tool_cleanup): Unlink it if required.
18876 (debug_objcopy): New function.
18877 (run_gcc): Handle early debug sections in the IL files by
18878 extracting them to separate files, partially linkin them and
18879 feeding the result back as result to the linker.
18880 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18881 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18882 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18883 sections into a separate segment.
18884 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18886 (darwin_asm_dwarf_section): Likewise.
18887 (darwin_asm_output_dwarf_offset): Likewise.
18888 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18890 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18891 Alan Hayward <alan.hayward@arm.com>
18892 David Sherwood <david.sherwood@arm.com>
18894 * read-md.h (md_reader::record_potential_iterator_use): Replace
18895 pointer argument with an rtx and an index.
18896 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18897 (apply_mode_iterator): Likewise.
18898 (apply_code_iterator): Likewise.
18899 (apply_int_iterator): Likewise.
18900 (apply_subst_iterator): Likewise.
18901 (record_iterator_use): Likewise.
18902 (record_attribute_use): Likewise.
18903 (md_reader::record_potential_iterator_use): Likewise. Update calls
18904 to record_iterator_use and apply_iterator.
18905 (iterator_use): Replace ptr with x and index.
18906 (attribute_use): Likewise.
18907 (apply_attribute_uses): Update calls to apply_iterator.
18908 (apply_iterators): Likewise. Update initialization of iterator_use.
18909 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18910 and record_potential_iterator_use.
18911 (rtx_reader::read_rtx_operand): Likewise.
18913 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18914 Alan Hayward <alan.hayward@arm.com>
18915 David Sherwood <david.sherwood@arm.com>
18917 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18920 2017-08-21 Richard Biener <rguenther@suse.de>
18922 PR middle-end/81884
18923 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18924 at struct end conservatively when comparing common bases.
18926 2017-08-21 Richard Biener <rguenther@suse.de>
18928 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18929 (mem_ref_in_stmt): Remove.
18930 (determine_max_movement): Use ref index to get at the reference.
18931 (invariantness_dom_walker::before_dom_children): Deal with
18932 lim data already initialized.
18933 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18935 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18937 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18938 (TARGET_ISA_ROUND): Ditto.
18939 (TARGET_ROUND): Ditto.
18940 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18941 * config/i386/i386.md: Ditto.
18942 * config/i386/sse.md: Ditto.
18943 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18944 with OPTION_MASK_ISA_SSE4_1.
18946 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18949 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18950 __builtin_ia32_lzcnt_u16.
18952 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18955 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18956 (rs6000_set_current_function): Rewrite function to use it.
18958 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18961 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18962 and TYPE_WARN_IF_NOT_ALIGN.
18963 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18964 (handle_warn_if_not_align): New.
18965 (place_union_field): Call handle_warn_if_not_align.
18966 (place_field): Call handle_warn_if_not_align.
18967 Copy TYPE_WARN_IF_NOT_ALIGN.
18968 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18969 (layout_type): Likewise.
18970 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18972 (tree_decl_common): Add warn_if_not_align.
18973 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18974 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18975 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18976 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18977 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18978 * doc/extend.texi: Document warn_if_not_aligned attribute.
18979 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18981 2017-08-17 Martin Liska <mliska@suse.cz>
18984 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18985 (get_data_dependence): Use it as pointer type.
18986 (distribute_loop): Likewise.
18988 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18990 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18991 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18992 (p8_vmrgow_<mode>_direct): New define_insn.
18993 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18994 handle endianness for vmrgew and vmrgow permute patterns.
18996 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18998 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18999 * config/rs6000/rs6000-cpus.def: Remove comment.
19000 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19001 (POWERPC_MASKS): Likewise.
19002 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19003 use of TARGET_VSX_TIMODE.
19004 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19005 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19006 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19007 (rs6000_option_override_internal): Remove dead code.
19008 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19009 (rs6000_legitimize_reload_address): Likewise.
19010 (rs6000_legitimate_address_p): Likewise.
19011 (rs6000_opt_masks): Delete "vsx-timode".
19012 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19013 from function comment.
19014 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19015 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19016 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19018 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19019 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19020 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19021 (VSX_TI): Likewise.
19023 (define_peephole2): Likewise.
19025 2017-08-17 Martin Sebor <msebor@redhat.com>
19028 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19029 past the end of an array.
19030 (test_pp_format): Add test cases.
19032 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19034 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19035 missing ECF_NOTHROW flags.
19037 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19040 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19041 operands residing in integer registers.
19042 (*vsx_le_perm_load_<mode>): Likewise.
19043 (*vsx_le_perm_store_<mode>): Likewise.
19044 (define_peephole2): Add peepholes to optimize the above.
19046 2017-08-17 Marek Polacek <polacek@redhat.com>
19048 PR middle-end/81814
19049 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19050 to mimic what shorten_compare did. Change the return type to bool.
19051 (fold_cond_expr_with_comparison): Update call to
19052 operand_equal_for_comparison_p.
19053 (fold_ternary_loc): Likewise.
19055 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19057 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19059 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19061 2017-08-17 Richard Biener <rguenther@suse.de>
19063 * tree-ssa-structalias.c (solve_graph): When propagating
19064 to successors update the graphs succ edges and avoid duplicate work.
19066 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19069 * config/i386/i386.c (ix86_option_override_internal): Save target
19070 specific options after ix86_stack_protector_guard_reg was changed.
19072 2017-08-17 Richard Biener <rguenther@suse.de>
19074 PR tree-optimization/81827
19075 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19076 (new_var_info): Initialize it conservatively.
19077 (get_call_vi): Mark register vars.
19078 (new_scalar_tmp_constraint_exp): Likewise.
19079 (handle_rhs_call): Likewise.
19080 (handle_const_call): Likewise.
19081 (create_function_info_for): Likewise.
19082 (solve_constraints): Sort varinfos to separate register from
19083 non-register vars to pack points-to solution bitmaps during
19086 2017-08-17 Marek Polacek <polacek@redhat.com>
19088 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19090 2017-08-17 Richard Biener <rguenther@suse.de>
19092 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19093 to true when overflow is undefined and we saturated the result.
19095 2017-08-17 Alan Modra <amodra@gmail.com>
19098 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19099 Don't use store multiple if only one reg needs saving.
19100 (interesting_frame_related_regno): New function.
19101 (rs6000_frame_related): Don't emit frame info for regs that
19103 (rs6000_emit_epilogue): Likewise.
19105 2017-08-16 Nathan Sidwell <nathan@acm.org>
19107 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19108 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19109 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19110 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19111 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19112 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19113 (verify_type): Adjust for TYPE_BINFO move.
19114 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19115 process TYPE_BINFO directly.
19116 (hash_tree): Likewise.
19117 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19119 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19122 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19124 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19126 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19129 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19130 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19131 Add ix86_binary_operator_ok to insn constraint.
19132 (*iordi_1_bts): Ditto.
19133 (*xordi_1_btc): Ditto.
19134 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19135 Update corresponding peephole2 pattern.
19139 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19141 PR tree-optimization/81832
19142 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19143 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19145 2017-08-16 Marek Polacek <polacek@redhat.com>
19148 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19149 regarding size_zero_node.
19151 2017-08-16 Martin Liska <mliska@suse.cz>
19154 * config.gcc: Respect previously set extra_objs in case
19157 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19159 PR tree-optimization/81835
19160 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19161 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19162 not depend on the phi.
19164 2017-08-16 Alan Modra <amodra@gmail.com>
19166 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19169 2017-08-16 Alan Modra <amodra@gmail.com>
19171 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19172 (save_reg_p): ..into this. Update all callers.
19173 (first_reg_to_save): Simplify.
19175 2017-08-16 Alan Modra <amodra@gmail.com>
19177 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19180 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19184 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19185 constant count if that count is less than 32.
19187 2017-08-15 Nathan Sidwell <nathan@acm.org>
19189 * gcc.c (execute): Emit friendlier message if inferior is killed
19190 by an external cause.
19192 2017-08-15 Richard Biener <rguenther@suse.de>
19194 PR tree-optimization/81790
19195 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19196 CONSTRUCTORs from simplifying and VN.
19198 2017-08-14 Martin Sebor <msebor@redhat.com>
19200 * builtin-attrs.def: Add comments.
19202 2017-08-14 Martin Sebor <msebor@redhat.com>
19205 * doc/extend.texi (attribute nonstring): Document new attribute.
19207 2017-08-14 Martin Sebor <msebor@redhat.com>
19210 * tree-diagnostic.c (default_tree_printer): Handle %G.
19211 * gimple-pretty-print.h (percent_G_format): Declare new function.
19212 * gimple-pretty-print.c (percent_G_format): Define.
19213 * tree-pretty-print.c (percent_K_format): Add argument.
19215 2017-08-14 Martin Sebor <msebor@redhat.com>
19217 PR translation/79998
19218 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19219 Remove a stray space.
19221 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19224 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19225 (*iordi_1_bts): Ditto.
19226 (*xordi_1_btc): Ditto.
19228 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19231 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19234 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19235 quoted strings, and make more translator-friendly.
19236 (darwin_rs6000_override_options): Likewise.
19237 (rs6000_option_override_internal): Likewise.
19238 (rs6000_return_in_memory): Fix overlong line.
19239 (init_cmulative_args): Use quoted strings, and make more
19240 translator-friendly.
19241 (rs6000_pass_by_reference): Fix overlong line.
19242 (def_builtin): Use quoted strings.
19243 (altivec_expand_predicate_builtin): Use quoted strings, and make
19244 more translator-friendly.
19245 (htm_expand_builtin): Use quoted strings.
19246 (cpu_expand_builtin): Use quoted strings, and make more
19247 translator-friendly.
19248 (altivec_expand_builtin): Likewise.
19249 (paired_expand_predicate_builtin): Likewise.
19250 (rs6000_invalid_builtin): Likewise.
19251 (builtin_function_type): Use quoted strings.
19252 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19253 more translator-friendly.
19254 (rs6000_trampoline_init): Likewise.
19255 (rs6000_handle_altivec_attribute): Likewise.
19256 (rs6000_inner_target_options): Use quoted strings.
19257 (rs6000_disable_incompatible_switches): Likewise.
19258 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19259 strings, and make more translator-friendly.
19260 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19262 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19264 PR tree-optimization/81799
19265 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19266 cond_expr to simple gimple operand.
19268 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19270 PR middle-end/46932
19271 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19273 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19277 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19278 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19279 TARGET_GASISR_PROLOGUES.
19280 * config/avr/avr.c (avr_option_override): Same.
19281 (avr_pass_pre_proep::execute): Same.
19283 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19286 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19287 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19289 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19291 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19292 define_insn_and_split. Split to a memory load from 0 in
19293 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19294 using PTR mode iterator.
19295 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19296 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19297 (*add_tp_<mode>): Redefine as define_insn_and_split.
19298 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19299 address space. Merge with *add_tp_x32 using PTR mode iterator.
19300 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19301 Split to an add with a memory load from 0 in
19302 DEFAULT_TLS_SEG_REG address space.
19304 2017-08-12 Andrew Pinski <apinski@cavium.com>
19306 * config/aarch64/aarch64-option-extensions.def (rdma):
19307 Fix feature string to what Linux prints out in /proc/cpuinfo.
19309 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19312 * dwarf2out.c (modified_type_die): For C typedef types that have
19313 an ultimate origin, process the ultimate origin instead of the
19315 (gen_typedef_die): Assert that input DECLs have no ultimate
19317 (gen_type_die_with_usage): For typedef variants that have an
19318 ultimate origin, just call gen_decl_die on the original DECL.
19319 (process_scope_var): Avoid creating DIEs for local typedefs and
19320 concrete static variables.
19322 2017-08-12 Alan Modra <amodra@gmail.com>
19326 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19327 match gnu-user.h startfile.
19328 (ENDFILE_LINUX_SPEC): Similarly.
19330 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19333 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19335 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19337 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19338 * config/aarch64/aarch64.md (mov<mode>): Change.
19339 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19340 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19341 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19343 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19345 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19346 for storage order barriers.
19348 2017-08-11 Martin Liska <mliska@suse.cz>
19351 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19352 variables of void type.
19354 2017-08-11 Martin Liska <mliska@suse.cz>
19356 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19357 TARGET_SUPPORTS_ALIASES.
19358 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19359 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19360 (optimize_weakref): Likewise.
19361 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19362 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19363 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19365 2017-08-11 Martin Liska <mliska@suse.cz>
19368 * config/i386/i386.c (make_resolver_func): Do complete
19369 refactoring of the function.
19371 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19374 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19375 * config/i386/i386.c (ix86_stack_protect_guard): Use
19376 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19377 * doc/invoke.texi (x86 Options): Document
19378 -mstack-protector-guard-symbol= option.
19380 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19382 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19383 * config/i386/i386.c (ix86_split_stack_guard): New function.
19384 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19385 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19386 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19387 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19388 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19389 (split_stack_space_check): Call ix86_split_stack_guard.
19391 2017-08-10 Martin Sebor <msebor@redhat.com>
19393 * print-tree.c (print_node): Print location using the established
19395 Replace spaces with colons.
19396 (debug_raw, debug): Ditto.
19398 2017-08-10 Martin Sebor <msebor@redhat.com>
19401 * pretty-print.c (pp_format): Correct the handling of %s precision.
19403 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19406 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19408 (ix86_finalize_stack_frame_flags): This. Also clear
19409 frame_pointer_needed if -fno-omit-frame-pointer is used without
19411 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19412 with ix86_finalize_stack_frame_flags.
19413 (ix86_expand_epilogue): Likewise.
19414 (ix86_expand_split_stack_prologue): Likewise.
19415 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19417 2017-08-10 Martin Liska <mliska@suse.cz>
19420 * c-attribs.c (handle_target_attribute):
19421 Report warning for an empty string argument of target attribute.
19423 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19426 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19428 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19429 or DECL_NONLOCAL labels.
19430 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19431 or DECL_NONLOCAL labels here.
19433 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19435 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19436 to indicate when early gimple folding has been disabled.
19437 (rs6000_gimple_fold_builtin): Add debug content.
19438 (rs6000_invalid_builtin): Fix whitespace.
19439 (rs6000_expand_builtin): Fix whitespace.
19440 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19442 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19445 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19446 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19448 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19450 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19451 (qdf24xx): Likewise.
19452 * config/aarch64/aarch64.md: Include falkor.md.
19453 * config/aarch64/falkor.md: New.
19455 2017-08-09 Marek Polacek <polacek@redhat.com>
19458 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19459 * diagnostic.c (emit_diagnostic): Add a comment.
19460 (emit_diagnostic_valist): New function.
19462 2017-08-09 Marek Polacek <polacek@redhat.com>
19465 * input.c (make_location): New overload.
19466 * input.h (make_location): Declare.
19468 2017-08-08 Alan Modra <amodra@gmail.com>
19469 H.J. Lu <hongjiu.lu@intel.com>
19472 * gcc.c (NO_PIE_SPEC): Delete.
19473 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19475 (LINK_PIE_SPEC): ..to here.
19476 (LINK_COMMAND_SPEC): Support -no-pie.
19477 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19478 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19480 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19481 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19482 (ENDFILE_CRTEND_SPEC): Similarly.
19484 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19487 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19488 (mstack-protector-guard-offset=): Ditto.
19489 * config/i386/i386.c (ix86_option_override): Handle
19490 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19492 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19493 ix86_stack_protect_guard_offset variables.
19494 (TARGET_STACK_PROTECT_GUARD): Always define.
19495 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19496 and -mstack-protector-guard-offset= options.
19498 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19500 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19501 boundary case for the last candidate.
19503 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19505 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19506 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19508 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19510 PR middle-end/19706
19511 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19512 * config/aarch64/aarch64-builtins.c
19513 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19514 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19515 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19517 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19518 Andrew Pinski <pinskia@gmail.com>
19520 PR middle-end/19706
19521 * internal-fn.def (XORSIGN): New.
19522 * optabs.def (xorsign_optab): New.
19523 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19524 (convert_expand_mult_copysign): New.
19525 (pass_optimize_widening_mul::execute): Call
19526 convert_expand_mult_copysign.
19528 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19530 PR tree-optimization/81354
19531 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19532 Insert on edges rather than explicitly creating landing pads.
19533 (analyze_candidates_and_replace): Commit edge inserts.
19535 2017-08-08 Richard Biener <rguenther@suse.de>
19537 PR middle-end/81719
19538 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19539 (expand_simple_operations): Also look through ADDR_EXPRs with
19540 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19542 2017-08-08 Richard Biener <rguenther@suse.de>
19544 PR tree-optimization/81723
19545 * tree-vect-slp.c (struct bst_traits): New hash traits.
19546 (bst_fail): New global.
19547 (vect_build_slp_tree_2): New worker, split out from ...
19548 (vect_build_slp_tree): ... this now wrapping it with using
19549 bst_fail set to cache SLP tree build fails. Properly handle
19551 (vect_analyze_slp_instance): Allocate and free bst_fail.
19553 2017-08-08 Martin Liska <mliska@suse.cz>
19556 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19557 LABEL_DECLs that can be from a different function.
19559 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19561 PR tree-optimization/81744
19562 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19563 loop's number of iterations.
19565 2017-08-08 Martin Liska <mliska@suse.cz>
19567 * asan.c: Include header files.
19568 * attribs.c (build_decl_attribute_variant): New function moved
19570 (build_type_attribute_qual_variant): Likewise.
19571 (cmp_attrib_identifiers): Likewise.
19572 (simple_cst_list_equal): Likewise.
19573 (omp_declare_simd_clauses_equal): Likewise.
19574 (attribute_value_equal): Likewise.
19575 (comp_type_attributes): Likewise.
19576 (build_type_attribute_variant): Likewise.
19577 (lookup_ident_attribute): Likewise.
19578 (remove_attribute): Likewise.
19579 (merge_attributes): Likewise.
19580 (merge_type_attributes): Likewise.
19581 (merge_decl_attributes): Likewise.
19582 (merge_dllimport_decl_attributes): Likewise.
19583 (handle_dll_attribute): Likewise.
19584 (attribute_list_equal): Likewise.
19585 (attribute_list_contained): Likewise.
19586 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19587 (lookup_attribute_by_prefix): Likewise.
19588 * bb-reorder.c: Include header files.
19589 * builtins.c: Likewise.
19590 * calls.c: Likewise.
19591 * cfgexpand.c: Likewise.
19592 * cgraph.c: Likewise.
19593 * cgraphunit.c: Likewise.
19594 * convert.c: Likewise.
19595 * dwarf2out.c: Likewise.
19596 * final.c: Likewise.
19597 * fold-const.c: Likewise.
19598 * function.c: Likewise.
19599 * gimple-expr.c: Likewise.
19600 * gimple-fold.c: Likewise.
19601 * gimple-pretty-print.c: Likewise.
19602 * gimple.c: Likewise.
19603 * gimplify.c: Likewise.
19604 * hsa-common.c: Likewise.
19605 * hsa-gen.c: Likewise.
19606 * internal-fn.c: Likewise.
19607 * ipa-chkp.c: Likewise.
19608 * ipa-cp.c: Likewise.
19609 * ipa-devirt.c: Likewise.
19610 * ipa-fnsummary.c: Likewise.
19611 * ipa-inline.c: Likewise.
19612 * ipa-visibility.c: Likewise.
19614 * lto-cgraph.c: Likewise.
19615 * omp-expand.c: Likewise.
19616 * omp-general.c: Likewise.
19617 * omp-low.c: Likewise.
19618 * omp-offload.c: Likewise.
19619 * omp-simd-clone.c: Likewise.
19620 * opts-global.c: Likewise.
19621 * passes.c: Likewise.
19622 * predict.c: Likewise.
19623 * sancov.c: Likewise.
19624 * sanopt.c: Likewise.
19625 * symtab.c: Likewise.
19626 * toplev.c: Likewise.
19627 * trans-mem.c: Likewise.
19628 * tree-chkp.c: Likewise.
19629 * tree-eh.c: Likewise.
19630 * tree-into-ssa.c: Likewise.
19631 * tree-object-size.c: Likewise.
19632 * tree-parloops.c: Likewise.
19633 * tree-profile.c: Likewise.
19634 * tree-ssa-ccp.c: Likewise.
19635 * tree-ssa-live.c: Likewise.
19636 * tree-ssa-loop.c: Likewise.
19637 * tree-ssa-sccvn.c: Likewise.
19638 * tree-ssa-structalias.c: Likewise.
19639 * tree-ssa.c: Likewise.
19640 * tree-streamer-in.c: Likewise.
19641 * tree-vectorizer.c: Likewise.
19642 * tree-vrp.c: Likewise.
19643 * tsan.c: Likewise.
19644 * ubsan.c: Likewise.
19645 * varasm.c: Likewise.
19646 * varpool.c: Likewise.
19647 * tree.c: Remove functions moved to attribs.[ch].
19648 * tree.h: Likewise.
19649 * config/aarch64/aarch64.c: Add attrs.h header file.
19650 * config/alpha/alpha.c: Likewise.
19651 * config/arc/arc.c: Likewise.
19652 * config/arm/arm.c: Likewise.
19653 * config/avr/avr.c: Likewise.
19654 * config/bfin/bfin.c: Likewise.
19655 * config/c6x/c6x.c: Likewise.
19656 * config/cr16/cr16.c: Likewise.
19657 * config/cris/cris.c: Likewise.
19658 * config/darwin.c: Likewise.
19659 * config/epiphany/epiphany.c: Likewise.
19660 * config/fr30/fr30.c: Likewise.
19661 * config/frv/frv.c: Likewise.
19662 * config/ft32/ft32.c: Likewise.
19663 * config/h8300/h8300.c: Likewise.
19664 * config/i386/winnt.c: Likewise.
19665 * config/ia64/ia64.c: Likewise.
19666 * config/iq2000/iq2000.c: Likewise.
19667 * config/lm32/lm32.c: Likewise.
19668 * config/m32c/m32c.c: Likewise.
19669 * config/m32r/m32r.c: Likewise.
19670 * config/m68k/m68k.c: Likewise.
19671 * config/mcore/mcore.c: Likewise.
19672 * config/microblaze/microblaze.c: Likewise.
19673 * config/mips/mips.c: Likewise.
19674 * config/mmix/mmix.c: Likewise.
19675 * config/mn10300/mn10300.c: Likewise.
19676 * config/moxie/moxie.c: Likewise.
19677 * config/msp430/msp430.c: Likewise.
19678 * config/nds32/nds32-isr.c: Likewise.
19679 * config/nds32/nds32.c: Likewise.
19680 * config/nios2/nios2.c: Likewise.
19681 * config/nvptx/nvptx.c: Likewise.
19682 * config/pa/pa.c: Likewise.
19683 * config/pdp11/pdp11.c: Likewise.
19684 * config/powerpcspe/powerpcspe.c: Likewise.
19685 * config/riscv/riscv.c: Likewise.
19686 * config/rl78/rl78.c: Likewise.
19687 * config/rx/rx.c: Likewise.
19688 * config/s390/s390.c: Likewise.
19689 * config/sh/sh.c: Likewise.
19690 * config/sol2.c: Likewise.
19691 * config/sparc/sparc.c: Likewise.
19692 * config/spu/spu.c: Likewise.
19693 * config/stormy16/stormy16.c: Likewise.
19694 * config/tilegx/tilegx.c: Likewise.
19695 * config/tilepro/tilepro.c: Likewise.
19696 * config/v850/v850.c: Likewise.
19697 * config/vax/vax.c: Likewise.
19698 * config/visium/visium.c: Likewise.
19699 * config/xtensa/xtensa.c: Likewise.
19701 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19704 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19705 constraints since the -mupper-regs-* switches have been
19707 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19708 into a vector from a double word element that was extracted from
19709 another vector, and eliminate extra XXPERMDI instructions.
19710 (vsx_concat_<mode>_2): Likewise.
19711 (vsx_concat_<mode>_3): Likewise.
19712 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
19713 concat to allow optimizing inserts from previous extracts.
19715 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
19717 * config/i386/i386.c (ix86_stack_protect_guard): Generate
19718 memory reference to a SSP offset in TLS address space.
19719 (ix86_print_operand) <case '@'>: Remove.
19720 (ix86_print_operand_punct_valid_p): Remove '@' code.
19721 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
19722 UNSPEC_SP_TLS_TEST.
19723 (stack_tls_protect_set_<mode>): Remove.
19724 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
19725 (stack_tls_protect_test_<mode>): Remove.
19726 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
19728 2017-08-07 Olivier Hainque <hainque@adacore.com>
19731 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
19733 2017-08-07 Douglas Rupp <rupp@adacore.com>
19735 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19736 variable was referenced as multidir in command.
19738 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19741 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19744 2017-08-07 Martin Liska <mliska@suse.cz>
19746 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19747 * config/rl78/rl78.c: Add include of attribs.h.
19748 * config/sh/sh.c: Likewise.
19749 * config/v850/v850.c: Likewise.
19751 2017-08-07 Tom de Vries <tom@codesourcery.com>
19753 PR middle-end/78266
19754 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19756 2017-08-07 Martin Liska <mliska@suse.cz>
19758 * config/mips/mips.c: Include attribs.h.
19760 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19763 * doc/invoke.texi: Document change in behvaior for -Ofast for
19766 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19768 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19770 (aarch64_pop_regs): Likewise.
19771 (aarch64_gen_load_pair): Likewise.
19772 (aarch64_save_callee_saves): Likewise.
19773 (aarch64_restore_callee_saves): Likewise.
19775 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19777 * config/i386/i386.c: Revert the last change.
19779 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19782 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19784 (ix86_finalize_stack_frame_flags): This. Also clear
19785 frame_pointer_needed if -fno-omit-frame-pointer is used without
19787 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19788 with ix86_finalize_stack_frame_flags.
19789 (ix86_expand_epilogue): Likewise.
19790 (ix86_expand_split_stack_prologue): Likewise.
19792 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19795 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19796 to P_AES for Westmere.
19798 2017-08-07 Jonathan Yong <10walls@gmail.com>
19800 * config/i386/mingw.opt (fset-stack-executable): Removed.
19801 * config/i386/cygming.opt (fset-stack-executable): Moved
19803 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19805 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19807 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19809 2017-08-07 Marek Polacek <polacek@redhat.com>
19811 PR middle-end/81737
19812 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19814 2017-08-07 Martin Liska <mliska@suse.cz>
19816 * attribs.h (canonicalize_attr_name): New function.
19817 (cmp_attribs): Move from c-format.c and adjusted.
19818 (is_attribute_p): Moved from tree.h.
19819 * tree-inline.c: Add new includes.
19820 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19821 (private_is_attribute_p): Remove.
19822 (private_lookup_attribute): Likewise.
19823 (private_lookup_attribute_by_prefix): Simplify.
19824 (remove_attribute): Use is_attribute_p.
19825 * tree.h: Remove removed declarations.
19827 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19829 PR middle-end/81698
19830 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19831 instead of computing it in the function. Formatting fix.
19832 (expand_case): Don't rely on default_edge being the first edge,
19833 clear it if removing it, pass default_edge to
19834 emit_case_dispatch_table.
19835 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19838 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19840 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19841 insn in the function, emit NOP after the insn.
19843 2017-08-06 Tom de Vries <tom@codesourcery.com>
19845 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19848 2017-08-06 Tom de Vries <tom@codesourcery.com>
19850 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19853 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19855 PR tree-optimization/57371
19856 * match.pd: New pattern.
19858 2017-08-04 Marek Polacek <polacek@redhat.com>
19860 PR middle-end/81695
19861 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19862 perform the computation in offset_int.
19864 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19866 PR tree-optimization/81136
19867 * tree-vectorizer.h: Include tree-hash-traits.h.
19868 (vec_base_alignments): New typedef.
19869 (vec_info): Add a base_alignments field.
19870 (vect_record_base_alignments): Declare.
19871 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19873 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19874 (create_data_ref): Add an is_conditional_in_stmt argument.
19875 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19876 the is_conditional_in_stmt field.
19877 (data_ref_loc): Add an is_conditional_in_stmt field.
19878 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19879 (find_data_references_in_stmt): Update call to create_data_ref.
19880 (graphite_find_data_references_in_stmt): Likewise.
19881 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19882 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19883 (vect_record_base_alignment): New function.
19884 (vect_record_base_alignments): Likewise.
19885 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19886 for nested statements even if we fail to compute a misalignment.
19887 Use pooled base alignments for unconditional references.
19888 (vect_find_same_alignment_drs): Compare base addresses instead
19890 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19891 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19893 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19895 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19896 Add an explicit name for the enum. Use auto_vec for slp_instances
19897 and grouped_stores.
19898 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19900 (_bb_vec_info): Add a constructor and destructor.
19901 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19902 (destroy_loop_vec_info): Delete.
19903 (vect_destroy_datarefs): Likewise.
19904 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19905 (vec_info::vec_info): New function.
19906 (vec_info::~vec_info): Likewise.
19907 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19908 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19909 destroy_loop_vec_info.
19910 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19911 (_loop_vec_info::_loop_vec_info): ...this.
19912 (destroy_loop_vec_info): Replace with...
19913 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19914 the stmt_vec_infos. Leave handling of vec_info information to its
19915 destructor. Remove explicit vector releases.
19916 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19917 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19918 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19919 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19920 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19921 (destroy_bb_vec_info): Replace with...
19922 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19923 information to its destructor.
19924 (vect_slp_analyze_bb_1): Use new and delete instead of
19925 new_bb_vec_info and destroy_bb_vec_info.
19926 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19929 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19931 * tree-data-ref.h (subscript): Add access_fn field.
19932 (data_dependence_relation): Add could_be_independent_p.
19933 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19934 (same_access_functions): Move to tree-data-ref.c.
19935 * tree-data-ref.c (ref_contains_union_access_p): New function.
19936 (access_fn_component_p): Likewise.
19937 (access_fn_components_comparable_p): Likewise.
19938 (dr_analyze_indices): Add a reference to access_fn_component_p.
19939 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19941 (constant_access_functions): Likewise.
19942 (add_other_self_distances): Likewise.
19943 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19944 (initialize_data_dependence_relation): Use XCNEW and remove
19945 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19946 of access functions that have the same type. Allow the
19947 subsequence to end with different bases in some circumstances.
19948 Record the chosen access functions in SUB_ACCESS_FN.
19949 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19950 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19951 (subscript_dependence_tester_1): Likewise dra and drb.
19952 (build_classic_dist_vector): Update calls accordingly.
19953 (subscript_dependence_tester): Likewise.
19954 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19955 DDR_COULD_BE_INDEPENDENT_P.
19956 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19957 comp_alias_ddrs instead of may_alias_ddrs.
19958 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19960 (vect_analyze_data_ref_dependence): Use it if
19961 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19962 distance vectors if that fails.
19963 (dependence_distance_ge_vf): New function.
19964 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19965 LOOP_VINFO_MAY_ALIAS_DDRS.
19967 2017-08-04 Richard Biener <rguenther@suse.de>
19969 PR middle-end/81705
19970 * fold-const.c (fold_binary_loc): Properly restrict
19971 minus_var0 && minus_var1 case when associating undefined overflow
19974 2017-08-04 Tom de Vries <tom@codesourcery.com>
19976 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19978 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19980 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19981 Don't start diagnostic messages with a capital letter.
19982 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19984 (rs6000_invalid_builtin): Likewise.
19985 (rs6000_trampoline_init): Likewise.
19987 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19990 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19991 after setting changeable df flags.
19993 2017-08-03 Richard Biener <rguenther@suse.de>
19995 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19996 up if the use is in USE - X.
19998 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20000 * toplev.c (dumpfile.h): New include.
20001 (internal_error_reentered): New static function. Use it...
20002 (internal_error_function): ...here to handle reentered internal_error.
20004 2017-08-03 Richard Biener <rguenther@suse.de>
20006 PR middle-end/81148
20007 * fold-const.c (split_tree): Add minus_var and minus_con
20008 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20009 here but always use minus_*.
20010 (associate_trees): Assert we never associate with MINUS_EXPR
20011 and NULL first operand. Do not recurse for PLUS_EXPR operands
20012 when associating as MINUS_EXPR either.
20013 (fold_binary_loc): Track minus_var and minus_con.
20015 2017-08-03 Tom de Vries <tom@codesourcery.com>
20018 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20019 ACCEL_COMPILER, apply finish_options on
20020 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20022 2017-08-03 Tom de Vries <tom@codesourcery.com>
20025 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20026 function_entry_patch_area_size > 0.
20028 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20031 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20032 instead of 10??LU, perform unit multiplication in wide_int,
20033 don't change alloc_object_size_limit if the limit is larger
20036 PR tree-optimization/81655
20037 PR tree-optimization/81588
20038 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20039 the case when ranges[i].low and high are 1 for unsigned type with
20042 PR middle-end/81052
20043 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20044 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20046 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20048 * tree-vrp.h: Add include guard.
20050 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20053 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20054 (ud2): New insn pattern.
20055 * config/i386/i386.c (ix86_expand_epilogue):
20056 For naked functions, generate ud2 instead of trap insn.
20058 2017-08-02 Marek Polacek <polacek@redhat.com>
20061 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20063 2017-08-02 Tom de Vries <tom@codesourcery.com>
20064 Cesar Philippidis <cesar@codesourcery.com>
20066 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20067 Add missing edge probabilities.
20069 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20071 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20072 Correct endianness.
20074 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20076 PR middle-end/79499
20077 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20078 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20079 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20080 prologue_seq sequences - if any.
20082 2017-08-02 Richard Biener <rguenther@suse.de>
20084 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20085 via vectors if supported, integer extracts via punning if supported
20086 or otherwise vector extracts.
20088 2017-08-02 Richard Biener <rguenther@suse.de>
20090 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20092 (bitmap_insert_into_set): ... this.
20094 2017-08-02 Richard Biener <rguenther@suse.de>
20096 PR tree-optimization/81633
20098 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20100 PR tree-optimization/71752
20101 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20103 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20105 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20106 (machine_function::call_ms2sysv_pad_out): Remove field.
20107 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20108 (ix86_compute_frame_layout): Likewise.
20110 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20113 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20114 attribute with interrupt attribute.
20116 2017-08-01 Andrew Pinski <apinski@cavium.com>
20118 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20119 BIT_INSERT_EXPR's operand 1
20120 to see if the types precision matches.
20122 2017-08-01 Martin Liska <mliska@suse.cz>
20124 PR middle-end/70140
20125 * builtins.c (expand_builtin_memcpy_args): Remove.
20126 (expand_builtin_memcpy): Call newly added function
20127 expand_builtin_memory_copy_args.
20128 (expand_builtin_memcpy_with_bounds): Likewise.
20129 (expand_builtin_mempcpy): Remove last argument.
20130 (expand_builtin_mempcpy_with_bounds): Likewise.
20131 (expand_builtin_memory_copy_args): New function created from
20132 expand_builtin_mempcpy_args with small modifications.
20133 (expand_builtin_mempcpy_args): Remove.
20134 (expand_builtin_stpcpy): Remove unused argument.
20135 (expand_builtin): Likewise.
20136 (expand_builtin_with_bounds): Likewise.
20138 2017-08-01 Martin Liska <mliska@suse.cz>
20141 Make mempcpy more optimal (PR middle-end/70140).
20143 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20146 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20147 __builtin_vec_cmpne verify both arguments are compatible vectors
20148 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20149 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20150 move computation of aligned to after checking the argument types.
20154 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20155 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20158 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20159 Doug Rupp <rupp@adacore.com>
20160 Olivier Hainque <hainque@adacore.com>
20162 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20163 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20165 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20166 for TARGET_OS_CPP_BUILTIN.
20167 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20168 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20170 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20171 passing required abi options to the assembler for EABI configurations.
20172 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20173 of .text.hot and .text.unlikely sections for kernel modules when
20174 using ARM style exceptions.
20175 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20176 options. Add EXTRA_CC1_SPEC.
20177 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20179 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20181 (ARM_TARGET2_DWARF_FORMAT): Define.
20182 * config/arm/t-vxworks: Adjust multilib control to removal of the
20183 Diab command line options.
20185 2017-08-01 Martin Liska <mliska@suse.cz>
20187 PR gcov-profile/81561
20188 * gcov.c (unblock): Make unblocking safe as we need to preserve
20189 index correspondence of blocks and block_lists.
20191 2017-08-01 Richard Biener <rguenther@suse.de>
20193 PR tree-optimization/81181
20194 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20195 (compute_antic): ... end of iteration here.
20197 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20199 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20200 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20201 (ftree-slp-vectorize): Likewise.
20202 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20203 can no longer be set independent of flag_tree_loop_vectorize.
20204 * omp-general.c (emp_max_vf): Likewise.
20205 * opts.c (enable_fdo_optimizations): Remove references to
20206 flag_tree_vectorize, these are now implicit.
20207 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20208 and leave it for the options machinery.
20210 2017-08-01 Martin Liska <mliska@suse.cz>
20212 PR middle-end/70140
20213 * builtins.c (expand_builtin_memcpy_args): Remove.
20214 (expand_builtin_memcpy): Call newly added function
20215 expand_builtin_memory_copy_args.
20216 (expand_builtin_memcpy_with_bounds): Likewise.
20217 (expand_builtin_mempcpy): Remove last argument.
20218 (expand_builtin_mempcpy_with_bounds): Likewise.
20219 (expand_builtin_memory_copy_args): New function created from
20220 expand_builtin_mempcpy_args with small modifications.
20221 (expand_builtin_mempcpy_args): Remove.
20222 (expand_builtin_stpcpy): Remove unused argument.
20223 (expand_builtin): Likewise.
20224 (expand_builtin_with_bounds): Likewise.
20226 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20229 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20230 print "ds:" only for immediates in generic address space.
20232 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20235 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20236 (ix86_function_ok_for_sibcall): Return false for naked functions.
20238 2017-08-01 Richard Biener <rguenther@suse.de>
20240 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20241 (compute_antic): Seed worklist with exit block predecessors.
20242 * cfganal.c (dfs_find_deadend): For a cycle return the source
20243 of the edge closing it.
20245 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20247 * config/aarch64/aarch64.c
20248 (aarch64_can_const_movi_rtx_p): Move 0 check.
20250 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20252 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20253 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20255 * match.pd: Ditto in address comparison pattern.
20257 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20259 PR tree-optimization/81627
20260 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20261 closed ssa form for store-store chain.
20263 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20265 PR tree-optimization/81620
20266 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20267 for store-store chain.
20269 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20271 PR tree-optimization/81588
20272 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20273 ranges[i].in_p, invert comparison code ccode. For >/>=,
20274 swap rhs1 and rhs2 and comparison code unconditionally,
20275 for </<= don't do that. Don't swap rhs1/rhs2 again if
20276 ranges[i].in_p, instead invert comparison code ccode if
20277 opcode or oe->rank is BIT_IOR_EXPR.
20280 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20281 a direct optab to conversion optab.
20282 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20283 with GET_MODE_INNER as last argument instead of optab_handler.
20284 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20285 vector extraction if possible and optab is available.
20286 * expr.c (store_constructor): Use convert_optab_handler instead
20287 of optab_handler. Use vector initialization from smaller
20288 vectors if possible and optab is available.
20289 * tree-vect-stmts.c (vectorizable_load): Likewise.
20290 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20291 now have two modes.
20292 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20293 of vec_init from half-sized vectors with the same element mode.
20294 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20295 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20296 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20297 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20298 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20299 after mode in gen_vec_extract* calls.
20300 (vec_extract<mode>): Renamed to ...
20301 (vec_extract<mode><ssescalarmodelower>): ... this.
20302 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20303 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20304 element mode after mode in gen_vec_init* calls.
20305 (VEC_INIT_HALF_MODE): New mode iterator.
20306 (vec_init<mode>): Renamed to ...
20307 (vec_init<mode><ssescalarmodelower>): ... this.
20308 (vec_init<mode><ssehalfvecmodelower>): New expander.
20309 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20310 (vec_extractv2sfsf): ... this.
20311 (vec_initv2sf): Renamed to ...
20312 (vec_initv2sfsf): ... this.
20313 (vec_extractv2si): Renamed to ...
20314 (vec_extractv2sisi): ... this.
20315 (vec_initv2si): Renamed to ...
20316 (vec_initv2sisi): ... this.
20317 (vec_extractv4hi): Renamed to ...
20318 (vec_extractv4hihi): ... this.
20319 (vec_initv4hi): Renamed to ...
20320 (vec_initv4hihi): ... this.
20321 (vec_extractv8qi): Renamed to ...
20322 (vec_extractv8qiqi): ... this.
20323 (vec_initv8qi): Renamed to ...
20324 (vec_initv8qiqi): ... this.
20325 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20326 (vec_init<mode>): Renamed to ...
20327 (vec_init<mode><VEC_base_l>): ... this.
20328 (vec_extract<mode>): Renamed to ...
20329 (vec_extract<mode><VEC_base_l>): ... this.
20330 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20331 (vec_initv2sfsf): ... this.
20332 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20333 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20334 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20335 element mode after mode in gen_vec_init* calls.
20336 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20337 (vec_init<mode><Vel>): ... this.
20338 (vec_extract<mode>): Renamed to ...
20339 (vec_extract<mode><Vel>): ... this.
20340 * config/aarch64/iterators.md (Vel): New mode attribute.
20341 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20342 Add element mode after mode in gen_vec_extract* calls.
20343 * config/s390/vector.md (non_vec_l): New mode attribute.
20344 (vec_extract<mode>): Renamed to ...
20345 (vec_extract<mode><non_vec_l>): ... this.
20346 (vec_init<mode>): Renamed to ...
20347 (vec_init<mode><non_vec_l>): ... this.
20348 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20349 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20351 * config/arm/iterators.md (V_elem_l): New mode attribute.
20352 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20353 (vec_extract<mode><V_elem_l>): ... this.
20354 (vec_extractv2di): Renamed to ...
20355 (vec_extractv2didi): ... this.
20356 (vec_init<mode>): Renamed to ...
20357 (vec_init<mode><V_elem_l>): ... this.
20358 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20359 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20360 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20361 Add element mode after gen_vec_extract* calls.
20362 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20363 (vec_init<mode><unitmode>): ... this.
20364 (vec_extract<mode>): Renamed to ...
20365 (vec_extract<mode><unitmode>): ... this.
20366 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20367 (vec_init<mode><unitmode>): ... this.
20368 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20369 (vec_initv2sfsf): ... this.
20370 (vec_extractv2sf): Renamed to ...
20371 (vec_extractv2sfsf): ... this.
20372 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20373 Add element mode after gen_vec_extract* calls.
20374 * config/mips/mips.md (unitmode): New mode iterator.
20375 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20376 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20377 * config/spu/spu.md (inner_l): New mode attribute.
20378 (vec_init<mode>): Renamed to ...
20379 (vec_init<mode><inner_l>): ... this.
20380 (vec_extract<mode>): Renamed to ...
20381 (vec_extract<mode><inner_l>): ... this.
20382 * config/sparc/sparc.md (veltmode): New mode iterator.
20383 (vec_init<VMALL:mode>): Renamed to ...
20384 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20385 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20386 (vec_initv2sisi): ... this.
20387 (vec_initv2sf): Renamed to ...
20388 (vec_initv2sfsf): ... this.
20389 (vec_extractv2sf): Renamed to ...
20390 (vec_extractv2sfsf): ... this.
20391 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20392 (vec_init<mode>): Renamed to ...
20393 (vec_init<mode><VEC_base_l>): ... this.
20394 (vec_extract<mode>): Renamed to ...
20395 (vec_extract<mode><VEC_base_l>): ... this.
20396 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20397 (vec_initv2sfsf): ... this.
20398 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20399 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20400 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20401 gen_vec_init* calls.
20403 2017-08-01 Richard Biener <rguenther@suse.de>
20405 PR tree-optimization/81297
20406 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20407 TREE_OVERFLOW from INTEGER_CSTs.
20409 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20411 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20413 2017-07-31 Carl Love <cel@us.ibm.com>
20415 * config/rs6000/rs6000-c: Add support for built-in functions
20416 vector signed char vec_xl_be (signed long long, signed char *);
20417 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20418 vector signed int vec_xl_be (signed long long, signed int *);
20419 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20420 vector signed long long vec_xl_be (signed long long, signed long long *);
20421 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20422 vector signed short vec_xl_be (signed long long, signed short *);
20423 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20424 vector double vec_xl_be (signed long long, double *);
20425 vector float vec_xl_be (signed long long, float *);
20426 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20427 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20428 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20430 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20431 (altivec_expand_builtin): Add switch statement to call
20432 altivec_expand_xl_be for each builtin.
20433 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20434 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20435 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20436 __builtin_vsx_le_be_v16qi.
20437 * doc/extend.texi: Update the built-in documentation file for the
20438 new built-in functions.
20440 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20443 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20445 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20447 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20449 * config.gcc: Add z14.
20450 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20451 CPU model numbers for z13s and z14.
20452 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20454 * config/s390/s390-opts.h (enum processor_type): Rename
20455 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20456 * config/s390/s390.c (processor_table): Add field for CPU name to
20457 be passed to Binutils.
20458 (s390_asm_output_machine_for_arch): Use the new field in
20459 processor_table for Binutils.
20460 (s390_expand_builtin): Replace arch12 with z14.
20461 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20462 (s390_get_sched_attrmask): Likewise.
20463 (s390_get_unit_mask): Likewise.
20464 * config/s390/s390.opt: Add z14 to processor_type enum.
20466 2017-07-31 Martin Jambor <mjambor@suse.cz>
20469 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20470 regardless of optimization level.
20472 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20473 Martin Liska <mliska@suse.cz>
20475 * predict.def: Remove old comment and adjust probability.
20476 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20477 PREDICT statements.
20479 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20482 * config/i386/i386.c (ix86_function_naked): New function.
20483 (ix86_can_use_return_insn_p): Return false for naked functions.
20484 (ix86_expand_prologue): Skip prologue for naked functions.
20485 (ix86_expand_epilogue): Skip epilogue for naked functions
20486 and emit trap instruction.
20487 (ix86_warn_func_return): New function.
20488 (ix86_attribute_table): Add "naked" attribute specification.
20489 (TARGET_WARN_FUNC_RETURN): Define.
20490 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20492 2017-07-31 Martin Liska <mliska@suse.cz>
20494 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20495 (dump_gimple_bb_header): Always dump BB info.
20496 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20498 2017-07-31 Martin Liska <mliska@suse.cz>
20501 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20502 also with current_function_decl non-null equality.
20504 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20507 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20508 change type to the element type, instead add eltype variable and
20509 use it where we are interested in the element type.
20511 PR tree-optimization/81603
20512 * ipa-polymorphic-call.c
20513 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20514 offset arithmetic in offset_int, bail out if the resulting bit offset
20515 doesn't fit into shwi.
20517 2017-07-31 Martin Liska <mliska@suse.cz>
20519 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20520 (gimplify_save_expr): Fix comment.
20522 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20525 * config/i386/i386.c (ix86_function_arg): Update arguments for
20527 (ix86_compute_frame_layout): Set the initial stack offset to
20528 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20529 INCOMING_FRAME_SP_OFFSET.
20530 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20531 stack before exception handler returns.
20532 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20533 the 'ERROR_CODE' for exception handler.
20535 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20537 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20538 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20539 (ASM_OUTPUT_REG_POP): Ditto.
20540 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20541 instead of asm_fprintf to output pure string.
20543 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20545 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20546 to imported_module_or_decl hook.
20547 (debug_nothing_tree_tree_tree_bool): Remove.
20548 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20549 * debug.c (do_nothing_debug_hooks): Use
20550 debug_nothing_tree_tree_tree_bool_bool instead of
20551 debug_nothing_tree_tree_tree_bool.
20552 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20553 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20554 * sdbout.c (sdb_debug_hooks): Likewise.
20555 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20556 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20558 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20559 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20560 attribute, don't add anything.
20562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20564 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20565 (fold_build2_stat_loc): Likewise.
20566 (fold_build3_stat_loc): Likewise.
20567 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20568 (fold_build1_loc): Remove macro.
20569 (fold_build2_loc): Likewise.
20570 (fold_build3_loc): Likewise.
20572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20574 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20575 (gimple_build_debug_bind_source_stat): Likewise.
20576 * gimple.h (gimple_build_debug_bind): Remove macro.
20577 (gimple_build_debug_bind_source): Likewise.
20579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20581 * bitmap.c (bitmap_alloc): Adjust.
20582 (bitmap_gc_alloc): Likewise.
20583 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20587 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20588 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20589 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20590 (bitmap_gc_alloc_stat): Likewise.
20591 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20595 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20596 * rtl.h (shallow_copy_rtx): Remove macro.
20598 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20600 * emit-rtl.c (gen_raw_REG): Adjust.
20601 * gengenrtl.c (gendef): Likewise.
20602 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20603 * rtl.h (rtx_alloc): Remove macro.
20605 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20607 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20608 (build_tree_list_stat): Likewise.
20609 * tree.h (build_tree_list): Remove macro.
20610 (build_tree_list_vec): Likewise.
20612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20614 * tree.c (make_vector_stat): Remove _stat from name.
20615 (build_vector_stat): Likewise.
20616 * tree.h (make_vector_stat): Remove macro.
20617 (build_vector_stat): Likewise.
20619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20621 * tree.h (build_var_debug_value): Remove prototype.
20623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20625 * tree.c (tree_cons_stat): Remove _stat from name.
20626 * tree.h (tree_cons): Remove macro.
20628 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20630 * tree.c (build_vl_exp_stat): Remove _stat from name.
20631 * tree.h (build_vl_exp): Remove macro.
20633 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20635 * tree.c (build_decl_stat): Remove _stat from name.
20636 * tree.h (build_decl): Remove macro.
20638 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20640 * gimple.c (gimple_build_with_ops_stat): Adjust.
20641 (gimple_alloc_stat): Remove _stat from name.
20642 * gimple.h (gimple_alloc): Remove macro.
20644 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20646 * tree.c (make_tree_vec_stat): Remove _stat from name.
20647 (grow_tree_vec_stat): Likewise.
20648 * tree.h (make_tree_vec_stat): Adjust prototype.
20649 (grow_tree_vec_stat): Likewise.
20650 (make_tree_vec): Remove macro.
20651 (grow_tree_vec): Likewise.
20653 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20655 * fold-const.c (fold_build1_stat_loc): Adjust.
20656 (fold_build2_stat_loc): Likewise.
20657 (fold_build3_stat_loc): Likewise.
20658 * tree.c (build0_stat): Remove _stat from name.
20659 (build1_stat): Likewise.
20660 (build2_stat): Likewise.
20661 (build3_stat): Likewise.
20662 (build4_stat): Likewise.
20663 (build5_stat): Likewise.
20664 * tree.h (build1_loc): Remove macro, and rename _stat function
20666 (build2_loc): Likewise.
20667 (build3_loc): Likewise.
20668 (build4_loc): Likewise.
20669 (build5_loc): Likewise.
20671 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20673 * tree.c (make_int_cst_stat): Remove _stat from name.
20674 * tree.h (make_int_cst_stat): Adjust prototype.
20675 (make_int_cst): Remove macro.
20677 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20679 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20680 * tree.h (make_tree_binfo_stat): Adjust prototype.
20681 (make_tree_binfo): Remove.
20683 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20685 * tree.c (copy_node_stat): Rename to copy_node.
20686 (build_distinct_type_copy): Adjust.
20687 * tree.h (copy_node_stat): Adjust prototype.
20688 (copy_node): Remove macro.
20690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20692 * tree.c (make_node_stat): rename to make_node.
20693 (build_tree_list_stat): Adjust.
20694 (build0_stat): Likewise.
20695 (build2_stat): Likewise.
20696 (build3_stat): Likewise.
20697 (build4_stat): Likewise.
20698 (build5_stat): Likewise.
20699 (build_decl_stat): Likewise.
20700 * tree.h (make_node_stat): Adjust prototype.
20701 (make_node): remove macro.
20703 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20705 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20706 (PPC_FEATURE2_SCV): Likewise.
20707 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20709 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20711 * config/aarch64/aarch64.c
20712 (aarch64_internal_mov_immediate): Add new special pattern.
20713 * config/aarch64/aarch64.md (*movdi_aarch64):
20714 Add reg/32bit const mov case.
20716 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20717 Richard Sandiford <richard.sandiford@linaro.org>
20719 * config/aarch64/aarch64.md (mov<mode>): Generalize.
20720 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20721 Add integer and movi cases.
20722 (movi-split-hf-df-sf split, fp16): New.
20723 (enabled): Added TARGET_FP_F16INST.
20724 * config/aarch64/iterators.md (GPF_HF): New.
20725 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
20727 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20729 * config/aarch64/aarch64.c
20730 (aarch64_simd_container_mode): Add prototype.
20731 (aarch64_expand_mov_immediate): Add HI support.
20732 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
20733 (aarch64_can_const_movi_rtx_p): New.
20734 (aarch64_preferred_reload_class):
20735 Remove restrictions of using FP registers for certain SIMD operations.
20736 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20737 (aarch64_valid_floating_const): Add integer move validation.
20738 (aarch64_simd_imm_scalar_p): Remove.
20739 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20740 (aarch64_legitimate_constant_p): Expand list of supported cases.
20741 * config/aarch64/aarch64-protos.h
20742 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20743 (aarch64_reinterpret_float_as_int): New.
20744 (aarch64_simd_imm_scalar_p): Remove.
20745 * config/aarch64/constraints.md (Uvi): New.
20746 (Dd): Split into Ds and new Dd.
20747 * config/aarch64/aarch64.md (*movsi_aarch64):
20749 (*movdi_aarch64): Add SIMD mov case.
20751 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20753 * tree-predcom.c: (struct chain): Handle store-store chain in which
20754 stores for elimination only store loop invariant values.
20755 (execute_pred_commoning_chain): Ditto.
20756 (prepare_initializers_chain_store_elim): Ditto.
20757 (prepare_finalizers): Ditto.
20758 (is_inv_store_elimination_chain): New function.
20759 (initialize_root_vars_store_elim_1): New function.
20761 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20763 * tree-predcom.c: Revise general description of the pass.
20764 (enum chain_type): New enum type for store elimination.
20765 (struct chain): New field supporting store elimination.
20766 (struct component): Ditto.
20767 (dump_chain): Dump store-stores chain.
20768 (release_chain): Release resources.
20769 (split_data_refs_to_components): Compute and create component
20770 contains only stores for elimination.
20771 (get_chain_last_ref_at): New function.
20772 (make_invariant_chain): Initialization.
20773 (make_rooted_chain): Specify chain type in parameter and record it.
20774 (add_looparound_copies): Skip for store-stores chain.
20775 (determine_roots_comp): Compute type of chain and pass it to
20777 (initialize_root_vars_store_elim_2): New function.
20778 (finalize_eliminated_stores): New function.
20779 (remove_stmt): Handle store for elimination.
20780 (execute_pred_commoning_chain): Execute predictive commoning on
20781 store-store chains.
20782 (determine_unroll_factor): Skip unroll for store-stores chain.
20783 (prepare_initializers_chain_store_elim): New function.
20784 (prepare_initializers_chain): Hanlde store-store chain.
20785 (prepare_finalizers_chain, prepare_finalizers): New function.
20786 (tree_predictive_commoning_loop): Return integer value indicating
20787 if loop is unrolled or lcssa form is corrupted.
20788 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20790 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20792 * tree-predcom.c (initialize_root): Delete.
20793 (execute_pred_commoning_chain): Initialize root vars and replace
20794 reference of non-combined chain directly, rather than call above
20797 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20799 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20800 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20802 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20804 * tree-predcom.c (struct chain): New field init_seq.
20805 (release_chain): Release init_seq.
20806 (prepare_initializers_chain): Record intialization stmts in above
20808 (insert_init_seqs): New function.
20809 (tree_predictive_commoning_loop): Call insert_init_seqs.
20811 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20813 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20815 2017-07-28 Richard Biener <rguenther@suse.de>
20817 * match.pd: Remove superfluous :c.
20818 * genmatch.c (simplify::id): Add member.
20819 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20821 (current_id): New global.
20822 (dt_node::parent): Move from ...
20823 (dt_operand::parent): ... here. Add for_id member.
20824 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20825 (decision_tree::find_node): Relax order requirement when
20826 merging DT_TRUE nodes to ones inbetween the current simplify
20827 and the one we try to merge with. Add diagnostic whenever
20828 we need to enforce pattern order by not merging.
20829 (decision_tree::insert): Set current_id.
20830 (decision_tree::print_node): Dump parent node and for_id.
20831 (parser::last_id): Add member.
20832 (parser::push_simplify): Assign unique id.
20833 (parser::parser): Initialize last_id.
20835 2017-07-28 Martin Liska <mliska@suse.cz>
20838 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20839 gimple_build_debug_bind.
20841 2017-07-28 Richard Biener <rguenther@suse.de>
20843 PR tree-optimization/81502
20844 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20846 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20848 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20849 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20850 for BIT_FIELD_REF args.
20851 * fold-const.c (make_bit_field_ref): Likewise.
20852 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20854 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20857 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20858 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20859 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20860 Or it into SANITIZER_UNDEFINED.
20861 * ubsan.c: Include gimple-fold.h and varasm.h.
20862 (ubsan_expand_ptr_ifn): New function.
20863 (instrument_pointer_overflow): New function.
20864 (maybe_instrument_pointer_overflow): New function.
20865 (instrument_object_size): Formatting fix.
20866 (pass_ubsan::execute): Call instrument_pointer_overflow
20867 and maybe_instrument_pointer_overflow.
20868 * internal-fn.c (expand_UBSAN_PTR): New function.
20869 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20870 * sanitizer.def (__ubsan_handle_pointer_overflow,
20871 __ubsan_handle_pointer_overflow_abort): New builtins.
20872 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20873 * internal-fn.def (UBSAN_PTR): New internal function.
20874 * opts.c (sanitizer_opts): Add pointer-overflow.
20875 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20876 * fold-const.c (build_range_check): Compute pointer range check in
20877 integral type if pointer arithmetics would be needed. Formatting
20880 2017-07-28 Martin Liska <mliska@suse.cz>
20883 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20884 parameters that are of a variable-length.
20886 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20888 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20890 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20891 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20892 (CRT_CALL_STATIC_FUNCTION): Likewise.
20893 (ASM_DEFAULT_SPEC): New define.
20894 (ASM_SPEC32): Likewise.
20895 (ASM_SPEC64): Likewise.
20896 (ASM_SPEC_COMMON): Likewise.
20897 (ASM_SPEC): Likewise.
20898 (INVALID_64BIT): Likewise.
20899 (LINK_OS_DEFAULT_SPEC): Likewise.
20900 (LINK_OS_SPEC32): Likewise.
20901 (LINK_OS_SPEC64): Likewise.
20902 (POWERPC_LINUX): Likewise.
20903 (PTRDIFF_TYPE): Likewise.
20904 (RESTORE_FP_PREFIX): Likewise.
20905 (RESTORE_FP_SUFFIX): Likewise.
20906 (SAVE_FP_PREFIX): Likewise.
20907 (SAVE_FP_SUFFIX): Likewise.
20908 (SIZE_TYPE): Likewise.
20909 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20910 (TARGET_64BIT): Likewise.
20911 (TARGET_64BIT): Likewise.
20912 (TARGET_AIX): Likewise.
20913 (WCHAR_TYPE_SIZE): Likewise.
20914 (WCHAR_TYPE): Undefine.
20915 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20916 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20917 (CPP_OS_RTEMS_SPEC): Delete.
20918 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20919 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20921 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20923 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20925 PR tree-optimization/81578
20926 * tree-parloops.c (build_new_reduction): Bail out if
20927 reduction_code isn't one of the standard OpenMP reductions.
20928 Move the details printing after that decision.
20930 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20932 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20933 related to reload_in_progress.
20934 (splat_input_operand): Likewise.
20935 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20937 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20939 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20940 (TARGET_INSTANTIATE_DECLS): Likewise.
20941 (legitimate_indexed_address_p): Delete reload_in_progress code.
20942 (rs6000_debug_legitimate_address_p): Likewise.
20943 (rs6000_eliminate_indexed_memrefs): Likewise.
20944 (rs6000_emit_le_vsx_store): Likewise.
20945 (rs6000_emit_move_si_sf_subreg): Likewise.
20946 (rs6000_emit_move): Likewise.
20947 (register_to_reg_type): Likewise.
20948 (rs6000_pre_atomic_barrier): Likewise.
20949 (rs6000_machopic_legitimize_pic_address): Likewise.
20950 (rs6000_allocate_stack_temp): Likewise.
20951 (rs6000_address_for_fpconvert): Likewise.
20952 (rs6000_address_for_altivec): Likewise.
20953 (rs6000_secondary_memory_needed_rtx): Delete function.
20954 (rs6000_check_sdmode): Likewise.
20955 (rs6000_alloc_sdmode_stack_slot): Likewise.
20956 (rs6000_instantiate_decls): Likewise.
20957 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20958 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20959 Delete reload_in_progress.
20960 (*vec_reload_and_plus_<mptrsize>): Likewise.
20961 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20962 (vsx_div_v2di): Likewise.
20963 (vsx_udiv_v2di): Likewise.
20965 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20967 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20968 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20969 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20970 (rs6000_debug_reg_global): Delete print of LRA status.
20971 (rs6000_option_override_internal): Delete dead LRA related code.
20972 (rs6000_lra_p): Delete function.
20973 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20975 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20977 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20978 * config/riscv/rtems.h: New file.
20980 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20981 Sudakshina Das <sudi.das@arm.com>
20983 * config/aarch64/aarch64.md
20984 (define_split for and<mode>3nr_compare): Move
20985 non aarch64_logical_operand to a register.
20986 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20987 register immediate operand to a register.
20988 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20990 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20992 PR middle-end/81564
20993 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20995 2017-07-27 Richard Biener <rguenther@suse.de>
20997 PR tree-optimization/81573
20998 PR tree-optimization/81494
20999 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21000 multi defuse cycle case.
21002 2017-07-27 Richard Biener <rguenther@suse.de>
21004 PR tree-optimization/81571
21005 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21008 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21010 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21011 earlier and only if MASK_FPU is set. Adjust formatting.
21013 2017-07-27 Martin Liska <mliska@suse.cz>
21015 * opt-functions.awk: Add validation of value of Init.
21016 * optc-gen.awk: Pass new argument.
21018 2017-07-27 Martin Liska <mliska@suse.cz>
21020 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21021 Fix wrong condition.
21023 2017-07-27 Martin Liska <mliska@suse.cz>
21025 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21026 BBs and edges seen by autoFDO.
21028 2017-07-27 Richard Biener <rguenther@suse.de>
21030 PR tree-optimization/81502
21031 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21032 with incompatible but same sized type.
21033 (execute_update_addresses_taken): Likewise.
21035 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21037 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21038 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21040 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21043 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21044 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21045 Change s_operand to memory_operand.
21047 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21049 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21050 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21051 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21052 Emit instructions rather than returning an expression. Handle TFmode
21053 and KFmode by casting to TImode.
21054 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21055 (rs6000_emit_le_vsx_store): Likewise.
21056 * config/rs6000/vsx.md (VSX_TI): New iterator.
21057 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21058 (*vsx_le_undo_permute_<mode>): Likewise.
21059 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21060 emit the split sequence.
21061 (*vsx_le_perm_store_<mode>): Likewise.
21063 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21065 PR tree-optimization/81555
21066 PR tree-optimization/81556
21067 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21068 if true, force CHANGED for the recursive invocation.
21069 (reassociate_bb): Remember original length of ops array, pass
21070 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21072 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21073 attributes for noipa attribute. For naked attribute use
21074 lookup_attribute first before lookup_attribute_spec.
21075 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21077 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21078 for functions with noipa attribute.
21079 (cgraph_externally_visible_p): Return true for functions with noipa
21081 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21082 for functions with noipa attribute.
21083 * doc/extend.texi: Document noipa function attribute.
21084 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21085 also for functions with noipa attribute.
21086 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21088 2017-07-26 Andrew Pinski <apinski@cavium.com>
21090 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21091 vec_unalign_load_cost and vec_unalign_store_cost.
21093 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21095 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21096 -mvsx-small-integer option.
21097 (ISA_3_0_MASKS_IEEE): Likewise.
21098 (OTHER_VSX_VECTOR_MASKS): Likewise.
21099 (POWERPC_MASKS): Likewise.
21100 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21101 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21102 code, only testing for DImode being allowed in non-VSX floating
21104 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21105 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21107 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21108 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21109 TARGET_P8_VECTOR test.
21110 (rs6000_secondary_reload_simple_move): Likewise.
21111 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21112 since TARGET_P9_VECTOR was already tested.
21113 (rs6000_opt_masks): Remove -mvsx-small-integer.
21114 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21115 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21117 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21118 test for TARGET_VEXTRACTUB was used, and that uses
21120 (p9 extract splitter): Likewise.
21121 (vsx_extract_<mode>_di_p9): Likewise.
21122 (vsx_extract_<mode>_store_p9): Likewise.
21123 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21124 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21125 the elimination of TARGET_VSX_SMALL_INTEGER.
21126 (vsx_extract_<mode>_p8): Likewise.
21127 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21128 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21129 (vsx_set_<mode>_p9): Likewise.
21130 (vsx_set_v4sf_p9): Likewise.
21131 (vsx_set_v4sf_p9_zero): Likewise.
21132 (vsx_insert_extract_v4sf_p9): Likewise.
21133 (vsx_insert_extract_v4sf_p9_2): Likewise.
21134 * config/rs6000/rs6000.md (sign extend splitter): Change
21135 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21136 (floatsi<mode>2_lfiwax_mem): Likewise.
21137 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21138 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21139 since a test for TARGET_P9_VECTOR was used.
21140 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21141 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21142 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21143 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21144 TARGET_P8_VECTOR test.
21145 (fix_trunc<mode>si2_stfiwx): Likewise.
21146 (fix_trunc<mode>si2_internal): Likewise.
21147 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21148 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21150 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21151 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21152 TARGET_P8_VECTOR test.
21153 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21154 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21155 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21157 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21158 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21159 since a test for TARGET_P9_VECTOR was used.
21160 (splitter for loading small constants): Likewise.
21162 2017-07-26 Andrew Pinski <apinski@cavium.com>
21164 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21167 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21170 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21171 (fp_valid_at): Likewise.
21173 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21175 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21176 (qdf24xx_addrcost_table): Likewise.
21177 (cortexa57_tunings): Update to use generic_branch_cost.
21178 (cortexa72_tunings): Likewise.
21179 (cortexa73_tunings): Likewise.
21180 (qdf24xx_tunings): Likewise.
21182 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21184 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21185 (thunderx2t99_branch_cost): Likewise.
21186 (cortexa35_tunings): Update to use generic_branch_cost.
21187 (cortexa53_tunings): Likewise.
21188 (cortexa57_tunings): Likewise.
21189 (cortexa72_tunings): Likewise.
21190 (cortexa73_tunings): Likewise.
21191 (thunderx2t99_tunings): Likewise.
21193 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21195 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21196 (sparc_option_override): Honour MASK_FSMULD.
21197 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21198 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21199 * config/sparc/sparc.opt (mfsmuld): New option.
21200 * doc/invoke.texi (mfsmuld): Document option.
21202 2017-07-26 Marek Polacek <polacek@redhat.com>
21204 PR middle-end/70992
21205 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21207 2017-07-26 Richard Biener <rguenther@suse.de>
21209 * gimple-match-head.c (do_valueize): Return OP if valueize
21211 (get_def): New helper to get at the def stmt of a SSA name
21212 if valueize allows.
21213 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21214 do_valueize to get at the def stmt.
21215 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21217 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21219 PR middle-end/46932
21220 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21222 2017-07-26 Martin Liska <mliska@suse.cz>
21225 * function.c (expand_function_start): Make expansion of
21226 nonlocal_goto_save_area after parm_birth_insn.
21228 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21230 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21231 from all CPU target flags enable members.
21233 2017-07-26 Richard Biener <rguenther@suse.de>
21235 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21236 (decision_tree::gen): Make 'type' const.
21237 (write_predicate): Likewise.
21239 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21241 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21242 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21243 (rs6000_option_override_internal): Likewise.
21244 (rs6000_expand_vector_set): Likewise.
21245 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21246 (TARGET_UPPER_REGS_SF): Likewise.
21247 (TARGET_UPPER_REGS_DI): Likewise.
21248 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21249 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21250 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21251 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21252 (Splitters for DI constants in Altivec registers): Likewise.
21253 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21254 (vsx_set_v4sf_p9): Likewise.
21255 (vsx_set_v4sf_p9_zero): Likewise.
21256 (vsx_insert_extract_v4sf_p9): Likewise.
21257 (vsx_insert_extract_v4sf_p9_2): Likewise.
21259 2017-07-25 Carl Love <cel@us.ibm.com>
21261 * doc/extend.texi: Update the built-in documentation file for the
21262 existing built-in functions
21263 vector signed char vec_cnttz (vector signed char);
21264 vector unsigned char vec_cnttz (vector unsigned char);
21265 vector signed short vec_cnttz (vector signed short);
21266 vector unsigned short vec_cnttz (vector unsigned short);
21267 vector signed int vec_cnttz (vector signed int);
21268 vector unsigned int vec_cnttz (vector unsigned int);
21269 vector signed long long vec_cnttz (vector signed long long);
21270 vector unsigned long long vec_cnttz (vector unsigned long long);
21272 2017-07-25 Andrew Pinski <apinski@cavium.com>
21274 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21275 accesses where the use is for the first operand of a BIT_INSERT.
21277 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21280 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21281 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21283 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21285 * config/i386/gstabs.h: Delete.
21286 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21288 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21290 * config/i386/i386.c (ix86_decompose_address): Do not check for
21291 register RTX when looking at index_reg or base_reg.
21292 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21294 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21296 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21297 to update EH info here.
21299 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21301 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21303 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21305 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21307 2017-07-25 Torsten Duwe <duwe@suse.de>
21309 * common.opt: Introduce -fpatchable-function-entry
21310 command line option, and its variables function_entry_patch_area_size
21311 and function_entry_patch_area_start.
21312 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21313 including a two-value parser.
21314 * target.def (print_patchable_function_entry): New target hook.
21315 * targhooks.h (default_print_patchable_function_entry): New function.
21316 * targhooks.c (default_print_patchable_function_entry): Likewise.
21317 * toplev.c (process_options): Switch off IPA-RA if
21318 patchable function entries are being generated.
21319 * varasm.c (assemble_start_function): Look at the
21320 patchable-function-entry command line switch and current
21321 function attributes and maybe generate NOP instructions by
21322 calling the print_patchable_function_entry hook.
21323 * doc/extend.texi: Document patchable_function_entry attribute.
21324 * doc/invoke.texi: Document -fpatchable_function_entry
21325 command line option.
21326 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21328 * doc/tm.texi: Re-generate.
21330 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21333 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21334 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21336 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21338 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21340 2017-07-25 Richard Biener <rguenther@suse.de>
21342 PR tree-optimization/81455
21343 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21344 not walk in cycles when looking for guards.
21346 2017-07-25 Richard Biener <rguenther@suse.de>
21348 PR tree-optimization/81529
21349 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21350 when optimizing backedge uses.
21352 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21354 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21356 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21357 to dl_section_ref. On AIX, append an expression to subtract
21358 the size of the section length to dl_section_ref.
21360 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21362 * configure.ac: If any of the config.* scripts fail, exit 1.
21363 * configure: Regenerate.
21365 2017-07-25 Richard Biener <rguenther@suse.de>
21367 PR middle-end/81546
21368 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21369 of immediate uses, be more verbose on errors.
21371 2017-07-25 Richard Biener <rguenther@suse.de>
21373 PR tree-optimization/81510
21374 * tree-vect-loop.c (vect_is_simple_reduction): When the
21375 reduction stmt is not inside the loop bail out.
21377 2017-07-25 Richard Biener <rguenther@suse.de>
21379 PR tree-optimization/81303
21380 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21381 profitability check against LOOP_VINFO_NITERSM1.
21383 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21385 * domwalk.c (cmp_bb_postorder): Simplify.
21386 (sort_bbs_postorder): New function. Use it...
21387 (dom_walker::walk): ...here to optimize common cases.
21389 2017-07-25 Martin Liska <mliska@suse.cz>
21392 * ipa-visibility.c (function_and_variable_visibility): Make the
21393 redirection just on target that supports aliasing.
21394 Fix GNU coding style.
21396 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21399 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21401 * config/arm/rtems.h: Likewise.
21402 * config/bfin/rtems.h: Likewise.
21403 * config/i386/rtemself.h: Likewise.
21404 * config/lm32/rtems.h: Likewise.
21405 * config/m32c/rtems.h: Likewise.
21406 * config/m68k/rtemself.h: Likewise.
21407 * config/microblaze/rtems.h: Likewise.
21408 * config/mips/rtems.h: Likewise.
21409 * config/moxie/rtems.h: Likewise.
21410 * config/nios2/rtems.h: Likewise.
21411 * config/powerpcspe/rtems.h: Likewise.
21412 * config/rs6000/rtems.h: Likewise.
21413 * config/rtems.h: Likewise.
21414 * config/sh/rtems.h: Likewise.
21415 * config/sh/rtemself.h: Likewise.
21416 * config/sparc/rtemself.h: Likewise.
21418 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21421 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21422 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21423 * tree-ssa-structalias.c (alias_get_name): Same.
21425 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21428 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21429 instructions if no du chain is found.
21431 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21433 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21435 2017-07-25 Richard Biener <rguenther@suse.de>
21437 PR middle-end/81505
21438 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21441 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21443 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21444 upper-regs options.
21445 (ISA_2_7_MASKS_SERVER): Likewise.
21446 (ISA_3_0_MASKS_IEEE): Likewise.
21447 (OTHER_P8_VECTOR_MASKS): Likewise.
21448 (OTHER_VSX_VECTOR_MASKS): Likewise.
21449 (POWERPC_MASKS): Likewise.
21450 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21451 duplicate list of options.
21452 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21453 explicit -mupper-regs options.
21454 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21455 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21456 alias for -mupper-regs-df.
21457 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21458 (rs6000_init_hard_regno_mode_ok): Likewise.
21459 (rs6000_option_override_internal): Likewise.
21460 (rs6000_opt_masks): Likewise.
21461 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21462 options in terms of whether -mvsx or -mpower8-vector was used.
21463 (TARGET_UPPER_REGS_DI): Likewise.
21464 (TARGET_UPPER_REGS_SF): Likewise.
21465 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21466 -mupper-regs-* options.
21468 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21470 * passes.c (emergency_dump_function): Print some empty lines and a
21471 header before the RTL dump.
21473 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21475 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21477 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21480 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21481 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21483 2017-07-24 Carl Love <cel@us.ibm.com>
21485 * config/rs6000/rs6000-c.c: Add support for built-in functions
21486 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21487 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21488 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21489 vec_extract_fp_from_shortl): Add defines for the two builtins.
21490 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21491 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21493 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21494 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21495 * doc/extend.texi: Update the built-in documentation file for the
21496 new built-in function.
21498 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21501 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21503 * doc/generic.texi: Likewise.
21504 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21505 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21507 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21509 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21510 (aarch64_mls_elt_merge<mode>): Likewise.
21512 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21514 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21515 having __cxa_atexit.
21517 2017-07-23 Michael Collison <michael.collison@arm.com>
21519 * config/arm/arm.c (arm_option_override): Deprecate
21520 use of -mstructure-size-boundary.
21521 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21522 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21524 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21527 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21528 Reduce cost estimate for direct moves.
21530 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21533 * config/i386/i386.c (ix86_option_override_internal): Disable
21534 BMI, BMI2 and TBM instructions for -m16.
21536 2017-07-21 Carl Love <cel@us.ibm.com>
21538 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21539 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21540 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21541 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21542 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21543 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21544 VMULOSW): New enum "unspec" values.
21545 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21546 altivec_vmulosw): New patterns.
21547 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21548 VMULOSW): Add definitions.
21550 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21552 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21553 (qdf24xx): Likewise.
21554 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21555 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21556 (AARCH64_FL_V8_1): Renumber.
21557 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21558 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21559 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21560 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21561 rdma to feature modifiers list.
21563 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21565 PR middle-end/56727
21566 * ipa-visibility (function_and_variable_visibility): Convert
21567 recursive PLT call to direct call if appropriate.
21569 2017-07-21 Andrew Pinski <apinski@cavium.com>
21571 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21572 operand 1 to see if the types precision matches.
21573 * fold-const.c (operand_equal_p): Likewise.
21575 2017-07-21 Richard Biener <rguenther@suse.de>
21577 PR tree-optimization/81303
21578 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21579 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21580 (vect_peeling_hash_get_lowest_cost): Adjust.
21581 (vect_enhance_data_refs_alignment): Likewise. Use
21582 vect_get_peeling_costs_all_drs to compute the penalty for no
21583 peeling to match up costs.
21585 2017-07-21 Richard Biener <rguenther@suse.de>
21587 PR tree-optimization/81500
21588 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21589 we didn't identify a reduction path.
21591 2017-07-21 Tom de Vries <tom@codesourcery.com>
21592 Cesar Philippidis <cesar@codesourcery.com>
21594 PR gcov-profile/81442
21595 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21598 2017-07-21 Tom de Vries <tom@codesourcery.com>
21601 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21603 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21604 nvptx_override_options_after_change.
21606 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21608 * dwarf2out.c (output_file_names): Avoid double testing for
21609 dwarf_version >= 5.
21611 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21613 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21615 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21617 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21619 (try_crossjump_to_edge): Do not punt on partitioned functions.
21621 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21623 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21624 Put all BBs reachable only via paths crossing cold region to cold
21626 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21628 2016-07-21 Richard Biener <rguenther@suse.de>
21630 PR tree-optimization/81303
21631 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21632 into account prologue and epilogue iterations when raising
21633 min_profitable_iters to sth at least covering one vector iteration.
21635 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21637 * config/arm/arm.c (arm_test_cpu_arch_dat):
21640 2017-07-20 Nathan Sidwell <nathan@acm.org>
21642 Remove TYPE_METHODS.
21643 * tree.h (TYPE_METHODS): Delete.
21644 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21645 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21646 (dbxout_type_methods): Scan TYPE_FIELDS.
21647 (dbxout_type): Don't check TYPE_METHODS here.
21648 * function.c (use_register_for_decl): Always ignore register for
21649 class types when not optimizing.
21650 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21651 * tree.c (free_lang_data_in_type): Stitch out member functions and
21652 templates from TYPE_FIELDS.
21653 (build_distinct_type_copy, verify_type_variant,
21654 verify_type): Member fns are on TYPE_FIELDS.
21655 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21656 * tree-pretty-print.c (dump_generic_node): Likewise.
21658 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21661 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21662 V2TImode and V4TImode.
21663 (ix86_expand_vector_extract): Likewise.
21664 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21665 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21666 (ssescalarmode): Handle V4TImode and V2TImode.
21667 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21668 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21669 (VEXTRACTI128_MODE): New mode iterator.
21670 (splitter for *vec_extractv?ti first element): New.
21671 (VEC_INIT_MODE): New mode iterator.
21672 (vec_init<mode>): Consolidate 3 expanders into one using
21673 VEC_INIT_MODE mode iterator.
21675 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21677 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21678 non_spilled_static_chain_regno_p.
21680 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21682 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21684 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21686 * bb-reorder.c (connect_traces): Allow copying of blocks within
21689 2017-07-20 Richard Biener <rguenther@suse.de>
21691 * gimple.h (gimple_phi_result): Add gphi * overload.
21692 (gimple_phi_result_ptr): Likewise.
21693 (gimple_phi_arg): Likewise. Adjust index assert to only
21694 allow actual argument accesses rather than all slots available
21696 (gimple_phi_arg_def): Add gphi * overload.
21697 * tree-phinodes.c (make_phi_node): Initialize only actual
21699 (resize_phi_node): Clear memory not covered by old node,
21700 do not initialize excess argument slots.
21701 (reserve_phi_args_for_new_edge): Initialize new argument slot
21704 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21706 PR tree-optimization/81388
21708 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21710 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21711 by removing computation of may_be_zero.
21713 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21714 Tom de Vries <tom@codesourcery.com>
21716 PR middle-end/81030
21717 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
21718 when gimple level profile disagrees with what RTL expander did.
21720 2017-07-20 Richard Biener <rguenther@suse.de>
21722 PR tree-optimization/61171
21723 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
21724 (vect_analyze_stmt): Add slp instance parameter.
21725 (vectorizable_reduction): Likewise.
21726 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
21727 (vect_is_simple_reduction): Deal with chains not detected
21728 as SLP reduction chain, specifically not properly associated
21729 chains containing a mix of plus/minus.
21730 (get_reduction_op): Remove.
21731 (get_initial_defs_for_reduction): Simplify, pass in whether
21732 this is a reduction chain, pass in the SLP node for the PHIs.
21733 (vect_create_epilog_for_reduction): Get the SLP instance as
21735 (vectorizable_reduction): Get the SLP instance as arg.
21736 During analysis remember the SLP node with the PHIs in the
21737 instance. Simplify getting at the vectorized reduction PHIs.
21738 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21739 through SLP instance.
21740 (vect_slp_analyze_operations): Likewise.
21741 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21742 (vect_transform_stmt): Likewise.
21744 2017-07-20 Tom de Vries <tom@codesourcery.com>
21746 PR tree-optimization/81489
21747 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21748 read of phi arg location to before loop that modifies phi.
21750 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21752 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21755 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21757 PR middle-end/81331
21758 * except.c (execute): Fix ordering issue.
21760 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21762 PR rtl-optimization/81423
21763 * combine.c (make_compound_operation_int): Don't try to optimize
21764 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21766 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21768 PR rtl-optimization/81423
21769 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21770 with a constant that is -1 in the truncated to mode.
21772 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21774 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21775 (determine_unlikely_bbs): ... here.
21776 * predict.h (propagate_unlikely_bbs_forward): Declare.
21777 * cfgexpand.c (pass_expand::execute): Use it.
21778 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21780 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21781 propagate_unlikely_bbs_forward.
21783 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21785 PR middle-end/81331
21786 * except.c (maybe_add_nop_after_section_switch): New function.
21789 2017-07-19 Tom de Vries <tom@codesourcery.com>
21791 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21793 2017-07-19 Tom de Vries <tom@codesourcery.com>
21795 * gimple.h (gimple_phi_arg): Make assert more strict.
21797 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21799 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21800 * config/rs6000/mmintrin.h: New file.
21801 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21803 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21805 PR tree-optimization/81346
21806 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21808 2017-07-19 Tom de Vries <tom@codesourcery.com>
21810 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21812 2017-07-19 Tom de Vries <tom@codesourcery.com>
21814 * config/nvptx/nvptx-modes.def: Add V2DImode.
21815 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21816 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21817 (nvptx_output_mov_insn): Handle lack of mov.b128.
21818 (nvptx_print_operand): Handle 'H' and 'L' codes.
21819 (nvptx_vector_mode_supported): Allow V2DImode.
21820 (nvptx_preferred_simd_mode): New function.
21821 (nvptx_data_alignment): New function.
21822 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21823 nvptx_preferred_simd_mode.
21824 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21826 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21828 2017-07-19 Tom de Vries <tom@codesourcery.com>
21830 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21831 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21832 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21833 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21834 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21835 (mov<VECIM>_insn): New define_insn.
21836 (define_expand "mov<VECIM>): New define_expand.
21838 2017-07-19 Tom de Vries <tom@codesourcery.com>
21840 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21842 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21844 PR tree-optimization/81346
21845 * fold-const.h (fold_div_compare, range_check_type): Declare.
21846 * fold-const.c (range_check_type): New function.
21847 (build_range_check): Use range_check_type.
21848 (fold_div_compare): No longer static, rewritten into
21849 a match.pd helper function.
21850 (fold_comparison): Don't call fold_div_compare here.
21851 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21852 as helper function.
21854 2017-07-19 Nathan Sidwell <nathan@acm.org>
21856 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21857 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21858 * tree.c (find_decls_types_r, verify_type): Use
21859 TYPE_{MIN,MAX}_VALUE_RAW.
21860 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21861 (hash_tree): Likewise.
21862 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21864 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21867 2017-07-18 Tom de Vries <tom@codesourcery.com>
21869 PR middle-end/81464
21870 * omp-expand.c (expand_omp_for_static_chunk): Handle
21871 equal-argument loop exit phi.
21873 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21876 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21877 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21878 operand 2 predicate.
21879 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21880 operand 2 predicate.
21881 (ror,rol -> rorx splitters): Use const_int_operand as
21882 operand 2 predicate.
21884 2017-06-18 Richard Biener <rguenther@suse.de>
21886 PR tree-optimization/81410
21887 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21888 the gap in the ! slp_perm SLP case after each group.
21890 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21892 PR middle-end/81463
21893 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21896 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21898 PR middle-end/81462
21899 * predict.c (set_even_probabilities): Cleanup; do not affect
21900 probabilities that are already known.
21901 (combine_predictions_for_bb): Call even when count is set.
21903 2017-07-18 Nathan Sidwell <nathan@acm.org>
21905 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21908 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21911 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21912 optimization for loop niter analysis.
21914 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21917 * config/avr/avr.c (avr_optimize_casesi): Don't use
21918 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21920 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21922 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21923 body_cost_vec from _vect_peel_extended_info.
21924 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21925 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21928 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21930 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21932 2017-07-18 Richard Biener <rguenther@suse.de>
21934 PR tree-optimization/80620
21935 PR tree-optimization/81403
21936 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21937 info when re-using a VN table entry.
21939 2017-07-18 Richard Biener <rguenther@suse.de>
21941 PR tree-optimization/81418
21942 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21943 vectype_in. Verify that with lane-reducing reduction operations
21944 we have a single def-use cycle.
21946 2017-07-17 Carl Love <cel@us.ibm.com>
21948 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21950 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21951 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21952 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21953 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21954 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21955 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21956 VMULOSW): New enum "unspec" values.
21957 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21958 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21959 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21960 altivec_vmulosw): New patterns.
21961 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21962 VMULOSW): Add definitions.
21964 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21966 * config/alpha/alpha.c: Include predict.h.
21968 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21970 * tree-vrp.c (compare_assert_loc): Fix comparison function
21971 to return predictable results.
21973 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21975 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21977 (subdi3): Likewise.
21978 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21979 * doc/invoke.texi (mexpand-adddi): Update text.
21981 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21983 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21984 that also clobbers the CC register. The old expand code is moved
21986 (*arc_clzsi2): ... here.
21987 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21988 the CC register. The old expand code is moved to ...
21989 (arc_ctzsi2): ... here.
21991 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21993 * config/arc/arc.opt (mindexed-loads): Use initial value
21994 TARGET_INDEXED_LOADS_DEFAULT.
21995 (mauto-modify-reg): Use initial value
21996 TARGET_AUTO_MODIFY_REG_DEFAULT.
21997 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21998 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21999 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22000 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22002 2017-07-17 Martin Liska <mliska@suse.cz>
22005 * opts.c (finish_options): Do not allow -fgnu-tm
22006 w/ -fsanitize={kernel-,}address. Say sorry.
22008 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22011 * tree-loop-distribution.c (classify_partition): Only assert on
22012 numer of iterations.
22013 (merge_dep_scc_partitions): Delete prameter. Update function call.
22014 (distribute_loop): Remove code handling loop with unknown niters.
22015 (pass_loop_distribution::execute): Skip loop with unknown niters.
22017 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22020 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22021 function sort_partitions_by_post_order.
22023 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22025 PR tree-optimization/81374
22026 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22027 the max index of basic blocks, rather than number of basic blocks.
22029 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22031 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22033 (arc_legitimate_pic_operand_p): Likewise.
22034 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22036 (arc_needs_pcl_p): Likewise.
22037 (arc_legitimate_pc_offset_p): Likewise.
22038 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22039 function is also used in constrains.md.
22040 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22041 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22042 PLUS. Only return true/false in known cases, otherwise assert.
22043 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22044 is already called in arc_legitimate_constant_p.
22045 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22047 (LEGITIMATE_PIC_OPERAND_P): Use
22048 arc_raw_symbolic_reference_mentioned_p function.
22049 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22054 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22055 Andrew Burgess <andrew.burgess@embecosm.com>
22057 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22058 (arc_return_address_register): New function.
22059 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22060 (arc_handle_fndecl_attribute): Add naked attribute.
22061 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22062 (TARGET_WARN_FUNC_RETURN): Likewise.
22063 (arc_allocate_stack_slots_for_args): New function.
22064 (arc_warn_func_return): Likewise.
22065 (machine_function): Change type fn_type.
22066 (arc_compute_function_type): Consider new naked function type,
22067 change function return type.
22068 (arc_must_save_register): Adapt to handle new
22069 arc_compute_function_type's return type.
22070 (arc_expand_prologue): Likewise.
22071 (arc_expand_epilogue): Likewise.
22072 (arc_return_address_regs): Delete.
22073 (arc_return_address_register): New function.
22074 (arc_epilogue_uses): Use above function.
22075 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22076 (arc_function_type): Change encoding, add naked type.
22077 (ARC_INTERRUPT_P): Change to handle the new encoding.
22078 (ARC_FAST_INTERRUPT_P): Likewise.
22079 (ARC_NORMAL_P): Define.
22080 (ARC_NAKED_P): Likewise.
22081 (arc_compute_function_type): Delete prototype.
22082 * config/arc/arc.md (in_ret_delay_slot): Use
22083 arc_return_address_register function.
22084 (simple_return): Likewise.
22085 (p_return_i): Likewise.
22087 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22089 PR tree-optimization/81428
22090 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22091 can't be built for those types.
22093 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22095 Remove stuff dead since r239246.
22097 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22098 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22099 (avr_inform_devices): Remove dead stuff.
22101 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22103 * config/arm/arm_neon.h: Fix softp typo.
22105 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22107 PR tree-optimization/81365
22108 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22109 aggregate moves onto bb predecessor edges, make sure there are no
22110 loads that could alias the lhs in between the start of bb and the
22113 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22116 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22117 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22118 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22120 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22122 PR tree-optimization/81396
22123 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22124 (init_symbolic_number): Initialize it to 1.
22125 (perform_symbolic_merge): Add n_ops from both operands into the new
22127 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22128 without base_addr as useless if they need more than one operation.
22129 (bswap_replace): Handle !bswap case for NULL base_addr.
22131 2017-07-17 Tom de Vries <tom@codesourcery.com>
22134 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22137 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22139 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22140 conditional builtin define __FIX_LEON3FT_B2BST.
22142 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22144 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22145 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22148 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22150 PR rtl-optimization/81424
22151 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22152 to remove potential trapping from operands if -fnon-call-exceptions.
22154 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22156 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22157 profile_proability for scalling.
22158 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22160 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22162 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22164 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22166 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22167 fixpoint arithmetics.
22169 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22171 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22172 fixpoint arithmetics.
22174 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22176 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22177 fixpoint arithmetics.
22179 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22181 * profile-count.h (profile_probability::from_reg_br_prob_note,
22182 profile_probability::to_reg_br_prob_note): New functions.
22183 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22184 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22185 * predict.c (probability_reliable_p): Update.
22186 (edge_probability_reliable_p): Update.
22187 (br_prob_note_reliable_p): Update.
22188 (invert_br_probabilities): Update.
22189 (add_reg_br_prob_note): New function.
22190 (combine_predictions_for_insn): Update.
22191 * asan.c (asan_clear_shadow): Update.
22192 * cfgbuild.c (compute_outgoing_frequencies): Update.
22193 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22194 (update_br_prob_note): Update.
22195 (rtl_verify_edges): Update.
22196 (purge_dead_edges): Update.
22197 (fixup_reorder_chain): Update.
22198 * emit-rtl.c (try_split): Update.
22199 * ifcvt.c (cond_exec_process_insns): Update.
22200 (cond_exec_process_if_block): Update.
22201 (dead_or_predicable): Update.
22202 * internal-fn.c (expand_addsub_overflow): Update.
22203 (expand_neg_overflow): Update.
22204 (expand_mul_overflow): Update.
22205 * loop-doloop.c (doloop_modify): Update.
22206 * loop-unroll.c (compare_and_jump_seq): Update.
22207 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22208 * predict.h: Update.
22209 * reorg.c (mostly_true_jump): Update.
22211 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22212 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22213 * config/arc/arc.c: (emit_unlikely_jump): Update.
22214 * config/arm/arm.c: (emit_unlikely_jump): Update.
22215 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22216 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22217 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22218 (ix86_print_operand): Update.
22219 (ix86_split_fp_branch): Update.
22220 (predict_jump): Update.
22221 * config/ia64/ia64.c (ia64_print_operand): Update.
22222 * config/mmix/mmix.c (mmix_print_operand): Update.
22223 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22224 (rs6000_expand_split_stack_prologue): Update.
22225 * config/rs6000/rs6000.c: Update.
22226 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22227 (s390_expand_vec_movstr): Update.
22228 (s390_expand_cs_tdsi): Update.
22229 (s390_expand_split_stack_prologue): Update.
22230 * config/sh/sh.c (sh_print_operand): Update.
22231 (expand_cbranchsi4): Update.
22232 (expand_cbranchdi4): Update.
22233 * config/sparc/sparc.c (output_v9branch): Update.
22234 * config/spu/spu.c (get_branch_target): Update.
22235 (ea_load_store_inline): Update.
22236 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22237 * config/tilepro/tilepro.c: Update.
22239 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22241 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22242 (gimplify_save_expr): Likewise.
22244 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22246 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22248 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22250 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22252 * timevar.def (TV_IPA_FNSUMMARY): Define.
22254 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22256 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22257 to back store errata sensitive sequence from being generated.
22258 (sqrtdf2_fix): Likewise.
22260 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22262 * tree-ssa-threadupdate.c (compute_path_counts,
22263 update_joiner_offpath_counts): Use profile_probability.
22265 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22268 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22270 * config/arm/arm-c.c (arm_cpu_builtins): Define
22271 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22273 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22275 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22276 array entries to represent __ieee128 versions of the
22277 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22278 scalar_extract_sig, and scalar_insert_exp built-in functions.
22279 (altivec_resolve_overloaded_builtin): Add special case handling
22280 for the __builtin_scalar_insert_exp function, as represented by
22281 the P9V_BUILTIN_VEC_VSIEDP constant.
22282 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22283 exponent support for __ieee128 argument.
22284 (VSESQP): Add scalar extract signature support for __ieee128
22286 (VSTDCNQP): Add scalar test negative support for __ieee128
22288 (VSIEQP): Add scalar insert exponent support for __int128 argument
22289 with __ieee128 result.
22290 (VSIEQPF): Add scalar insert exponent support for __ieee128
22291 argument with __ieee128 result.
22292 (VSTDCQP): Add scalar test data class support for __ieee128
22294 (VSTDCNQP): Add overload support for scalar test negative with
22295 __ieee128 argument.
22296 (VSTDCQP): Add overload support for scalar test data class
22297 __ieee128 argument.
22298 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22299 UNSPEC_VSX_SXSIGDP.
22300 (UNSPEC_VSX_SIEXPQP): New constant.
22301 (xsxexpqp): New insn for VSX scalar extract exponent quad
22303 (xsxsigqp): New insn for VSX scalar extract significand quad
22305 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22306 precision with floating point argument.
22307 (xststdcqp): New expand for VSX scalar test data class quad
22309 (xststdcnegqp): New expand for VSX scalar test negative quad
22311 (xststdcqp): New insn to match expansions for VSX scalar test data
22312 class quad precision and VSX scalar test negative quad precision.
22313 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22314 special case operand checking to enforce that second operand of
22315 VSX scalar test data class with quad precision argument is a 7-bit
22317 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22318 prototypes and descriptions of __ieee128 versions of
22319 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22320 scalar_test_data_class, and scalar_test_neg built-in functions.
22322 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22324 PR tree-optimization/81162
22325 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22326 replace a negate with an add.
22328 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22330 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22332 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22334 * config/arm/arm-c.c (arm_cpu_builtins): Define
22335 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22337 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22339 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22340 (armv8-r): Set ARM Cortex-R52 as default CPU.
22341 * config/arm/arm-tables.opt: Regenerate.
22342 * config/arm/arm-tune.md: Regenerate.
22343 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22345 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22346 extension for -mcpu=cortex-r52.
22348 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22350 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22351 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22352 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22353 (fp-armv8): Define it as FP_ARMv8 only.
22354 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22355 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22357 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22358 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22359 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22360 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22361 than TARGET_FPU_ARMV8.
22362 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22363 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22364 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22366 * config/arm/neon.md (neon_vrint): Likewise.
22367 (neon_vcvt): Likewise.
22368 (neon_<fmaxmin_op><mode>): Likewise.
22369 (<fmaxmin><mode>3): Likewise.
22370 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22371 * config/arm/predicates.md (arm_cond_move_operator): Check against
22372 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22374 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22376 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22377 to top of function.
22379 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22381 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22382 loop in comment with memset.
22384 2017-07-14 Martin Liska <mliska@suse.cz>
22386 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22387 * dwarf2out.c (is_java): Remove the function.
22388 (output_pubname): Remove usage of the function.
22389 (lower_bound_default): Remove usage of DW_LANG_Java.
22390 (gen_compile_unit_die): Likewise.
22391 * gcc.c: Remove compiler defaults for .java and .zip files.
22392 * gimple-expr.c (remove_suffix): Change as there's no longer
22393 extension than 4-letter one.
22394 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22395 (gimplify_save_expr): Likewise.
22396 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22397 as it's possible even for other languages than Java.
22398 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22399 * lto-opts.c (lto_write_options): Remove reference to Java.
22400 * opts.c (strip_off_ending): Update file extension handling.
22401 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22402 * tree-eh.c (lower_resx): Likewise.
22403 * tree.c (free_lang_data_in_type): Remove dead code.
22404 (find_decls_types_r): Likewise.
22405 (build_common_builtin_nodes): Remove Java from a comment.
22406 (verify_type): Remove dead code.
22407 * varasm.c (assemble_external): Remove Java from a comment.
22409 2017-07-14 Martin Liska <mliska@suse.cz>
22411 * opts.c (finish_options): Add quotes.
22412 (common_handle_option): Likewise.
22414 2017-07-14 Martin Liska <mliska@suse.cz>
22416 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22417 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22418 Remove N_SO_PASCAL.
22419 * dwarf2out.c (lower_bound_default): Do not handle
22421 (gen_compile_unit_die): Likewise.
22422 * gcc.c: Remove default extension binding for GNU Pascal.
22423 * stmt.c: Remove Pascal language from a comment.
22424 * xcoffout.c: Likewise.
22426 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22429 * diagnostic-show-locus.c (fixit_cmp): New function.
22430 (layout::layout): Sort m_fixit_hints.
22431 (column_range::column_range): Assert that the values are valid.
22432 (struct char_span): New struct.
22433 (correction::overwrite): New method.
22434 (struct source_line): New struct.
22435 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22436 calls in terms of classes source_line and char_span, and
22437 correction::overwrite.
22438 (selftest::test_overlapped_fixit_printing_2): New function.
22439 (selftest::diagnostic_show_locus_c_tests): Call it.
22441 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22443 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22444 early if there is no lhs.
22446 2017-07-13 Martin Liska <mliska@suse.cz>
22448 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22449 (gen_reference_type_die): Likewise.
22450 * stor-layout.c: Remove Pascal-related comment.
22452 2017-07-13 Martin Liska <mliska@suse.cz>
22454 * opts.c (finish_options): Add quotes to error messages.
22455 (parse_sanitizer_options): Likewise.
22457 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22459 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22461 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22463 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22465 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22467 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22469 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22471 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22474 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22475 provides the hardware capability bits, define the macro
22476 __BUILTIN_CPU_SUPPORTS__.
22477 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22478 if GLIBC does not provide the hardware capability bits. Add a
22479 gcc_unreachable call if the built-in cpu function is neither
22480 __builtin_cpu_is nor __builtin_cpu_supports.
22481 (rs6000_get_function_versions_dispatcher): Change the warning
22482 that an old GLIBC is used which does not export the capability
22483 bits to be an error.
22484 * doc/extend.texi (target_clones attribute): Document the
22485 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22486 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22487 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22488 the macros defined by GCC if the newer GLIBC is available.
22490 2017-07-12 Jeff Law <law@redhat.com>
22492 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22493 remaining includes slightly.
22494 * config/riscv/riscv-builtins.c: Include profile-count.h.
22496 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22499 * config/avr/avr.c (avr_set_current_function): In diagnostic
22500 messages: Quote keywords and (parts of) identifiers.
22501 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22504 2017-07-12 Carl Love <cel@us.ibm.com>
22506 * config/rs6000/rs6000-c.c: Add support for built-in functions
22507 vector bool char vec_revb (vector bool char);
22508 vector bool short vec_revb (vector short char);
22509 vector bool int vec_revb (vector bool int);
22510 vector bool long long vec_revb (vector bool long long);
22511 * doc/extend.texi: Update the built-in documentation file for the
22512 new built-in functions.
22514 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22516 * config/s390/s390.md: Remove movcc splitter.
22518 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22520 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22521 load/store on condition.
22523 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22526 * config/avr/avr.c (avr_encode_section_info)
22527 [progmem && !TREE_READONLY]: Error if progmem object needs
22530 2017-07-11 Michael Collison <michael.collison@arm.com>
22532 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22535 2017-07-11 Carl Love <cel@us.ibm.com>
22537 * config/rs6000/rs6000-c.c: Add support for builtins
22538 vector unsigned int vec_parity_lsbb (vector signed int);
22539 vector unsigned int vec_parity_lsbb (vector unsigned int);
22540 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22541 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22542 vector unsigned long long vec_parity_lsbb (vector signed long long);
22543 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22544 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22545 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22546 * doc/extend.texi: Update the built-in documentation file for the
22547 new built-in functions.
22549 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22551 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22552 (layout::m_primary_loc): New field.
22553 (layout::layout): Initialize new field. Move location filtering
22554 logic from here to...
22555 (layout::maybe_add_location_range): ...this new method. Add
22556 support for filtering to just the lines already specified by other
22558 (layout::will_show_line_p): New method.
22559 (gcc_rich_location::add_location_if_nearby): New method.
22560 (selftest::test_add_location_if_nearby): New test function.
22561 (selftest::diagnostic_show_locus_c_tests): Call it.
22562 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22565 2017-07-11 Tom de Vries <tom@codesourcery.com>
22567 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22568 (bb_first_real_insn): New function.
22569 (nvptx_single): Add extra initialization of broadcasted condition
22572 2017-07-11 Nathan Sidwell <nathan@acm.org>
22574 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22576 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22578 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22579 Binutils doc as TEXI will mess them up.
22580 * doc/invoke.texi (AVR Options): Same here.
22582 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22584 * config/sparc/sparc.opt (mfix-ut700): New option.
22585 (mfix-gr712rc): Likewise.
22586 (sparc_fix_b2bst): New variable.
22587 * doc/invoke.texi (SPARC options): Document them.
22588 (ARM options): Fix warnings.
22589 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22590 instructions to prevent sequences that can trigger the store-store
22591 errata for certain LEON3FT processors.
22592 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22593 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22594 * config/sparc/sparc.md (fix_b2bst): New attribute.
22595 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22597 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22600 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22602 (*rsqrtsf2_sse): Ditto.
22604 (div<mode>3): Macroize insn from divdf3 and divsf3
22605 using MODEF mode iterator.
22607 2017-07-10 Martin Sebor <msebor@redhat.com>
22609 PR tree-optimization/80397
22610 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22611 instead of testing for equality to INTEGER_TYPE.
22613 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22615 * config.gcc: Remove uclibc from arc target spec.
22617 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22619 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22621 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22624 * lto-wrapper.c (remove_option): New function.
22625 (merge_and_complain): Merge PIC/PIE options more realistically.
22627 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22629 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22633 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22634 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22635 * config.in: Regenerate.
22636 * configure: Regenerate.
22637 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22638 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22639 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22640 (TARGET_GASISR_PROLOGUES): ...target mask.
22641 * common/config/avr/avr-common.c
22642 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22643 Set -mgas-isr-prologues.
22644 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22645 INSERT_PASS_BEFORE for it.
22646 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22647 * config/avr/avr.c (avr_option_override)
22648 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22649 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22650 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22651 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22652 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22654 (make_avr_pass_pre_proep): New function.
22655 (emit_push_sfr) <treg>: Add argument to function and use it
22656 instead of TMP_REG.
22657 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22658 and set machine->gasisr.yes.
22659 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22660 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22661 __gcc_isr.n_pushed to .L__stack_usage.
22662 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22663 (avr_asm_final_postscan_insn): ...this new static function.
22664 * config/avr/avr.h (machine_function)
22665 <is_no_gccisr, use_L__stack_usage>: New fields.
22666 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22667 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22668 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22669 (gasisr, *gasisr): New expander and insn.
22670 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22671 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22672 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22674 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22676 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22679 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22681 Move jump-tables out of .text again.
22684 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22685 (ASM_OUTPUT_ADDR_VEC): New function.
22686 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22687 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22688 INSN_ADDRESSes as asm comment.
22689 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22690 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22691 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22692 * config/avr/avr.md (*tablejump): Adjust comment.
22693 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22694 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22696 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22697 (avr_output_addr_vec): New proto.
22698 (avr_log_t) <insn_addresses>: New field.
22700 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22703 * config/i386/i386.c (ix86_function_arg_advance): Set
22704 outgoing_args_on_stack to true if there are outgoing arguments
22706 (ix86_function_arg): Likewise.
22707 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22708 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22709 * config/i386/i386.h (machine_function): Add
22710 outgoing_args_on_stack.
22712 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
22714 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22715 supporting pthreds.
22716 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
22718 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
22720 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
22721 (REAL_H): Remove $(MACHMODE_H).
22722 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
22724 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
22725 $(MACHMODE_H) and double-int.h.
22726 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
22728 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
22731 2017-07-07 Andrew Pinski <apinski@cavium.com>
22733 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
22734 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22736 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22738 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22739 Add warning if GCC was not configured to link against a GLIBC that
22740 exports the hardware capability bits.
22741 (make_resolver_func): Make resolver function private and not a
22742 COMDAT function. Create the name with clone_function_name instead
22743 of make_unique_name.
22746 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22747 correct operand in doing the split.
22749 2017-07-07 Carl Love <cel@us.ibm.com>
22751 * config/rs6000/rs6000-c: Add support for built-in function
22752 vector unsigned short vec_pack_to_short_fp32 (vector float,
22754 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22755 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22756 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22757 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22758 (convert_4f32_8i16): Add define_expand.
22759 * doc/extend.texi: Update the built-in documentation file for the
22760 new built-in function.
22762 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22764 * config/sparc/m8.md: New file.
22765 * config/sparc/sparc.md: Include m8.md.
22767 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22769 * config/sparc/sparc.opt: New option -mvis4b.
22770 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22771 (sparc_option_override): Handle VIS4B.
22772 (enum sparc_builtins): Define
22773 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22774 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22775 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22776 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22777 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22778 (check_constant_argument): New function.
22779 (sparc_vis_init_builtins): Define builtins
22780 __builtin_vis_dictunpack{8,16,32},
22781 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22782 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22783 __builtin_vis_fpcmpde{8,16,32}shl and
22784 __builtin_vis_fpcmpur{8,16,32}shl.
22785 (sparc_expand_builtin): Check that the constant operands to
22786 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22787 constant and in range.
22788 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22790 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22791 (SPARC_IMM5_P): Likewise.
22792 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22793 (enabled): Handle vis4b.
22794 (UNSPEC_DICTUNPACK): New unspec.
22795 (UNSPEC_FPCMPSHL): Likewise.
22796 (UNSPEC_FPUCMPSHL): Likewise.
22797 (UNSPEC_FPCMPDESHL): Likewise.
22798 (UNSPEC_FPCMPURSHL): Likewise.
22799 (cpu_feature): New CPU feature `vis4b'.
22800 (dictunpack{8,16,32}): New insns.
22801 (FPCSMODE): New mode iterator.
22802 (fpcscond): New code iterator.
22803 (fpcsucond): Likewise.
22804 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22805 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22806 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22807 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22808 * config/sparc/constraints.md: Define constraints `q' for unsigned
22809 2-bit integer constants and `t' for unsigned 5-bit integer
22811 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22813 (imm5_operand_dictunpack16): Likewise.
22814 (imm5_operand_dictunpack32): Likewise.
22815 (imm2_operand): Likewise.
22816 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22817 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22818 ditunpack* and fpcmp*shl builtins.
22820 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22822 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22823 * config.in: Add HAVE_AS_SPARC6 define.
22824 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22826 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22828 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22829 (CPP_CPU_SPEC): Handle m8.
22830 (ASM_CPU_SPEC): Likewise.
22831 * config/sparc/sparc-opts.h (enum processor_type): Add
22833 * config/sparc/sparc.c (m8_costs): New struct.
22834 (sparc_option_override): Handle TARGET_CPU_m8.
22835 (sparc32_initialize_trampoline): Likewise.
22836 (sparc64_initialize_trampoline): Likewise.
22837 (sparc_issue_rate): Likewise.
22838 (sparc_register_move_cost): Likewise.
22839 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22840 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22841 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22842 (CPP_CPU_SPEC): Handle M8.
22843 (ASM_CPU_SPEC): Likewise.
22844 (AS_M8_FLAG): Define.
22845 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22846 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22847 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22849 * configure: Regenerate.
22850 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22853 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22855 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22857 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22858 ("*movdi_insn_sp32"): Do not set v3pipe.
22859 ("*movsi_insn"): Likewise.
22860 ("*movdi_insn_sp64"): Likewise.
22861 ("*movsf_insn"): Likewise.
22862 ("*movdf_insn_sp32"): Likewise.
22863 ("*movdf_insn_sp64"): Likewise.
22864 ("*zero_extendsidi2_insn_sp64"): Likewise.
22865 ("*sign_extendsidi2_insn"): Likewise.
22866 ("*mov<VM32:mode>_insn"): Likewise.
22867 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22868 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22869 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22870 ("<vlop:code><VL:mode>3"): Likewise.
22871 ("*not_<vlop:code><VL:mode>3"): Likewise.
22872 ("*nand<VL:mode>_vis"): Likewise.
22873 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22874 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22875 ("one_cmpl<VL:mode>2"): Likewise.
22876 ("faligndata<VM64:mode>_vis"): Likewise.
22877 ("alignaddrsi_vis"): Likewise.
22878 ("alignaddrdi_vis"): Likweise.
22879 ("alignaddrlsi_vis"): Likewise.
22880 ("alignaddrldi_vis"): Likewise.
22881 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22882 ("bmaskdi_vis"): Likewise.
22883 ("bmasksi_vis"): Likewise.
22884 ("bshuffle<VM64:mode>_vis"): Likewise.
22885 ("cmask8<P:mode>_vis"): Likewise.
22886 ("cmask16<P:mode>_vis"): Likewise.
22887 ("cmask32<P:mode>_vis"): Likewise.
22888 ("pdistn<P:mode>_vis"): Likewise.
22889 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22891 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22893 * config/sparc/sparc.md ("subtype"): New insn attribute.
22894 ("*wrgsr_sp64"): Set insn subtype.
22895 ("*rdgsr_sp64"): Likewise.
22896 ("alignaddrsi_vis"): Likewise.
22897 ("alignaddrdi_vis"): Likewise.
22898 ("alignaddrlsi_vis"): Likewise.
22899 ("alignaddrldi_vis"): Likewise.
22900 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22901 ("fexpand_vis"): Likewise.
22902 ("fpmerge_vis"): Likewise.
22903 ("faligndata<VM64:mode>_vis"): Likewise.
22904 ("bshuffle<VM64:mode>_vis"): Likewise.
22905 ("cmask8<P:mode>_vis"): Likewise.
22906 ("cmask16<P:mode>_vis"): Likewise.
22907 ("cmask32<P:mode>_vis"): Likewise.
22908 ("fchksm16_vis"): Likewise.
22909 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22910 ("fmean16_vis"): Likewise.
22911 ("fp<plusminus_insn>64_vis"): Likewise.
22912 ("<plusminus_insn>v8qi3"): Likewise.
22913 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22914 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22915 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22916 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22917 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22918 ("*movqi_insn"): Likewise.
22919 ("*movhi_insn"): Likewise.
22920 ("*movsi_insn"): Likewise.
22921 ("movsi_pic_gotdata_op"): Likewise.
22922 ("*movdi_insn_sp32"): Likewise.
22923 ("*movdi_insn_sp64"): Likewise.
22924 ("movdi_pic_gotdata_op"): Likewise.
22925 ("*movsf_insn"): Likewise.
22926 ("*movdf_insn_sp32"): Likewise.
22927 ("*movdf_insn_sp64"): Likewise.
22928 ("*zero_extendhisi2_insn"): Likewise.
22929 ("*zero_extendqihi2_insn"): Likewise.
22930 ("*zero_extendqisi2_insn"): Likewise.
22931 ("*zero_extendqidi2_insn"): Likewise.
22932 ("*zero_extendhidi2_insn"): Likewise.
22933 ("*zero_extendsidi2_insn_sp64"): Likewise.
22934 ("ldfsr"): Likewise.
22935 ("prefetch_64"): Likewise.
22936 ("prefetch_32"): Likewise.
22937 ("tie_ld32"): Likewise.
22938 ("tie_ld64"): Likewise.
22939 ("*tldo_ldub_sp32"): Likewise.
22940 ("*tldo_ldub1_sp32"): Likewise.
22941 ("*tldo_ldub2_sp32"): Likewise.
22942 ("*tldo_ldub_sp64"): Likewise.
22943 ("*tldo_ldub1_sp64"): Likewise.
22944 ("*tldo_ldub2_sp64"): Likewise.
22945 ("*tldo_ldub3_sp64"): Likewise.
22946 ("*tldo_lduh_sp32"): Likewise.
22947 ("*tldo_lduh1_sp32"): Likewise.
22948 ("*tldo_lduh_sp64"): Likewise.
22949 ("*tldo_lduh1_sp64"): Likewise.
22950 ("*tldo_lduh2_sp64"): Likewise.
22951 ("*tldo_lduw_sp32"): Likewise.
22952 ("*tldo_lduw_sp64"): Likewise.
22953 ("*tldo_lduw1_sp64"): Likewise.
22954 ("*tldo_ldx_sp64"): Likewise.
22955 ("*mov<VM32:mode>_insn"): Likewise.
22956 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22957 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22959 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22961 * config/sparc/sparc.md ("type"): New insn type viscmp.
22962 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22964 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22965 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22966 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22967 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22969 ("n7_vis_logical_11cycle"): Likewise.
22970 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22971 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22972 * config/sparc/niagara.md ("niag_vis"): Likewise.
22973 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22974 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22976 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22978 * config/sparc/sparc.md: New instruction type `bmask'.
22979 (bmaskdi_vis): Use the `bmask' type.
22980 (bmasksi_vis): Likewise.
22981 * config/sparc/ultra3.md (us3_array): Likewise.
22982 * config/sparc/niagara7.md (n7_array): Likewise.
22983 * config/sparc/niagara4.md (n4_array): Likewise.
22984 * config/sparc/niagara2.md (niag2_vis): Likewise.
22985 (niag3_vis): Likewise.
22986 * config/sparc/niagara.md (niag_vis): Likewise.
22988 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22990 * ipa-comdats.c: Remove optimize check from gate.
22991 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22992 for functions not optimized.
22993 (ipa_fn_summary_read): Skip optimize check.
22994 (ipa_fn_summary_write): Likewise.
22995 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22997 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22999 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23000 (check_callers): Give up on caller which is not optimized.
23001 (inline_small_functions): Likewise.
23002 (ipa_inline): Do not give up when not optimizing.
23003 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23004 away unoptimizes cdtors.
23005 (whole_program_function_and_variable_visibility): Do
23006 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23007 * ipa.c (process_references): Do not check optimize.
23008 (symbol_table::remove_unreachable_nodes): Update optimize check.
23009 (set_writeonly_bit): Update optimize check.
23010 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23011 (pass_ipa_single_use::gate): Remove.
23013 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23015 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23016 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23017 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23018 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23019 permute_load, permute_store, adjust_extract, adjust_splat,
23020 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23021 replace_swap_with_copy, dump_swap_insn_table,
23022 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23023 recombine_lvx_pattern, recombine_stvx_pattern,
23024 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23025 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23026 to file rs6000-p8swap.c.
23027 * config/rs6000/rs6000-p8swap.c: New file.
23028 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23029 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23030 and rs6000*-*-* targets.
23032 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23034 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23036 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23038 * lto-wrapper.c (merge_and_complain): Do not merge
23039 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23040 fsigned_zeros, ftrapping_math, fwrapv.
23041 (append_compiler_options): Do not track these options.
23042 (append_linker_options): Likewie
23044 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23046 * cgraphunit.c (cgraph_node::finalize_function): When
23047 !flag_toplevel_reorde set no_reorder flag.
23048 (varpool_node::finalize_decl): Likewise.
23049 (symbol_table::compile): Drop no toplevel reorder path.
23051 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23053 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23054 edges; zero probability is not better than uninitialized.
23056 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23058 * asan.h (asan_sanitize_allocas_p): Declare.
23059 * asan.c (asan_sanitize_allocas_p): New function.
23060 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23061 (handle_builtin_alloca): Likewise.
23062 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23063 if !asan_sanitize_allocas_p.
23064 * params.def (asan-instrument-allocas): Add new option.
23065 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23066 * opts.c (common_handle_option): Disable allocas sanitization for
23069 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23071 * asan.c: Include gimple-fold.h.
23072 (get_last_alloca_addr): New function.
23073 (handle_builtin_stackrestore): Likewise.
23074 (handle_builtin_alloca): Likewise.
23075 (asan_emit_allocas_unpoison): Likewise.
23076 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23077 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23078 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23079 (instrument_builtin_call): Pass gimple iterator to
23080 get_mem_refs_of_builtin_call.
23081 (last_alloca_addr): New global.
23082 * asan.h (asan_emit_allocas_unpoison): Declare.
23083 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23084 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23085 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23086 if function calls alloca.
23087 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23088 * gimple-fold.h (replace_call_with_value): Declare.
23089 * internal-fn.c: Include asan.h.
23090 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23091 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23093 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23095 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23096 (C_SELFTEST_FLAGS): New.
23097 (CPP_SELFTEST_FLAGS): New.
23098 (SELFTEST_DEPS): New, from deps of s-selftest.
23099 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23100 (CPP_SELFTEST_DEPS): New.
23101 (selftest): Add dependency on s-selftest-c++.
23102 (s-selftest): Rename to...
23103 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23104 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23105 than SELFTEST_FLAGS.
23106 (selftest-gdb): Rename to...
23107 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23109 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23110 (selftest-valgrind): Rename to...
23111 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23113 (selftest-valgrind): Reintroduce as an alias for
23114 selftest-c-valgrind.
23115 (s-selftest-c++): New.
23116 (selftest-c++-gdb): New.
23117 (selftest-c++-valgrind): New.
23119 2017-07-06 Olivier Hainque <hainque@adacore.com>
23121 * gcc.c (process_command): When deciding if undefined variables
23122 should be ignored when processing specs, accept "gcc -v" as well.
23124 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23126 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23127 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23129 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23131 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23132 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23133 * config/arm/arm-tables.opt: Regenerate.
23134 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23136 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23138 2017-07-06 Carl Love <cel@us.ibm.com>
23140 * ChangeLog: Clean up from mid air collision
23142 2017-07-06 Carl Love <cel@us.ibm.com>
23144 * config/rs6000/rs6000-c.c: Add support for built-in functions
23145 vector signed int vec_subc (vector signed int, vector signed int);
23146 vector signed __int128 vec_subc (vector signed __int128,
23147 vector signed __int128);
23148 vector unsigned __int128 vec_subc (vector unsigned __int128,
23149 vector unsigned __int128);
23150 vector signed int vec_sube (vector signed int, vector signed int,
23151 vector signed int);
23152 vector unsigned int vec_sube (vector unsigned int,
23153 vector unsigned int,
23154 vector unsigned int);
23155 vector signed __int128 vec_sube (vector signed __int128,
23156 vector signed __int128,
23157 vector signed__int128);
23158 vector unsigned __int128 vec_sube (vector unsigned __int128,
23159 vector unsigned __int128,
23160 vector unsigned __int128);
23161 vector signed int vec_subec (vector signed int, vector signed int,
23162 vector signed int);
23163 vector unsigned int vec_subec (vector unsigned int,
23164 vector unsigned int,
23165 vector unsigned int);
23166 vector signed __int128 vec_subec (vector signed __int128,
23167 vector signed __int128,
23168 vector signed__int128);
23169 vector unsigned __int128 vec_subec (vector unsigned __int128,
23170 vector unsigned __int128,
23171 vector unsigned __int128);
23172 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23173 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23174 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23175 BU_ALTIVEC_OVERLOAD_X definitions.
23176 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23177 * doc/extend.texi: Update the built-in documentation file for the new
23178 built-in functions.
23180 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23183 * diagnostic-show-locus.c (layout::layout): Use start and finish
23184 spelling location for the start and finish of each range.
23185 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23186 Add unused aspect param.
23187 * input.c (expand_location_1): Add "aspect" param, and use it
23188 to access the correct part of the location.
23189 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23191 (expand_location_to_spelling_point): Likewise.
23192 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23193 param, and pass it to expand_location_1.
23195 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23197 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23198 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23199 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23200 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23201 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23202 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23203 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23204 _mm_maskz_getmant_ss): New intrinsics.
23205 (__builtin_ia32_getexpss128_mask): Changed to ...
23206 __builtin_ia32_getexpss128_round ... this.
23207 (__builtin_ia32_getexpsd128_mask): Changed to ...
23208 __builtin_ia32_getexpsd128_round ... this.
23209 * config/i386/i386-builtin-types.def
23210 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23211 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23212 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23213 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23214 __builtin_ia32_getmantss_mask_round): New builtins.
23215 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23216 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23217 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23218 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23219 * config/i386/sse.md
23220 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23221 avx512f_sgetexp<mode><mask_scalar_name>
23222 <round_saeonly_scalar_name> ... this.
23223 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23224 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23225 vgetexp<ssescalarmodesuffix>
23226 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23227 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23228 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23229 avx512f_vgetmant<mode><mask_scalar_name>
23230 <round_saeonly_scalar_name> ... this.
23231 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23232 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23233 vgetmant<ssescalarmodesuffix>
23234 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23235 %0<mask_scalar_operand4>, %1, %2
23236 <round_saeonly_scalar_mask_op4>, %3} ... this.
23237 * config/i386/subst.md (mask_scalar_operand4,
23238 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23239 round_saeonly_scalar_nimm_predicate): New subst attributes.
23241 2017-07-06 Julia Koval <julia.koval@intel.com>
23243 * config/i386/i386.c (ix86_erase_embedded_rounding):
23244 Remove code for old rounding pattern.
23246 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23248 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23250 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23252 * doc/sourcebuild.texi (Test Directives, Variants of
23253 dg-require-support): Add documentation for dg-require-stack-check.
23255 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23257 * config/i386/subst.md (mask_scalar, round_scalar,
23258 round_saeonly_scalar): New meta-templates.
23259 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23260 round_scalar_mask_operand3, round_scalar_mask_op3,
23261 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23262 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23263 round_saeonly_scalar_constraint,
23264 round_saeonly_scalar_prefix): New subst attribute.
23265 * config/i386/sse.md
23266 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23267 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23268 <round_scalar_name> ... this.
23269 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23270 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23271 <round_scalar_name> ... this.
23272 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23273 <sse>_vm<code><mode>3<mask_scalar_name>
23274 <round_saeonly_scalar_name> ... this.
23275 (v<plusminus_mnemonic><ssescalarmodesuffix>
23276 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23277 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23278 v<plusminus_mnemonic><ssescalarmodesuffix>
23279 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23280 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23281 (v<multdiv_mnemonic><ssescalarmodesuffix>
23282 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23283 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23284 v<multdiv_mnemonic><ssescalarmodesuffix>
23285 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23286 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23287 (v<maxmin_float><ssescalarmodesuffix>
23288 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23289 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23290 v<maxmin_float><ssescalarmodesuffix>
23291 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23292 %0<mask_scalar_operand3>, %1, %<iptr>2
23293 <round_saeonly_scalar_mask_op3>} ... this.
23295 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23297 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23298 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23300 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23301 Alan Hayward <alan.hayward@arm.com>
23302 David Sherwood <david.sherwood@arm.com>
23304 * combine.c (simplify_if_then_else): Remove "enum" before
23306 * compare-elim.c (can_eliminate_compare): Likewise.
23307 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23309 (aarch64_lookup_simd_builtin_type): Likewise.
23310 (aarch64_simd_builtin_type): Likewise.
23311 (aarch64_init_simd_builtin_types): Likewise.
23312 (aarch64_simd_expand_args): Likewise.
23313 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23315 (aarch64_reverse_mask): Likewise.
23316 (aarch64_simd_emit_reg_reg_move): Likewise.
23317 (aarch64_gen_adjusted_ldpstp): Likewise.
23318 (aarch64_ccmp_mode_to_code): Likewise.
23319 (aarch64_operands_ok_for_ldpstp): Likewise.
23320 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23321 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23323 (aarch64_min_divisions_for_recip_mul): Likewise.
23324 (aarch64_reassociation_width): Likewise.
23325 (aarch64_get_condition_code_1): Likewise.
23326 (aarch64_simd_emit_reg_reg_move): Likewise.
23327 (aarch64_simd_attr_length_rglist): Likewise.
23328 (aarch64_reverse_mask): Likewise.
23329 (aarch64_operands_ok_for_ldpstp): Likewise.
23330 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23331 (aarch64_gen_adjusted_ldpstp): Likewise.
23332 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23334 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23335 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23336 (arm_lookup_simd_builtin_type): Likewise.
23337 (arm_simd_builtin_type): Likewise.
23338 (arm_init_simd_builtin_types): Likewise.
23339 (arm_expand_builtin_args): Likewise.
23340 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23341 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23342 (ft32_setup_incoming_varargs): Likewise.
23343 (ft32_function_arg): Likewise.
23344 (ft32_function_arg_advance): Likewise.
23345 (ft32_pass_by_reference): Likewise.
23346 (ft32_arg_partial_bytes): Likewise.
23347 (ft32_valid_pointer_mode): Likewise.
23348 (ft32_addr_space_pointer_mode): Likewise.
23349 (ft32_addr_space_legitimate_address_p): Likewise.
23350 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23352 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23353 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23354 (iamcu_alignment): Likewise.
23355 (canonicalize_vector_int_perm): Likewise.
23356 (ix86_noce_conversion_profitable_p): Likewise.
23357 (ix86_mpx_bound_mode): Likewise.
23358 (ix86_operands_ok_for_move_multiple): Likewise.
23359 * config/microblaze/microblaze-protos.h
23360 (microblaze_expand_conditional_branch_reg): Likewise.
23361 * config/microblaze/microblaze.c
23362 (microblaze_expand_conditional_branch_reg): Likewise.
23363 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23365 (rs6000_reassociation_width): Likewise.
23366 (rs6000_invalid_binary_op): Likewise.
23367 (fusion_p9_p): Likewise.
23368 (emit_fusion_p9_load): Likewise.
23369 (emit_fusion_p9_store): Likewise.
23370 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23372 (riscv_hard_regno_mode_ok_p): Likewise.
23373 (riscv_address_insns): Likewise.
23374 (riscv_split_symbol): Likewise.
23375 (riscv_legitimize_move): Likewise.
23376 (riscv_function_value): Likewise.
23377 (riscv_hard_regno_nregs): Likewise.
23378 (riscv_expand_builtin): Likewise.
23379 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23380 (riscv_build_integer): Likewise.
23381 (riscv_split_integer): Likewise.
23382 (riscv_legitimate_constant_p): Likewise.
23383 (riscv_cannot_force_const_mem): Likewise.
23384 (riscv_regno_mode_ok_for_base_p): Likewise.
23385 (riscv_valid_base_register_p): Likewise.
23386 (riscv_valid_offset_p): Likewise.
23387 (riscv_valid_lo_sum_p): Likewise.
23388 (riscv_classify_address): Likewise.
23389 (riscv_legitimate_address_p): Likewise.
23390 (riscv_address_insns): Likewise.
23391 (riscv_load_store_insns): Likewise.
23392 (riscv_force_binary): Likewise.
23393 (riscv_split_symbol): Likewise.
23394 (riscv_force_address): Likewise.
23395 (riscv_legitimize_address): Likewise.
23396 (riscv_move_integer): Likewise.
23397 (riscv_legitimize_const_move): Likewise.
23398 (riscv_legitimize_move): Likewise.
23399 (riscv_address_cost): Likewise.
23400 (riscv_subword): Likewise.
23401 (riscv_output_move): Likewise.
23402 (riscv_canonicalize_int_order_test): Likewise.
23403 (riscv_emit_int_order_test): Likewise.
23404 (riscv_function_arg_boundary): Likewise.
23405 (riscv_pass_mode_in_fpr_p): Likewise.
23406 (riscv_pass_fpr_single): Likewise.
23407 (riscv_pass_fpr_pair): Likewise.
23408 (riscv_get_arg_info): Likewise.
23409 (riscv_function_arg): Likewise.
23410 (riscv_function_arg_advance): Likewise.
23411 (riscv_arg_partial_bytes): Likewise.
23412 (riscv_function_value): Likewise.
23413 (riscv_pass_by_reference): Likewise.
23414 (riscv_setup_incoming_varargs): Likewise.
23415 (riscv_print_operand): Likewise.
23416 (riscv_elf_select_rtx_section): Likewise.
23417 (riscv_save_restore_reg): Likewise.
23418 (riscv_for_each_saved_reg): Likewise.
23419 (riscv_register_move_cost): Likewise.
23420 (riscv_hard_regno_mode_ok_p): Likewise.
23421 (riscv_hard_regno_nregs): Likewise.
23422 (riscv_class_max_nregs): Likewise.
23423 (riscv_memory_move_cost): Likewise.
23424 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23425 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23426 (rl78_addr_space_address_mode): Likewise.
23427 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23429 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23430 (rs6000_reassociation_width): Likewise.
23431 (rs6000_invalid_binary_op): Likewise.
23432 (fusion_p9_p): Likewise.
23433 (emit_fusion_p9_load): Likewise.
23434 (emit_fusion_p9_store): Likewise.
23435 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23436 (ok_for_simple_move_operands): Likewise.
23437 (ok_for_simple_move_strict_operands): Likewise.
23438 (ok_for_simple_arith_logic_operands): Likewise.
23439 (visium_legitimize_reload_address): Likewise.
23440 (visium_select_cc_mode): Likewise.
23441 (output_cbranch): Likewise.
23442 (visium_split_double_move): Likewise.
23443 (visium_expand_copysign): Likewise.
23444 (visium_expand_int_cstore): Likewise.
23445 (visium_expand_fp_cstore): Likewise.
23446 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23447 (visium_function_arg): Likewise.
23448 (visium_function_arg_advance): Likewise.
23449 (visium_libcall_value): Likewise.
23450 (visium_setup_incoming_varargs): Likewise.
23451 (visium_legitimate_constant_p): Likewise.
23452 (visium_legitimate_address_p): Likewise.
23453 (visium_legitimize_address): Likewise.
23454 (visium_secondary_reload): Likewise.
23455 (visium_register_move_cost): Likewise.
23456 (visium_memory_move_cost): Likewise.
23457 (prepare_move_operands): Likewise.
23458 (ok_for_simple_move_operands): Likewise.
23459 (ok_for_simple_move_strict_operands): Likewise.
23460 (ok_for_simple_arith_logic_operands): Likewise.
23461 (visium_function_value_1): Likewise.
23462 (rtx_ok_for_offset_p): Likewise.
23463 (visium_legitimize_reload_address): Likewise.
23464 (visium_split_double_move): Likewise.
23465 (visium_expand_copysign): Likewise.
23466 (visium_expand_int_cstore): Likewise.
23467 (visium_expand_fp_cstore): Likewise.
23468 (visium_split_cstore): Likewise.
23469 (visium_select_cc_mode): Likewise.
23470 (visium_split_cbranch): Likewise.
23471 (output_cbranch): Likewise.
23472 (visium_print_operand_address): Likewise.
23473 * expmed.c (flip_storage_order): Likewise.
23474 * expmed.h (emit_cstore): Likewise.
23475 (flip_storage_order): Likewise.
23476 * genrecog.c (validate_pattern): Likewise.
23477 * hsa-gen.c (gen_hsa_addr): Likewise.
23478 * internal-fn.c (expand_arith_overflow): Likewise.
23479 * ira-color.c (allocno_copy_cost_saving): Likewise.
23480 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23481 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23482 (process_invariant_for_inheritance): Likewise.
23483 * lra-eliminations.c (move_plus_up): Likewise.
23484 * omp-low.c (lower_oacc_reductions): Likewise.
23485 * simplify-rtx.c (simplify_subreg): Likewise.
23486 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23487 (TARGET_CHKP_BOUND_MODE): Likewise..
23488 * targhooks.c (default_chkp_bound_mode): Likewise.
23489 (default_setup_incoming_vararg_bounds): Likewise.
23490 * targhooks.h (default_chkp_bound_mode): Likewise.
23491 (default_setup_incoming_vararg_bounds): Likewise.
23492 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23493 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23494 (have_whole_vector_shift): Likewise.
23495 * tree-vect-stmts.c (vectorizable_load): Likewise.
23496 * doc/tm.texi: Regenerate.
23498 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23500 Graceful degrade if Binutils PR21472 is not available.
23503 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23504 .rodata in flash test fails.
23505 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23506 * confgure: Regenerate.
23507 * config.in: Regenerate.
23508 * config/avr/avr.c (avr_asm_named_section)
23509 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23510 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23511 (avr_asm_init_sections): Same.
23513 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23515 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23516 (fma<VH:mode>4_intrinsic): Likewise.
23517 (*fmsub<VCVTF:mode>4): Likewise.
23518 (*fmsub<VH:mode>4_intrinsic): Likewise.
23520 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23523 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23524 Don't depend on "optimize > 0".
23525 (out_movhi_r_mr, out_movqi_mr_r): Same.
23526 (out_movhi_mr_r, out_movqi_r_mr): Same.
23527 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23528 io_address_operand on "optimize > 0".
23530 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23532 * tree-loop-distribution.c: Add general explanantion on the pass.
23533 (generate_loops_for_partition): Mark distributed loop.
23534 (pg_add_dependence_edges): New parameter. Handle alias data
23535 dependence specially and record it in the parameter if asked.
23536 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23537 (init_partition_graph_vertices, add_partition_graph_edge): New.
23538 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23539 (free_partition_graph_vdata, build_partition_graph): New.
23540 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23541 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23542 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23543 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23544 (version_for_distribution_p, finalize_partitions): New.
23545 (distribute_loop): Handle alias data dependence specially. Factor
23546 out loop fusion code as functions and call these functions.
23548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23550 * tree-loop-distribution.c (classify_partition): New parameter and
23551 better handle reduction statement.
23552 (rdg_build_partitions): Revise comment.
23553 (distribute_loop): Compute statements in all partitions and pass it
23554 to classify_partition.
23556 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23558 * tree-loop-distribution.c (enum partition_type): New.
23559 (struct partition): New field type.
23560 (partition_merge_into): Add parameter. Update partition type.
23561 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23562 (build_rdg_partition_for_vertex): Compute partition type.
23563 (rdg_build_partitions): Dump partition type.
23564 (distribute_loop): Update calls to partition_merge_into.
23566 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23568 * tree-loop-distribution.c (struct ddr_hasher): New.
23569 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23571 (classify_partition): Call get_data_dependence.
23572 (pg_add_dependence_edges): Ditto.
23573 (distribute_loop): Release data dependence hash table.
23575 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23577 * tree-loop-distribution.c (ref_base_address): Delete.
23578 (similar_memory_accesses): Rename ...
23579 (share_memory_accesses): ... to this. Check if partitions access
23580 the same memory reference.
23581 (distribute_loop): Call share_memory_accesses.
23583 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23585 * tree-loop-distribution.c (struct partition): New field recording
23586 its data reference.
23587 (partition_alloc, partition_free): Init and release data refs.
23588 (partition_merge_into): Merge data refs.
23589 (build_rdg_partition_for_vertex): Collect data refs for partition.
23590 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23592 (distribute_loop): Remve data refs from vertice data of partition
23595 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23597 * tree-loop-distribution.c (params.h): Include header file.
23598 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23599 (datarefs_vec): New global var.
23600 (create_rdg_vertices): Use datarefs_vec directly.
23601 (free_rdg): Don't free data references.
23602 (build_rdg): Update use. Don't free data references.
23603 (distribute_loop): Compute global variable for data references.
23604 Bail out if there are too many data references.
23606 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23608 * tree-loop-distribution.c (loop_nest): New global var.
23609 (build_rdg): Use loop directly, rather than loop nest.
23610 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23612 (distribute_loop): Compute global variable loop nest. Update use.
23614 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23616 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23617 (partition_merge_into): New parameter. Dump reason for fusion.
23618 (distribute_loop): Update use of partition_merge_into.
23620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23622 * tree-loop-distribution.c (bb_top_order_index): New.
23623 (bb_top_order_index_size, bb_top_order_cmp): New.
23624 (stmts_from_loop): Use topological order.
23625 (pass_loop_distribution::execute): Compute and release topological
23626 order for basic blocks.
23628 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23630 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23633 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23635 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23636 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23637 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23638 * internal-fn.def (LOOP_DIST_ALIAS): New.
23639 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23640 (fold_loop_internal_call): ... this.
23641 (vect_loop_dist_alias_call): New function.
23642 (set_uid_loop_bbs): Call fold_loop_internal_call.
23643 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23646 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23649 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23650 Require dead FLAGS_REG at the beginning of a peephole.
23652 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23655 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23656 arguments in the call to __builtin_ia32_sbb_u32.
23657 (_subborrow_u64): Swap _X and _Y arguments in the call to
23658 __builtin_ia32_sbb_u64.
23660 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23663 * tree-vrp.c (compare_assert_loc): Turn into a function template
23664 with stable template parameter. Only test if a->e is NULL,
23665 !a->e == !b->e has been verified already. Use e == NULL or
23666 e != NULL instead of e or ! e tests. If stable is true, don't use
23667 iterative_hash_expr, on the other side allow a or b or both NULL
23668 and sort the NULLs last.
23669 (process_assert_insertions): Sort using compare_assert_loc<false>
23670 instead of compare_assert_loc, later sort using
23671 compare_assert_loc<true> before calling process_assert_insertions_for
23672 in a loop. Use break instead of continue once seen NULL pointer.
23674 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23676 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23677 Cortex-R7 and Cortex-R8 processors.
23679 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23681 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23682 uninitialized while src is not.
23684 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23686 * common/config/arm/arm-common.c: Adjust include path for
23688 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23689 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23690 (arm-cpu-data.h): Likewise.
23691 (arm-cpu-cdata.h): Likewise.
23692 * config/arm/arm-cpu.h: Delete.
23693 * config/arm/arm-cpu-cdata.h: Delete.
23694 * config/arm/arm-cpu-data.h: Delete.
23696 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23698 * config/arm/arm-cpus.in (cortex-a55): New.
23699 (cortex-a75): Likewise.
23700 (cortex-a75.cortex-a55): Likewise.
23701 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23703 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23704 * config/arm/arm-cpu-cdata.h: Regenerate.
23705 * config/arm/arm-cpu-data.h: Regenerate.
23706 * config/arm/arm-cpu.h: Regenerate.
23707 * config/arm/arm-tables.opt: Regenerate.
23708 * config/arm/arm-tune.md: Regenerate.
23710 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23712 * haifa-sched.c (sched_create_recovery_edges): Update profile.
23714 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23716 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
23719 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
23721 PR tree-optimization/81292
23722 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
23723 full_string_p, also call adjust_related_strinfos if the adjustment
23724 is simple, otherwise invalidate related strinfos.
23726 2017-07-04 Martin Liska <mliska@suse.cz>
23729 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
23730 newly created variable as DECL_IGNORED_P.
23732 2017-07-04 Martin Liska <mliska@suse.cz>
23735 * ipa-inline.c (inline_small_functions):
23736 Use xstrdup_for_dump.
23738 2017-07-04 Tom de Vries <tom@codesourcery.com>
23740 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23742 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23745 * config/darwin.c (darwin_function_switched_text_sections):
23748 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23750 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23752 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23754 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23756 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23758 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23759 min_profitable_iters, and th as inclusive lower bounds.
23760 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23761 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23762 for min_profitable_iters and min_profitable_estimate.
23763 (vect_transform_loop): Treat th as an inclusive lower bound.
23764 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23766 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23769 * config/darwin.c (darwin_function_switched_text_sections):
23770 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23771 in two pieces, and suppress the use of buf.
23773 2017-07-03 Nathan Sidwell <nathan@acm.org>
23775 * hash-table.h (hash_table_mod1): Fix indentation.
23777 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23779 PR middle-end/81290
23780 * predict.c (force_edge_cold): Be more careful about propagation
23782 * profile-count.h (profile_probability::guessed,
23783 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23785 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23787 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23789 * doc/invoke.texi (rcpc architecture extension): Document it.
23791 2017-07-03 Richard Biener <rguenther@suse.de>
23793 PR tree-optimization/60510
23794 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23795 the scalar reduction PHI and use it.
23796 (vectorizable_reduction): Properly guard the single_defuse_cycle
23797 path for non-SLP reduction chains where we cannot use it.
23798 Rework reduc_def/index and vector type deduction. Rework
23799 vector operand gathering during reduction op code-gen.
23800 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23801 chains dissolve the chain and leave it to non-SLP reduction
23804 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23806 * tree-data-ref.h (dr_alignment): Declare.
23807 * tree-data-ref.c (dr_alignment): New function.
23808 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23809 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23811 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23813 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23815 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23816 and base_misalignment fields.
23817 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23818 * tree-data-ref.c: Include builtins.h.
23819 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23820 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23821 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23822 * tree-vect-data-refs.c: Include tree-cfg.h.
23823 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23824 fields instead of calculating an alignment here.
23825 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23826 innermost_loop_behavior fields.
23828 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23830 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23832 (DR_STEP_ALIGNMENT): New macro.
23833 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23834 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23835 (create_data_ref): Print it.
23836 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23837 to tell whether the step preserves vector (mis)alignment.
23838 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23839 Move the check for an integer step and generalise to all INTEGER_CST.
23840 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23841 Print the outer step alignment.
23843 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23845 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23846 with offset_alignment.
23847 (DR_ALIGNED_TO): Delete.
23848 (DR_OFFSET_ALIGNMENT): New macro.
23849 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23850 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23851 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23852 (create_data_ref): Likewise.
23853 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23854 (vect_analyze_data_refs): Likewise.
23855 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23856 creating dummy innermost behavior.
23858 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23860 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23861 with a "innermost_loop_behavior *" and refeence tree.
23862 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23863 (create_data_ref): Update call accordingly.
23864 * tree-predcom.c (find_looparound_phi): Likewise.
23866 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23868 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23869 fields with dr_wrt_vec_loop.
23870 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23871 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23872 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23873 (vect_dr_behavior): New function.
23874 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23875 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23876 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23877 track whether the step preserves the misalignment.
23878 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23879 Use vect_dr_behavior.
23880 (vect_setup_realignment): Update call accordingly.
23881 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23882 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23883 call to vect_create_addr_base_for_vector_ref.
23884 (vect_create_cond_for_align_checks): Likewise.
23885 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23886 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23887 (vect_recog_mask_conversion_pattern): Likewise.
23888 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23889 (new_stmt_vec_info): Remove redundant zeroing.
23891 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23893 * common/config/arm/arm-common.c (arm_be8_option): New function.
23894 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23895 (ISA_ARMv6): Add isa_bit_be8.
23896 * config/arm/arm.h (arm_be8_option): Add prototype.
23897 (BE8_SPEC_FUNCTION): New define.
23898 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23899 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23900 (mlittle-endian): Similarly.
23901 (mbe8, mbe32): New options.
23902 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23903 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23905 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23907 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23909 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23911 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23912 (cleanup_tree_cfg_bb): Use it.
23913 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23915 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23917 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23920 * loop-doloop.c (add_test): Update profile.
23922 2017-07-03 Martin Liska <mliska@suse.cz>
23925 * sanopt.c (rewrite_usage_of_param): New function.
23926 (sanitize_rewrite_addressable_params): Likewise.
23927 (pass_sanopt::execute): Call rewrite_usage_of_param.
23929 2017-07-03 Richard Biener <rguenther@suse.de>
23931 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23932 back to using VIEW_CONVERT_EXPR.
23934 2017-07-03 Martin Liska <mliska@suse.cz>
23937 * doc/extend.texi: Document when a resolver function is
23938 generated for target_clones.
23940 2017-07-03 Martin Liska <mliska@suse.cz>
23942 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23943 and shadow memory of auto variables which are subject of
23944 use-after-scope sanitization.
23945 (asan_expand_mark_ifn): Add do set only when is_poison.
23947 2016-07-03 Richard Biener <rguenther@suse.de>
23949 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23951 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23952 (vectorizable_reduction): Perform reduction PHI creation when
23953 visiting a reduction PHI and adjust and simplify code generation
23954 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23955 (vect_transform_loop): Visit reduction PHIs.
23956 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23957 defs into the SLP tree.
23958 (vect_build_slp_tree): Reduction defs terminate the recursion.
23959 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23961 (vect_get_vec_defs_for_stmt_copy): Export.
23962 (vect_get_vec_defs): Likewise.
23963 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23965 (vect_get_vec_defs_for_stmt_copy): Declare.
23966 (vect_get_vec_defs): Likewise.
23968 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23970 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23971 parameter with a "loop" parameter and use it instead of the
23972 loop containing DR_STMT. Don't check simple_iv when doing
23973 BB analysis. Describe the two analysis modes in the comment.
23975 2017-07-03 Tom de Vries <tom@codesourcery.com>
23977 PR tree-optimization/69468
23978 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23979 (find_same_succ_bb): Handle ignore_edge_flags.
23981 2017-07-03 Tom de Vries <tom@codesourcery.com>
23983 PR tree-optimization/81192
23984 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23986 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23988 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23990 2017-07-03 Tom de Vries <tom@codesourcery.com>
23992 PR tree-optimization/81192
23993 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23994 BB_SAME_SUCC (bb) == NULL.
23996 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23998 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24001 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24003 * dumpfile.c: Include profile-count.h
24004 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24006 (insert_cond_bb): Update profile.
24007 * tree-cfg.h (insert_cond_bb): Update prototype.
24008 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24009 * tree-dump.c: Do not include tree-cfg.
24011 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24013 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24015 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24017 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24020 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24022 * tree-complex.c (expand_complex_div_wide): update profile.
24024 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24025 Alan Hayward <alan.hayward@arm.com>
24026 David Sherwood <david.sherwood@arm.com>
24028 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24029 (CORETYPES_H): New define.
24030 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24031 (insn-modes-inline.h, s-modes-inline-h): New rules.
24032 (generated_files): Add insn-modes-inline.h.
24033 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24034 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24035 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24036 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24037 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24038 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24039 (build/gencodes.o, build/genconditions.o): Likewise.
24040 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24041 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24042 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24043 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24044 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24045 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24046 * coretypes.h: Include everything up to real.h for generators.
24047 Include insn-modes.h first. Include wide-int-print.h after
24048 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24049 * machmode.h: Don't include insn-modes.h here.
24050 * function-tests.c: Remove includes of signop.h, machmode.h,
24051 double-int.h and wide-int.h.
24053 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24055 * optc-save-gen.awk: Likewise.
24056 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24057 * godump.c: Remove include of wide-int-print.h.
24058 * pretty-print.h: Likewise.
24059 * wide-int-print.cc: Likewise.
24060 * wide-int.cc: Likewise.
24061 * hash-map-tests.c: Remove include of signop.h.
24062 * hash-set-tests.c: Likewise.
24063 * rtl-tests.c: Likewise.
24064 * mkconfig.sh: Remove include of machmode.h.
24065 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24067 (emit_insn_modes_inline_h): ...this new function. Emit the code
24068 into an insn-modes-inline.h header file, adding appropriate
24069 include guards and end comments.
24070 (emit_insn_modes_c_header): Remove include of machmode.h.
24071 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24073 (main): Handle -i flag and call emit_insn_modes_inline_h when
24076 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24078 * tree-ssa-strlen.c (strinfo): Rename the length field to
24079 nonzero_chars. Add a full_string_p field.
24080 (compare_nonzero_chars, zero_length_string_p): New functions.
24081 (get_addr_stridx): Add an offset_out parameter.
24082 Use compare_nonzero_chars.
24083 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24084 (new_strinfo): Update after above changes to strinfo.
24085 (set_endptr_and_length): Set full_string_p.
24086 (get_string_length): Update after above changes to strinfo.
24087 (unshare_strinfo): Update call to new_strinfo.
24088 (maybe_invalidate): Likewise.
24089 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24090 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24091 as a uhwi instead of an shwi. Update after above changes to
24092 strinfo and new_strinfo.
24093 (zero_length_string): Assert that chainsi contains full strings.
24094 Use zero_length_string_p. Update call to new_strinfo.
24095 (adjust_related_strinfos): Update after above changes to strinfo.
24096 Copy full_string_p from origsi.
24097 (adjust_last_stmt): Use zero_length_string_p.
24098 (handle_builtin_strlen): Update after above changes to strinfo and
24099 new_strinfo. Install the lhs as the string length if the previous
24100 entry didn't describe a full string.
24101 (handle_builtin_strchr): Update after above changes to strinfo
24103 (handle_builtin_strcpy): Likewise.
24104 (handle_builtin_strcat): Likewise.
24105 (handle_builtin_malloc): Likewise.
24106 (handle_pointer_plus): Likewise.
24107 (handle_builtin_memcpy): Likewise. Track nonzero characters
24108 that aren't necessarily followed by a nul terminator.
24109 (handle_char_store): Likewise.
24111 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24113 PR tree-optimization/80769
24114 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24115 for malloc and calloc. Document the new invariant that all related
24116 strinfos have delayed lengths or none do.
24117 (verify_related_strinfos): Move earlier in file.
24118 (set_endptr_and_length): New function, split out from...
24119 (get_string_length): ...here. Also set the lengths of related
24121 (zero_length_string): Assert that chainsi has known (rather than
24123 (adjust_related_strinfos): Likewise.
24125 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24127 PR tree-optimization/81136
24128 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24129 assert that two references with the same misalignment have the same
24130 compile-time misalignment if those compile-time misalignments
24133 2017-07-01 Andi Kleen <ak@linux.intel.com>
24135 * print-tree.c (print_node): Print all attributes.
24137 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24139 * cfg.c (scale_bbs_frequencies): New function.
24140 * cfg.h (scale_bbs_frequencies): Declare it.
24141 * cfgloopanal.c (single_likely_exit): Cleanup.
24142 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24144 (scale_loop_profile): Likewise.
24145 (loop_version): Likewise.
24146 (create_empty_loop_on_edge): Update.
24147 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24148 scale_loop_frequencies, scale_loop_profile, loopify,
24149 loop_version): Update prototypes.
24150 * modulo-sched.c (sms_schedule): Update.
24151 * predict.c (unlikely_executed_edge_p): Also check probability.
24152 (probably_never_executed_edge_p): Fix typo.
24153 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24154 * tree-parloops.c (gen_parallel_loop): Update.
24155 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24156 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24157 * tree-ssa-loop-split.c (split_loop): Update.
24158 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24159 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24160 (vect_loop_versioning): Update.
24161 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24163 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24165 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24167 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24169 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24170 to keep profile consistent.
24172 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24174 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24175 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24176 * profile-count.h (max_safe_multiplier): Make unsigned.
24177 (profile_count::guessed_zero): New.
24179 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24181 * bb-reorder.c (fix_up_crossing_landing_pad,
24182 fix_crossing_conditional_branches): Use make_single_succ_edge
24183 to keep profile consistent.
24185 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24187 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24190 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24193 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24194 the right scopes, make sure cond_jump isn't preserved between multiple
24195 iterations. Search for fallthru edge whenever there are 3+ edges and
24196 use find_fallthru_edge for it.
24198 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24200 Patch by Alexander Monakov <amonakov@ispras.ru>
24201 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24202 probabilities consistently.
24204 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24206 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24207 emit_cmp_and_jump_insns.
24209 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24212 * tree-inline.c (expand_call_inline): Combine profile statuses.
24214 2017-06-30 Andrew Pinski <apinski@cavium.com>
24216 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24217 fold_stmt returned true.
24219 2017-06-30 Nathan Sidwell <nathan@acm.org>
24221 * ggc.h (empty_string): Delete.
24222 * cfgexpand.c (expand_asm_stmt): Use plain "".
24223 * optabs.c (expand_asm_memory_barrier): Likewise.
24224 * stringpool.c (empty_string): Delete.
24225 (digit_vector, digit_string): Delete.
24226 (ggc_alloc_string): Use plain "", don't optimize single digit
24227 strings. Use ggc_alloc_atomic.
24229 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24231 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24232 comparison set and one other set, use the cost of the non-comparison
24235 2017-06-30 Nathan Sidwell <nathan@acm.org>
24237 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24240 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24242 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24243 loops. Remove now unneeded calls to gimple_switch_set_label() that
24244 just set removed labels to NULL_TREE.
24246 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24248 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24249 (set_range_info): ...here. Only call set_range_info_raw if domain
24251 (set_nonzero_bits): Call set_range_info_raw.
24252 * tree-ssanames.h (set_range_info_raw): New.
24254 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24257 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24258 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24259 of nonimmediate_operand and <store_mask_constraint> instead of m
24260 for the input operand. For V8FI iterator, always split if input
24261 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24262 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24263 <store_mask_predicate> instead of register_operand and
24264 <store_mask_constraint> instead of v for the input operand. Make
24265 sure both operands aren't MEMs for if not <mask_applied>.
24267 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24269 * lto-wrapper.c (copy_file) Close both file descriptors before
24272 2017-06-30 Martin Liska <mliska@suse.cz>
24275 * multiple_target.c (create_dispatcher_calls): Make ifunc
24276 also for function that don't have calls or are not referenced.
24278 2017-06-30 Richard Biener <rguenther@suse.de>
24280 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24281 analyze the first scalar stmt. Move vector type computation
24282 for the BB case here from ...
24283 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24284 live operation processing in the SLP case properly.
24286 2017-06-30 Richard Biener <rguenther@suse.de>
24288 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24290 2017-06-30 Martin Liska <mliska@suse.cz>
24293 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24294 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24296 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24298 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24301 2017-06-30 Richard Biener <rguenther@suse.de>
24303 PR tree-optimization/81249
24304 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24305 condition reduction result to original scalar type.
24307 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24309 * profile-count.h (enum profile_quality): Fix typos and whitespace
24312 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24314 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24315 type for branch probabilities.
24317 2017-06-29 Julian Brown <julian@codesourcery.com>
24318 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24320 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24321 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24322 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24323 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24325 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24327 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24328 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24329 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24330 CC usage from generic code to here.
24331 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24332 CC usage into the target macros.
24334 2017-06-29 Maya Rashish <coypu@sdf.org>
24336 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24339 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24341 * arm/arm-builtins.c: Include profile-count.h
24342 * except.c (sjlj_emit_function_enter): Use
24343 profile_probability::unlikely.
24345 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24347 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24348 and tocrel_offset be pointer args rather than implicitly using
24350 (legitimate_constant_pool_address_p, rs6000_emit_move,
24351 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24352 tocrel_offset and use in toc_relative_expr_p call.
24353 (print_operand, print_operand_address): Use static tocrel_base_oac
24354 and tocrel_offset_oac.
24355 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24358 2017-06-29 Maya Rashish <coypu@sdf.org>
24360 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24362 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24364 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24365 objects, take into account only the alignment of 'op0' and 'mode1' if
24368 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24370 * ccmp.c (ccmp_tree_comparison_p): New function.
24371 (ccmp_candidate_p): Update to use above function.
24372 (get_compare_parts): New function.
24373 (expand_ccmp_next): Update to use new functions.
24374 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24376 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24377 take mode as argument.
24378 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24379 * expr.c (expand_expr_real_1): Pass mode as argument.
24381 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24383 * combine.c (combine_instructions): Print insns to dump_file, together
24386 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24388 * asan.c (asan_emit_stack_protection): Update.
24389 (create_cond_insert_point): Update.
24390 * auto-profile.c (afdo_propagate_circuit): Update.
24391 * basic-block.h (struct edge_def): Turn probability to
24392 profile_probability.
24393 (EDGE_FREQUENCY): Update.
24394 * bb-reorder.c (find_traces_1_round): Update.
24395 (better_edge_p): Update.
24396 (sanitize_hot_paths): Update.
24397 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24398 (make_single_succ_edge): Update.
24399 (check_bb_profile): Update.
24400 (dump_edge_info): Update.
24401 (update_bb_profile_for_threading): Update.
24402 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24403 probabilitycount to 0.
24404 * cfgbuild.c (compute_outgoing_frequencies): Update.
24405 * cfgcleanup.c (try_forward_edges): Update.
24406 (outgoing_edges_match): Update.
24407 (try_crossjump_to_edge): Update.
24408 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24409 (expand_gimple_tailcall): Update.
24410 (construct_init_block): Use make_single_succ_edge.
24411 (construct_exit_block): Use make_single_succ_edge.
24412 * cfghooks.c (verify_flow_info): Update.
24413 (redirect_edge_succ_nodup): Update.
24414 (split_edge): Update.
24415 (account_profile_record): Update.
24416 * cfgloopanal.c (single_likely_exit): Update.
24417 * cfgloopmanip.c (scale_loop_profile): Update.
24418 (set_zero_probability): Remove.
24419 (duplicate_loop_to_header_edge): Update.
24420 * cfgloopmanip.h (loop_version): Update prototype.
24421 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24422 (force_nonfallthru_and_redirect): Update.
24423 (update_br_prob_note): Update.
24424 (rtl_verify_edges): Update.
24425 (purge_dead_edges): Update.
24426 (rtl_lv_add_condition_to_bb): Update.
24427 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24428 * cgraphunit.c (init_lowered_empty_function): Update.
24429 (cgraph_node::expand_thunk): Update.
24430 * cilk-common.c: Include profile-count.h
24431 * dojump.c (inv): Remove.
24432 (jumpifnot): Update.
24433 (jumpifnot_1): Update.
24434 (do_jump_1): Update.
24436 (do_jump_by_parts_greater_rtx): Update.
24437 (do_compare_rtx_and_jump): Update.
24438 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24439 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24440 * dwarf2cfi.c: Include profile-count.h
24441 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24442 (sjlj_emit_dispatch_table): Likewise.
24443 * explow.c: Include profile-count.h
24444 * expmed.c (emit_store_flag_force): Update.
24445 (do_cmp_and_jump): Update.
24446 * expr.c (compare_by_pieces_d::generate): Update.
24447 (compare_by_pieces_d::finish_mode): Update.
24448 (emit_block_move_via_loop): Update.
24449 (store_expr_with_bounds): Update.
24450 (store_constructor): Update.
24451 (expand_expr_real_2): Update.
24452 (expand_expr_real_1): Update.
24453 * expr.h (try_casesi, try_tablejump): Update prototypes.
24454 * gimple-pretty-print.c (dump_probability): Update.
24455 (dump_profile): New.
24456 (dump_gimple_label): Update.
24457 (dump_gimple_bb_header): Update.
24458 * graph.c (draw_cfg_node_succ_edges): Update.
24459 * hsa-gen.c (convert_switch_statements): Update.
24460 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24461 (find_if_case_1): Update.
24462 (find_if_case_2): Update.
24463 * internal-fn.c (expand_arith_overflow_result_store): Update.
24464 (expand_addsub_overflow): Update.
24465 (expand_neg_overflow): Update.
24466 (expand_mul_overflow): Update.
24467 (expand_vector_ubsan_overflow): Update.
24468 * ipa-cp.c (good_cloning_opportunity_p): Update.
24469 * ipa-split.c (split_function): Use make_single_succ_edge.
24470 * ipa-utils.c (ipa_merge_profiles): Update.
24471 * loop-doloop.c (add_test): Update.
24472 (doloop_modify): Update.
24473 * loop-unroll.c (compare_and_jump_seq): Update.
24474 (unroll_loop_runtime_iterations): Update.
24475 * lra-constraints.c (lra_inheritance): Update.
24476 * lto-streamer-in.c (input_cfg): Update.
24477 * lto-streamer-out.c (output_cfg): Update.
24478 * mcf.c (adjust_cfg_counts): Update.
24479 * modulo-sched.c (sms_schedule): Update.
24480 * omp-expand.c (expand_omp_for_init_counts): Update.
24481 (extract_omp_for_update_vars): Update.
24482 (expand_omp_ordered_sink): Update.
24483 (expand_omp_for_ordered_loops): Update.
24484 (expand_omp_for_generic): Update.
24485 (expand_omp_for_static_nochunk): Update.
24486 (expand_omp_for_static_chunk): Update.
24487 (expand_cilk_for): Update.
24488 (expand_omp_simd): Update.
24489 (expand_omp_taskloop_for_outer): Update.
24490 (expand_omp_taskloop_for_inner): Update.
24491 * omp-simd-clone.c (simd_clone_adjust): Update.
24492 * optabs.c (expand_doubleword_shift): Update.
24493 (expand_abs): Update.
24494 (emit_cmp_and_jump_insn_1): Update.
24495 (expand_compare_and_swap_loop): Update.
24496 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24497 * predict.c (predictable_edge_p): Update.
24498 (edge_probability_reliable_p): Update.
24499 (set_even_probabilities): Update.
24500 (combine_predictions_for_insn): Update.
24501 (combine_predictions_for_bb): Update.
24502 (propagate_freq): Update.
24503 (estimate_bb_frequencies): Update.
24504 (force_edge_cold): Update.
24505 * profile-count.c (profile_count::dump): Add missing space into dump.
24506 (profile_count::debug): Add newline.
24507 (profile_count::differs_from_p): Explicitly convert to unsigned.
24508 (profile_count::stream_in): Update.
24509 (profile_probability::dump): New member function.
24510 (profile_probability::debug): New member function.
24511 (profile_probability::differs_from_p): New member function.
24512 (profile_probability::differs_lot_from_p): New member function.
24513 (profile_probability::stream_in): New member function.
24514 (profile_probability::stream_out): New member function.
24515 * profile-count.h (profile_count_quality): Rename to ...
24516 (profile_quality): ... this one.
24517 (profile_probability): New.
24518 (profile_count): Update.
24519 * profile.c (compute_branch_probabilities): Update.
24520 * recog.c (peep2_attempt): Update.
24521 * sched-ebb.c (schedule_ebbs): Update.
24522 * sched-rgn.c (find_single_block_region): Update.
24523 (compute_dom_prob_ps): Update.
24524 (schedule_region): Update.
24525 * sel-sched-ir.c (compute_succs_info): Update.
24526 * stmt.c (struct case_node): Update.
24527 (do_jump_if_equal): Update.
24528 (get_outgoing_edge_probs): Update.
24529 (conditional_probability): Update.
24530 (emit_case_dispatch_table): Update.
24531 (expand_case): Update.
24532 (expand_sjlj_dispatch_table): Update.
24533 (emit_case_nodes): Update.
24534 * targhooks.c: Update.
24535 * tracer.c (better_p): Update.
24536 (find_best_successor): Update.
24537 * trans-mem.c (expand_transaction): Update.
24538 * tree-call-cdce.c: Update.
24539 * tree-cfg.c (gimple_split_edge): Upate.
24540 (move_sese_region_to_fn): Upate.
24541 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24542 * tree-eh.c (lower_resx): Upate.
24543 (cleanup_empty_eh_move_lp): Upate.
24544 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24545 * tree-inline.c (copy_edges_for_bb): Update.
24546 (copy_cfg_body): Update.
24547 * tree-parloops.c (gen_parallel_loop): Update.
24548 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24549 (gimple_gen_time_profiler): Update.
24550 * tree-ssa-dce.c (remove_dead_stmt): Update.
24551 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24552 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24553 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24554 (unloop_loops): Update.
24555 (try_peel_loop): Update.
24556 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24557 * tree-ssa-loop-split.c (connect_loops): Update.
24558 (split_loop): Update.
24559 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24560 (hoist_guard): Update.
24561 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24562 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24563 (value_replacement): Update.
24564 * tree-ssa-reassoc.c (branch_fixup): Update.
24565 * tree-ssa-tail-merge.c (replace_block_by): Update.
24566 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24567 (create_edge_and_update_destination_phis): Update.
24568 (compute_path_counts): Update.
24569 (recompute_probabilities): Update.
24570 (update_joiner_offpath_counts): Update.
24571 (freqs_to_counts_path): Update.
24572 (duplicate_thread_path): Update.
24573 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24574 (struct switch_conv_info): Update.
24575 (gen_inbound_check): Update.
24576 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24577 (vect_do_peeling): Update.
24578 (vect_loop_versioning): Update.
24579 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24580 (optimize_mask_stores): Update.
24581 * ubsan.c (ubsan_expand_null_ifn): Update.
24582 * value-prof.c (gimple_divmod_fixed_value): Update.
24583 (gimple_divmod_fixed_value_transform): Update.
24584 (gimple_mod_pow2): Update.
24585 (gimple_mod_pow2_value_transform): Update.
24586 (gimple_mod_subtract): Update.
24587 (gimple_mod_subtract_transform): Update.
24588 (gimple_ic): Update.
24589 (gimple_stringop_fixed_value): Update.
24590 (gimple_stringops_transform): Update.
24591 * value-prof.h: Update.
24593 2017-06-29 Carl Love <cel@us.ibm.com>
24595 * config/rs6000/rs6000-c.c: Add support for built-in functions
24596 vector signed int vec_signed (vector float);
24597 vector signed long long vec_signed (vector double);
24598 vector signed int vec_signed2 (vector double, vector double);
24599 vector signed int vec_signede (vector double);
24600 vector signed int vec_signedo (vector double);
24601 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24602 instruction generator.
24603 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24604 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24605 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24607 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24608 vunsignede_v2df): Add define_expands.
24609 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24610 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24611 VEC_UNSIGNEDO): Add definitions.
24612 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24613 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24614 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24615 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24616 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24617 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24618 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24619 vec_unsignede, vec_unsignedo): Add builtin defines.
24620 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24622 * doc/extend.texi: Update the built-in documentation file for the
24623 new built-in functions.
24625 2017-06-29 Richard Biener <rguenther@suse.de>
24627 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24628 reduction chains to LOOP_VINFO_REDUCTIONS.
24629 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24630 SLP reductions after processing reduction chains.
24632 2017-06-29 Nathan Sidwell <nathan@acm.org>
24634 * builtins.c (fold_builtin_FUNCTION): Use
24635 lang_hooks.decl_printable_name.
24637 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24639 PR middle-end/81194
24640 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24641 with only one label.
24642 * stmt.c (expand_case): Assert NCASES is greater than one.
24644 2017-06-29 Richard Biener <rguenther@suse.de>
24646 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24648 (group_case_labels): Likewise.
24649 (find_taken_edge): Push sanity checking on val to workers...
24650 (find_taken_edge_cond_expr): ... here
24651 (find_taken_edge_switch_expr): ... and here, handle cases
24652 with just a default label.
24653 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24654 (group_case_labels): Likewise.
24655 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24656 group_case_labels does anything cleanup the CFG again.
24658 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24660 PR tree-optimization/81196
24661 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24662 exit condition comparing two IVs.
24664 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24666 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24667 profile to the dummy entry at the end of the list of architectures.
24668 * config/arm/arm-cpu-cdata.h: Regenerated.
24670 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24671 Michael Collison <michael.collison@arm.com>
24674 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24676 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24677 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24678 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24679 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24680 of shift when the shift amount is masked with constant equal to
24681 the size of the mode.
24682 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24685 2017-06-29 Martin Liska <mliska@suse.cz>
24687 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24689 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24692 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24695 (edge_growth_cache_entry::edge_growth_cache_entry): New
24697 (reset_edge_growth_cache): Update to use constructor.
24699 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24701 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24702 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24703 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24705 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24707 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24708 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24710 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24712 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
24713 (*-linux-uclibc*): Add t-uclibc tmake_file.
24714 * config/t-musl: New.
24715 * config/t-uclibc: New.
24717 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
24719 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
24721 (gen_comm_data): Emit architectural setting of arch_prof.
24722 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
24724 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
24725 (armv8-m.base, armv8-m.main): Likewise.
24726 * arm-protos.h (arm_build_target): Add profile field.
24727 (arch_option): Likewise.
24728 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
24730 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
24731 arm_active_target.profile.
24733 2017-06-28 Richard Biener <rguenther@suse.de>
24735 PR middle-end/81227
24736 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24737 TYPE_OVERFLOW_WRAPS.
24738 * match.pd (negate_expr_p): Likewise.
24739 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24740 fold_build2, not fold_binary.
24742 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24744 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24745 Convert memory address to Pmode.
24746 (aarch64_print_operand): Assert MEM operands are always Pmode.
24748 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24751 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24752 Remove redundant if.
24753 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24754 * config/arm/aarch-common-protos.h
24755 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24756 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24758 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24761 * multiple_target.c (create_dispatcher_calls): Set the default
24762 clone to be static, not public.
24764 2017-06-28 Richard Biener <rguenther@suse.de>
24766 * tree-vect-loop.c (vectorizable_reduction): Move special
24767 cond reduction IV var creation ...
24768 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24769 parameter. Use STMT_VINFO_VECTYPE.
24770 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24773 2017-06-28 Martin Liska <mliska@suse.cz>
24776 * ipa-visibility.c (non_local_p): Handle visibility.
24778 2017-06-28 Martin Liska <mliska@suse.cz>
24781 * common.opt: Add IntegerRange to various options.
24782 * opt-functions.awk (integer_range_info): New function.
24783 * optc-gen.awk: Add integer_range_info to cl_options struct.
24784 * opts-common.c (decode_cmdline_option): Handle
24785 CL_ERR_INT_RANGE_ARG.
24786 (cmdline_handle_error): Likewise.
24787 * opts.c (print_filtered_help): Show valid interval in
24788 when --help is provided.
24789 * opts.h (struct cl_option): Add range_min and range_max fields.
24790 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24792 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24794 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24795 (x * C EQ/NE y * C): New transformation.
24797 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24799 * genmultilib (combination_space): Accept '+' in option names.
24801 2017-06-28 Martin Liska <mliska@suse.cz>
24804 * asan.c (instrument_derefs): Bail out inner references
24805 that are hard register variables.
24807 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24810 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24811 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24813 2017-06-28 Richard Biener <rguenther@suse.de>
24815 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24816 (vect_get_slp_defs): Adjust.
24817 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24819 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24821 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24822 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24823 (vectorizable_reduction): Adjust.
24824 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24826 (vect_get_slp_defs): Likewise.
24827 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24828 (vectorizable_bswap): Adjust.
24829 (vectorizable_call): Likewise.
24830 (vectorizable_conversion): Likewise.
24831 (vectorizable_assignment): Likewise.
24832 (vectorizable_shift): Likewise.
24833 (vectorizable_operation): Likewise.
24834 (vectorizable_store): Likewise.
24835 (vectorizable_condition): Likewise.
24836 (vectorizable_comparison): Likewise.
24838 2017-06-28 Michael Collison <michael.collison@arm.com>
24841 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24843 (arm_gen_movmemqi): Removed unused variable 'i'.
24844 Convert 'for' loop into 'while' loop.
24845 (arm_expand_prologue): Remove last unnecessary set of insn.
24846 (thumb_pop): Remove unused variable 'pushed_words'.
24847 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24849 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24851 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24852 * config/s390/s390-protos.h: Add prototype of
24853 s390_rel_address_ok_p.
24854 * config/s390/s390.c (s390_got_symbol): New function.
24855 (s390_rel_address_ok_p): New function.
24856 (legitimize_pic_address): Use s390_rel_address_ok_p.
24857 (s390_load_got): Use s390_got_symbol.
24858 (s390_option_override): Issue error if
24859 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24860 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24862 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24864 2017-06-27 Andrew Pinski <apinski@cavium.com>
24866 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24867 (X * copysign (1.0, X)): New pattern.
24868 (X * copysign (1.0, -X)): New pattern.
24869 (copysign (-1.0, CST)): New pattern.
24871 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24873 * genmultilib (combination_space): Remove variable.
24874 Validate reuse rules against regular expression for any sequence
24875 of multilib options in any order.
24877 2017-06-27 Michael Collison <michael.collison@arm.com>
24879 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24880 call aarch64_split_simd_combine.
24881 * (aarch64_combine_internal<mode>): Delete pattern.
24882 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24883 Allow register and subreg operands.
24885 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24887 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24888 specific need, just fallback on defaults.
24889 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24891 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24892 Olivier Hainque <hainque@adacore.com>
24894 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24896 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24897 targets. Pick a default if no particular attempt applied.
24898 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24901 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24903 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24904 (x86_64-wrs-vxworks7): Likewise.
24906 2017-06-27 Marek Polacek <polacek@redhat.com>
24909 * ubsan.c (instrument_null): Check get_base_address's result for null.
24911 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24913 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24915 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24917 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24918 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24919 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24920 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24921 New function types.
24922 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24923 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24924 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24925 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24926 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24927 BUILT_IN_FEUPDATEENV): New builtins.
24928 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24929 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24930 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24931 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24933 (builtin_structptr_types): Adjust size.
24934 * tree.c (builtin_structptr_types): Add four entries.
24936 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24937 Olivier Hainque <hainque@adacore.com>
24939 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24940 (TLS_SYM): New local macro, forcing reference to __tls__ on
24941 link command lines for VxWorks 7 RTPs, triggering initialization
24943 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24944 OS features TLS support, true for RTPs on VxWorks 7.
24945 * config/vxworks.c (vxworks_override_options): Setup emutls
24948 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24950 * predict.c (test_prediction_value_range): Use -1U instead of -1
24951 to avoid narrowing conversion warning.
24952 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24953 to avoid narrowing conversion warning.
24954 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24956 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24958 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24960 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24961 64bit configurations.
24962 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24963 (SIZE_TYPE): Likewise.
24964 * config/vxworks.c (vxworks_emutls_var_fields): Use
24965 long_unsigned_type_node instead of unsigned_type_node as the offset
24966 field type, which is "pointer" mode in emutls.c.
24968 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24971 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24973 PR middle-end/81207
24974 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24975 gimple_vuse copying separately from gimple_vdef copying.
24977 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24979 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24982 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24983 Olivier Hainque <hainque@adacore.com>
24985 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24986 for all vxworks7 targets.
24987 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24988 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24989 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24990 variations for VX6/VX7 and 32/64bits later on in ...
24991 (VXWORKS_LIB_SPEC): Leverage new macros.
24992 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24993 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24995 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24997 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24998 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25000 2017-06-26 Carl Love <cel@us.ibm.com>
25002 * config/rs6000/rs6000-c.c: Add support for built-in functions
25003 vector bool char vec_reve (vector bool char);
25004 vector signed char vec_reve (vector signed char);
25005 vector unsigned char vec_reve (vector unsigned char);
25006 vector bool int vec_reve (vector bool int);
25007 vector signed int vec_reve (vector signed int);
25008 vector unsigned int vec_reve (vector unsigned int);
25009 vector bool long long vec_reve (vector bool long long);
25010 vector signed long long vec_reve (vector signed long long);
25011 vector unsigned long long vec_reve (vector unsigned long long);
25012 vector bool short vec_reve (vector bool short);
25013 vector signed short vec_reve (vector signed short);
25014 vector double vec_reve (vector double);
25015 vector float vec_reve (vector float);
25016 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25017 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25018 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25019 (altivec_vreve): New pattern.
25020 * config/rs6000/altivec.h (vec_reve): New define.
25021 * doc/extend.texi (vec_rev): Update the built-in documentation file
25022 for the new built-in functions.
25024 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25026 PR tree-optimization/71815
25027 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25029 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25031 (slsr_process_phi): Likewise.
25032 (replace_uncond_cands_and_profitable_phis): Don't replace a
25033 multiply candidate with a stride of 1 (copy or cast).
25034 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25036 (lowest_cost_path): Likewise.
25037 (total_savings): Likewise.
25039 2017-06-26 Richard Biener <rguenther@suse.de>
25042 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25043 Use def_builtin_pure for all gather builtins.
25045 2017-06-26 Richard Biener <rguenther@suse.de>
25047 PR tree-optimization/81203
25048 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25049 non-dominating BBs.
25051 2017-06-26 Marek Polacek <polacek@redhat.com>
25054 * doc/invoke.texi: Document -Wmultistatement-macros.
25056 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25058 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25059 arm_neon_ok_no_float_abi effective target.
25061 2017-06-26 Richard Biener <rguenther@suse.de>
25063 PR tree-optimization/80928
25064 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25065 (copy_bbs): Set BB_DUPLICATED flag early.
25066 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25068 (execute_on_shrinking_pred): Likewise.
25069 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25070 BB_DUPLICATED blocks.
25071 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25072 iterate over all PHIs considering removal of *gsi.
25074 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25076 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25079 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25081 * config/rs6000/rs6000-string.c: (expand_block_clear,
25082 do_load_for_compare, select_block_compare_mode,
25083 compute_current_alignment, expand_block_compare,
25084 expand_strncmp_align_check, expand_strn_compare,
25085 expand_block_move, rs6000_output_load_multiple)
25086 Move functions related to string/block move/compare
25087 to a separate file.
25088 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25089 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25090 for this function which is now used in two files.
25091 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25092 * config.gcc: Add rs6000-string.o to extra_objs for
25093 targets powerpc*-*-* and rs6000*-*-*.
25095 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25098 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25099 32-bit, since indexed is not valid for DImode.
25100 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25101 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25102 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25103 (define_peephole2 for Altivec d-form store): Likewise.
25106 * multiple_target.c (create_dispatcher_calls): Only create the
25107 dispatcher call if the function is the default clone of a
25108 versioned function.
25110 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25112 PR middle-end/80902
25113 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25114 a call, force the call to not be a tail call.
25116 2017-06-23 Jeff Law <law@redhat.com>
25118 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25121 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25123 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25124 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25125 handling for early expansion of vector shifts (sl,sr,sra,rl).
25126 (builtin_function_type): Add vector shift right instructions
25127 to the unsigned argument list.
25129 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25131 rtl-optimizatoin/79286
25132 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25133 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25134 trap. PIC register plus a const unspec without offset can never trap.
25136 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25138 * tree.h (builtin_structptr_type): New type.
25139 (builtin_structptr_types): Declare new array.
25140 * tree.c (builtin_structptr_types): New array.
25141 (free_lang_data, build_common_tree_nodes): Use it.
25143 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25146 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25149 2017-06-22 Matt Turner <mattst88@gmail.com>
25151 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25152 Lake models to skylake case. Assume skylake for unknown
25153 models with clflushopt.
25155 2017-06-22 Jeff Law <law@redhat.com>
25157 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25158 frame sizes that do not satisfy aarch64_uimm12_shift.
25160 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25162 * profile-count.h (apply_probability,
25163 apply_scale, probability_in): Fix checks for zero.
25165 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25167 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25168 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25170 2016-06-22 Richard Biener <rguenther@suse.de>
25172 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25173 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25174 REDUC_MAX_EXPR support.
25175 (vectorizable_reduction): Likewise.
25176 (vect_create_epilog_for_reduction): Likewise.
25178 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25180 * match.pd (A / (1 << B) -> A >> B): New.
25181 * generic-match-head.c: Include optabs-tree.h.
25182 * gimple-match-head.c: Likewise.
25183 * optabs-tree.h (target_supports_op_p): New.
25184 * optabs-tree.c (target_supports_op_p): New.
25186 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25188 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25189 $gcc_cv_ld --help output.
25190 (gcc_cv_ld_demangle): Likewise.
25191 (gcc_cv_ld_eh_frame_hdr): Likewise.
25192 (gcc_cv_ld_pie): Likewise.
25193 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25194 (gcc_cv_ld_buildid): Likewise.
25195 (gcc_cv_ld_sysroot): Likewise.
25196 (ld_bndplt_support): Likewise.
25197 (ld_pushpopstate_support): Likewise.
25198 * configure: Regenerate.
25199 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25201 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25204 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25205 operands indexes to avoid gap between operands and match_dups.
25207 2017-06-21 Andrew Pinski <apinski@cavium.com>
25209 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25210 Increment Arith_shift and Arith_shift_reg by 1.
25211 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25213 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25214 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25215 (aarch64_strip_extend): Add new argument and test for it.
25216 (aarch64_cheap_mult_shift_p): New function.
25217 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25218 add a cost if it is true.
25219 Update calls to aarch64_strip_extend.
25220 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25222 2017-06-21 Andrew Pinski <apinski@cavium.com>
25224 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25226 (thunderxt88): Likewise.
25227 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25228 (thunderx_prefetch_tune): New variable.
25229 (thunderx2t99_prefetch_tune): Update for the correct values.
25230 (thunderxt88_tunings): New variable.
25231 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25232 generic_prefetch_tune.
25233 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25235 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25237 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25238 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25239 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25240 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25241 (aarch64_atomic_cas<mode>, GPI): Likewise.
25243 2017-06-21 Martin Liska <mliska@suse.cz>
25245 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25246 statements on cold and hot labels.
25247 * predict.c (tree_estimate_probability_bb): Remove the
25248 prediction from this place.
25250 2017-06-21 Martin Liska <mliska@suse.cz>
25252 PR tree-optimization/79489
25253 * gimplify.c (maybe_add_early_return_predict_stmt): New
25255 (gimplify_return_expr): Call the function.
25256 * predict.c (tree_estimate_probability_bb): Remove handling
25258 * predict.def: Update comment about early return predictor.
25259 * gimple-predict.h (is_gimple_predict): New function.
25260 * predict.def: Change default value of early return to 66.
25261 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25263 * passes.def: Put pass_strip_predict_hints to the beginning of
25266 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25268 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25269 FUNCTION_DECL declarations.
25270 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25272 (dwaf2out_decl): Likewise.
25273 * godump.c (go_early_global_decl): Skip call to the real debug hook
25274 for FUNCTION_DECL declarations.
25275 * passes.c (rest_of_decl_compilation): Skip call to the
25276 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25277 -fdump-go-spec is passed.
25279 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25281 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25282 Reorder for compactness.
25283 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25284 (def_builtin_pure, def_builtin_pure2): New functions.
25285 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25287 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25289 * match.pd (nop_convert): New predicate.
25290 ((A +- CST1) +- CST2): Allow some NOP conversions.
25292 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25295 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25296 with ctors/dtors if GOVD_SHARED is set.
25298 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25300 * config/aarch64/aarch64.md (movti_aarch64):
25301 Emit mov rather than orr.
25302 (movtf_aarch64): Likewise.
25303 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25304 Emit mov rather than orr.
25306 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25308 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25309 Swap alternatives, make integer dup more expensive.
25311 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25313 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25314 Return true for non-tls symbols.
25316 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25318 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25319 (cortex-a75): Likewise.
25320 (cortex-a75.cortex-a55): Likewise.
25321 * config/aarch64/aarch64-tune.md: Regenerate.
25322 * doc/invoke.texi (-mtune): Document new values for -mtune.
25324 2017-06-21 Tom de Vries <tom@codesourcery.com>
25326 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25327 stack_size feature.
25328 (Effective-Target Keywords, Other attributes): Suggest using
25329 dg-add-options stack_size feature to get stack limit in stack_size
25330 effective target documentation.
25332 2017-06-21 Julian Brown <julian@codesourcery.com>
25333 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25335 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25336 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25337 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25339 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25340 attribute type list for neon_multiply.
25341 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25342 attribute type list for neon_multiply.
25343 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25344 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25345 attribute type list for neon_multiply.
25346 * config/arm/types.md (crypto_pmull): Add.
25347 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25348 attribute type list.
25350 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25352 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25355 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25357 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25358 to make sure not to dereference a NULL cost_classes_ptr pointer.
25360 2017-06-20 Carl Love <cel@us.ibm.com>
25362 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25363 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25364 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25365 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25366 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25367 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25368 VMULOSW): New enum "unspec" values.
25369 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25370 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25371 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25372 altivec_vmulosw): New patterns.
25373 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25374 VMULOSW): Add definitions.
25376 2017-06-20 Julia Koval <julia.koval@intel.com>
25378 * config/i386/i386.c: Fix rounding expand for new pattern.
25379 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25381 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25383 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25384 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25386 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25388 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25391 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25393 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25394 architecture, then by implementer ID.
25395 * config/aarch64/aarch64-tune.md: Regenerate.
25397 2017-06-20 Richard Biener <rguenther@suse.de>
25399 PR middle-end/81097
25400 * fold-const.c (split_tree): Fold to type before negating.
25402 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25404 * diagnostic-show-locus.c
25405 (selftest::test_fixit_deletion_affecting_newline): New function.
25406 (selftest::diagnostic_show_locus_c_tests): Call it.
25408 2017-06-20 Andreas Schwab <schwab@suse.de>
25411 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25414 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25416 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25417 __ARM_FEATURE_COPROC according to support.
25419 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25421 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25422 Rewritten to avoid overflow for > 32-bit pointers.
25425 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25426 by removing enum keyword.
25427 (ubsan_type_descriptor): Likewise. Formatting fix.
25430 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25431 splitter): Require TARGET_SSE2 in the condition.
25433 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25436 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25437 for doing vector set of SFmode on ISA 3.0.
25438 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25439 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25441 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25442 SFmode value into a V4SF variable that was extracted from another
25443 V4SF variable without converting the element to double precision
25444 and back to single precision vector format.
25445 (vsx_insert_extract_v4sf_p9_2): Likewise.
25447 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25449 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25450 in UWHI to avoid undefined overflow.
25453 * ubsan.h (enum ubsan_encode_value_phase): New.
25454 (ubsan_encode_value): Change second argument to
25455 enum ubsan_encode_value_phase with default value of
25456 UBSAN_ENCODE_VALUE_GENERIC.
25457 * ubsan.c (ubsan_encode_value): Change second argument to
25458 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25459 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25460 create_tmp_var_raw instead of create_tmp_var and use a
25462 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25463 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25464 ubsan_encode_value callers.
25467 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25468 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25469 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25471 2017-06-19 Richard Biener <rguenther@suse.de>
25473 PR middle-end/81118
25474 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25475 estimates if we changed anything.
25477 2017-06-19 Richard Biener <rguenther@suse.de>
25479 PR tree-optimization/80887
25480 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25481 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25482 simplified lookups, then reset mprts_hook.
25483 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25485 (try_to_simplify): Likewise.
25487 2017-06-19 Martin Liska <mliska@suse.cz>
25490 * gimplify.c (gimplify_switch_expr):
25491 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25493 2017-06-19 Martin Liska <mliska@suse.cz>
25495 * doc/install.texi: Document that PGO runs in 4 stages.
25497 2017-06-19 Martin Liska <mliska@suse.cz>
25500 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25501 to dispatcher function name.
25502 * multiple_target.c (replace_function_decl): New function.
25503 (create_dispatcher_calls): Redirect both edges and references.
25505 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25507 * profile-count.c (profile_count::dump): Dump quality.
25508 (profile_count::differs_from_p): Update for unsigned val.
25509 * profile-count.h (profile_count_quality): New enum.
25510 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25512 2017-06-19 Richard Biener <rguenther@suse.de>
25514 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25515 struct function as arg.
25516 (estimate_numbers_of_iterations): Export overload with loop arg.
25517 (free_numbers_of_iterations_estimates_loop): Use an overload of
25518 free_numbers_of_iterations_estimates instead.
25519 * tree-cfg.c (remove_bb): Adjust.
25520 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25521 * tree-parloops.c (gen_parallel_loop): Likewise.
25522 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25524 (tree_unroll_loops_completely): Likewise.
25525 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25526 Use an overload instead and export.
25527 (estimated_loop_iterations): Adjust.
25528 (max_loop_iterations): Likewise.
25529 (likely_max_loop_iterations): Likewise.
25530 (estimate_numbers_of_iterations): Take struct function as arg
25532 (loop_exits_before_overflow): Adjust.
25533 (free_numbers_of_iterations_estimates_loop): Use an overload.
25534 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25535 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25537 2017-06-19 Richard Biener <rguenther@suse.de>
25540 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25541 RANGE_EXPR conservatively.
25543 2017-06-16 Carl Love <cel@us.ibm.com>
25545 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25546 definitions for vec_float, vec_float2, vec_floato,
25547 vec_floate built-ins.
25548 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25549 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25551 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25552 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25553 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25554 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25555 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25556 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25557 vec_floato): Add builtin defines.
25558 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25559 Update the built-in documentation file for the new built-in
25562 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25564 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25565 (mthumb): Mark as the negative of -marm.
25567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25569 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25571 (ARM Options, -mtune): Document that this accepts the same
25572 extension options as -mcpu.
25573 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25577 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25578 permitted extensions.
25580 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25582 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25583 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25584 (armv8-m.main): Add option +nodsp.
25585 * config/arm/arm-cpu-cdata.h: Regenerated.
25587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25589 * config/arm/t-fuchsia: New file.
25590 * config.gcc (arm*-*-fuchsia*): Use it.
25592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25594 * config/arm/t-symbian: Rewrite for new option infrastructure.
25596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25598 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25599 (MULTILIB_REQUIRED): Likewise.
25601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25603 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25604 (MULTILIB_RESUE): Likewise.
25605 (MULTILIB_MATCHES): Likewise.
25606 (MULTLIB_REQUIRED): Likewise.
25608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25610 * config/arm/t-rtems: Rewrite for new option framework.
25612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25614 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25615 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25616 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25617 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25618 * config/arm/t-multilib: ... here.
25619 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25620 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25621 armv7-a and armv8*-a when A-profile libraries have not been built.
25622 * config/arm/t-rmprofile: Rewrite.
25624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25626 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25627 with a backslash. Remove the backslash after substituting unescaped
25629 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25633 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25634 the driver to pass through the default setting of -mfloat-abi.
25635 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25637 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25638 (all_feat_combs): New rule.
25639 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25641 * config/arm/t-aprofile: Rewrite.
25643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25645 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25646 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25647 fpu is not specified by the user/command-line.
25648 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25649 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25650 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25651 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25652 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25653 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25657 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25658 * config/arm/t-arm-elf: Rewritten.
25660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25662 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25663 have some floating-point instructions.
25664 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25665 (TARGET_MAYBE_HARD_FLOAT): New macro.
25666 * config/arm/arm-builtins.c (arm_init_builtins): Use
25667 TARGET_MAYBE_HARD_FLOAT.
25668 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25672 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25673 (configargs.h): Include it.
25674 (arm_print_hint_for_fpu_option): New function.
25675 (arm_parse_fpu_option): New function.
25676 (candidate_extension): New class.
25677 (arm_canon_for_multilib): New function.
25678 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25679 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25680 (ARCH_CANONICAL_SPECS): New macro.
25681 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25683 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25685 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25686 are set after handling multilib fragments. Set target_cpu_default2
25689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25691 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25693 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25694 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25696 (check_arch): Likewise.
25697 * config/arm/arm.c (arm_configure_build_target): Handle
25698 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25699 options in the default.
25701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25703 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25704 when an option is an alias of another.
25705 * config/arm/parsecpu.awk (optalias): New parser token.
25706 (gen_comm_data): Mark non-alias options as such. Emit entries
25707 for extension aliases.
25708 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25709 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25710 (armv6kz, armv6zk, armv6t2): Likewise.
25711 (armv7): Make vfpv3-d16 an alias.
25712 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
25714 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
25715 Sort in canonical order.
25716 (armv8-a): Sort in canonical order.
25717 (armv8.1-a, armv8.2-a): Likewise.
25718 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
25720 (cortex-a9): Sort in canonical order.
25721 * config/arm/arm.c (selftests.h): Include it.
25722 (arm_test_cpu_arch_data): New function.
25723 (arm_run_self_tests): New function.
25724 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
25725 (targetm): Move declaration to the end of the file.
25726 * arm-cpu-cdata.h: Regenerated.
25728 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25730 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
25731 call to target_mode_check describing the type of option passed.
25732 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
25733 (arm_target_thumb_only): Use arm_parse_arch_option_name or
25734 arm_parse_cpu_option_name to match parameters against list of
25736 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25737 arm_arch_core_flags data structure.
25738 * config/arm/arm-cpu_cdata.h: Regenerated.
25740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25742 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25744 (arm_print_hint_for_cpu_option): Likewise.
25745 (arm_print_hint_for_arch_option): Likewise.
25746 (arm_parse_cpu_option_name): Likewise.
25747 (arm_parse_arch_option_name): Likewise.
25748 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25749 of entries in the all_fpus list.
25750 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25751 (arm_parse_cpu_option_name): Declare.
25752 (arm_parse_arch_option_name): Declare.
25753 (arm_parse_option_features): Declare.
25754 (arm_intialize_isa): Declare.
25755 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25757 (gen_comm_data): ... here. Make definitions non-static.
25758 * config/arm/arm-cpu-data.h: Regenerated.
25759 * config/arm/arm-cpu-cdata.h: Regenerated.
25761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25763 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25764 (cpu_arch_extension): New structure.
25765 (cpu_arch_option, arch_option, cpu_option): New structures.
25766 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25767 architecture types.
25768 (gen_data): Generate new format data tables.
25769 * config/arm/arm.c (cpu_tune): New structure.
25770 (cpu_option, processors): Delete.
25771 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25772 (arm_print_hint_for_cpu_option): ... this and ...
25773 (arm_print_hint_for_arch_option): ... this.
25774 (arm_parse_arch_cpu_name): Delete. Replace with ...
25775 (arm_parse_cpu_option_name): ... this and ...
25776 (arm_parse_arch_option_name): ... this.
25777 (arm_unrecognized_feature): Change type of target parameter to
25779 (arm_parse_arch_cpu_features): Delete. Replace with ...
25780 (arm_parse_option_features): ... this.
25781 (arm_configure_build_target): Rework to use new configuration data
25783 (arm_print_tune_info): Rework for new configuration data tables.
25784 * config/arm/arm-cpu-data.h: Regenerated.
25785 * config/arm/arm-cpu.h: Regenerated.
25787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25789 * Makefile.in (OBJS): Move sbitmap.o from here ...
25790 (OBJS-libcommon): ... to here.
25792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25794 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25795 (ISA_ALL_CRYPTO): New macro.
25796 (ISA_ALL_SIMD): New macro
25797 (ISA_ALL_FP): New macro.
25798 * config/arm/arm.c (fpu_bitlist): Update initializer.
25799 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25801 (arm9e): Add fpu. Add option for nofp
25802 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25803 (arm926ej-s, arm1026ej-s): Likewise.
25804 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25805 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25806 neon-fp16, neon-vfpv4, nofp and nosimd.
25807 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25808 (cortex-a8): Add fpu. Add option for nofp.
25809 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25810 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25811 (cortex-r4f): Add fpu.
25812 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25813 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25815 (cortex-r8): Likewise.
25816 (cortex-m4): Add fpu. Add option for nofp.
25817 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25818 (cortex-a17.cortex-a7): Likewise.
25819 (cortex-a32): Add fpu. Add options for crypto and nofp.
25820 (cortex-a35, cortex-a53): Likewise.
25821 (cortex-a57): Add fpu. Add option for crypto.
25822 (cortex-a72, cortex-a73): Likewise.
25823 (exynos-m1): Likewise.
25824 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25825 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25826 (cortex-m33): Add fpu. Add option for nofp.
25827 * config/arm/arm-cpu-cdata.h: Regenerated
25828 * config/arm/arm-cpu-data.h: Regenerated.
25830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25832 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25833 (armv5te, armv5tej): Likewise.
25834 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25835 (armv7): Add options fp and vfpv3-d16.
25836 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25837 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25839 (armv7ve): Likewise.
25840 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25841 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25842 (armv8-a): Add nocrypto option.
25843 (armv8.1-a, armv8.2-a): Likewise.
25844 (armv8-m.main): add options fp, fp.dp and nofp.
25846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25848 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25850 (armv8-a+crc): Delete.
25851 (armv8.1-a): Add options simd, crypto and nofp.
25852 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25853 (armv8.2-a+fp16): Delete.
25854 (armv8-m.main): Add option dsp.
25855 (armv8-m.main+dsp): Delete.
25856 (cortex-a8): Add fpu. Add nofp option.
25857 (cortex-a9): Add fpu. Add nofp and nosimd options.
25858 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25859 link to main cpu and architecture data structures.
25860 (gen_comm_data): Only put isa attributes from the main architecture
25862 (option): New statement for architecture and CPU entries.
25863 * arm.c (struct cpu_option): New structure.
25864 (struct processors): Add entry for options.
25865 (arm_unrecognized_feature): New function.
25866 (arm_parse_arch_cpu_name): Ignore any characters after the first
25868 (arm_parse_arch_cpu_feature): New function.
25869 (arm_configure_build_target): Separate out any CPU and architecture
25870 features and parse separately. Don't error out if -mfpu=auto is
25871 used with only an architecture string.
25872 (arm_print_asm_arch_directives): New function.
25873 (arm_file_start): Call it.
25874 * config/arm/arm-cpu-cdata.h: Regenerated.
25875 * config/arm/arm-cpu-data.h: Likewise.
25876 * config/arm/arm-tables.opt: Likewise.
25878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25880 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25881 assembler when it is not -mfpu=auto.
25883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25885 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25886 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25887 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25888 (ASM_CPU_SPEC): Rewrite.
25889 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25890 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25891 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25892 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25893 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25894 copied string is NUL-terminated. Also strip any characters prefixed
25896 (arm_rewrite_selected_arch): New function.
25897 (arm_rewrite_march): New function.
25899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25901 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25902 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25903 (march, mcpu, mtune): Convert to string-based options.
25904 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25905 (arm_parse_arch_cpu_name): New function.
25906 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25907 identify selected architecture or CPU.
25908 (arm_option_save): New function.
25909 (TARGET_OPTION_SAVE): Redefine.
25910 (arm_option_restore): Restore string options.
25911 (arm_option_print): Print string options.
25913 2017-06-16 Martin Sebor <msebor@redhat.com>
25915 PR tree-optimization/80933
25916 PR tree-optimization/80934
25917 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25918 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25919 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25920 (gimple_fold_builtin): Call them.
25922 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25924 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25925 as unlikely; update profile.
25927 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25929 * predict.c (force_edge_cold): Handle declaring edges impossible
25932 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25934 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25936 (try_unroll_loop_completely): Fix reporting.
25938 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25940 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25942 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25945 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25946 if given a non-constant argument for an intrinsic which requires a
25949 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25951 * profile.c (compare_freqs): New function.
25952 (branch_prob): Sort edge list.
25953 (find_spanning_tree): Assume that the list is priority sorted.
25955 2017-06-16 Richard Biener <rguenther@suse.de>
25957 PR tree-optimization/81090
25958 * passes.def (pass_record_bounds): Remove.
25959 * tree-pass.h (make_pass_record_bounds): Likewise.
25960 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25961 make_pass_record_bounds): Likewise.
25962 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25963 not free niter estimates at the beginning but at the end.
25964 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25966 2017-06-16 Richard Biener <rguenther@suse.de>
25968 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25969 initializer to workaround ICE in host GCC 4.8.
25971 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25973 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25975 (clone_inlined_nodes): Update.
25977 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25979 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25980 prefetch settings, and enable prefetching by default at -O3.
25982 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25984 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25985 Set flag_prefetch_loop_arrays according to tuning data.
25987 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25989 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25990 New tune structure.
25991 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25992 [Unrelated to main purpose of the patch] Place the pointer field last
25993 to enable type checking errors when tune structure are wrongly merged.
25994 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25995 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25996 (thunderx2t99_prefetch_tune): New tune constants.
25997 (tune_params *_tunings): Update all tunings (no functional change).
25998 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25999 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26000 from tunings structures.
26002 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26005 * ubsan.c (instrument_null): Add T argument, use it instead
26006 of computing it based on IS_LHS.
26007 (instrument_object_size): Likewise.
26008 (pass_ubsan::execute): Adjust instrument_null and
26009 instrument_object_size callers to pass gimple_get_lhs or
26010 gimple_assign_rhs1 result to it. Use instrument_null instead of
26011 calling get_base_address and instrument_mem_ref. Handle
26012 aggregate call arguments for object-size sanitization.
26014 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26016 PR tree-optimization/81089
26017 * tree-vrp.c (is_masked_range_test): Validate operands of
26020 2017-06-15 Martin Sebor <msebor@redhat.com>
26023 * dumpfile.c (dump_register): Avoid calling memset to initialize
26024 a class with a default ctor.
26025 * gcc.c (struct compiler): Remove const qualification.
26026 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26027 * hash-table.h: Ditto.
26028 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26030 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26031 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26033 * params.h (struct param_info): Make struct members non-const.
26034 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26035 with default initialization.
26036 * vec.h (vec_copy_construct, vec_default_construct): New helper
26038 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26039 with vec_copy_construct.
26040 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26041 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26042 * doc/invoke.texi (-Wclass-memaccess): Document.
26044 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26046 * emit-rtl.h (is_leaf): Update comment about local
26047 register allocator.
26049 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26052 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26053 for a variable to have a section before checking if the section has a
26055 Set section to.persistent if persistent attribute is set.
26056 Warn if .persistent attribute is used on an automatic variable.
26058 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26060 PR rtl-optimization/80474
26061 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26063 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26065 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26068 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26070 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26071 (casesi): Emit bounds checking as RTL.
26072 (casesi_internal_mips16_<mode>): Remove bounds checking.
26074 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26076 * config/xtensa/xtensa.c (xtensa_option_override): Append
26077 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26078 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26079 xtensa_doloop_hooks): Define unconditionally.
26080 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26082 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26083 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26084 for it in xtensa_option_override.
26085 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26086 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26088 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26090 * doc/cppopts.texi: Document '-' special value to -MF.
26092 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26094 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26095 (cortex_a53_fconst): Likewise.
26096 (cortex_a53_fpmul): Likewise.
26097 (cortex_a53_f_load_64): Likewise.
26098 (cortex_a53_f_load_many): Likewise.
26099 (cortex_a53_advsimd_alu): Likewise.
26100 (cortex_a53_advsimd_alu_q): Likewise.
26101 (cortex_a53_advsimd_mul): Likewise.
26102 (cortex_a53_advsimd_mul_q): Likewise.
26103 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26104 Add missing fmul, r2f_cvt and fconst cases.
26106 2017-06-14 Richard Biener <rguenther@suse.de>
26108 PR middle-end/81088
26109 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26111 (fold_binary_loc): When associating do not treat pre-existing
26112 TREE_OVERFLOW on literal constants as a reason to allow
26113 TREE_OVERFLOW on associated literal constants.
26115 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26117 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26118 (MASK_FEATURES): New macro.
26119 * config/sparc/sparc.c (sparc_option_override): Remove the special
26120 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26122 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26124 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26125 a division of 0 if non-call exceptions are enabled.
26127 2017-06-14 Andrew Pinski <apinski@cavium.com>
26128 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26131 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26132 Improve vector initialization code gen for only variable case.
26134 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26136 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26138 2017-06-14 Richard Biener <rguenther@suse.de>
26140 PR tree-optimization/81083
26141 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26144 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26146 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26147 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26148 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26149 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26150 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26151 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26153 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26155 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26156 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26158 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26160 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26162 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26164 * config/rs6000/t-rtems: Don't handle SPE.
26166 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26168 * config/rs6000/t-linux: Don't handle SPE.
26170 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26172 * config/rs6000/eabispe.h: Delete file.
26174 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26176 * config/rs6000/t-spe: Delete file.
26178 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26180 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26181 (rs6000_legitimate_offset_address_p): Return false for anything in
26182 V2SImode or V2SFmode.
26184 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26186 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26187 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26188 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26190 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26191 (rs6000_legitimate_offset_address_p): Ditto.
26192 (rs6000_emit_move): Ditto.
26193 (rs6000_init_builtins): Remove V4HI_type_node.
26195 2017-06-13 Martin Liska <mliska@suse.cz>
26198 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26199 (gate_asan): Likewise.
26200 * asan.h (asan_no_sanitize_address_p): Remove the function.
26201 (sanitize_flags_p): New function.
26202 * builtins.def: Fix coding style.
26203 * common.opt: Use renamed enum value.
26204 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26205 * doc/extend.texi: Document no_sanitize attribute.
26206 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26207 to SANITIZE_UNDEFINED_NONDEFAULT.
26208 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26209 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26210 Use sanitize_flags_p.
26211 * gimplify.c (gimplify_function_tree): Likewise.
26212 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26213 * opts.c (parse_no_sanitize_attribute): New function.
26214 (common_handle_option): Use renamed enum value.
26215 * opts.h (parse_no_sanitize_attribute): Declare.
26216 * tree.c (sanitize_flags_p): New function.
26217 * tree.h: Declared here.
26218 * tsan.c: Use sanitize_flags_p.
26219 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26220 (instrument_mem_ref): Likewise.
26221 (instrument_bool_enum_load): Likewise.
26222 (do_ubsan_in_current_function): Remove the function.
26223 (pass_ubsan::execute): Use sanitize_flags_p.
26224 * ubsan.h: Remove do_ubsan_in_current_function
26225 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26226 (dump_function_to_file): Use it here.
26228 2017-06-13 Martin Jambor <mjambor@suse.cz>
26230 PR tree-optimization/80803
26231 PR tree-optimization/81063
26232 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26233 (propagate_subaccesses_across_link): Enqueue subtree whenever
26234 necessary instead of relying on the caller.
26236 2017-06-13 Martin Jambor <mjambor@suse.cz>
26238 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26239 that have a first_link.
26240 (sort_and_splice_var_accesses): Do not check first_link before
26242 (subtree_mark_written_and_enqueue): Likewise.
26243 (propagate_all_subaccesses): Likewise and do not stop at first
26244 parent with a first_link.
26246 2017-06-13 Martin Jambor <mjambor@suse.cz>
26248 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26251 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26253 * match.pd: New pattern.
26255 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26257 * tree-vrp.c (is_masked_range_test): New function.
26258 (register_edge_assert_for): Determine ranges for
26261 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26263 PR tree-optimization/67328
26264 * fold-const.c (maskable_range_p): New function.
26265 (build_range_check): Generate bittests if possible.
26267 2017-06-13 Martin Liska <mliska@suse.cz>
26269 * gimple-pretty-print.c (dump_probability): Add new argument.
26270 (dump_edge_probability): Dump both probability and count.
26271 (dump_gimple_label): Likewise.
26272 (dump_gimple_bb_header): Likewise.
26274 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26277 * config/avr/avr-devices.c: Fix indentation.
26278 * config/avr/gen-avr-mmcu-specs.c: Dito.
26280 2017-06-13 Richard Biener <rguenther@suse.de>
26282 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26283 instead get vector type from stmt_info.
26284 (vectorizable_reduction): Adjust. Remove dead code.
26286 2017-06-13 Richard Biener <rguenther@suse.de>
26288 PR middle-end/81065
26289 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26290 case of C * (x * C2 + C3).
26291 (fold_addr_of_array_ref_difference): Properly fold index difference.
26293 2017-06-12 David S. Miller <davem@davemloft.net>
26296 * config/sparc/sparc.md (return expander): Emit frame blockage if
26297 function uses alloca.
26299 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26301 * combine.c (make_field_assignment): Check len rather than the mode
26302 precision when calling force_to_mode.
26304 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26306 Support multilibs and devices that see flash in RAM address range.
26309 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26310 (avr_mcu_t) <flash_pm_offset>: New field.
26311 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26312 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26313 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26314 (AVR_TINY_PM_OFFSET): Remove macro.
26315 * config/avr/avr.opt (-mshort-calls): New option.
26316 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26317 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26318 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26319 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26320 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26321 instead of avr_arch->have_jmp_call.
26322 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26323 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26324 avr_arch->flash_pm_offset to define.
26325 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26326 new field flash_pm_offset. Add entry for avrxmega3.
26327 (avr_texinfo): Add entry for avrxmega3.
26328 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26329 attiny212, attiny214,
26330 attiny412, attiny414, attiny416, attiny417,
26331 attiny814, attiny816, attiny817,
26332 attiny1614, attiny1616, attiny1617,
26333 attiny3214, attiny3216, attiny3217.
26334 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26335 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26336 (avr_print_operand_address) [AVR_TINY]: Same.
26337 (avr_asm_init_sections) <readonly_data_section>: Only patch
26338 callback if avr_arch->flash_pm_offset = 0.
26339 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26340 for rodata if avr_arch->flash_pm_offset != 0.
26341 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26342 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26343 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26344 (m_options): Append opt_rcall.
26345 (m_dirnames): Append dir_rcall.
26346 * config/avr/t-multilib: Regenerate.
26348 * configure.ac [target=avr]: Check whether avrxmega3 default
26349 linker description file works as needed.
26350 * configure: Regenerate.
26351 * doc/avr-mmcu.texi: Regenerate.
26352 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26353 <__AVR_ARCH__>: Document avrxmega3 and 103.
26354 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26355 <__AVR_SHORT_CALLS__>: Document it.
26356 <__AVR_PM_BASE_ADDRESS__>: Document it.
26357 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26358 (AVR Variable Attributes) <progmem>: Document this is
26359 not needed for avrxmega3.
26360 (AVR Named Address Spaces) <__flash>: Dito.
26362 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26364 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26366 2017-06-12 Doug Rupp <rupp@adacore.com>
26368 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26369 Append vxworks-stdint.h to the tm_file list.
26370 * config/vxworks-stdint.h: New file.
26372 2017-06-12 Martin Liska <mliska@suse.cz>
26374 PR tree-optimization/81041
26375 * tree-profile.c (gimple_gen_ic_func_profiler):
26376 Create an extra BB in profile-generate
26377 (gimple_gen_time_profiler): Likewise.
26379 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26381 PR tree-optimization/81003
26382 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26383 (update_range_test): Use it instead of force_gimple_operand_gsi.
26385 2017-06-12 Richard Biener <rguenther@suse.de>
26387 PR tree-optimization/81053
26388 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26389 with backedge value not defined in loop. Simplify def stmt
26392 2017-06-11 Tom de Vries <tom@codesourcery.com>
26395 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26397 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26398 nvptx_cannot_force_const_mem.
26400 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26402 * opts.c (finish_options): Move test for flag_split_stack after
26403 it has been initialized.
26405 2017-06-11 Jason Merrill <jason@redhat.com>
26407 * tree.h (id_equal): New.
26408 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26409 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26410 instead of strcmp of IDENTIFIER_POINTER.
26412 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26414 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26415 (mark_all_inlined_calls_cdtor): Fix formating.
26416 (inline_transform): Rescale profile before inlining.
26418 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26420 * cgraph.h (cgraph_edge::clone): Update prototype.
26421 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26422 (cgraph_node::create_clone): Update.
26423 (cgraph_node::create_version_clone): Update.
26424 * tree-inline.c (copy_bb): Update.
26425 (expand_call_inline): Update.
26427 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26429 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26430 factored out from ...
26431 (rs6000_emit_prologue): ... here.
26433 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26435 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26436 factored out from ...
26437 (rs6000_emit_prologue): ... here.
26439 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26441 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26443 (handle_missing_profiles): Fix computation of tp_first_run.
26444 (counts_to_freqs): Do not touch freqs when count is 0.
26446 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26448 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26451 2017-06-10 Tom de Vries <tom@codesourcery.com>
26453 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26454 attributes): Document signal effective target.
26456 2017-06-10 Tom de Vries <tom@codesourcery.com>
26458 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26459 Document effective target stack_size.
26461 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26463 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26464 to the edit_context if they can be auto-applied.
26466 2017-06-9 Ian Lance Taylor <iant@golang.org>
26468 * opts.c (finish_options): If -fsplit-stack, disable implicit
26469 -forder-blocks-and-partition.
26470 * doc/invoke.texi (Optimize Options): Document that when using
26471 -fsplit-stack -forder-blocks-and-partition is not implicitly
26474 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26476 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26477 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26478 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26479 * builtins.def (abort, trap, unreachable): Declare cold.
26480 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26481 * tree-core.h (ECF_COLD): New.
26482 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26483 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26485 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26487 * predict.c (unlikely_executed_stmt_p): Cleanup.
26489 2017-06-09 Richard Biener <rguenther@suse.de>
26491 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26492 model if the ref is always written to.
26494 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26496 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26498 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26500 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26503 2017-06-09 Tom de Vries <tom@codesourcery.com>
26506 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26507 "target cannot support label values" when encountering LABEL_REF.
26509 2017-06-09 Martin Liska <mliska@suse.cz>
26511 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26512 (gimple_gen_ic_func_profiler): Emit direct comparison
26513 of __gcov_indirect_call_callee with NULL.
26514 (gimple_gen_time_profiler): Change probability from
26515 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26517 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26519 * profile.c (edge_gcov_counts): Turn to pointer.
26520 (compute_branch_probabilities, compute_branch_probabilities): Update.
26521 (branch_prob): Do not clear edge_gcov_count.
26522 * profile.h (edge_gcov_counts): Turn to pointer.
26523 (edge_gcov_count): Update.
26525 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26527 * gimple.h (gimple_check_failed): Mark cold.
26529 2017-06-09 Richard Biener <rguenther@suse.de>
26531 PR tree-optimization/66623
26532 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26533 refactor check_reduction into two parts, properly computing
26534 whether we have to check reduction validity for outer loop
26537 2017-06-09 Richard Biener <rguenther@suse.de>
26539 PR tree-optimization/79483
26540 * graphite-scop-detection.c (order): New global.
26541 (get_order): Compute bb to order mapping that satisfies code
26542 generation constraints.
26543 (cmp_pbbs): New helper.
26544 (build_scops): Start domwalk at entry block, sort generated
26547 2017-06-09 Richard Biener <rguenther@suse.de>
26549 PR middle-end/81007
26550 * ipa-polymorphic-call.c
26551 (ipa_polymorphic_call_context::restrict_to_inner_class):
26552 Skip FIELD_DECLs with error_mark_node type.
26553 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26556 2017-06-09 Martin Liska <mliska@suse.cz>
26558 * predict.c (struct branch_predictor): New struct.
26559 (test_prediction_value_range): New test.
26560 (predict_c_tests): New function.
26561 * selftest-run-tests.c (selftest::run_tests): Run the function.
26562 * selftest.h: Declare new tests.
26564 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26567 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26568 gen_add3_insn did not fail.
26569 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26570 r0, construct that number in a temporary reg and add that reg to r0.
26571 If asked to put the result in r0 as well, fail.
26573 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26575 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26576 for early expansion of vec_eqv.
26578 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26580 PR middle-end/81005
26581 * ubsan.c (instrument_null): Avoid pointless code temporary.
26582 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26584 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26588 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26591 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26594 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26596 * predict.c (unlikely_executed_edge_p): Move ahead.
26597 (probably_never_executed_edge_p): Use it.
26599 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26601 PR middle-end/79988
26602 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26603 gimple_call_builtin_p call.
26605 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26607 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26608 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26609 rtl_check_failed_type2, rtl_check_failed_code1,
26610 rtl_check_failed_code2, rtl_check_failed_code_mode,
26611 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26612 rtvec_check_failed_bounds, rtl_check_failed_flag,
26613 _fatal_insn_not_found, _fatal_insn): Likewise.
26614 * tree.h (tree_contains_struct_check_failed,
26615 tree_check_failed, tree_not_check_failed,
26616 tree_class_check_failed, tree_range_check_failed,
26617 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26618 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26619 tree_operand_check_failed, omp_clause_check_failed,
26620 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26623 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26625 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26626 flag_branch_probabilities.
26627 * ipa-inline.c (edge_badness): Likewise.
26628 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26629 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26630 * predict.c (maybe_hot_frequency_p): Likewise.
26631 (probably_never_executed): Likewise.
26632 * sched-ebb.c (schedule_ebbs): Likewise.
26633 * sched-rgn.c (find_single_block_region): Likewise.
26634 * tracer.c (tail_duplicate): Likewise.
26636 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26638 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26639 longer requires x_flag_profile_use.
26641 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26643 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26644 instead of flag_reorder_blocks_and_partition.
26645 * dbxout.c (dbxout_function_end): Likewise.
26646 * dwarf2out.c (gen_subprogram_die): Likewise.
26647 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26648 * hw-doloop.c (reorg_loops): Likewise.
26649 * varasm.c (assemble_start_function,
26650 assemble_end_function): Likewise.
26651 (decide_function_section): Do not check for
26652 flag_reorder_blocks_and_partition.
26654 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26656 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26658 (chkp_get_hard_register_fake_addr_expr): Ditto.
26659 (chkp_build_addr_expr): Add check for hard reg case.
26660 (chkp_parse_array_and_component_ref): Ditto.
26661 (chkp_find_bounds_1): Ditto.
26662 (chkp_process_stmt): Don't generate bounds store for
26665 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26667 * predict.c (maybe_hot_bb_p): Do not check profile status.
26668 (maybe_hot_edge_p): Likewise.
26669 (probably_never_executed): Check for zero counts even if profile
26671 (unlikely_executed_edge_p): New function.
26672 (unlikely_executed_stmt_p): New function.
26673 (unlikely_executed_bb_p): New function.
26674 (set_even_probabilities): Use unlikely predicates.
26675 (combine_predictions_for_bb): Likewise.
26676 (predict_paths_for_bb): Likewise.
26677 (predict_paths_leading_to_edge): Likewise.
26678 (determine_unlikely_bbs): New function.
26679 (estimate_bb_frequencies): Use it.
26680 (compute_function_frequency): Use zero counts even if profile is
26682 * profile-count.h: Fix typo.
26684 2017-08-08 Julia Koval <julia.koval@intel.com>
26686 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26687 _mm512_mask_cvtsepi16_storeu_epi8,
26688 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26689 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26690 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26691 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26692 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26693 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26694 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26695 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26696 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26697 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26698 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26699 __builtin_ia32_pmovuswb256mem_mask,
26700 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26701 __builtin_ia32_pmovwb512mem_mask): New builtins.
26703 2017-08-08 Julia Koval <julia.koval@intel.com>
26705 PR target/73350,80862
26706 * config/i386/subst.md (round): Fix round pattern.
26707 * config/i386/i386.c (ix86_erase_embedded_rounding):
26708 Fix erasing rounding for the fixed pattern.
26710 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26712 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
26714 2017-06-08 Martin Liska <mliska@suse.cz>
26716 PR gcov-profile/80911
26717 * gcov.c (block_info::block_info): New constructor.
26719 2017-06-07 Carl Love <cel@us.ibm.com>
26721 * config/rs6000/rs6000-c: The return type of the following
26722 built-in functions was implemented as int not long long. Fix sign
26723 of return value for the unsigned version of vec_mulo and vec_mule.
26724 vector unsigned long long vec_bperm (vector unsigned long long,
26725 vector unsigned char)
26726 vector signed long long vec_mule (vector signed int,
26728 vector unsigned long long vec_mule (vector unsigned int,
26729 vector unsigned int)
26730 vector signed long long vec_mulo (vector signed int,
26732 vector unsigned long long vec_mulo (vector unsigned int,
26733 vector unsigned int)
26734 * doc/extend.texi: Fix the documentation for the built-in
26737 2017-06-07 Carl Love <cel@us.ibm.com>
26740 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26743 2017-06-07 Carl Love <cel@us.ibm.com>
26745 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26746 support, Generate doublehv for signed int/float for BE case only.
26748 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26750 * doc/invoke.texi (mcx16): Rewrite.
26752 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26754 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26755 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26756 *mov<mode>_softfloat, and an anonymous splitter): Use
26757 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26759 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26761 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26763 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26764 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26765 (rs6000_debug_reg_global): Adjust.
26766 (rs6000_init_hard_regno_mode_ok): Adjust.
26767 (rs6000_dbx_register_number): Adjust.
26768 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26769 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26770 Remove SPE_ACC and SPEFSCR.
26771 (REG_ALLOC_ORDER): Ditto.
26772 (FRAME_POINTER_REGNUM): Change to 111.
26773 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26774 (REG_CLASS_NAMES): Ditto.
26775 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26776 (REGISTER_NAMES): Ditto.
26777 (ADDITIONAL_REG_NAMES): Ditto.
26778 (rs6000_reg_names): Ditto.
26779 * config/rs6000/rs6000.md: Renumber some register number
26782 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26784 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26786 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26787 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26789 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26790 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26791 Delete the SPE high registers.
26792 (REG_ALLOC_ORDER): Ditto.
26793 (enum reg_class): Remove SPE_HIGH_REGS.
26794 (REG_CLASS_NAMES): Ditto.
26795 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26796 (REGISTER_NAMES): Ditto.
26797 (rs6000_reg_names): Ditto.
26798 * doc/tm.texi.in: Remove SPE as example.
26799 * doc/tm.texi: Regenerate.
26801 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26803 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26804 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26805 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26806 * config/rs6000/rs6000.md (type): Remove "brinc".
26808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26810 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26811 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26812 * config/rs6000/linuxspe.h: Delete file.
26813 * config/rs6000/rs6000.md: Don't include spe.md.
26814 * config/rs6000/spe.h: Delete file.
26815 * config/rs6000/spe.md: Delete file.
26816 * config/rs6000/t-rs6000: Remove spe.md.
26818 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26820 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26821 (reg_or_none500mem_operand): Delete.
26822 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26823 instead of reg_or_none500mem_operand.
26825 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26827 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26828 handling of SPE flags.
26829 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26831 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26833 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26835 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26836 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26837 paired_divv2sf3): Similar.
26838 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26839 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26840 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26842 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26843 Rename the paired_* instruction patterns.
26844 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26846 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26847 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26848 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26849 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26850 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26851 PAIRED_VECTOR_MODE.
26852 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26853 (spe_func_has_64bit_regs_p): Delete.
26854 (spe_expand_predicate_builtin): Delete.
26855 (spe_expand_evsel_builtin): Delete.
26856 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26857 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26858 (invalid_e500_subreg): Delete.
26859 (rs6000_legitimize_address): Always force_reg op2 as well, for
26860 paired single memory accesses.
26861 (rs6000_member_type_forces_blk): Delete.
26862 (rs6000_spe_function_arg): Delete.
26863 (rs6000_expand_unop_builtin): Delete SPE handling.
26864 (rs6000_expand_binop_builtin): Ditto.
26865 (spe_expand_stv_builtin): Delete.
26866 (bdesc_2arg_spe): Delete.
26867 (spe_expand_builtin): Delete.
26868 (spe_expand_predicate_builtin): Delete.
26869 (spe_expand_evsel_builtin): Delete.
26870 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26871 (spe_init_builtins): Delete.
26872 (spe_func_has_64bit_regs_p): Delete.
26873 (savres_routine_name): Delete "info" parameter. Adjust callers.
26874 (rs6000_emit_stack_reset): Ditto.
26875 (rs6000_dwarf_register_span): Delete.
26876 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26877 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26878 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26880 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26882 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26883 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26884 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26885 mulv2sf3, divv2sf3): Delete expanders.
26887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26889 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26893 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26894 * config/rs6000/rs6000.c: Ditto.
26896 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26898 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26899 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26900 comparison_operator.
26902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26904 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26905 * config/rs6000/rs6000.opt: Ditto.
26906 * config/rs6000/t-rtems: Ditto.
26908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26910 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26911 TARGET_E500_SINGLE by 0, simplify.
26912 * config/rs6000/rs6000.c: Ditto.
26913 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26914 (spe_build_register_parallel): Delete.
26915 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26916 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26917 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26918 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26919 (E500_CONVERT): Delete.
26920 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26922 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26924 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26925 * config/rs6000/dfp.md: Ditto.
26926 (negdd2, *negdd2_fpr): Merge.
26927 (absdd2, *absdd2_fpr): Merge.
26928 (negtd2, *negtd2_fpr): Merge.
26929 (abstd2, *abstd2_fpr): Merge.
26930 * config/rs6000/e500.h: Delete file.
26931 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26932 TARGET_FPRS by 1 and simplify.
26933 * config/rs6000/rs6000-c.c: Ditto.
26934 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26935 TARGET_DF_SPE by 0.
26936 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26938 * config/rs6000/rs6000.md: Ditto.
26939 (floatdidf2, *floatdidf2_fpr): Merge.
26940 (move_from_CR_gt_bit): Delete.
26941 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26942 (E500_CR_IOR_COMPARE): Delete.
26943 (All patterns that require !TARGET_FPRS): Delete.
26944 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26946 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26948 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26950 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26952 * graphds.c (add_edge): Intitialize edge's attached data.
26953 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26954 pointer parameter. Call pointed function on each edge during
26955 graph traversing. Skip traversing the edge when the function
26957 (graphds_dfs, graphds_scc): Ditto.
26958 (for_each_edge): New parameter. Pass the new parameter to callback
26960 * graphds.h (skip_edge_callback): New function pointer type.
26961 (graphds_dfs, graphds_scc): New function pointer parameter.
26962 (graphds_edge_callback, for_each_edge): New parameter.
26964 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26966 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26967 out code checking if runtime alias check is possible to below ...
26968 Call the new function.
26969 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26970 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26972 2017-06-07 Marek Polacek <polacek@redhat.com>
26975 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26976 TYPE_OVERFLOW_WRAPS check.
26978 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26980 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26981 if versioning is required.
26982 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26983 peeling with the check for versioning.
26985 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26987 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26988 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26989 Set true to new parameter if new ssa variable is defined.
26990 (vect_gen_vector_loop_niters): Refactor. Set range information
26991 for the new vector loop bound variable.
26992 (vect_do_peeling): Ditto.
26994 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26996 * tree-affine.c (ssa.h): Include header file.
26997 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26998 has wrapping overflow behavior.
27000 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27002 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27004 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27006 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27007 (tree_to_aff_combination): ... here.
27009 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27011 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27012 reg_pressure model function.
27013 (ivopts_global_cost_for_size): Delete.
27014 (determine_set_costs, iv_ca_recount_cost): Call new model function
27015 ivopts_estimate_reg_pressure.
27017 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27019 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27020 expensive than udiv. Remove floating point cases from mod.
27022 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27024 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27025 Increase idiv cost.
27027 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27029 * config/aarch64/aarch64.md
27030 (copysignsf3): Fix mask generation.
27032 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27034 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27036 (class dump_manager): Add register_dumps method.
27037 * dumpfile.c: Include langhooks.h.
27038 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27039 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27040 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27041 (dump_manager::dump_register): Start with 512 entries instead of 32.
27042 (dump_manager::register_dumps): New method.
27043 * toplev.c (general_init): Instead of invoking register_dumps
27044 langhook, invoke register_dumps method on the dump manager.
27045 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27048 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27050 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27051 Say that # requires an associated define_split to exist, and that
27052 the define_split must be suitable for use after register allocation.
27054 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27056 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27057 (compute_outgoing_frequencies): Also initialize zero counts.
27058 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27059 around loops; preserve more of profile when nothing changes.
27061 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27063 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27065 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27066 * config/arm/arm-cpu-cdata.h: Regenerate.
27067 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27068 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27069 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27070 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27071 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27073 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27075 * config/arm/t-rmprofile: Likewise.
27076 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27078 2017-06-06 David S. Miller <davem@davemloft.net>
27081 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27082 blockage if function uses alloca.
27084 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27086 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27087 New "uid" fields to hold pretty-print IDs of group and ref.
27088 Memory references are now identified as <group_id>:<ref_id>
27089 instead of using [random] addresses.
27090 (dump_mem_details): Simplify, no functional change.
27091 (dump_mem_ref): Simplify and make output more concise.
27092 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27093 (find_or_create_group): Initialize group uid.
27094 (record_ref): Initialize ref uid. Improve debug output.
27095 (prune_group_by_reuse, should_issue_prefetch_p,)
27096 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27097 (mark_nontemporal_store, determine_loop_nest_reuse):
27098 Improve debug output.
27100 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27102 * dbgcnt.def (prefetch): New debug counter.
27103 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27104 (schedule_prefetches): Stop issueing prefetches if debug counter
27107 2017-06-06 Tom de Vries <tom@codesourcery.com>
27109 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27110 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27112 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27114 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27115 Use aarch64_reg_or_zero predicate for operand 4.
27116 (aarch64_compare_and_swap<mode> define_insn_and_split):
27117 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27118 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27120 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27122 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27123 (arm_compute_save_core_reg_mask): This.
27124 (thumb1_compute_save_reg_mask): Rename into ...
27125 (thumb1_compute_save_core_reg_mask): This.
27126 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27127 (arm_compute_frame_layout): Likewise.
27129 2017-06-06 Richard Biener <rguenther@suse.de>
27131 PR tree-optimization/80974
27132 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27133 keep or clear leaders SSA info.
27135 2017-06-06 Tom de Vries <tom@codesourcery.com>
27137 * config/nvptx/nvptx.c (split_mode_p): New function.
27138 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27140 2017-06-06 Tom de Vries <tom@codesourcery.com>
27142 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27144 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27147 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27150 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27152 * shrink-wrap.c (handle_simple_exit): Update profile.
27153 (try_shrink_wrapping): Upate profile.
27155 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27157 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27158 (tree_guess_outgoing_edge_probabilities): New.
27159 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27160 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27162 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27164 * ipa-split.c (split_function): Initialize return bb profile.
27166 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27168 * profile.c (compute_branch_probabilities): Also initialize
27169 EXIT_BLOCK profile.
27171 2017-06-06 Richard Biener <rguenther@suse.de>
27173 PR tree-optimization/80928
27174 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27175 (vect_analyze_loop_operations): Properly guard analysis for
27177 (vect_transform_loop): Likewise.
27178 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27179 (vect_model_induction_cost): Do not cost for pure SLP.
27180 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27181 of induction in inner loop vectorization.
27182 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27183 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27184 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27186 (vect_analyze_slp_cost_1): Cost induction.
27187 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27188 (vect_get_slp_vect_defs): Likewise.
27189 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27190 (vect_transform_stmt): Handle SLP reductions.
27191 * tree-vectorizer.h (vectorizable_induction): Adjust.
27193 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27195 * config/rs6000/rs6000.c (make_resolver_func): Update
27196 init_lowered_empty_function call.
27198 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27200 * doc/invoke.texi: Document the -fprofile-abs-path option.
27201 * common.opt (fprofile-abs-path): New option.
27202 * gcov-io.h (gcov_write_filename): Declare.
27203 * gcov-io.c (gcov_write_filename): New function.
27204 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27205 * profile.c (output_location): Likewise.
27207 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27209 * shring-wrap.c: Revert accidental commit.
27211 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27213 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27215 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27217 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27219 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27220 profile in callgraph edge.
27221 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27222 (apply_scale): Likewise.
27223 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27224 Also scale profile when inlining function with zero profile.
27225 (initialize_cfun): Update exit block profile even when it is zero.
27226 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27227 when profile is read.
27229 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27231 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27232 (CLONE_*): New constants to define the processors we can generate
27233 code for with the target_clone attribute.
27234 (rs6000_clone_map): New array to identify which clone processors
27235 the current program is running on.
27236 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27237 target_clone attribute.
27238 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27239 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27240 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27241 (cpu_expand_builtin): Add support for target_clone attribute.
27242 (rs6000_valid_attribute_p): Allow "default" attribute.
27243 (get_decl_name): New debug function to simplify printing the
27244 current function name in debugging statements.
27245 (rs6000_clone_priority): New functions to support the target_clone
27246 attribute, and be able to generate code to switch between ISA 2.05
27247 through ISA 3.0 (power6 through power9).
27248 (rs6000_compare_version_priority): Likewise.
27249 (rs6000_get_function_versions_dispatcher): Likewise.
27250 (make_resolver_func): Likewise.
27251 (add_condition_to_bb): Likewise.
27252 (dispatch_function_versions): Likewise.
27253 (rs6000_generate_version_dispatcher_body): Likewise.
27254 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27255 (fusion_gpr_load_p): Fix a spacing issue.
27256 * doc/extend.texi (Common Function Attributes): Document that the
27257 PowerPC supports the target_clone attribute.
27259 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27261 * config/arm/arm.h: explain F symbol found in description of ARM
27262 register allocation in its legend.
27264 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27266 * config/mips/frame-header-opt.c: Include profile-count.h.
27267 * config/riscv/riscv.c: Include profile-count.h
27269 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27271 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27273 (sm_set_flag_if_changed): Add bbs field.
27274 (execute_sm_if_changed_flag_set): Pass BBS.
27275 (execute_sm): Update.
27277 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27279 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27282 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27284 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27285 (peephole2): New peephole2 to emit the above.
27286 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27288 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27290 * config/aarch64/aarch64.c (define_peephole2 above
27291 *sub_<shift>_<mode>): New peephole.
27293 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27295 * config/i386/i386.c (make_resolver_func): Update.
27296 * Makefile.in: Add profile-count.h and profile-count.o
27297 * auto-profile.c (afdo_indirect_call): Update to new API.
27298 (afdo_set_bb_count): Update.
27299 (afdo_propagate_edge): Update.
27300 (afdo_propagate_circuit): Update.
27301 (afdo_calculate_branch_prob): Update.
27302 (afdo_annotate_cfg): Update.
27303 * basic-block.h: Include profile-count.h
27304 (struct edge_def): Turn count to profile_count.
27305 (struct basic_block_def): Likewie.
27306 (REG_BR_PROB_BASE): Move to profile-count.h
27307 (RDIV): Move to profile-count.h
27308 * bb-reorder.c (max_entry_count): Turn to profile_count.
27309 (find_traces): Update.
27310 (rotate_loop):Update.
27311 (connect_traces):Update.
27312 (sanitize_hot_paths):Update.
27313 * bt-load.c (migrate_btr_defs): Update.
27314 * cfg.c (RDIV): Remove.
27315 (init_flow): Use alloc_block.
27316 (alloc_block): Uninitialize count.
27317 (unchecked_make_edge): Uninitialize count.
27318 (check_bb_profile): Update.
27319 (dump_edge_info): Update.
27320 (dump_bb_info): Update.
27321 (update_bb_profile_for_threading): Update.
27322 (scale_bbs_frequencies_int): Update.
27323 (scale_bbs_frequencies_gcov_type): Update.
27324 (scale_bbs_frequencies_profile_count): New.
27325 * cfg.h (update_bb_profile_for_threading): Update.
27326 (scale_bbs_frequencies_profile_count): Declare.
27327 * cfgbuild.c (compute_outgoing_frequencies): Update.
27328 (find_many_sub_basic_blocks): Update.
27329 * cfgcleanup.c (try_forward_edges): Update.
27330 (try_crossjump_to_edge): Update.
27331 * cfgexpand.c (expand_gimple_tailcall): Update.
27332 (construct_exit_block): Update.
27333 * cfghooks.c (verify_flow_info): Update.
27334 (dump_bb_for_graph): Update.
27335 (split_edge): Update.
27336 (make_forwarder_block): Update.
27337 (duplicate_block): Update.
27338 (account_profile_record): Update.
27339 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27340 (get_estimated_loop_iterations): Update.
27341 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27342 (single_likely_exit): Update.
27343 * cfgloopmanip.c (scale_loop_profile): Update.
27345 (set_zero_probability): Update.
27346 (lv_adjust_loop_entry_edge): Update.
27347 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27348 (purge_dead_edges): Update.
27349 (rtl_account_profile_record): Update.
27350 * cgraph.c (cgraph_node::create): Uninitialize count.
27351 (symbol_table::create_edge): Uninitialize count.
27352 (cgraph_update_edges_for_call_stmt_node): Update.
27353 (cgraph_edge::dump_edge_flags): Update.
27354 (cgraph_node::dump): Update.
27355 (cgraph_edge::maybe_hot_p): Update.
27356 * cgraph.h: Include profile-count.h
27357 (create_clone), create_edge, create_indirect_edge): Update.
27358 (cgraph_node): Turn count to profile_count.
27359 (cgraph_edge0: Likewise.
27360 (make_speculative, clone): Update.
27361 (create_edge): Update.
27362 (init_lowered_empty_function): Update.
27363 * cgraphclones.c (cgraph_edge::clone): Update.
27364 (duplicate_thunk_for_node): Update.
27365 (cgraph_node::create_clone): Update.
27366 * cgraphunit.c (cgraph_node::analyze): Update.
27367 (cgraph_node::expand_thunk): Update.
27368 * final.c (dump_basic_block_info): Update.
27369 * gimple-streamer-in.c (input_bb): Update.
27370 * gimple-streamer-out.c (output_bb): Update.
27371 * graphite.c (print_global_statistics): Update.
27372 (print_graphite_scop_statistics): Update.
27373 * hsa-brig.c: Include basic-block.h.
27374 * hsa-dump.c: Include basic-block.h.
27375 * hsa-gen.c (T sum_slice): Update.
27376 (convert_switch_statements):Update.
27377 * hsa-regalloc.c: Include basic-block.h.
27378 * ipa-chkp.c (chkp_produce_thunks): Update.
27379 * ipa-cp.c (struct caller_statistics): Update.
27380 (init_caller_stats): Update.
27381 (gather_caller_stats): Update.
27382 (ipcp_cloning_candidate_p): Update.
27383 (good_cloning_opportunity_p): Update.
27384 (get_info_about_necessary_edges): Update.
27385 (dump_profile_updates): Update.
27386 (update_profiling_info): Update.
27387 (update_specialized_profile): Update.
27388 (perhaps_add_new_callers): Update.
27389 (decide_about_value): Update.
27390 (ipa_cp_c_finalize): Update.
27391 * ipa-devirt.c (struct odr_type_warn_count): Update.
27392 (struct decl_warn_count): Update.
27393 (struct final_warning_record): Update.
27394 (possible_polymorphic_call_targets): Update.
27395 (ipa_devirt): Update.
27396 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27397 * ipa-icf.c (sem_function::merge): Update.
27398 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27399 * ipa-inline.c (compute_uninlined_call_time): Update.
27400 (compute_inlined_call_time): Update.
27401 (want_inline_small_function_p): Update.
27402 (want_inline_self_recursive_call_p): Update.
27403 (edge_badness): Update.
27404 (lookup_recursive_calls): Update.
27405 (recursive_inlining): Update.
27406 (inline_small_functions): Update.
27407 (dump_overall_stats): Update.
27408 (dump_inline_stats): Update.
27409 * ipa-profile.c (ipa_profile_generate_summary): Update.
27410 (ipa_propagate_frequency): Update.
27411 (ipa_profile): Update.
27412 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27413 * ipa-utils.c (ipa_merge_profiles): Update.
27414 * loop-doloop.c (doloop_modify): Update.
27415 * loop-unroll.c (report_unroll): Update.
27416 (unroll_loop_runtime_iterations): Update.
27417 * lto-cgraph.c (lto_output_edge): Update.
27418 (lto_output_node): Update.
27419 (input_node): Update.
27420 (input_edge): Update.
27421 (merge_profile_summaries): Update.
27422 * lto-streamer-in.c (input_cfg): Update.
27423 * lto-streamer-out.c (output_cfg): Update.
27424 * mcf.c (create_fixup_graph): Update.
27425 (adjust_cfg_counts): Update.
27426 (sum_edge_counts): Update.
27427 * modulo-sched.c (sms_schedule): Update.
27428 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27429 * predict.c (maybe_hot_count_p): Update.
27430 (probably_never_executed): Update.
27431 (dump_prediction): Update.
27432 (combine_predictions_for_bb): Update.
27433 (propagate_freq): Update.
27434 (handle_missing_profiles): Update.
27435 (counts_to_freqs): Update.
27436 (rebuild_frequencies): Update.
27437 (force_edge_cold): Update.
27438 * predict.h: Include profile-count.h
27439 (maybe_hot_count_p, counts_to_freqs): UPdate.
27440 * print-rtl-function.c: Do not include cfg.h
27441 * print-rtl.c: Include basic-block.h
27442 * profile-count.c: New file.
27443 * profile-count.h: New file.
27444 * profile.c (is_edge_inconsistent): Update.
27445 (correct_negative_edge_counts): Update.
27446 (is_inconsistent): Update.
27447 (set_bb_counts): Update.
27448 (read_profile_edge_counts): Update.
27449 (compute_frequency_overlap): Update.
27450 (compute_branch_probabilities): Update; Initialize and deinitialize
27452 (branch_prob): Update.
27453 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27454 (edge_gcov_count): New.
27455 (bb_gcov_count): New.
27456 * shrink-wrap.c (try_shrink_wrapping): Update.
27457 * tracer.c (better_p): Update.
27458 * trans-mem.c (expand_transaction): Update.
27459 (ipa_tm_insert_irr_call): Update.
27460 (ipa_tm_insert_gettmclone_call): Update.
27461 * tree-call-cdce.c: Update.
27462 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27463 (gimple_duplicate_sese_tail): Update.
27464 (gimple_account_profile_record): Update.
27465 (execute_fixup_cfg): Update.
27466 * tree-inline.c (copy_bb): Update.
27467 (copy_edges_for_bb): Update.
27468 (initialize_cfun): Update.
27469 (freqs_to_counts): Update.
27470 (copy_cfg_body): Update.
27471 (expand_call_inline): Update.
27472 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27473 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27474 (try_unroll_loop_completely): Update.
27475 (try_peel_loop): Update.
27476 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27477 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27478 * tree-ssa-loop-split.c (connect_loops): Update.
27479 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27480 * tree-ssa-reassoc.c (branch_fixup): Update.
27481 * tree-ssa-tail-merge.c (replace_block_by): Update.
27482 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27483 (compute_path_counts): Update.
27484 (update_profile): Update.
27485 (recompute_probabilities): Update.
27486 (update_joiner_offpath_counts): Update.
27487 (estimated_freqs_path): Update.
27488 (freqs_to_counts_path): Update.
27489 (clear_counts_path): Update.
27490 (ssa_fix_duplicate_block_edges): Update.
27491 (duplicate_thread_path): Update.
27492 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27493 (struct switch_conv_info): Update.
27494 * tree-tailcall.c (decrease_profile): Update.
27495 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27496 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27497 * value-prof.c (check_counter): Update.
27498 (gimple_divmod_fixed_value): Update.
27499 (gimple_mod_pow2): Update.
27500 (gimple_mod_subtract): Update.
27501 (gimple_ic_transform): Update.
27502 (gimple_stringop_fixed_value): Update.
27503 * value-prof.h (gimple_ic): Update.
27505 2017-06-02 Carl Love <cel@us.ibm.com>
27507 * config/rs6000/rs6000-c: Add support for built-in functions
27508 vector double vec_doublee (vector signed int);
27509 vector double vec_doublee (vector unsigned int);
27510 vector double vec_doublee (vector float);
27511 vector double vec_doubleh (vector signed int);
27512 vector double vec_doubleh (vector unsigned int);
27513 vector double vec_doubleh (vector float);
27514 vector double vec_doublel (vector signed int);
27515 vector double vec_doublel (vector unsigned int);
27516 vector double vec_doublel (vector float);
27517 vector double vec_doubleo (vector signed int);
27518 vector double vec_doubleo (vector unsigned int);
27519 vector double vec_doubleo (vector float);.
27520 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27521 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27523 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27524 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27525 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27527 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27528 vec_doublel, vec_doubleh.
27529 * doc/extend.texi: Update the built-in documentation file for the
27530 new built-in functions.
27532 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27535 * ipa-inline-analysis.c (free_growth_caches): Set
27536 edge_removal_hook_holder to NULL after removing it.
27538 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27540 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27541 comparision with zero.
27543 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27544 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27545 for early expansion of vec_min and vec_max builtins.
27546 (builtin_function_type): Add min/max unsigned variants to those
27547 identified as having unsigned arguments.
27549 2017-06-02 Olivier Hainque <hainque@adacore.com>
27551 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27553 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27555 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27556 Use VALL_F16 iterator rather than VALL.
27558 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27560 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27561 Emit CBNZ inside loop when doing a strong exchange and comparing
27562 against zero. Generate the CC flags after the loop.
27564 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27566 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27567 (dl_section_ref): New.
27568 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27569 On AIX, append an expression to subtract the size of the
27570 section length to dl_section_ref.
27572 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27574 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27575 for early expansion of vector absolute builtins.
27577 2017-06-02 Richard Biener <rguenther@suse.de>
27579 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27580 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27582 2017-06-02 Richard Biener <rguenther@suse.de>
27584 PR tree-optimization/80948
27585 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27586 stmt order as well.
27588 2017-06-02 Richard Biener <rguenther@suse.de>
27590 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27592 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27593 for inductions as relevant.
27595 2017-06-02 Richard Biener <rguenther@suse.de>
27597 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27598 (vectorizable_induction): ... this. Remove dead code.
27600 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27602 * builtins. (expand_builtin_alloca): Remove second parameter and
27603 infer its value from the first parameter instead.
27604 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27606 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27608 PR rtl-optimization/80903
27609 * loop-doloop.c (add_test): Unshare sequence.
27611 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27613 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27615 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27617 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27619 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27620 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27621 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27622 (xlogue_layout::get_stub_rtx): Make static.
27623 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27624 (xlogue_layout::compute_stub_managed_regs): Rename to...
27625 (xlogue_layout::count_stub_managed_regs): ...this.
27626 (xlogue_layout::is_stub_managed_reg): New function.
27627 (xlogue_layout::m_stub_names): Rename to...
27628 (xlogue_layout::s_stub_names): ...this, make static.
27629 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27630 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27631 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27632 xlogue_layout::s_stub_names): Instantiate statics.
27633 (stub_managed_regs): Remove.
27634 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27635 (disable_call_ms2sysv_xlogues): Rename to...
27636 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27637 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27639 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27640 change after reload_completed.
27641 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27643 (ix86_expand_prologue): Likewise.
27644 (ix86_expand_epilogue): Likewise.
27645 (ix86_expand_split_stack_prologue): Likewise.
27646 (ix86_compute_frame_layout): Remove frame parameter ...
27647 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27648 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27650 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27651 (ix86_frame): Move from here ...
27652 * config/i386/i386.h (ix86_frame): ... to here.
27653 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27654 complete ix86_frame data structure instead. Remove some_ld_name.
27656 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27658 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27659 symbols that hold a DECL_VALUE_EXPR.
27661 2017-06-01 Martin Jambor <mjambor@suse.cz>
27663 PR tree-optimization/80898
27664 * tree-sra.c (process_subtree_disqualification): Removed.
27665 (disqualify_candidate): Do not acll
27666 process_subtree_disqualification.
27667 (subtree_mark_written_and_enqueue): New function.
27668 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27669 RHS has been disqualified and re-queue LHS if necessary. Apart
27670 from that, ignore disqualified RHS.
27672 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27674 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27675 address fetch for z10 or later.
27677 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27679 * config/arc/arc.md (tst_movb): Add guard when splitting.
27681 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27683 * config/arc/arc.c (arc_can_eliminate): Test against
27684 arc_frame_pointer_needed.
27686 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27688 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27689 to prevent store reordering.
27690 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27691 (type): Add block type.
27692 (stack_tie): Define special instruction to be used in
27695 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27697 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27698 constraint. It is not valid for the pattern.
27699 (noncommutative_binary_comparison): Likewise.
27701 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27703 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27706 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27708 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27709 be used by the reg-alloc.
27711 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27713 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
27714 reg-alloc when having mul64 or mul32x16 instructions.
27715 (mulsidi3): Likewise.
27716 (umulsidi3): Likewise.
27717 (mulsi32x16): New pattern.
27718 (mulsi64): Likewise.
27719 (mulsidi64): Likewise.
27720 (umulsidi64): Likewise.
27721 (MUL32x16_REG): Define.
27722 (mul64_600): Use MUL32x16_REG.
27723 (mac64_600): Likewise.
27724 (umul64_600): Likewise.
27725 (umac64_600): Likewise.
27727 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27729 * config/arc/arc.md (mulsi3_700): Make it commutative.
27731 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
27733 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27735 (*sign_extendsidi2_insn): Likewise for movstosw.
27737 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27739 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27740 the type of the input discriminant value. Convert the
27741 discriminant value of signedness vary.
27743 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27745 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27746 Add to -Wall section.
27748 2017-06-01 Richard Biener <rguenther@suse.de>
27750 PR middle-end/66313
27751 * fold-const.c (fold_plusminus_mult_expr): If the factored
27752 factor may be zero use a wrapping type for the inner operation.
27753 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27754 and handle moved defs.
27755 (process_assignment): Properly guard the unary op case. Return a
27756 tri-state indicating that moving the stmt before the call may allow
27757 to continue. Pass through to_move.
27758 (find_tail_calls): Handle moving unrelated defs before
27761 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27764 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27765 splitter result in the canonical way.
27767 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27769 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27770 also for 32bit target. Update insn attributes.
27771 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27773 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27775 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27776 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27777 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27778 (_mm_maskz_min_ss): New intrinsics.
27780 2017-05-31 Martin Liska <mliska@suse.cz>
27782 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27783 Change comment style to one we normally use.
27784 (vectorizable_reduction): Likewise.
27785 (vectorizable_induction): Likewise.
27786 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27787 (vectorizable_call): Likewise.
27788 (vectorizable_simd_clone_call): Likewise.
27789 (vectorizable_conversion): Likewise.
27790 (vectorizable_assignment): Likewise.
27791 (vectorizable_shift): Likewise.
27792 (vectorizable_operation): Likewise.
27793 (vectorizable_store): Likewise.
27794 (vectorizable_load): Likewise.
27795 * tree-vectorizer.h: Likewise.
27797 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27799 * passes.c (emergency_dump_function): New.
27800 * tree-pass.h (emergency_dump_function): Declare.
27801 * plugin.c (plugins_internal_error_function): Remove.
27802 * plugin.h (plugins_internal_error_function): Remove declaration.
27803 * toplev.c (internal_error_function): New static function. Use it...
27804 (general_init): ...here.
27806 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27808 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27809 (arc_rtx_costs): Add costs for new patterns.
27810 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27811 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27813 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27815 * tree-ssa-strlen.c (get_next_strinfo): New function.
27816 (get_stridx_plus_constant): Use it.
27817 (zero_length_string): Likewise.
27818 (adjust_related_strinfos): Likewise.
27819 (adjust_last_stmt): Likewise.
27821 2017-05-31 Richard Biener <rguenther@suse.de>
27824 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27825 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27827 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27829 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27830 loop_vinfo argument and use of dependence distance vectors.
27831 Check instead whether the two references differ only in their
27832 initial value and assume that they have the same alignment if the
27833 difference is a multiple of the vector alignment.
27834 (vect_analyze_data_refs_alignment): Update call accordingly.
27836 2017-05-31 Martin Liska <mliska@suse.cz>
27839 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27841 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27843 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27844 (create_intersect_range_checks): Move from ...
27845 * tree-data-ref.c (create_intersect_range_checks_index)
27846 (create_intersect_range_checks): ... to here.
27847 (create_runtime_alias_checks): New function factored from ...
27848 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27849 here. Call above function.
27850 * tree-data-ref.h (create_runtime_alias_checks): New function.
27852 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27854 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27855 segment length for dr_b and compute it in wide_int.
27857 2017-05-31 Richard Biener <rguenther@suse.de>
27859 PR tree-optimization/80906
27860 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27861 and pass through iv_map.
27862 (copy_bb_and_scalar_dependences): Adjust.
27863 (translate_pending_phi_nodes): Likewise.
27864 (copy_loop_close_phi_args): Handle code-generating IVs instead
27867 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27869 * diagnostic-color.c (color_dict): Add "type-diff".
27870 (parse_gcc_colors): Update comment.
27871 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27872 -fdiagnostics-show-template-tree and -fno-elide-type.
27873 (GCC_COLORS): Add type-diff to example.
27875 (-fdiagnostics-show-template-tree): New.
27876 (-fno-elide-type): New.
27877 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27878 the pp_format_decoder callback. Call any m_format_postprocessor's
27880 (pretty_printer::pretty_printer): Initialize
27881 m_format_postprocessor.
27882 (pretty_printer::~pretty_printer): Delete any
27883 m_format_postprocessor.
27884 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27885 (class format_postprocessor): New class.
27886 (struct pretty_printer::format_decoder): Document the new parameters.
27887 (struct pretty_printer::m_format_postprocessor): New field.
27888 * tree-diagnostic.c (default_tree_printer): Update for new
27889 bool and const char ** params.
27890 * tree-diagnostic.h (default_tree_printer): Likewise.
27892 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27894 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27895 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27896 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27897 rs6000_gen_cell_microcode code.
27898 (rs6000_final_prescan_insn): Delete.
27899 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27900 "warn-cell-microcode" entries.
27901 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27902 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27903 throughout. Change cc_reg_not_micro_cr0_operand to
27904 cc_reg_not_cr0_operand throughout.
27905 (*extendhi<mode>2_noload): Delete.
27906 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27907 (mwarn-cell-microcode): Delete.
27908 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27909 -mgen-cell-microcode and -mwarn-cell-microcode.
27911 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27914 * config/i386/constraints.md (Yd): New constraint.
27916 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27917 and (?Yd, r) alternatives. Update insn attributes.
27918 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27919 and (?*Yd, r) alternatives. Update insn attributes.
27920 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27922 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27924 * gimplify.c (gimplify_modify_expr): Don't create a
27925 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27928 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27930 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27932 2017-05-30 Richard Biener <rguenther@suse.de>
27934 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27935 and reduc_def fields.
27936 (STMT_VINFO_REDUC_TYPE): New define.
27937 (STMT_VINFO_REDUC_DEF): Likewise.
27938 (vect_force_simple_reduction): Adjust prototype.
27939 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27940 (vect_is_simple_reduction): Remove check_reduction argument.
27941 (vect_force_simple_reduction): Adjust and set
27942 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27943 (vectorizable_reduction): Do not re-do reduction analysis
27944 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27945 * tree-parloops.c (gather_scalar_reductions): Adjust.
27947 2017-05-30 Richard Biener <rguenther@suse.de>
27949 PR middle-end/80901
27950 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27953 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27955 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27956 Introduce unknown_misalignment parameter and remove vf.
27957 (vect_peeling_hash_get_lowest_cost):
27958 Pass unknown_misalignment parameter.
27959 (vect_enhance_data_refs_alignment):
27960 Fix unsupportable data ref treatment.
27962 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27964 * tree-vect-data-refs.c (vect_get_data_access_cost):
27965 Workaround for SLP handling.
27966 (vect_enhance_data_refs_alignment):
27967 Compute costs for doing no peeling at all, compare to the best
27968 peeling costs so far and avoid peeling if cheaper.
27970 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27972 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27973 Return peeling info and set costs to zero for unlimited cost
27975 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27976 with unknown misalignment. Compute and costs for unknown
27977 misalignment, compare them to the costs for known misalignment
27978 and choose the cheapest for peeling.
27980 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27982 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27983 (vect_get_peeling_costs_all_drs): Create function.
27984 (vect_peeling_hash_get_lowest_cost):
27985 Use vect_get_peeling_costs_all_drs.
27986 (vect_peeling_supportable): Create function.
27987 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27989 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27991 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27992 DR_HAS_NEGATIVE_STEP.
27993 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27994 (vect_enhance_data_refs_alignment): Use.
27995 (vect_duplicate_ssa_name_ptr_info): Use.
27996 * tree-vectorizer.h (dr_misalignment): Use.
27997 (known_alignment_for_access_p): Use.
27999 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28002 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28004 (has_section_name): New function.
28006 2017-05-30 Martin Liska <mliska@suse.cz>
28009 * auto-profile.c (get_function_decl_from_block): Fix
28012 2017-05-30 Richard Biener <rguenther@suse.de>
28014 PR middle-end/80876
28015 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28017 2017-05-30 Martin Liska <mliska@suse.cz>
28019 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28020 * dumpfile.h (struct dump_file_info): Remove ctors.
28022 2017-05-30 Martin Liska <mliska@suse.cz>
28024 * predict.def: Fix GNU coding style.
28026 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28028 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28029 Mark 'to' argument with ATTRIBUTE_UNUSED.
28031 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28033 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28034 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28035 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28038 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28040 * doc/install.texi (Options specification): Restore entry of
28041 --enable-sjlj-exceptions.
28043 2017-05-27 Michael Eager <eager@eagercon.com>
28046 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28048 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28050 * config/microblaze/microblaze.h
28051 (FIXED_REGISTERS): Update in macro.
28052 (CALL_USED_REGISTERS): Update in macro.
28054 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28056 * doc/install.texi: Add links to macOS binary distributions.
28058 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28062 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28064 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28066 2017-05-26 Martin Liska <mliska@suse.cz>
28068 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28070 2017-05-26 Martin Liska <mliska@suse.cz>
28072 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28073 always leading ';; '.
28074 (dump_bb_info): Likewise.
28075 (brief_dump_cfg): Likewise.
28076 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28077 * dumpfile.c: Remove usage of TDF_VERBOSE.
28078 * dumpfile.h (enum dump_kind): Likewise.
28079 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28080 * print-tree.c (debug_verbose): Remove.
28081 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28082 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28083 * tree-diagnostic.c (default_tree_printer): Replace
28084 TDF_DIAGNOSTIC with TDF_SLIM.
28086 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28088 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28089 in parameter loop, rather than loop_vinfo.
28090 (create_intersect_range_checks): Ditto.
28091 (vect_create_cond_for_alias_checks): Update call to above functions.
28093 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28095 PR tree-optimization/80815
28096 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28097 for merging runtime alias checks. Handle negative DR_STEPs.
28099 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28101 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28103 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28104 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28105 out code pruning runtime alias checks.
28106 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28107 factored out from above.
28108 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28110 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28112 (prune_runtime_alias_test_list): New decalaration.
28114 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28116 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28117 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28118 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28119 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28120 (operator==, comp_dr_with_seg_len_pair): Ditto.
28121 (vect_prune_runtime_alias_test_list): Ditto.
28123 2017-05-26 Martin Liska <mliska@suse.cz>
28126 * params.def: Bound partial-inlining-entry-probability param.
28128 2017-05-26 Marek Polacek <polacek@redhat.com>
28131 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28134 2017-05-26 Richard Biener <rguenther@suse.de>
28136 PR tree-optimization/80842
28137 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28140 2017-05-26 Richard Biener <rguenther@suse.de>
28142 PR tree-optimization/80844
28143 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28145 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28147 * doc/md.texi (Machine Constraints): Update x86 family
28148 machine constraints section to match 'config/i386/constraints.md'.
28150 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28152 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28154 2017-05-25 Nathan Sidwell <nathan@acm.org>
28156 * doc/invoke.texi (--enable-languages): Update documentation.
28158 2017-05-25 Martin Liska <mliska@suse.cz>
28160 * dumpfile.c: Add TDF_FOLDING.
28161 * dumpfile.h (enum dump_kind): Likewise.
28162 * genmatch.c (dt_simplify::gen_1): Use it.
28164 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28166 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28168 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28170 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28171 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28173 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28175 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28176 * match.pd (X == C): Rewrite it here.
28177 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28178 with_certain_nonzero_bits2): New predicates.
28179 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28181 2017-05-24 Nathan Sidwell <nathan@acm.org>
28183 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28186 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28189 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28191 * config/powerpcspe: New port. Files are copied from the rs6000
28192 port, with "rs6000" in filenames replaced by "powerpcspe".
28194 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28196 PR rtl-optimization/80754
28197 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28199 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28201 * config/sparc/sparc.md (length): Return the correct value for -mflat
28202 sibcalls to match output_sibcall.
28204 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28208 * config/rs6000/rs6000.c (struct machine_function): Add new field
28210 (rs6000_get_separate_components): Init that field, use it.
28211 (rs6000_components_for_bb): Use the field.
28213 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28215 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28217 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28219 PR middle-end/80823
28220 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28222 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28225 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28226 address against address_operand predicate.
28227 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28229 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28231 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28232 parameters passed indirectly.
28234 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28236 * config/i386/i386.md (*movdi_internal): Remove SSE4
28237 alternative 18 (?r, *v). Update insn attributes.
28238 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28239 Update insn attributes.
28240 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28241 Update insn attributes.
28242 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28243 alternative 1 (r, v). Remove isa attribute.
28244 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28245 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28246 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28248 2017-05-23 Tom de Vries <tom@codesourcery.com>
28250 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28253 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28255 * cgraphunit.c (symbol_table::process_new_functions): Update.
28256 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28257 (inline_generate_summary): Rename to ...
28258 (ipa_fn_summary_generate): ... this one.
28259 (inline_read_summary): Rename to ...
28260 (ipa_fn_summary_read): ... this one.
28261 (inline_write_summary): Rename to ...
28262 (ipa_fn_summary_write): ... this one.
28263 (inline_free_summary): Rename to ...
28264 (ipa_free_fn_summary): ... this one.
28265 (pass_data_local_fn_summary, pass_local_fn_summary,
28266 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28267 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28268 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28269 make_pass_ipa_fn_summary): New.
28270 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28271 inline_write_summary, inline_free_summary): Remove.
28272 (ipa_free_fn_summary) : New.
28273 * ipa-inline.c (ipa_inline): Update.
28274 (pass_ipa_inline): Do not generate summaries.
28275 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28277 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28278 and add pass_ipa_fn_summary.
28279 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28281 (make_pass_inline_parameters): Remove.
28283 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28285 * omp-low.c (struct omp_context): Remove "default_kind" member.
28288 * omp-offload.c (execute_oacc_device_lower): Remove the
28289 parallelism dimensions function attributes for unparallelized
28290 OpenACC kernels constructs.
28292 2017-05-23 Martin Liska <mliska@suse.cz>
28294 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28296 (cgraph_edge::make_speculative): Likewise.
28297 (cgraph_edge::resolve_speculation): Likewise.
28298 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28299 (cgraph_node::dump): Likewise.
28300 * cgraph.h: Likewise.
28301 * cgraphunit.c (analyze_functions): Likewise.
28302 (symbol_table::compile): Likewise.
28303 * ipa-cp.c (print_all_lattices): Likewise.
28304 (determine_versionability): Likewise.
28305 (initialize_node_lattices): Likewise.
28306 (ipcp_verify_propagated_values): Likewise.
28307 (estimate_local_effects): Likewise.
28308 (update_profiling_info): Likewise.
28309 (create_specialized_node): Likewise.
28310 (perhaps_add_new_callers): Likewise.
28311 (decide_about_value): Likewise.
28312 (decide_whether_version_node): Likewise.
28313 (identify_dead_nodes): Likewise.
28314 (ipcp_store_bits_results): Likewise.
28315 * ipa-devirt.c (dump_targets): Likewise.
28316 (ipa_devirt): Likewise.
28317 * ipa-icf.c (sem_item::dump): Likewise.
28318 (sem_function::equals): Likewise.
28319 (sem_variable::equals): Likewise.
28320 (sem_item_optimizer::read_section): Likewise.
28321 (sem_item_optimizer::execute): Likewise.
28322 (congruence_class::dump): Likewise.
28323 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28324 (dump_inline_summary): Likewise.
28325 (estimate_node_size_and_time): Likewise.
28326 (inline_analyze_function): Likewise.
28327 * ipa-inline-transform.c (inline_call): Likewise.
28328 * ipa-inline.c (report_inline_failed_reason): Likewise.
28329 (want_early_inline_function_p): Likewise.
28330 (edge_badness): Likewise.
28331 (update_edge_key): Likewise.
28332 (inline_small_functions): Likewise.
28333 * ipa-profile.c (ipa_profile): Likewise.
28334 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28335 (ipa_make_edge_direct_to_target): Likewise.
28336 (remove_described_reference): Likewise.
28337 (ipa_impossible_devirt_target): Likewise.
28338 (propagate_controlled_uses): Likewise.
28339 (ipa_print_node_params): Likewise.
28340 (ipcp_transform_function): Likewise.
28341 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28342 (propagate_pure_const): Likewise.
28343 * ipa-reference.c (generate_summary): Likewise.
28344 (read_write_all_from_decl): Likewise.
28345 (propagate): Likewise.
28346 (ipa_reference_read_optimization_summary): Likewise.
28347 * ipa-utils.c (ipa_merge_profiles): Likewise.
28348 * ipa.c (walk_polymorphic_call_targets): Likewise.
28349 (symbol_table::remove_unreachable_nodes): Likewise.
28350 (ipa_single_use): Likewise.
28351 * passes.c (execute_todo): Likewise.
28352 * predict.c (drop_profile): Likewise.
28353 * symtab.c (symtab_node::get_dump_name): New function.
28354 (symtab_node::dump_name): Likewise.
28355 (symtab_node::dump_asm_name): Likewise.
28356 (symtab_node::dump_references): Likewise.
28357 (symtab_node::dump_referring): Likewise.
28358 (symtab_node::dump_base): Likewise.
28359 (symtab_node::debug_symtab): Likewise.
28360 * tree-sra.c (convert_callers_for_node): Likewise.
28361 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28362 * value-prof.c (init_node_map): Likewise.
28364 2017-05-23 Martin Liska <mliska@suse.cz>
28366 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28367 and symtab_node::debug_symtab to symbol_table::debug.
28368 * cgraphunit.c (analyze_functions): Use the renamed function.
28369 (symbol_table::compile): Likewise.
28370 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28371 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28372 * passes.c (execute_todo): Likewise.
28373 * symtab.c (symbol_table::dump): New function.
28374 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28376 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28378 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28379 that nonconst implies exec.
28381 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28383 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28384 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28385 (inline_edge_summary_vec): Turn into ...
28386 (ipa_call_summaries): ... this one.
28387 (redirect_to_unreachable, edge_set_predicate,
28388 evaluate_properties_for_edge, inline_summary_alloc,
28389 reset_ipa_call_summary, reset_inline_summary,
28390 inline_summary_t::duplicate): Update.
28391 (inline_edge_duplication_hook): Turn to ...
28392 (ipa_call_summary_t::duplicate): ... this one.
28393 (inline_edge_removal_hook): Turn to ...
28394 (ipa_call_summary_t::remove): ... this one.
28395 (dump_inline_edge_summary): Turn to ...
28396 (dump_ipa_call_summary): ... this one.
28397 (estimate_function_body_sizes): Update.
28398 (inline_update_callee_summaries): Update.
28399 (remap_edge_change_prob): Update.
28400 (remap_edge_summaries): Update.
28401 (inline_merge_summary): Update.
28402 (do_estimate_edge_time): Update.
28403 (inline_generate_summary): Update.
28404 (inline_read_section): Update.
28405 (inline_read_summary): Update.
28406 (inline_free_summary): Update.
28407 * ipa-inline.c (can_inline_edge_p): Update.
28408 (compute_inlined_call_time): Update.
28409 (want_inline_small_function_p): Update.
28410 (edge_badness): Update.
28411 (early_inliner): Update.
28412 * ipa-inline.h (inline_edge_summary): Turn to ...
28413 (ipa_call_summary): ... this one.
28414 (ipa_call_summary_t): New class.
28415 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28416 (ipa_call_summaries): New.
28417 (inline_edge_summary): Remove.
28418 (estimate_edge_growth): Update.
28419 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28420 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28421 * ipa-split.c (execute_split_functions): Update.
28422 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28424 2017-05-23 Tom de Vries <tom@codesourcery.com>
28426 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28427 attributes): Document rdrand effective target.
28429 2017-05-23 Tom de Vries <tom@codesourcery.com>
28431 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28432 attributes): Sort alphabetically.
28434 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28436 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28438 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28441 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28442 V2DF/V2DI splat into two separate patterns, one that handles
28443 registers, and the other that only handles memory. Drop support
28444 for splatting from a GPR on ISA 2.07 and then splitting the
28445 splat into direct move and splat.
28446 (vsx_splat_<mode>_reg): Likewise.
28447 (vsx_splat_<mode>_mem): Likewise.
28449 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28451 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28453 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28455 PR middle-end/80809
28456 * omp-low.c (finish_taskreg_remap): New function.
28457 (finish_taskreg_scan): If unit size of ctx->record_type
28458 is non-constant, unshare the size expression and replace
28459 decls in it with possible outer var refs.
28461 PR middle-end/80809
28462 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28463 GOVD_SHARED rather than GOVD_PRIVATE with it.
28464 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28465 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28467 PR middle-end/80853
28468 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28469 as last argument to build_outer_var_ref for pointer bases of array
28470 section reductions.
28472 2017-05-19 Martin Sebor <msebor@redhat.com>
28474 * print-tree.c (print_node): Print DECL_READ_P flag.
28476 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28478 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28479 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28480 * cgraph.c: Likewise.
28481 * cgraphunit.c: Likewise.
28482 * gengtype.c: Likewise.
28483 * ipa-cp.c: Likewise.
28484 * ipa-devirt.c: Likewise.
28485 * ipa-icf.c: Likewise.
28486 * ipa-predicate.c: Likewise.
28487 * ipa-profile.c: Likewise.
28488 * ipa-prop.c: Likewise.
28489 * ipa-split.c: Likewise.
28491 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28492 edge_predicate_pool, dump_inline_hints,
28493 inline_summary::account_size_time, redirect_to_unreachable,
28494 edge_set_predicate, set_hint_predicate,
28495 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28496 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28497 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28498 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28499 ipa_call_summary_t::remove, initialize_growth_caches,
28500 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28501 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28502 mark_modified, unmodified_parm_1, unmodified_parm,
28503 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28504 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28505 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28506 will_be_nonconstant_predicate, record_modified_bb_info,
28507 get_minimal_bb, record_modified, param_change_prob,
28508 phi_result_unknown_predicate, predicate_for_phi_result,
28509 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28510 estimate_function_body_sizes, compute_inline_parameters,
28511 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28512 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28513 inline_update_callee_summaries, remap_edge_change_prob,
28514 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28515 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28516 inline_analyze_function, inline_summary_t::insert,
28517 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28518 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28519 inline_free_summary): Move to ipa-fnsummary.h
28520 (predicate_t): Remove.
28521 * ipa-fnsummary.c: New file.
28522 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28523 (enum inline_hints_vals, inline_hints, agg_position_info,
28524 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28525 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28526 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28527 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28528 inline_read_summary, inline_write_summary, inline_free_summary,
28529 inline_analyze_function, initialize_inline_failed,
28530 inline_merge_summary, inline_update_overall_summary,
28531 compute_inline_parameters): Move to ipa-fnsummary.h
28532 * ipa-fnsummary.h: New file.
28533 * ipa-inline-transform.h: Include ipa-inline.h.
28534 * ipa-inline.c: LIkewise.
28536 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28538 * ipa-inline.c (edge_badness): Use inlined_time instead of
28539 inline_summaries->get.
28541 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28543 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28545 2017-05-22 Nathan Sidwell <nathan@acm.org>
28547 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28548 (fdump-lang): Document 'raw' option.
28549 * dumpfile.h (TDI_tu): Delete.
28550 * dumpfile.c (dump_files): Remove translation-unit.
28551 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28553 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28555 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28556 command option from $(AWK) call.
28557 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28558 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28559 [FORMAT]: Remove handling of variable.
28560 * config/avr/t-multilib: Regenerate.
28562 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28564 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28566 (dump_inline_summary): Do not print self_time.
28567 (estimate_function_body_sizes): Do not set self_time.
28568 (compute_inline_parameters): Likewise.
28569 (inline_read_section, inline_write_summary): Do not stream self_time.
28570 * ipa-inline.h (inline_summary): Drop self_time.
28572 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28574 * ipa-inline-analysis.c (account_size_time): Rename to ...
28575 (inline_summary::account_size_time): ... this one.
28576 (reset_ipa_call_summary): Turn to ...
28577 (ipa_call_summary::reset): ... this one.
28578 (reset_inline_summary): Turn to ...
28579 (inline_summary::reset): ... this one.
28580 (inline_summary_t::remove): Update.
28581 (inline_summary_t::duplicate): Update.
28582 (ipa_call_summary_t::remove): Update.
28583 (dump_inline_summary): Update.
28584 (estimate_function_body_sizes): Update.
28585 (compute_inline_parameters): Update.
28586 (estimate_node_size_and_time): Update.
28587 (inline_merge_summary): Update.
28588 (inline_update_overall_summary): Update.
28589 (inline_read_section): Update.
28590 (inline_write_summary): Update.
28591 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28592 add account_size_time and reset member functions.
28593 (ipa_call_summary): Add reset function.
28594 * ipa-predicate.h (predicate::operator &): Constify.
28596 2017-05-22 Richard Biener <rguenther@suse.de>
28598 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28600 2017-05-19 Jason Merrill <jason@redhat.com>
28602 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28604 2017-05-19 Marek Polacek <polacek@redhat.com>
28607 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28608 TYPE_OVERFLOW_WRAPS checks.
28610 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28612 * tree-core.h (enum omp_clause_default_kind): Add
28613 "OMP_CLAUSE_DEFAULT_PRESENT".
28614 * tree-pretty-print.c (dump_omp_clause): Handle it.
28615 * gimplify.c (enum gimplify_omp_var_data): Add
28616 "GOVD_MAP_FORCE_PRESENT".
28617 (gimplify_adjust_omp_clauses_1): Map it to
28618 "GOMP_MAP_FORCE_PRESENT".
28619 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28621 * gimplify.c (oacc_default_clause): Clarify.
28623 2017-05-19 Nathan Sidwell <nathan@acm.org>
28625 LANG_HOOK_REGISTER_DUMPS
28626 * toplev.c (general_init): Call register dump lang hook.
28627 * doc/invoke.texi: Document -fdump-lang option family.
28628 * dumpfile.c (dump_files): Remove class dump here.
28629 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28630 * dumpfile.h (tree_dump_index): Remove TDI_class.
28631 * langhooks-def.h (lhd_register_dumps): Declare.
28632 (LANG_HOOKS_REGISTER_DUMPS): Define.
28633 (LANG_HOOKS_INITIALIZER): Add it.
28634 * langhooks.c (lhd_register_dumps): Define.
28635 * langhooks.h (struct lang_hooks): Add register_dumps.
28637 2017-05-19 Nathan Sidwell <nathan@acm.org>
28639 * context.h (context::set_passes): New.
28640 * context.c (context::context): Do not create pass manager.
28641 * toplev.c (general_init): Create pass manager here.
28643 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28645 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28646 use this splitter if two add or or instructions would also work for
28647 the constant we want to generate.
28649 2017-05-19 Richard Biener <rguenther@suse.de>
28652 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28653 predicate evaluation.
28655 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28657 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28659 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28662 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28664 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28665 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28666 (inline_edge_summary_vec): Turn into ...
28667 (ipa_call_summaries): ... this one.
28668 (redirect_to_unreachable, edge_set_predicate,
28669 evaluate_properties_for_edge, inline_summary_alloc,
28670 reset_ipa_call_summary, reset_inline_summary,
28671 inline_summary_t::duplicate): Update.
28672 (inline_edge_duplication_hook): Turn to ...
28673 (ipa_call_summary_t::duplicate): ... this one.
28674 (inline_edge_removal_hook): Turn to ...
28675 (ipa_call_summary_t::remove): ... this one.
28676 (dump_inline_edge_summary): Turn to ...
28677 (dump_ipa_call_summary): ... this one.
28678 (estimate_function_body_sizes): Update.
28679 (inline_update_callee_summaries): Update.
28680 (remap_edge_change_prob): Update.
28681 (remap_edge_summaries): Update.
28682 (inline_merge_summary): Update.
28683 (do_estimate_edge_time): Update.
28684 (inline_generate_summary): Update.
28685 (inline_read_section): Update.
28686 (inline_read_summary): Update.
28687 (inline_free_summary): Update.
28688 * ipa-inline.c (can_inline_edge_p): Update.
28689 (compute_inlined_call_time): Update.
28690 (want_inline_small_function_p): Update.
28691 (edge_badness): Update.
28692 (early_inliner): Update.
28693 * ipa-inline.h (inline_edge_summary): Turn to ...
28694 (ipa_call_summary): ... this one.
28695 (ipa_call_summary_t): New class.
28696 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28697 (ipa_call_summaries): New.
28698 (inline_edge_summary): Remove.
28699 (estimate_edge_growth): Update.
28700 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28701 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28702 * ipa-split.c (execute_split_functions): Update.
28703 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28705 2017-05-19 Richard Biener <rguenther@suse.de>
28707 PR middle-end/80764
28708 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28710 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28712 * config/rs6000/rs6000.c (struct machine_function): Add field
28713 fpr_is_wrapped_separately.
28714 (rs6000_get_separate_components): Use 64 components. Handle the
28715 new FPR components.
28716 (rs6000_components_for_bb): Handle the FPR components.
28717 (rs6000_emit_prologue_components): Handle the FPR components.
28718 (rs6000_emit_epilogue_components): Handle the FPR components.
28719 (rs6000_set_handled_components): Handle the FPR components.
28720 (rs6000_emit_prologue): Don't output prologue code for those FPRs
28721 that are already separately shrink-wrapped.
28722 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
28723 that are already separately shrink-wrapped.
28725 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28728 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
28731 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
28732 (define_peephole2 for Altivec d-form load): Add peepholes to catch
28733 cases where the register allocator uses a move and an offsettable
28734 memory operation to/from a FPR register on ISA 2.06/2.07.
28735 (define_peephole2 for Altivec d-form store): Likewise.
28737 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28740 * config/i386/mmx.md (*mov<mode>_internal): Enable
28741 alternatives 11, 12, 13 and 14 also for 32bit targets.
28742 Remove alternatives 15, 16, 17 and 18.
28743 * config/i386/sse.md (vec_concatv2di): Change
28744 alternative (!x, *y) to (x, ?!*Yn).
28746 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28748 * dumpfile.h (enum dump_kind): Remove stray comma.
28750 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28752 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28753 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28754 predicate::num_conditions
28755 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28756 (CHANGED): turn into predicate::changed.
28757 (agg_position_info): Move to ipa-predicate.h
28758 (add_condition, predicate::add_clause, predicate::operator &=,
28759 predicate::or_with, predicate::evaluate, predicate::probability,
28760 dump_condition, dump_clause, predicate::dump,
28761 predicate::remap_after_duplication, predicate::remap_after_inlining,
28762 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28763 (evaluate_conditions_for_known_args): Update.
28764 (set_cond_stmt_execution_predicate): Update.
28765 * ipa-inline.h: Include ipa-predicate.h
28766 (condition, inline_param_summary, conditions, agg_position_info,
28767 predicate): Move to ipa-predicate.h
28768 * ipa-predicate.c: New file.
28769 * ipa-predicate.h: New file.
28771 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28773 * final.c (leaf_function_p): Check we are not in a sequence.
28775 2017-05-18 Martin Liska <mliska@suse.cz>
28777 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28778 * dumpfile.c (dump_register): Use new enum dump_kind.
28779 (get_dump_file_name): Likewise.
28780 (dump_enable_all): Likewise.
28781 (dump_switch_p_1): Likewise.
28782 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28783 * dumpfile.h (enum dump_kind): New enum type.
28784 (struct dump_file_info): Create constructor and
28785 format fields and comments.
28786 * passes.c (pass_manager::register_one_dump_file):
28788 * statistics.c (statistics_early_init): Likewise.
28789 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28790 TDF_TREE with TDF_SLIM.
28791 (gather_memory_references_ref): Likewise.
28793 2017-05-18 Martin Liska <mliska@suse.cz>
28795 * vec.h (struct vnull): Use it.
28797 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28799 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28800 (true_predicate, false_predicate, true_predicate_p,
28801 false_predicate_p): Remove.
28802 (single_cond_predicate, not_inlined_predicate): Turn to member function
28804 (add_condition): Update.
28805 (add_clause): Turn to...
28806 (predicate::add_clause): ... this one; update; allow passing NULL
28808 (and_predicates): Turn to ...
28809 (predicate::operator &=): ... this one.
28810 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28811 (or_predicates): Turn to ...
28812 (predicate::or_with): ... this one.
28813 (evaluate_predicate): Turn to ...
28814 (predicate::evaluate): ... this one.
28815 (predicate_probability): Turn to ...
28816 (predicate::probability): ... this one.
28817 (dump_condition): Update.
28818 (dump_predicate): Turn to ...
28819 (predicate::dump): ... this one.
28820 (account_size_time): Update.
28821 (edge_set_predicate): Update.
28822 (set_hint_predicate): UPdate.
28823 (evaluate_conditions_for_known_args): Update.
28824 (evaluate_properties_for_edge): Update.
28825 (remap_predicate_after_duplication): Turn to...
28826 (predicate::remap_after_duplication): ... this one.
28827 (remap_hint_predicate_after_duplication): Update.
28828 (inline_summary_t::duplicate): UPdate.
28829 (dump_inline_edge_summary): Update.
28830 (dump_inline_summary): Update.
28831 (set_cond_stmt_execution_predicate): Update.
28832 (set_switch_stmt_execution_predicate): Update.
28833 (compute_bb_predicates): Update.
28834 (will_be_nonconstant_expr_predicate): Update.
28835 (will_be_nonconstant_predicate): Update.
28836 (phi_result_unknown_predicate): Update.
28837 (predicate_for_phi_result): Update.
28838 (array_index_predicate): Update.
28839 (estimate_function_body_sizes): Update.
28840 (estimate_node_size_and_time): Update.
28841 (estimate_ipcp_clone_size_and_time): Update.
28842 (remap_predicate): Rename to ...
28843 (predicate::remap_after_inlining): ... this one.
28844 (remap_hint_predicate): Update.
28845 (inline_merge_summary): Update.
28846 (inline_update_overall_summary): Update.
28847 (estimate_size_after_inlining): Update.
28848 (read_predicate): Rename to ...
28849 (predicate::stream_in): ... this one.
28850 (read_inline_edge_summary): Update.
28851 (write_predicate): Rename to ...
28852 (predicate::stream_out): ... this one.
28853 (write_inline_edge_summary): Update.
28854 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28855 (clause_t): Turn to uint32_t
28856 (predicate): Turn to class; implement constructor and operators
28858 (size_time_entry): Update.
28859 (inline_summary): Update.
28860 (inline_edge_summary): Update.
28862 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28864 * fold-const.c (fold_binary_loc): Move transformation...
28865 * match.pd (C - X CMP X): ... here.
28867 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28869 * config/sparc/sparc.c (sparc_option_override): Set function
28870 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28871 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28873 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28875 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28877 2017-05-18 Marek Polacek <polacek@redhat.com>
28880 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28881 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28882 gimple_assign_load_p.
28884 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28886 PR middle-end/80692
28887 * real.c (do_compare): Give decimal_do_compare preference over
28888 comparing just the signs.
28890 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28892 * doc/md.texi (Canonicalization of Instructions): Describe the
28893 canonical form of instructions that inherently set a condition
28896 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28898 PR middle-end/80775
28899 * tree-cfg.c: Move deletion of unreachable case statements to after
28900 the merging of consecutive case labels.
28902 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28904 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28905 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28906 restoring of callee-saved registers.
28908 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28910 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28911 * config/visium/visium.c (single_set_and_flags): Likewise.
28912 * config/visium/visium.md (Substitutions): Likewise.
28914 2017-05-17 Martin Liska <mliska@suse.cz>
28916 * cfg.c: Introduce dump_flags_t type and
28917 use it instead of int type.
28919 * cfghooks.c: Likewise.
28920 * cfghooks.h (struct cfg_hooks): Likewise.
28921 * cfgrtl.c: Likewise.
28922 * cfgrtl.h: Likewise.
28923 * cgraph.c (cgraph_node::get_body): Likewise.
28924 * coretypes.h: Likewise.
28925 * domwalk.c: Likewise.
28926 * domwalk.h: Likewise.
28927 * dumpfile.c (struct dump_option_value_info): Likewise.
28928 (dump_enable_all): Likewise.
28929 (dump_switch_p_1): Likewise.
28930 (opt_info_switch_p): Likewise.
28931 * dumpfile.h (enum tree_dump_index): Likewise.
28932 (struct dump_file_info): Likewise.
28933 * genemit.c: Likewise.
28934 * generic-match-head.c: Likewise.
28935 * gengtype.c (open_base_files): Likewise.
28936 * gimple-pretty-print.c: Likewise.
28937 * gimple-pretty-print.h: Likewise.
28938 * graph.c (print_graph_cfg): Likewise.
28939 * graphite-scop-detection.c (dot_all_sese): Likewise.
28940 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28941 * loop-unroll.c (report_unroll): Likewise.
28942 * passes.c (pass_manager::register_one_dump_file): Likewise.
28943 * print-tree.c: Likewise.
28944 * statistics.c: Likewise.
28945 * tree-cfg.c: Likewise.
28946 * tree-cfg.h: Likewise.
28947 * tree-dfa.c: Likewise.
28948 * tree-dfa.h: Likewise.
28949 * tree-dump.c (dump_function): Likewise.
28950 * tree-dump.h (struct dump_info): Likewise.
28951 * tree-pretty-print.c: Likewise.
28952 * tree-pretty-print.h: Likewise.
28953 * tree-ssa-live.c: Likewise.
28954 * tree-ssa-live.h: Likewise.
28955 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28956 * tree-vect-loop.c: Likewise.
28957 * tree-vect-slp.c: Likewise.
28959 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28960 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28962 PR tree-optimization/80457
28963 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28964 of all arguments to a statement as scalar_to_vec operations.
28965 (vectorizable_call): Adjust call to vect_model_simple_cost for
28967 (vectorizable_conversion): Likewise.
28968 (vectorizable_assignment): Likewise.
28969 (vectorizable_shift): Likewise.
28970 (vectorizable_operation): Likewise.
28971 (vectorizable_comparison): Likewise.
28972 (vect_is_simple_cond): Record the def types for operands.
28973 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28974 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28975 for statement argument count.
28977 2017-05-16 Carl Love <cel@us.ibm.com>
28979 * config/rs6000/rs6000-c: Add support for built-in functions
28980 vector unsigned long long vec_bperm (vector unsigned long long,
28981 vector unsigned char)
28982 vector signed long long vec_mule (vector signed int,
28984 vector unsigned long long vec_mule (vector unsigned int,
28985 vector unsigned int)
28986 vector signed long long vec_mulo (vector signed int,
28988 vector unsigned long long vec_mulo (vector unsigned int,
28989 vector unsigned int)
28990 vector signed char vec_sldw (vector signed char,
28991 vector signed char,
28993 vector unsigned char vec_sldw (vector unsigned char,
28994 vector unsigned char,
28996 vector signed short vec_sldw (vector signed short,
28997 vector signed short,
28999 vector unsigned short vec_sldw (vector unsigned short,
29000 vector unsigned short,
29002 vector signed int vec_sldw (vector signed int,
29005 vector unsigned int vec_sldw (vector unsigned int,
29006 vector unsigned int,
29008 vector signed long long vec_sldw (vector signed long long,
29009 vector signed long long,
29011 vector unsigned long long vec_sldw (vector unsigned long long,
29012 vector unsigned long long,
29014 * config/rs6000/rs6000-c: Add support for built-in functions
29015 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29016 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29017 * doc/extend.texi: Update the built-in documentation for the
29018 new built-in functions.
29020 2017-05-16 Marek Polacek <polacek@redhat.com>
29024 * tree.c (save_expr): Don't fold the expression.
29026 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29028 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29029 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29030 and (?*y,m). Update insn attributes.
29032 2017-05-16 Martin Liska <mliska@suse.cz>
29034 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29035 flags argument of print_gimple_stmt, print_gimple_expr,
29036 print_generic_stmt and print_generic_expr.
29037 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29038 * coretypes.h: Likewise.
29039 * except.c (dump_eh_tree): Likewise.
29040 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29041 * gimple-pretty-print.h: Likewise.
29042 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29043 (backprop::push_to_worklist): Likewise.
29044 (backprop::pop_from_worklist): Likewise.
29045 (backprop::process_use): Likewise.
29046 (backprop::intersect_uses): Likewise.
29047 (note_replacement): Likewise.
29048 * gimple-ssa-store-merging.c
29049 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29050 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29051 (pass_store_merging::execute): Likewise.
29052 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29053 (ssa_base_cand_dump_callback): Likewise.
29054 (dump_incr_vec): Likewise.
29055 (replace_refs): Likewise.
29056 (replace_mult_candidate): Likewise.
29057 (create_add_on_incoming_edge): Likewise.
29058 (create_phi_basis): Likewise.
29059 (insert_initializers): Likewise.
29060 (all_phi_incrs_profitable): Likewise.
29061 (introduce_cast_before_cand): Likewise.
29062 (replace_one_candidate): Likewise.
29063 * gimplify.c (gimplify_expr): Likewise.
29064 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29065 (set_rename): Likewise.
29066 (rename_uses): Likewise.
29067 (copy_loop_phi_nodes): Likewise.
29068 (add_close_phis_to_merge_points): Likewise.
29069 (copy_loop_close_phi_args): Likewise.
29070 (copy_cond_phi_args): Likewise.
29071 (graphite_copy_stmts_from_block): Likewise.
29072 (translate_pending_phi_nodes): Likewise.
29073 * graphite-poly.c (print_pdr): Likewise.
29074 (dump_gbb_cases): Likewise.
29075 (dump_gbb_conditions): Likewise.
29076 (print_scop_params): Likewise.
29077 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29078 (build_cross_bb_scalars_use): Likewise.
29079 (gather_bbs::before_dom_children): Likewise.
29080 * hsa-dump.c (dump_hsa_immed): Likewise.
29081 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29082 (get_replacement_map): Likewise.
29083 * ipa-inline-analysis.c (dump_condition): Likewise.
29084 (estimate_function_body_sizes): Likewise.
29085 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29086 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29087 * ipa-prop.c (ipa_dump_param): Likewise.
29088 (ipa_print_node_jump_functions_for_edge): Likewise.
29089 (ipa_modify_call_arguments): Likewise.
29090 (ipa_modify_expr): Likewise.
29091 (ipa_dump_param_adjustments): Likewise.
29092 (ipa_dump_agg_replacement_values): Likewise.
29093 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29094 * ipa-pure-const.c (check_stmt): Likewise.
29095 (pass_nothrow::execute): Likewise.
29096 * ipa-split.c (execute_split_functions): Likewise.
29097 * omp-offload.c (dump_oacc_loop_part): Likewise.
29098 (dump_oacc_loop): Likewise.
29099 * trans-mem.c (tm_log_emit): Likewise.
29100 (tm_memopt_accumulate_memops): Likewise.
29101 (dump_tm_memopt_set): Likewise.
29102 (dump_tm_memopt_transform): Likewise.
29103 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29104 (print_loop): Likewise.
29105 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29106 (chkp_gather_checks_info): Likewise.
29107 (chkp_get_check_result): Likewise.
29108 (chkp_remove_check_if_pass): Likewise.
29109 (chkp_use_outer_bounds_if_possible): Likewise.
29110 (chkp_reduce_bounds_lifetime): Likewise.
29111 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29112 (chkp_mark_completed_bounds): Likewise.
29113 (chkp_register_incomplete_bounds): Likewise.
29114 (chkp_mark_invalid_bounds): Likewise.
29115 (chkp_maybe_copy_and_register_bounds): Likewise.
29116 (chkp_build_returned_bound): Likewise.
29117 (chkp_get_bound_for_parm): Likewise.
29118 (chkp_build_bndldx): Likewise.
29119 (chkp_get_bounds_by_definition): Likewise.
29120 (chkp_generate_extern_var_bounds): Likewise.
29121 (chkp_get_bounds_for_decl_addr): Likewise.
29122 * tree-chrec.c (chrec_apply): Likewise.
29123 * tree-data-ref.c (dump_data_reference): Likewise.
29124 (dump_subscript): Likewise.
29125 (dump_data_dependence_relation): Likewise.
29126 (analyze_overlapping_iterations): Likewise.
29127 * tree-inline.c (expand_call_inline): Likewise.
29128 (tree_function_versioning): Likewise.
29129 * tree-into-ssa.c (dump_defs_stack): Likewise.
29130 (dump_currdefs): Likewise.
29131 (dump_names_replaced_by): Likewise.
29132 (dump_update_ssa): Likewise.
29133 (update_ssa): Likewise.
29134 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29135 * tree-parloops.c (build_new_reduction): Likewise.
29136 (try_create_reduction_list): Likewise.
29137 (ref_conflicts_with_region): Likewise.
29138 (oacc_entry_exit_ok_1): Likewise.
29139 (oacc_entry_exit_single_gang): Likewise.
29140 * tree-pretty-print.h: Likewise.
29141 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29142 (get_scalar_evolution): Likewise.
29143 (add_to_evolution): Likewise.
29144 (get_loop_exit_condition): Likewise.
29145 (analyze_evolution_in_loop): Likewise.
29146 (analyze_initial_condition): Likewise.
29147 (analyze_scalar_evolution): Likewise.
29148 (instantiate_scev): Likewise.
29149 (number_of_latch_executions): Likewise.
29150 (gather_chrec_stats): Likewise.
29151 (final_value_replacement_loop): Likewise.
29152 (scev_const_prop): Likewise.
29153 * tree-sra.c (dump_access): Likewise.
29154 (disqualify_candidate): Likewise.
29155 (create_access): Likewise.
29156 (reject): Likewise.
29157 (maybe_add_sra_candidate): Likewise.
29158 (create_access_replacement): Likewise.
29159 (analyze_access_subtree): Likewise.
29160 (analyze_all_variable_accesses): Likewise.
29161 (sra_modify_assign): Likewise.
29162 (initialize_constant_pool_replacements): Likewise.
29163 (find_param_candidates): Likewise.
29164 (decide_one_param_reduction): Likewise.
29165 (replace_removed_params_ssa_names): Likewise.
29166 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29167 * tree-ssa-copy.c (dump_copy_of): Likewise.
29168 (copy_prop_visit_cond_stmt): Likewise.
29169 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29170 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29171 (record_equivalences_from_stmt): Likewise.
29172 * tree-ssa-dse.c (compute_trims): Likewise.
29173 (delete_dead_call): Likewise.
29174 (delete_dead_assignment): Likewise.
29175 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29176 (forward_propagate_into_cond): Likewise.
29177 (pass_forwprop::execute): Likewise.
29178 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29179 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29181 (move_computations_worker): Likewise.
29182 (execute_sm): Likewise.
29183 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29184 (remove_exits_and_undefined_stmts): Likewise.
29185 (remove_redundant_iv_tests): Likewise.
29186 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29187 (adjust_iv_update_pos): Likewise.
29188 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29189 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29190 (value_replacement): Likewise.
29191 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29192 * tree-ssa-pre.c (print_pre_expr): Likewise.
29193 (get_representative_for): Likewise.
29194 (create_expression_by_pieces): Likewise.
29195 (insert_into_preds_of_block): Likewise.
29196 (eliminate_insert): Likewise.
29197 (eliminate_dom_walker::before_dom_children): Likewise.
29198 (eliminate): Likewise.
29199 (remove_dead_inserted_code): Likewise.
29200 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29201 * tree-ssa-reassoc.c (get_rank): Likewise.
29202 (eliminate_duplicate_pair): Likewise.
29203 (eliminate_plus_minus_pair): Likewise.
29204 (eliminate_not_pairs): Likewise.
29205 (undistribute_ops_list): Likewise.
29206 (eliminate_redundant_comparison): Likewise.
29207 (update_range_test): Likewise.
29208 (optimize_range_tests_var_bound): Likewise.
29209 (optimize_vec_cond_expr): Likewise.
29210 (rewrite_expr_tree): Likewise.
29211 (rewrite_expr_tree_parallel): Likewise.
29212 (linearize_expr): Likewise.
29213 (break_up_subtract): Likewise.
29214 (linearize_expr_tree): Likewise.
29215 (attempt_builtin_powi): Likewise.
29216 (attempt_builtin_copysign): Likewise.
29217 (transform_stmt_to_copy): Likewise.
29218 (transform_stmt_to_multiply): Likewise.
29219 (dump_ops_vector): Likewise.
29220 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29221 (print_scc): Likewise.
29222 (set_ssa_val_to): Likewise.
29223 (visit_reference_op_store): Likewise.
29224 (visit_use): Likewise.
29225 (sccvn_dom_walker::before_dom_children): Likewise.
29226 (run_scc_vn): Likewise.
29227 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29229 (expr_hash_elt::print): Likewise.
29230 (const_and_copies::pop_to_marker): Likewise.
29231 (const_and_copies::record_const_or_copy_raw): Likewise.
29232 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29233 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29234 (dump_predicates): Likewise.
29235 (find_uninit_use): Likewise.
29236 (warn_uninitialized_phi): Likewise.
29237 (pass_late_warn_uninitialized::execute): Likewise.
29238 * tree-ssa.c (verify_vssa): Likewise.
29239 (verify_ssa): Likewise.
29240 (maybe_optimize_var): Likewise.
29241 * tree-vrp.c (dump_value_range): Likewise.
29242 (dump_all_value_ranges): Likewise.
29243 (dump_asserts_for): Likewise.
29244 (register_edge_assert_for_2): Likewise.
29245 (vrp_visit_cond_stmt): Likewise.
29246 (vrp_visit_switch_stmt): Likewise.
29247 (vrp_visit_stmt): Likewise.
29248 (vrp_visit_phi_node): Likewise.
29249 (simplify_cond_using_ranges_1): Likewise.
29250 (fold_predicate_in): Likewise.
29251 (evrp_dom_walker::before_dom_children): Likewise.
29252 (evrp_dom_walker::push_value_range): Likewise.
29253 (evrp_dom_walker::pop_value_range): Likewise.
29254 (execute_early_vrp): Likewise.
29256 2017-05-16 Richard Biener <rguenther@suse.de>
29258 * dwarf2out.c (loc_list_from_tree_1): Do not create
29259 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29261 2017-05-16 Richard Biener <rguenther@suse.de>
29263 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29265 (note_variable_value_in_expr): If we resolved the decl ref
29266 do not push to the stack.
29268 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29270 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29271 operations in fast-math mode.
29272 (vaddq_f16): Likewise.
29273 (vmul_f16): Likewise.
29274 (vmulq_f16): Likewise.
29275 (vsub_f16): Likewise.
29276 (vsubq_f16): Likewise.
29277 * config/arm/neon.md (add<mode>3): New.
29279 (fma:<VH:mode>3): New. Also remove outdated comment.
29282 2017-05-16 Martin Liska <mliska@suse.cz>
29286 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29287 (odr_types_equivalent_p): Likewise.
29289 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29291 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29293 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29296 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29297 non-interunit SSE move alternatives with '?'.
29298 (zero-extendsidi peephole2): New peephole to skip intermediate
29299 general register in SSE zero-extend sequence.
29301 2017-05-15 Jeff Law <law@redhat.com>
29303 * reorg.c (relax_delay_slots): Create a new variable to hold
29304 the temporary target rather than clobbering TARGET_LABEL.
29306 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29307 missing argument to extract_bit_field call.
29308 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29310 2017-05-15 Martin Liska <mliska@suse.cz>
29313 * gcc.c (process_command): Do not allow empty argument of -o option.
29315 2017-05-15 Renlin Li <renlin.li@arm.com>
29317 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29318 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29319 * config/aarch64/constraints.md (Usf): Add long call check.
29320 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29321 (call_value): Likewise.
29322 (sibcall): Likewise.
29323 (sibcall_value): Likewise.
29325 (call_value_insn): New.
29326 (sibcall_insn): Update rtx pattern.
29327 (sibcall_value_insn): Likewise.
29328 (call_internal): Remove.
29329 (call_value_internal): Likewise.
29330 (sibcall_internal): Likewise.
29331 (sibcall_value_internal): Likewise.
29332 (call_reg): Likewise.
29333 (call_symbol): Likewise.
29334 (call_value_reg): Likewise.
29335 (call_value_symbol): Likewise.
29337 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29340 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29342 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29344 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29345 compatible with CCGOCmode and with CCZmode.
29347 2017-05-14 Martin Sebor <msebor@redhat.com>
29349 PR middle-end/77671
29350 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29351 (gimple_fold_builtin_snprintf): Same.
29352 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29353 (gimple_fold_builtin_snprintf): Same.
29354 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29355 of character types.
29356 (is_call_safe): New function.
29357 (try_substitute_return_value): Call it.
29358 (try_simplify_call): New function.
29359 (pass_sprintf_length::handle_gimple_call): Call it.
29361 2017-05-14 Martin Sebor <msebor@redhat.com>
29363 PR middle-end/80669
29364 * builtins.c (expand_builtin_stpncpy): Simplify.
29366 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29368 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29369 * config/i386/i386.h
29370 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29371 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29372 (struct machine_function): Add new members call_ms2sysv,
29373 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29374 (struct machine_frame_state): New fields sp_realigned and
29375 sp_realigned_offset.
29376 * config/i386/i386.c
29377 (enum xlogue_stub): New enum.
29378 (enum xlogue_stub_sets): New enum.
29379 (class xlogue_layout): New class.
29380 (struct ix86_frame): New fields stack_realign_allocate_offset,
29381 stack_realign_offset and outlined_save_offset. Modify comments to
29382 detail stack layout when using out-of-line stubs.
29383 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29384 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29385 -mcall-ms2sysv-xlogues.
29386 (stub_managed_regs): New static variable.
29387 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29388 registers managed by out-of-line stub.
29389 (disable_call_ms2sysv_xlogues): New function.
29390 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29391 m->call_ms2sysv when appropriate and compute frame layout for
29393 (sp_valid_at, fp_valid_at): New inline functions.
29394 (choose_basereg): New function.
29395 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29397 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29398 Use align parameter of choose_baseaddr to generated aligned SSE movs
29400 (pro_epilogue_adjust_stack): Modify to track
29401 machine_frame_state::sp_realigned.
29402 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29403 (ix86_nsaved_sseregs): Likewise.
29404 (ix86_emit_save_regs): Likewise.
29405 (ix86_emit_save_regs_using_mov): Likewise.
29406 (ix86_emit_save_sse_regs_using_mov): Likewise.
29407 (get_scratch_register_on_entry): Likewise.
29408 (gen_frame_set): New function.
29409 (gen_frame_load): Likewise.
29410 (gen_frame_store): Likewise.
29411 (emit_outlined_ms2sysv_save): Likewise.
29412 (emit_outlined_ms2sysv_restore): Likewise.
29413 (ix86_expand_prologue): Modify stack re-alignment code and call
29414 emit_outlined_ms2sysv_save when appropriate.
29415 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29416 parameter rtx_insn *insn, which allows the function to be used to only
29417 generate the notes.
29418 (ix86_expand_epilogue): Modify validity checks of frame and stack
29419 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29420 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29421 * config/i386/predicates.md
29422 (save_multiple): New predicate.
29423 (restore_multiple): Likewise.
29424 * config/i386/sse.md
29425 (save_multiple<mode>): New pattern.
29426 (save_multiple_realign<mode>): Likewise.
29427 (restore_multiple<mode>): Likewise.
29428 (restore_multiple_and_return<mode>): Likewise.
29429 (restore_multiple_leave_return<mode>): Likewise.
29430 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29432 2017-05-14 Julia Koval <julia.koval@intel.com>
29434 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29435 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29436 (__builtin_ia32_xsetbv): New builtins.
29437 * config/i386/i386.c (ix86_expand_special_args_builtin):
29439 (ix86_expand_builtin): Special expand for new intrinsics.
29440 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29441 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29442 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29444 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29446 * cfganal.c (inverted_post_order_compute): Change argument type
29448 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29449 * df-core.c (rest_of_handle_df_initialize): Adjust.
29450 (rest_of_handle_df_finish): Likewise.
29451 (df_analyze_1): Likewise.
29452 (df_analyze): Likewise.
29453 (loop_inverted_post_order_compute): Change argument to be a vec *.
29454 (df_analyze_loop): Adjust.
29455 (df_get_n_blocks): Likewise.
29456 (df_get_postorder): Likewise.
29457 * df.h (struct df_d): Change field to be a vec.
29458 * lcm.c (compute_laterin): Adjust.
29459 (compute_available): Likewise.
29460 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29461 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29462 * tree-ssa-pre.c (compute_antic): Likewise.
29464 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29466 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29467 (depth_first_search::depth_first_search): Change structure init
29468 function to this constructor.
29469 (depth_first_search::add_bb): Rename function to this member.
29470 (depth_first_search::execute): Likewise.
29471 (flow_dfs_compute_reverse_finish): Adjust.
29473 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29475 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29476 (longest_simple_path): Likewise.
29477 * shrink-wrap.c (spread_components): Likewise.
29478 (disqualify_problematic_components): Likewise.
29479 (emit_common_heads_for_components): Likewise.
29480 (emit_common_tails_for_components): Likewise.
29481 (insert_prologue_epilogue_for_components): Likewise.
29483 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29485 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29488 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29490 * df-core.c (df_set_blocks): Start using auto_bitmap.
29491 (df_compact_blocks): Likewise.
29492 * df-problems.c (df_rd_confluence_n): Likewise.
29493 * df-scan.c (df_insn_rescan_all): Likewise.
29494 (df_process_deferred_rescans): Likewise.
29495 (df_update_entry_block_defs): Likewise.
29496 (df_update_exit_block_uses): Likewise.
29497 (df_entry_block_bitmap_verify): Likewise.
29498 (df_exit_block_bitmap_verify): Likewise.
29499 (df_scan_verify): Likewise.
29500 * lra-constraints.c (lra_constraints): Likewise.
29501 (undo_optional_reloads): Likewise.
29502 (lra_undo_inheritance): Likewise.
29503 * lra-remat.c (calculate_gen_cands): Likewise.
29504 (do_remat): Likewise.
29505 * lra-spills.c (assign_spill_hard_regs): Likewise.
29506 (spill_pseudos): Likewise.
29507 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29508 (bitmap_set_subtract_values): Likewise.
29510 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29512 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29513 management with auto_bitmap.
29514 (fix_inter_tick): Likewise.
29515 (fix_recovery_deps): Likewise.
29516 * ira.c (add_store_equivs): Likewise.
29517 (find_moveable_pseudos): Likewise.
29518 (split_live_ranges_for_shrink_wrap): Likewise.
29519 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29520 (rtx_reuse_manager::seen_def_p): Likewise.
29521 (rtx_reuse_manager::set_seen_def): Likewise.
29522 * print-rtl.h (class rtx_reuse_manager): Likewise.
29524 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29526 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29528 (migrate_btr_def): Likewise.
29529 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29530 * df-core.c (loop_post_order_compute): Likewise.
29531 (loop_inverted_post_order_compute): Likewise.
29532 * hsa-common.h: Likewise.
29533 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29534 * init-regs.c (initialize_uninitialized_regs): Likewise.
29535 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29536 (inline_small_functions): Likewise.
29537 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29538 * ira.c (combine_and_move_insns): Likewise.
29539 (build_insn_chain): Likewise.
29540 * loop-invariant.c (find_invariants): Likewise.
29541 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29542 * predict.c (tree_predict_by_opcode): Likewise.
29543 (predict_paths_leading_to): Likewise.
29544 (predict_paths_leading_to_edge): Likewise.
29545 (estimate_loops_at_level): Likewise.
29546 (estimate_loops): Likewise.
29547 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29548 (spread_components): Likewise.
29549 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29550 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29551 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29552 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29553 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29554 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29555 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29556 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29557 (mark_threaded_blocks): Likewise.
29558 (thread_through_all_blocks): Likewise.
29559 * tree-ssa.c (verify_ssa): Likewise.
29560 (execute_update_addresses_taken): Likewise.
29561 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29565 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29567 (post_order_compute): Likewise.
29568 (inverted_post_order_compute): Likewise.
29569 (pre_and_rev_post_order_compute_fn): Likewise.
29571 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29573 * genrecog.c (int_set::int_set): Explicitly construct our
29574 auto_vec base class.
29575 * vec.h (auto_vec::auto_vec): New constructor.
29577 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29579 * bitmap.h (class auto_bitmap): New constructor taking
29580 bitmap_obstack * argument.
29582 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29584 * bitmap.h (class auto_bitmap): Change type of m_bits to
29585 bitmap_head, and adjust ctor / dtor and member operators.
29587 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29589 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29590 when returned register mode doesn't match original mode.
29592 2017-05-12 Jeff Law <law@redhat.com>
29593 Jakub Jelinek <jakub@redhat.com>
29595 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29596 we look for cc setter after the compare-elim changes.
29597 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29598 within the vector to match what compare-elim now expects.
29599 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29600 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29602 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29603 after the compare-elim changes.
29604 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29605 the vector to match what compare-elim now expects.
29606 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29607 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29608 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29609 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29610 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29612 * config/visium/visium.c (single_set_and_flags): Fix where
29613 we look for cc setter after the compare-elim changes.
29614 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29615 with the vector to match what compare-elim now expects.
29616 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29617 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29618 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29619 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29620 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29621 (neg<mode>2_insn_set_overflow): Likewise.
29623 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29625 PR middle-end/79794
29626 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29627 maybe_expand_insn call, set ops[0].target. If still set after call,
29628 set alt_rtl. Add extra arg to recursive calls.
29629 (extract_bit_field): Add alt_rtl argument. Pass to
29631 * expmed.h (extract_bit_field): Fix prototype.
29632 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29633 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29634 to extract_bit_field_calls.
29635 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29636 Pass alt_rtl to extract_bit_field calls.
29637 * calls.c (store_unaligned_arguments_into_psuedos)
29638 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29639 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29641 * optabs.h (struct expand_operand): Add target bitfield.
29643 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29645 * compare-elim.c (try_eliminate_compare): Canonicalize
29646 operation with embedded compare to
29647 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29648 (set (reg) (operation)].
29650 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29652 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29655 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29656 cost of adding a carry flag for ADC instruction.
29657 [case MINUS]: Ignore the cost of subtracting a carry flag
29658 for SBB instruction.
29660 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29662 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29664 * config/rs6000/bmiintrin.h: New file.
29665 * config/rs6000/bmi2intrin.h: New file.
29666 * config/rs6000/x86intrin.h: New file.
29668 2017-05-12 Jeff Law <law@redhat.com>
29670 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29673 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29675 PR middle-end/80707
29676 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29678 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29680 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29681 early expansion of vector divide builtins.
29682 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29683 builtins identified as having unsigned arguments.
29685 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29687 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29688 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29689 expansion of vector logical operations (and, andc, or, xor,
29692 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29694 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29695 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29697 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29699 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29700 early GIMPLE expansion of vector multiplies.
29702 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29704 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29705 TARGET_HAVE_MOVT conditional.
29706 (movt splitter): Likewise.
29708 2017-05-12 Richard Biener <rguenther@suse.de>
29710 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29711 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29712 Fold all stmts not inplace.
29714 2017-05-12 Richard Biener <rguenther@suse.de>
29716 PR tree-optimization/80713
29717 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
29718 inserted_exprs bit for not removed stmts.
29720 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
29722 PR middle-end/69921
29723 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
29724 parallelized" attribute for parallelized OpenACC kernels.
29725 * omp-offload.c (execute_oacc_device_lower): Use it.
29727 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
29728 Set "oacc kernels" attribute.
29729 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
29730 parameter. Adjust all users.
29731 (oacc_fn_attrib_kernels_p): Remove function.
29732 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
29733 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
29734 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29735 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29736 assert "oacc kernels" attribute is set.
29738 2017-05-11 Carl Love <cel@us.ibm.com>
29740 * config/rs6000/rs6000-c: Add support for built-in functions
29741 vector unsigned char vec_popcnt (vector signed char)
29742 vector unsigned char vec_popcnt (vector unsigned char)
29743 vector unsigned short vec_popcnt (vector signed short)
29744 vector unsigned short vec_popcnt (vector unsigned short)
29745 vector unsigned int vec_popcnt (vector signed int)
29746 vector unsigned int vec_popcnt (vector unsigned int)
29747 vector unsigned long long vec_popcnt (vector signed long long)
29748 vector unsigned long long vec_popcnt (vector unsigned long long)
29749 vector signed long long vec_slo (vector signed long long,
29750 vector signed char)
29751 vector signed long long vec_slo (vector signed long long,
29752 vector unsigned char)
29753 vector unsigned long long vec_slo (vector unsigned long long,
29754 vector signed char)
29755 vector unsigned long long vec_slo (vector unsigned long long,
29756 vector unsigned char)
29757 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29758 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29759 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29760 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29761 * doc/extend.texi: Update the built-in documentation file for the
29762 new built-in functions.
29764 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29766 * attribs.h (sorted_attr_string): Move machine independent
29767 functions for target clone support from the i386 port to common
29768 code. Rename ix86_function_versions to common_function_versions.
29769 Rename make_name to make_unique_name.
29770 (common_function_versions): Likewise.
29771 (make_unique_name): Likewise.
29772 (make_dispatcher_decl): Likewise.
29773 (is_function_default_version): Likewise.
29774 * attribs.c (attr_strcmp): Likewise.
29775 (sorted_attr_string): Likewise.
29776 (common_function_versions): Likewise.
29777 (make_unique_name): Likewise.
29778 (make_dispatcher_decl): Likewise.
29779 (is_function_default_version): Likewise.
29780 * config/i386/i386.c (attr_strcmp): Likewise.
29781 (sorted_attr_string): Likewise.
29782 (ix86_function_versions): Likewise.
29783 (make_name): Likewise.
29784 (make_dispatcher_decl): Likewise.
29785 (is_function_default_version): Likewise.
29786 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29788 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29791 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29792 Account for direct move costs for vec_construct of integer
29795 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29798 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29799 (UNSPEC_STX_ATOMIC): Ditto.
29800 (loaddi_via_sse): New insn.
29801 (storedi_via_sse): Ditto.
29802 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29803 Update corresponding peephole2 patterns.
29804 (atomic_storedi_fpu): Ditto.
29806 2017-05-11 Julia Koval <julia.koval@intel.com>
29808 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29809 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29811 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29812 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29813 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29815 2017-05-11 Nathan Sidwell <nathan@acm.org>
29817 * graphite-poly.c: Include dumpfile.h.
29819 * dumpfle.h (dump_function): Declare here ...
29820 * tree-dump.h (dump_function): ... not here.
29821 * dumpfile.c: #include tree-cfg.h.
29822 (dump_function): Move here from ...
29823 * tree-dump.c (dump_function): ... here.
29824 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29825 * graphite-poly.c: Don't include tree-dump.h.
29826 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29827 * print-tree.c: Likewise.
29828 * stor-layout.c: Likewise.
29829 * tree-nested.c: Likewise.
29831 * dumpfile.c (dump_start): Use TDF_FLAGS.
29832 (dump_enable_all): Fix TDF_KIND check thinko.
29834 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29836 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29837 array entries to represent two legal parameterizations of the
29838 overloaded __builtin_cmpb function, as represented by the
29839 P6_OV_BUILTIN_CMPB constant.
29840 (altivec_resolve_overloaded_builtin): Add special case handling
29841 for the __builtin_cmpb function, as represented by the
29842 P6_OV_BUILTIN_CMPB constant.
29843 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29844 (BU_P6_64BIT_2): New macro.
29845 (BU_P6_OVERLOAD_2): New macro
29846 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29847 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29848 (CMPB): Add overload support to represent both 32-bit and 64-bit
29849 compare-bytes function.
29850 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29851 support for TARGET_CMPB.
29852 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29853 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29854 documentation of the __builtin_cmpb overloaded built-in function.
29856 2017-05-11 Richard Biener <rguenther@suse.de>
29858 PR tree-optimization/80705
29859 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29860 bases are not vectorizable.
29862 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29864 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29865 when counting register pressure.
29867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29869 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29870 (iv_ca_more_deps): Renamed to ...
29871 (iv_ca_compare_deps): ... this.
29872 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29874 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29876 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29878 (determine_group_iv_costs): ... here.
29879 (find_inv_vars_cb): Record inv var if it's not recorded before.
29881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29883 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29884 (get_shiftadd_cost): Ditto.
29886 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29888 * tree-ssa-address.c: Include header file.
29889 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29891 (add_to_parts): Refactor.
29892 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29893 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29898 PR tree-optimization/53090
29899 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29901 (extract_cond_operands): Detect condition with IV on both sides
29902 and return COMP_IV_EXPR_2.
29903 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29904 (rewrite_use_compare): Simplify by removing call to function
29905 extract_cond_operands.
29907 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29909 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29910 (extract_cond_operands): Detect condition comparing against non-
29911 invariant bound and return appropriate enum value.
29912 (find_interesting_uses_cond): Update use of extract_cond_operands.
29913 Handle its return value accordingly.
29914 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29916 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29918 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29919 nonlinear iv_use computation in loop invariant sensitive way.
29921 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29923 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29924 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29926 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29928 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29929 (dump_cand): Support iv_cand.inv_exprs.
29930 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29932 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29935 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29937 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29939 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29940 as local function. Include necessary header files.
29941 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29945 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29947 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29949 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29950 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29951 RSHIFT_EXPR and BIT_NOT_EXPR.
29953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29955 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29956 (adjust_setup_cost): New parameter supporting round up adjustment.
29957 (struct address_cost_data): Delete.
29958 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29959 (split_address_cost, ptr_difference_cost): Delete.
29960 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29961 (struct ainc_cost_data): New struct.
29962 (get_address_cost_ainc): New function.
29963 (get_address_cost, get_computation_cost): Reimplement.
29964 (determine_group_iv_cost_address): Record inv_expr for all uses of
29966 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29967 (iv_ca_has_deps): Reimplemented to ...
29968 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29970 (iv_ca_extend): Call iv_ca_more_deps.
29972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29974 * tree-ssa-address.c (struct mem_address): Move to header file.
29975 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29976 * tree-ssa-address.h (struct mem_address): Move from C file.
29977 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29979 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29981 * tree-affine.h (aff_combination_type): New interface.
29982 (aff_combination_zero_p): Remove static.
29983 (aff_combination_const_p): New interface.
29984 (aff_combination_singleton_var_p): New interfaces.
29986 2017-05-11 Richard Biener <rguenther@suse.de>
29988 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29989 Skip unreachable blocks and destinations.
29990 (eliminate): Move stmt removal and fixup ...
29991 (fini_eliminate): ... here. Skip inserted exprs.
29992 (pass_pre::execute): Move fini_pre after fini_eliminate.
29993 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29994 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29995 PRE to get rid of dead code that has invalid SSA form and
29996 split critical edges again.
29998 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30000 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30002 2017-05-11 Richard Biener <rguenther@suse.de>
30004 * passes.c (execute_function_todo): Verify loops if they are
30005 said to be up-to-date.
30006 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30007 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30009 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30012 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30013 handle calling assemble_external ourself.
30016 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30017 modes with zero size. Enhance comment.
30019 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30021 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30022 built-ins for vec_xl and vec_xst with short and char pointer
30025 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30027 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30028 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30029 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30030 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30031 (_mm_maskz_min_round_ss): New intrinsics.
30032 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30033 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30034 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30035 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30036 (__builtin_ia32_minss_mask_round): New builtins.
30037 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30038 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30039 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30041 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30042 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30044 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30047 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30049 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30050 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30051 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30052 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30053 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30054 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30055 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30056 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30057 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30058 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30059 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30060 (__builtin_ia32_mulss_mask_round): New builtins.
30061 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30062 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30063 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30065 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30066 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30068 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30071 2017-05-10 Julia Koval <julia.koval@intel.com>
30073 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30074 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30075 (_mm256_setr_m128i): New intrinsics.
30077 2017-05-10 Julia Koval <julia.koval@intel.com>
30079 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30080 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30081 (_mm_maskz_rcp14_ss): New intrinsics.
30082 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30083 (__builtin_ia32_rcp14ss_mask): New builtins.
30084 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30086 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30088 PR tree-optimization/51513
30089 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30090 (assert_unreachable_fallthru_edge_p): Use it.
30091 (group_case_labels_stmt): Likewise.
30092 * tree-cfg.h: Prototype it.
30093 * stmt.c: Include cfghooks.h and tree-cfg.h.
30094 (emit_case_dispatch_table) <gap_label>: New local variable.
30095 Use it to fill dispatch table gaps.
30096 Test for default_label before updating probabilities.
30097 (expand_case) <default_label>: Remove unneeded initialization.
30098 Test for unreachable default case statement and remove its edge.
30099 Set default_label accordingly.
30100 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30102 2017-05-10 Carl Love <cel@us.ibm.com>
30104 * config/rs6000/rs6000-c: Add support for built-in functions
30105 vector signed char vec_neg (vector signed char)
30106 vector signed short int vec_neg (vector short int)
30107 vector signed int vec_neg (vector signed int)
30108 vector signed long long vec_neg (vector signed long long)
30109 vector float vec_neg (vector float)
30110 vector double vec_neg (vector double)
30111 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30113 * config/rs6000/altivec.h: Add define for vec_neg
30114 * doc/extend.texi: Update the built-in documentation for the
30115 new built-in functions.
30117 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30119 PR tree-optimization/77644
30120 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30122 2017-05-10 Nathan Sidwell <nathan@acm.org>
30124 * dumpfile.h (TDI_lang_all): New.
30125 (TDF_KIND): New. Renumber others
30126 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30128 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30130 (get_dump_file_name): Adjust suffix generation.
30131 (dump_enable_all): Use TDF_KIND.
30132 * doc/invoke.texi (-fdump-lang-all): Document.
30134 * dumpfile.h: Tabify.
30136 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30139 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30140 Move member access before delete.
30142 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30144 * tree-inline.c (expand_call_inline): Split block at stmt
30147 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30150 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30151 are now unused after splitting mov{sf,sd}_hardfloat.
30152 (f32_lr2): Likewise.
30153 (f32_lm): Likewise.
30154 (f32_lm2): Likewise.
30155 (f32_li): Likewise.
30156 (f32_li2): Likewise.
30157 (f32_lv): Likewise.
30158 (f32_sr): Likewise.
30159 (f32_sr2): Likewise.
30160 (f32_sm): Likewise.
30161 (f32_sm2): Likewise.
30162 (f32_si): Likewise.
30163 (f32_si2): Likewise.
30164 (f32_sv): Likewise.
30165 (f32_dm): Likewise.
30166 (f32_vsx): Likewise.
30167 (f32_av): Likewise.
30168 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30169 For movsf, order stores so the VSX stores occur before the GPR
30170 store which encourages the register allocator to use a traditional
30171 FPR instead of a GPR. For movsd, order the stores so that the GPR
30172 store comes before the VSX stores to allow the power6 to work.
30173 This is due to the power6 not having a 32-bit integer store
30174 instruction from a FPR.
30175 (movsf_hardfloat): Likewise.
30176 (movsd_hardfloat): Likewise.
30178 2017-05-09 Martin Sebor <msebor@redhat.com>
30180 PR translation/80280
30181 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30184 PR translation/80280
30185 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30186 data member added in r247778.
30187 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30189 2017-05-09 Nathan Sidwell <nathan@acm.org>
30191 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30193 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30196 2017-05-09 Marek Polacek <polacek@redhat.com>
30198 * doc/invoke.texi: Fix typo.
30200 2017-05-09 Richard Biener <rguenther@suse.de>
30202 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30203 (vrp_val_is_min): Likewise.
30204 (set_value_range_to_value): Likewise.
30205 (set_value_range_to_nonnegative): Likewise.
30206 (gimple_assign_nonzero_p): Likewise.
30207 (gimple_stmt_nonzero_p): Likewise.
30208 (vrp_int_const_binop): Likewise. Remove unreachable case.
30209 (adjust_range_with_scev): Adjust comments.
30210 (compare_range_with_value): Likewise.
30211 (extract_range_from_phi_node): Likewise.
30212 (test_for_singularity): Likewise.
30214 2017-05-09 Richard Biener <rguenther@suse.de>
30216 * tree-vrp.c (get_single_symbol): Add assert that we don't
30217 get overflowed constants as invariant part.
30218 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30219 checks. Use wi::cmp instead of recursing for integer constants.
30220 (compare_values): Just ignore whether we assumed undefined
30221 overflow instead of failing the compare.
30222 (extract_range_for_var_from_comparison_expr): Add comment before the
30223 TREE_NO_WARNING sets.
30224 (test_for_singularity): Likewise.
30225 (extract_range_from_comparison): Do not disable optimization
30226 when we assumed undefined overflow.
30227 (extract_range_basic): Remove init of unused var.
30229 2017-05-09 Richard Biener <rguenther@suse.de>
30231 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30232 (extract_range_from_multiplicative_op_1): Adjust.
30233 (extract_range_from_binary_expr_1): Use int_const_binop.
30235 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30238 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30239 rs6000_store_data_bypass_p in seven define_bypass directives and
30240 in several comments.
30241 * config/rs6000/rs6000-protos.h: Add prototype for
30242 rs6000_store_data_bypass_p function.
30243 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30244 function implements slightly different (rs6000-specific) semantics
30245 than store_data_bypass_p, returning false rather than aborting
30246 with assertion error when arguments do not satisfy the
30247 requirements of store data bypass.
30248 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30249 rs6000_store_data_bypass_p.
30251 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30253 * config/xtensa/xtensa-protos.h
30254 (xtensa_initial_elimination_offset): New declaration.
30255 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30256 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30257 macro definition, add case for FRAME_POINTER_REGNUM when
30258 FRAME_GROWS_DOWNWARD.
30259 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30260 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30261 xtensa_initial_elimination_offset.
30263 2017-05-08 Nathan Sidwell <nathan@acm.org>
30265 * doc/invoke.texi: Alphabetize -fdump options.
30267 2017-05-08 Martin Sebor <msebor@redhat.com>
30269 PR translation/80280
30270 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30272 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30274 * target.def (compute_frame_layout): New optional target hook.
30275 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30276 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30277 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30279 * reload1.c (verify_initial_elim_offsets): Likewise.
30280 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30281 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30282 (arm_get_frame_offsets): Split up into this ...
30283 (arm_compute_frame_layout): ... and this function.
30285 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30287 * config/aarch64/constraints.md (Usa): New constraint.
30288 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30290 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30292 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30293 with_multilib_list after it has been checked.
30295 2017-05-08 Richard Biener <rguenther@suse.de>
30297 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30298 (bitmap_set_subtract_values): Likewise.
30300 2017-05-08 Richard Biener <rguenther@suse.de>
30302 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30303 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30305 (gimple_stmt_nonzero_warnv_p): Rename to ...
30306 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30308 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30309 (extract_range_basic): Adjust, do not disable propagation on
30310 strict overflow sensitive simplification.
30311 (vrp_visit_cond_stmt): Likewise.
30313 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30315 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30316 body size unconditionally.
30318 2017-05-07 Jeff Law <law@redhat.com>
30321 2017-05-06 Jeff Law <law@redhat.com>
30322 PR tree-optimization/78496
30323 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30326 PR tree-optimization/78496
30327 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30328 (simplify_stmt_using_ranges): Call it.
30329 (vrp_dom_walker::before_dom_children): Extract equivalences
30330 from an ASSERT_EXPR with an equality comparison against a
30333 2017-05-06 Jeff Law <law@redhat.com>
30335 PR tree-optimization/78496
30336 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30339 PR tree-optimization/78496
30340 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30341 (simplify_stmt_using_ranges): Call it.
30342 (vrp_dom_walker::before_dom_children): Extract equivalences
30343 from an ASSERT_EXPR with an equality comparison against a
30346 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30348 * lra-constraints.c (lra_copy_reg_equiv): New function.
30349 (split_reg): Use it to copy equivalence information from the
30350 original register to the spill register.
30352 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30354 PR rtl-optimization/75964
30355 * simplify-rtx.c (simplify_const_relational_operation): Remove
30356 invalid handling of comparisons of integer ABS.
30358 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30360 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30361 initialize to zero.
30362 (init_regs): Remove declaration.
30363 (function_arg_advance_32): Initialize error_p as boolean variable.
30365 2017-05-05 Nathan Sidwell <nathan@acm.org>
30367 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30368 lines. Use for (;;).
30370 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30372 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30373 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30374 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30375 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30376 VF=2 that require versioning.
30378 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30380 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30383 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30385 * diagnostic.h (diagnostic_override_option_index): Convert from
30386 macro to inline function.
30388 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30390 * diagnostic.c (last_module_changed_p): New function.
30391 (set_last_module): New function.
30392 (diagnostic_report_current_module): Convert macro usage to
30393 the above functions.
30394 * diagnostic.h (diagnostic_context::last_module): Strengthen
30395 from const line_map * to const line_map_ordinary *.
30396 (diagnostic_last_module_changed): Delete macro.
30397 (diagnostic_set_last_module): Delete macro.
30399 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30401 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30402 with diagnostic_report_diagnostic.
30403 (diagnostic_n_impl_richloc): Likewise.
30404 * diagnostic.h (report_diagnostic): Delete macro.
30405 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30406 with diagnostic_report_diagnostic.
30407 * substring-locations.c (format_warning_va): Likewise.
30409 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30411 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30412 save/restor of format_spec. Move option-printing code to...
30413 (print_option_information): ...this new function, and
30414 reimplement by simply printing to the pretty_printer,
30415 rather than appending to the format string.
30417 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30419 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30420 handling logic into...
30421 (update_effective_level_from_pragmas): ...this new function.
30423 2017-05-04 Andrew Waterman <andrew@sifive.com>
30425 * config/riscv/riscv.opt (mstrict-align): New option.
30426 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30427 (SLOW_UNALIGNED_ACCESS): Define.
30428 (riscv_slow_unaligned_access): Declare.
30429 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30431 (riscv_slow_unaligned_access): New variable.
30432 (rocket_tune_info): Set slow_unaligned_access to true.
30433 (optimize_size_tune_info): Set slow_unaligned_access to false.
30434 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30435 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30436 (riscv_option_override): Set riscv_slow_unaligned_access.
30437 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30439 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30441 * config/riscv/riscv.md: Unify indentation.
30443 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30448 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30450 (extendsi<mode>2): Add support for doing sign extension via
30451 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30452 don't have ISA 3.0 instructions.
30453 (extendsi<mode>2 splitter): Likewise.
30454 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30455 generate the normal insns since SImode can now go in vector
30456 registers. Disallow the special UNSPECs needed for previous
30457 machines to hide SImode being used. Add new insns
30458 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30459 (fix_trunc<mode>si2_stfiwx): Likewise.
30460 (fix_trunc<mode>si2_internal): Likewise.
30461 (fixuns_trunc<mode>si2): Likewise.
30462 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30463 (fctiw<u>z_<mode>_smallint): Likewise.
30464 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30465 of floating point to 32-bit integer from doing a direct move to
30466 the GPR registers to do a store.
30467 (fctiwz_<mode>): Break long line.
30469 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30471 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30472 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30473 (addr_list, addr_offset_valid_p): New.
30474 (split_address_groups): Check offset validity with above function.
30475 (gt-tree-ssa-loop-ivopts.h): Include header file.
30477 2017-05-05 Nathan Sidwell <nathan@acm.org>
30479 * config.gcc (arm*-*-*): Add missing 'fi'.
30481 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30483 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30484 included in -fopt-info does not matter.
30485 * doc/optinfo.texi (-fopt-info): Fix description of default
30486 behavour. Explicitly say order of options included in -fopt-info
30489 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30491 * config.gcc: Allow combinations of aprofile and rmprofile values for
30492 --with-multilib-list.
30493 * config/arm/t-multilib: New file.
30494 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30495 variables. Remove setting of ISA and floating-point ABI in
30496 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30497 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30498 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30499 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30501 * config/arm/t-rmprofile: Likewise except for the matches changes.
30502 * doc/install.texi (--with-multilib-list): Document the combination of
30503 aprofile and rmprofile values and warn about pitfalls in doing that.
30505 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30507 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30508 (movdi_aarch64): Likewise.
30510 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30512 PR tree-optimization/80632
30513 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30515 (build_arrays): Initialize it for virtual phis.
30516 (fix_phi_nodes): Use it for virtual phis.
30518 PR tree-optimization/80558
30519 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30520 [x, y] op z into [x op, y op z] for op & or | if conditions
30523 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30524 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30527 * config/arm/arm.md (use_literal_pool): Remove.
30528 (64-bit immediate split): No longer takes cost into consideration
30529 if arm_disable_literal_pool is enabled.
30530 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30531 used when arm_disable_literal_pool is enabled.
30532 (arm_max_const_double_inline_cost): Remove use of
30533 arm_disable_literal_pool.
30534 (push_minipool_fix): Add assert.
30535 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30536 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30537 (no_literal_pool_sf_immediate): New.
30539 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30541 PR tree-optimization/80613
30542 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30543 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30545 2017-05-05 Richard Biener <rguenther@suse.de>
30547 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30549 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30551 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30552 of this flag from insn conditions due to removal from r247495.
30554 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30556 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30558 (arm_early_store_addr_dep_ptr): Likewise.
30559 * config/arm/aarch-common-protos.h
30560 (arm_early_load_addr_dep_ptr): Add prototype.
30561 (arm_early_store_addr_dep_ptr): Likewise.
30562 * config/arm/cortex-a53.md: Add new bypasses.
30564 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30566 * tree.c (next_type_uid): Change type to unsigned.
30567 (type_hash_canon): Decrement back next_type_uid if
30568 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30569 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30572 2017-05-04 Martin Sebor <msebor@redhat.com>
30574 * builtins.c: Fix a trivial typo in a comment.
30576 PR middle-end/79234
30577 * builtins.c (check_sizes): Adjust to handle reading past the end.
30578 Avoid printing excessive upper bound of ranges. Use %E to print
30579 tree nodes instead of converting them to %wu.
30580 (expand_builtin_memchr): New function.
30581 (compute_dest_size): Rename...
30582 (compute_objsize): ...to this.
30583 (expand_builtin_memcpy): Adjust.
30584 (expand_builtin_mempcpy): Adjust.
30585 (expand_builtin_strcat): Adjust.
30586 (expand_builtin_strcpy): Adjust.
30587 (check_strncat_sizes): Adjust.
30588 (expand_builtin_strncat): Adjust.
30589 (expand_builtin_strncpy): Adjust and simplify.
30590 (expand_builtin_memset): Adjust.
30591 (expand_builtin_bzero): Adjust.
30592 (expand_builtin_memcmp): Adjust.
30593 (expand_builtin): Handle memcmp.
30594 (maybe_emit_chk_warning): Check strncat just once.
30596 2017-05-04 Martin Sebor <msebor@redhat.com>
30598 PR preprocessor/79214
30599 PR middle-end/79222
30600 PR middle-end/79223
30601 * builtins.c (check_sizes): Add inlining context and issue
30602 warnings even when -Wno-system-headers is set.
30603 (check_strncat_sizes): Same.
30604 (expand_builtin_strncat): Same.
30605 (expand_builtin_memmove): New function.
30606 (expand_builtin_stpncpy): Same.
30607 (expand_builtin): Handle memmove and stpncpy.
30609 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30611 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30612 which is not used any more.
30614 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30616 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30618 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30620 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30621 (cortexa53_tunings): Likewise.
30622 (cortexa57_tunings): Likewise.
30623 (cortexa72_tunings): Likewise.
30624 (cortexa73_tunings): Likewise.
30626 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30628 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30629 Set loop alignment to 8.
30631 2017-05-04 Martin Sebor <msebor@redhat.com>
30633 PR translation/80280
30634 * builtins.c (expand_builtin_object_size): Add missing quoting to
30635 %D and like directives.
30636 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30637 (hsa_type_for_tree_type): Same.
30638 (verify_function_arguments): Same.
30639 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30640 * varasm.c (get_section): Same.
30643 2017-05-04 Martin Sebor <msebor@redhat.com>
30645 PR translation/80280
30646 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30648 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30650 * config/aarch64/aarch64.c (generic_addrcost_table):
30651 Change HI/TI mode setting.
30653 2017-05-04 Martin Jambor <mjambor@suse.cz>
30655 PR tree-optimization/80622
30656 * tree-sra.c (comes_initialized_p): New function.
30657 (build_accesses_from_assign): Only set write lazily when
30658 comes_initialized_p is false.
30659 (analyze_access_subtree): Use comes_initialized_p.
30660 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30661 instead of testing for PARM_DECL.
30663 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30665 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30666 constraint on operand 0 to allow more general addressing modes.
30667 Adjust output template.
30668 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30670 * config/aarch64/aarch64-protos.h
30671 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30672 * config/aarch64/constraints.md (Dp): New address constraint.
30673 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30676 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30678 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30679 update use of estimate_ipcp_clone_size_and_time.
30680 (estimate_local_effects): Update use of
30681 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30682 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30683 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30684 Return nonspecialized time.
30686 2017-05-04 Richard Biener <rguenther@suse.de>
30688 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30689 for the last VUSE which def dominates the PHI. Directly call
30691 (get_continuation_for_phi_1): Remove.
30693 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30695 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30696 to explain the use of truncating division. Cap the number of
30697 iterations to the maximum given by nb_iterations_upper_bound,
30700 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30702 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30703 * configure: Regenerate.
30704 * config.in: Regenerate.
30705 * config/i386/driver-mingw32.c: new file.
30706 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30707 * config.host: Link driver-mingw32.o on MinGW host.
30708 * doc/install.texi: Document new --enable-mingw-wildcard configure
30711 2017-05-04 Marek Polacek <polacek@redhat.com>
30713 PR tree-optimization/80612
30714 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
30716 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30717 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
30719 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
30720 (movt splitter): Likewise.
30721 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
30722 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
30723 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
30724 block for Thumb-1 with MOVT.
30725 (thumb2_legitimate_address_p): Move code block ...
30726 (can_avoid_literal_pool_for_label_p): ... into this new function.
30727 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
30729 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
30730 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
30731 "M-profile targets with the MOVT instruction".
30733 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30735 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30736 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30737 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30739 2017-05-04 Martin Liska <mliska@suse.cz>
30741 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30742 variable cond_code.
30744 2017-05-04 Richard Biener <rguenther@suse.de>
30746 * tree.c (array_at_struct_end_p): Handle arrays at struct
30747 end with flexarrays more conservatively. Refactor and treat
30748 arrays of arrays or aggregates more strict. Fix
30749 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30750 * tree.c (array_at_struct_end_p): Adjust prototype.
30751 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30752 * gimple-fold.c (get_range_strlen): Likewise.
30753 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30755 2017-05-04 Richard Biener <rguenther@suse.de>
30757 PR tree-optimization/31130
30758 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30760 (supports_overflow_infinity): Likewise.
30761 (is_negative_overflow_infinity): Likewise.
30762 (is_positive_overflow_infinity): Likewise.
30763 (is_overflow_infinity): Likewise.
30764 (stmt_overflow_infinity): Likewise.
30765 (overflow_infinity_range_p): Likewise.
30766 (usable_range_p): Remove as always returning true.
30767 (make_overflow_infinity): Remove.
30768 (negative_overflow_infinity): Likewise.
30769 (positive_overflow_infinity): Likewise.
30770 (avoid_overflow_infinity): Likewise.
30771 (set_value_range): Adjust accordingly.
30772 (set_value_range_to_nonnegative): Likewise, remove now unused
30773 overflow_infinity arg.
30774 (vrp_operand_equal_p): Adjust.
30775 (update_value_range): Likewise.
30776 (range_int_cst_singleton_p): Likewise.
30777 (operand_less_p): Likewise.
30778 (compare_values_warnv): Likewise.
30779 (extract_range_for_var_from_comparison_expr): Likewise.
30780 (vrp_int_const_binop): Likewise.
30781 (zero_nonzero_bits_from_vr): Likewise.
30782 (extract_range_from_multiplicative_op_1): Likewise.
30783 (extract_range_from_binary_expr_1): Likewise.
30784 (extract_range_from_unary_expr): Likewise.
30785 (extract_range_from_comparison): Likewise.
30786 (extract_range_basic): Likewise.
30787 (adjust_range_with_scev): Likewise.
30788 (compare_ranges): Likewise.
30789 (compare_range_with_value): Likewise.
30790 (dump_value_range): Likewise.
30791 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30793 (simplify_cond_using_ranges): Adjust.
30795 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30797 * brig-builtins.def: Added a builtin for class_f64.
30798 * builtin-types.def: Added a builtin type needed by class_f64.
30800 2017-05-03 Jason Merrill <jason@redhat.com>
30802 * timevar.def: Add TV_CONSTEXPR.
30804 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30806 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30808 2017-05-03 Martin Jambor <mjambor@suse.cz>
30810 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30811 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30812 * ipa-cp.c (ipcp_propagate_stage): Do not call
30813 ipa_update_after_lto_read.
30814 * ipa-inline.c (ipa_inline): Likewise.
30816 2017-05-03 Martin Jambor <mjambor@suse.cz>
30818 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30819 tag. Added a default constructor and a destructor.
30820 (ipa_edge_args_sum_t): New class;
30821 (ipa_edge_args_sum): Declare.
30822 (ipa_edge_args_vector): Remove declaration.
30823 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30824 (ipa_free_edge_args_substructures): Remove declaration.
30825 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30826 (ipa_edge_args_info_available_for_edge_p): Likewise.
30827 * ipa-prop.c (ipa_edge_args_vector): Removed.
30828 (edge_removal_hook_holder): Likewise.
30829 (edge_duplication_hook_holder): Likewise.
30830 (ipa_edge_args_sum): New variable.
30831 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30832 ipa_edge_args_vector.
30833 (ipa_free_edge_args_substructures): Likewise.
30834 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30835 ipa_edge_args_vector.
30836 (ipa_edge_removal_hook): Turned into method
30837 ipa_edge_args_sum_t::remove.
30838 (ipa_edge_duplication_hook): Turned into method
30839 ipa_edge_args_sum_t::duplicate.
30840 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30841 registering edge hooks.
30842 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30843 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30844 ipa_edge_args_sum instead of ipa_edge_args_vector.
30845 * ipa-profile.c (ipa_profile): Likewise.
30847 2017-05-03 Martin Jambor <mjambor@suse.cz>
30849 * symbol-summary.h (function_summary): New method exists.
30850 (function_summary::symtab_removal): Deallocate through release.
30851 (call_summary): New class.
30852 (gt_ggc_mx): New overload.
30853 (gt_pch_nx): Likewise.
30854 (gt_pch_nx): Likewise.
30856 2017-05-03 Jeff Law <law@redhat.com>
30858 PR tree-optimization/78496
30859 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30860 from simplify_cond_using_ranges. Split off code to walk
30861 backwards through casts into ...
30862 (simplify_cond_using_ranges_2): New function.
30863 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30864 (execute_vrp): After identifying jump threads, call
30865 simplify_cond_using_ranges_2.
30867 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30870 * ipa-inline.h (inline_summary): Add ctor.
30871 (create_ggc): Do not use ggc_cleared_alloc.
30873 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30874 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30876 * gcc.c (handle_braces): Support escaping in switch matching
30878 * doc/invoke.texi (Spec Files): Document it.
30879 Remove superfluous @code markup in items.
30881 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30883 * diagnostic-show-locus.c (struct column_range): New struct.
30884 (get_affected_columns): New function.
30885 (get_printed_columns): New function.
30886 (struct correction): New struct.
30887 (correction::ensure_capacity): New function.
30888 (correction::ensure_terminated): New function.
30889 (struct line_corrections): New struct.
30890 (line_corrections::~line_corrections): New dtor.
30891 (line_corrections::add_hint): New function.
30892 (layout::print_trailing_fixits): Reimplement in terms of the new
30894 (selftest::test_overlapped_fixit_printing): New function.
30895 (selftest::diagnostic_show_locus_c_tests): Call it.
30897 2017-05-03 Nathan Sidwell <nathan@acm.org>
30899 Canonicalize canonical type hashing
30900 * tree.h (type_hash_canon_hash): Declare.
30901 * tree.c (type_hash_list, attribute_hash_list): Move into
30902 type_hash_canon_hash.
30903 (build_type_attribute_qual_variant): Break out hash code calc into
30904 type_hash_canon_hash.
30905 (type_hash_canon_hash): New. Generic type hash computation.
30906 (build_range_type_1, build_array_type_1, build_function_type,
30907 build_method_type_directly, build_offset_type, build_complex_type,
30908 make_vector_type): Call it.
30910 2017-05-03 Richard Biener <rguenther@suse.de>
30912 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30913 When all DRs have unknown misaligned do not always peel
30914 when there is a store but apply the same costing model as if
30915 there were only loads.
30917 2017-05-03 Richard Biener <rguenther@suse.de>
30920 PR tree-optimization/80492
30921 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30922 compare_base_decls returning dont-know properly.
30924 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30926 * config/arm/iterators.md (CCSI): New mode iterator.
30927 (arch): New mode attribute.
30928 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30929 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30930 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30931 code iterator for success result mode.
30932 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30933 the corresponding new insn generators.
30935 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30938 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30939 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30941 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30943 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30944 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30945 (DDR_A): Wrap DDR argument in brackets.
30946 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30947 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30948 (DDR_REVERSED_P): Likewise.
30950 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30952 PR tree-optimization/79472
30953 * tree-switch-conversion.c (struct switch_conv_info): Add
30954 contiguous_range and default_case_nonstandard fields.
30955 (collect_switch_conv_info): Compute contiguous_range and
30956 default_case_nonstandard fields, don't clear final_bb if
30957 contiguous_range and only the default case doesn't have the required
30959 (check_all_empty_except_final): Set default_case_nonstandard instead
30960 of failing if contiguous_range and the default case doesn't have empty
30962 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30963 and only the default case doesn't have the required constants. Skip
30965 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30966 if default_case_nonstandard.
30967 (build_constructors): Build constant 1 just once. Assert that default
30968 values aren't inserted in between cases if contiguous_range. Skip
30970 (build_arrays): Skip virtual phis.
30971 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30972 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30973 Handle virtual phis.
30974 (gen_inbound_check): Handle default_case_nonstandard case.
30975 (process_switch): Adjust check_final_bb caller. Call
30976 gather_default_values with the first non-default case instead of
30977 default case if default_case_nonstandard.
30979 2017-05-02 Nathan Sidwell <nathan@acm.org>
30981 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30982 check. Fix formatting.
30984 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30986 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30987 errors when comparing specialized and unspecialized times.
30989 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30991 * diagnostic-show-locus.c
30992 (layout::should_print_annotation_line_p): Make private.
30993 (layout::print_annotation_line): Make private.
30994 (layout::annotation_line_showed_range_p): Make private.
30995 (layout::show_ruler): Make private.
30996 (layout::print_source_line): Make private. Pass in line and
30997 line_width, rather than calling location_get_source_line. Drop
30999 (layout::print_leading_fixits): New method.
31000 (layout::print_any_fixits): Rename to...
31001 (layout::print_trailing_fixits): ...this, and make private.
31002 Don't print newline fixits.
31003 (diagnostic_show_locus): Move logic for printing one row into...
31004 (layout::print_line): ...this new function. Move the
31005 location_get_source_line call and error-handling from
31006 print_source_line to here. Call print_leading_fixits, and rename
31007 print_any_fixits to print_trailing_fixits.
31008 (selftest::test_fixit_insert_containing_newline): Update now that
31009 newlines are partially supported.
31010 (selftest::test_fixit_insert_containing_newline_2): New test.
31011 (selftest::test_fixit_replace_containing_newline): Update comments.
31012 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31013 * edit-context.c (class added_line): New class.
31014 (class edited_line): Describe newline handling in comment.
31015 (edited_line::actually_edited_p): New method.
31016 (edited_line::print_content): Delete redundant decl.
31017 (edited_line::m_predecessors): New field.
31018 (edited_file::print_content): Call edited_line::print_content.
31019 (edited_file::print_diff): Update to support newlines.
31020 (edited_file::print_diff_hunk): Likewise.
31021 (edited_file::print_run_of_changed_lines): New function.
31022 (edited_file::print_diff_line): Convert to...
31023 (print_diff_line): ...this.
31024 (edited_file::get_effective_line_count): New function.
31025 (edited_line::edited_line): Initialize new field m_predecessors.
31026 (edited_line::~edited_line): Clean up m_predecessors.
31027 (edited_line::apply_fixit): Handle newlines.
31028 (edited_line::get_effective_line_count): New function.
31029 (edited_line::print_content): New function.
31030 (edited_line::print_diff_lines): New function.
31031 (selftest::test_applying_fixits_insert_containing_newline): New
31033 (selftest::test_applying_fixits_replace_containing_newline): New
31035 (selftest::insert_line): New function.
31036 (selftest::test_applying_fixits_multiple_lines): Add example of
31038 (selftest::edit_context_c_tests): Call the new tests.
31040 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31042 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31043 parameter cand. Update dump information.
31044 (get_computation_cost): Update uses.
31046 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31048 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31049 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31050 (get_computation_at, rewrite_use_address): Update use of
31051 get_computation_aff.
31053 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31055 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31056 (get_computation): Delete.
31057 (get_computation_cost): Implement like get_computation_cost_at.
31058 Use get_computation_at.
31059 (get_computation_cost_at): Delete.
31060 (rewrite_use_nonlinear_expr): Use get_computation_at.
31061 (rewrite_use_compare, remove_unused_ivs): Ditto.
31063 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31065 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31067 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31069 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31070 (ivopts_global_cost_for_size): Rename parameter and update uses.
31071 (iv_ca_recount_cost): Update uses.
31072 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31073 candidates seperately in n_invs and n_cands.
31074 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31076 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31078 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31079 (find_inv_vars_cb): New.
31080 (find_depends): Renamed to ...
31081 (find_inv_vars): ... this.
31082 (add_candidate_1, force_var_cost): Call find_inv_vars.
31083 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31085 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31087 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31088 inv_vars. Add inv_exprs.
31089 (struct iv_cand): Rename depends_on to inv_vars.
31090 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31091 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31092 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31093 (dump_cand): Dump inv_vars.
31094 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31095 (record_invariant, find_depends, add_candidate_1): Ditto.
31096 (set_group_iv_cost, force_var_cost): Ditto.
31097 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31098 (get_computation_cost_at, get_computation_cost): Ditto.
31099 (determine_group_iv_cost_generic): Ditto.
31100 (determine_group_iv_cost_address): Ditto.
31101 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31102 (determine_group_iv_costs): Ditto.
31103 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31104 (iv_ca_set_remove_invariants): Renamed to ...
31105 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31106 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31107 (iv_ca_set_add_invariants): Renamed to ...
31108 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31109 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31110 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31111 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31112 (create_new_ivs): Remove useless dump.
31114 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31116 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31118 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31119 (iv_ca_set_no_cp, create_new_iv): Ditto.
31121 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31123 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31125 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31127 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31128 function tree_check2.
31130 2017-05-02 Martin Liska <mliska@suse.cz>
31132 * doc/gcov.texi: Add missing preposition.
31133 * gcov.c (function_info::function_info): Properly fill up
31134 all member variables.
31136 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31138 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31140 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31142 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31144 2017-05-02 Martin Liska <mliska@suse.cz>
31147 * lto-streamer-in.c (lto_read_tree_1): Remove
31148 LTO_STREAMER_DEBUG.
31149 * lto-streamer.c (struct tree_hash_entry): Likewise.
31150 (struct tree_entry_hasher): Likewise.
31151 (tree_entry_hasher::hash): Likewise.
31152 (tree_entry_hasher::equal): Likewise.
31153 (lto_streamer_init): Likewise.
31154 (lto_orig_address_map): Likewise.
31155 (lto_orig_address_get): Likewise.
31156 (lto_orig_address_remove): Likewise.
31157 * lto-streamer.h: Likewise.
31158 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31159 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31161 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31163 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31164 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31165 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31166 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31167 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31168 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31169 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31170 (mm_maskz_sub_ss): New intrinsics.
31171 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31172 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31173 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31174 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31175 (__builtin_ia32_subss_mask_round): New builtins.
31176 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31177 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31178 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31180 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31181 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31183 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31186 2017-05-02 Martin Jambor <mjambor@suse.cz>
31188 PR tree-optimization/78687
31189 * tree-sra.c (access): New field parent.
31190 (process_subtree_disqualification): New function.
31191 (disqualify_candidate): Call it.
31192 (build_accesses_from_assign): Reset write flag if creating an
31194 (build_access_subtree): Fill in parent field and also prpagate
31195 down grp_write flag.
31196 (create_artificial_child_access): New parameter set_grp_write, set
31197 grp_write to its value.
31198 (propagate_subaccesses_across_link): Also propagate grp_write flag
31200 (propagate_all_subaccesses): Push the closest parent back to work
31201 queue if add_access_to_work_queue returned true.
31203 2017-05-02 Richard Biener <rguenther@suse.de>
31205 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31206 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31207 -fstrict-overflow documentation.
31208 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31209 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31210 flag_strict_overflow.
31211 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31212 * lto-opts.c (lto_write_options): Do not stream it.
31213 * lto-wrapper.c (merge_and_complain): Do not handle it.
31214 * opts.c (default_options_table): Do not set -fstrict-overflow.
31215 (finish_options): Likewise do not clear it when sanitizing.
31216 * simplify-rtx.c (simplify_const_relational_operation): Do not
31217 test flag_strict_overflow.
31219 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31221 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31222 using enabled attribute.
31223 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31224 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31225 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31226 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31227 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31228 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31229 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31230 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31231 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31232 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31234 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31236 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31238 2017-05-02 Richard Biener <rguenther@suse.de>
31240 PR tree-optimization/80591
31242 2017-04-10 Richard Biener <rguenther@suse.de>
31244 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31247 2017-05-02 Richard Biener <rguenther@suse.de>
31249 PR tree-optimization/80549
31250 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31251 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31252 headers if they do not have a preheader.
31254 2017-05-02 Martin Liska <mliska@suse.cz>
31257 * common.opt: Fix typo.
31258 * doc/invoke.texi: Likewise.
31260 2017-05-01 Jan Beulich <jbeulich@suse.com>
31262 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31263 swapping, add (x,x,m,x,n) alternative.
31265 2017-05-01 Nathan Sidwell <nathan@acm.org>
31267 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31268 unnecessary unadjusted_alignment check.
31270 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31273 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31275 * gimplify.c (gimplify_cilk_detach): New function.
31276 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31277 * tree-core.h: Document EXPR_CILK_SPAWN.
31278 * tree.h (EXPR_CILK_SPAWN): Define.
31280 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31282 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31283 to use new fixit_hint representation, using the "replace" logic.
31284 (get_line_span_for_fixit_hint): Likewise.
31285 (layout::print_any_fixits): Likewise.
31286 (selftest::test_one_liner_many_fixits): Rename to...
31287 (selftest::test_one_liner_many_fixits_1): ...this, and update
31288 comment and expected output to reflect that the multiple fix-it
31289 hints are now consolidated into one insertion.
31290 (selftest::test_one_liner_many_fixits_2): New test.
31291 (selftest::test_diagnostic_show_locus_one_liner): Update for
31293 (selftest::test_fixit_consolidation): Update for fix-it API
31295 * diagnostic.c (print_parseable_fixits): Likewise.
31296 * edit-context.c (edited_line::m_line_events): Convert from
31297 auto_vec <line_event *> to auto_vec <line_event>.
31298 (class line_event): Convert from abstract base class to a concrete
31299 class, taking over the role of replace_event.
31300 (class insert_event): Delete.
31301 (class replace_event): Rename to class line_event. Convert to
31303 (edit_context::add_fixits): Reimplement.
31304 (edit_context::apply_insert): Delete.
31305 (edit_context::apply_replace): Rename to...
31306 (edit_context::apply_fixit): ...this. Convert to half-open range.
31307 (edited_file::apply_insert): Delete.
31308 (edited_file::apply_replace): Rename to...
31309 (edited_file::apply_fixit): ...this.
31310 (edited_line::~edited_line): Drop deletion of events.
31311 (edited_line::apply_insert): Delete.
31312 (edited_line::apply_replace): Rename to...
31313 (edited_line::apply_fixit): ...this. Convert to half-open range.
31314 Update for change to type of m_line_events.
31315 * edit-context.h (edit_context::apply_insert): Delete.
31316 (edit_context::apply_replace): Rename to...
31317 (edit_context::apply_fixit): ...this.
31319 2017-05-01 Martin Sebor <msebor@redhat.com>
31321 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31324 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31327 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31328 __get_cpuid_max returns 0.
31329 (__get_cpuid_count): Ditto.
31331 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31333 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31334 replacement expression is another instance of one of its arguments.
31336 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31339 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31340 check for stack push/pop autoinc.
31341 * config/i386/i386.c (ix86_agi_dependent): Return false
31342 if the only reason why modified_in_p returned true is that
31343 addr is SP based and set_insn is a push or pop.
31345 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31347 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31350 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31353 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31354 (account_size_time): Use two predicates - exec_pred and
31356 (evaluate_conditions_for_known_args): Compute both clause and
31358 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31359 (inline_summary_t::duplicate): Update.
31360 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31362 (compute_inline_parameters): Likewise.
31363 (estimate_edge_size_and_time): Update caluclation of time.
31364 (estimate_node_size_and_time): Compute both time and nonspecialized
31366 (estimate_ipcp_clone_size_and_time): Update.
31367 (inline_merge_summary): Update.
31368 (do_estimate_edge_time): Update.
31369 (do_estimate_edge_size): Update.
31370 (do_estimate_edge_hints): Update.
31371 (inline_read_section, inline_write_summary): Stream both new predicates.
31372 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31374 (compute_inlined_call_time): Cleanup.
31375 (big_speedup_p): Update.
31376 (edge_badness): Update.
31377 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31378 (size_time_entry): Replace predicate by exec_predicate and
31379 nonconst_predicate.
31380 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31381 (estimate_edge_time): Return also nonspec_time.
31382 (reset_edge_growth_cache): Update.
31384 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31386 PR rtl-optimization/80491
31387 * ifcvt.c (noce_process_if_block): When looking for x setter
31388 with missing else_bb, don't check only the insn right before
31389 cond_earliest, but look for the last insn that x is modified in
31390 within the same bb.
31392 PR rtl-optimization/80491
31393 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31395 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31397 PR tree-optimization/80487
31398 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31400 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31402 PR tree-optimization/79697
31403 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31404 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31405 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31407 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31408 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31410 2017-04-28 Martin Sebor <msebor@redhat.com>
31412 PR tree-optimization/80523
31413 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31414 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31416 (maybe_warn, format_directive, parse_directive): Use new functions.
31417 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31419 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31421 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31423 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31425 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31426 target_header_dir): Set correctly.
31427 * configure: Regenerated.
31428 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31429 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31430 instead of SYSTEM_HEADER_DIR.
31432 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31434 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31435 (estimate_local_effects): Likewise.
31436 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31437 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31438 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31439 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31440 do_estimate_edge_time, estimate_edge_time): Likewise.
31441 * ipa-inline-analysis.c (estimate_node_size_and_time,
31442 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31443 (estimate_time_after_inlining): Remove.
31445 2017-04-28 Martin Liska <mliska@suse.cz>
31447 * doc/gcov.texi: Enhance documentation of gcov.
31449 2017-04-28 Martin Liska <mliska@suse.cz>
31451 * doc/gcov.texi: Sort options in alphabetic order.
31452 * doc/gcov-dump.texi: Likewise.
31453 * doc/gcov-tool.texi: Likewise.
31454 * gcov.c (print_usage): Likewise.
31455 * gcov-dump.c (print_usage): Likewise.
31456 * gcov-tool.c (print_merge_usage_message): Likewise.
31457 (print_rewrite_usage_message): Likewise.
31458 (print_overlap_usage_message): Likewise.
31460 2017-04-28 Martin Liska <mliska@suse.cz>
31462 PR gcov-profile/53915
31463 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31465 2017-04-28 Martin Liska <mliska@suse.cz>
31467 PR gcov-profile/79891
31468 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31469 is marked by compiler as living on a line.
31470 (get_cycles_count): Remove usage of the union.
31471 (output_intermediate_file): Likewise.
31472 (find_source): Fix GNU coding style.
31473 (accumulate_line_counts): Remove old non-all block mode.
31474 (output_lines): Remove usage of the union.
31475 * profile.c (output_location): Include all BBs, even if
31476 belonging to a same line (and file) as a previous BB.
31478 2017-04-28 Martin Liska <mliska@suse.cz>
31480 * gcov.c (process_args): Handle new argument 'w'.
31481 (read_graph_file): Assign ID to BBs.
31482 (output_branch_count): Display BB # if verbose flag is set.
31483 (output_lines): Likewise for arcs.
31484 (print_usage): Add '--verbose' option help.
31485 * doc/gcov.texi: Document --verbose (-w) option.
31487 2017-04-28 Martin Liska <mliska@suse.cz>
31489 * gcov.c (struct block_location_info): New struct.
31490 (process_file): Fill up the new structure.
31491 (read_graph_file): Replace usage of encoding by the newly added
31493 (add_line_counts): Likewise.
31494 (accumulate_line_counts): Remove usage of the union.
31495 (function_info::function_info): New function.
31496 (function_info::~function_info): Likewise.
31497 (process_file): Call delete instead of release_function.
31498 (release_function): Release the function.
31499 (release_structures): Call delete instead of release_function.
31500 (solve_flow_graph): Replace usage of num_blocks.
31501 (find_exception_blocks): Likewise.
31502 (output_lines): Fix GNU coding style.
31504 2017-04-28 Martin Liska <mliska@suse.cz>
31507 * coverage.c (coverage_remove_note_file): New function.
31508 * coverage.h: Declare the function.
31509 * toplev.c (finalize): Clean if an error has been seen.
31511 2017-04-28 Martin Liska <mliska@suse.cz>
31513 PR gcov-profile/80031
31514 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31515 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31516 * gcov.c (read_graph_file): Read just number of blocks.
31517 * profile.c (branch_prob): Do not stream 0 flags per a basic
31520 2017-04-28 Martin Liska <mliska@suse.cz>
31522 * gcov-dump.c (tag_*): Add new argument to declarations.
31523 (dump_gcov_file): Likewise.
31524 (tag_blocks): Add and use new argument depth.
31525 (tag_arcs): Likewise.
31526 (tag_lines): Likewise.
31527 (tag_counters): Likewise.
31528 (tag_summary): Likewise.
31529 (dump_working_sets): Use depth to do a proper indentation.
31531 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31534 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31535 * symtab.c (symtab_node::debug_symtab): Move definition here.
31537 2017-04-28 Richard Biener <rguenther@suse.de>
31539 * lto-streamer.h (LTO_major_version): Bump to 7.
31541 2017-04-28 Richard Biener <rguenther@suse.de>
31543 * tree-vrp.c (assert_info): New struct.
31544 (add_assert_info): New helper.
31545 (register_edge_assert_for_2): Refactor to add asserts to a vector
31547 (register_edge_assert_for_1): Likewise.
31548 (register_edge_assert_for): Likewise.
31549 (finish_register_edge_assert_for): New helper actually registering
31550 asserts where live on edge.
31551 (find_conditional_asserts): Adjust.
31552 (find_switch_asserts): Likewise.
31553 (evrp_dom_walker::try_find_new_range): Generalize.
31554 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31556 2017-04-27 Marek Polacek <polacek@redhat.com>
31559 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31560 arg10 and arg11 to itype.
31562 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31564 * doc/extend.texi (Object Size Checking): Improve grammar.
31566 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31569 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31570 that the logic for permitting reciprocal estimates matches that
31573 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31576 * tree.c (type_cache_hasher::equal): Only compare
31577 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31578 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31579 non-aggregate element types.
31580 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31581 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31583 2017-04-27 Richard Biener <rguenther@suse.de>
31585 PR middle-end/80533
31586 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31587 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31588 keeping a reference to a trailing array.
31590 2017-04-27 Richard Biener <rguenther@suse.de>
31592 PR middle-end/80539
31593 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31594 being in loop-closed SSA form conservatively.
31595 (chrec_fold_multiply_poly_poly): Likewise.
31597 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31599 PR middle-end/79665
31600 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31601 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31603 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31606 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31607 (aarch64_function_arg_alignment): Return unsigned int again, but still
31608 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31609 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31610 Don't emit -Wpsabi note.
31611 (aarch64_function_arg_boundary): Likewise.
31612 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31615 2017-04-26 Nathan Sidwell <nathan@acm.org>
31617 * tree.h (crc32_unsigned_n): Declare.
31618 (crc32_unsigned, crc32_unsigned): Make inline.
31619 * tree.c (crc32_unsigned_bits): Replace with ...
31620 (crc32_unsigned_n): ... this.
31621 (crc32_unsigned, crc32_byte): Remove.
31622 (crc32_string): Remove unnecessary braces.
31624 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31626 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31627 * ipa-inline-analysis.c (MAX_TIME): Remove.
31628 (account_size_time): Use sreal for time.
31629 (dump_inline_summary): Update.
31630 (estimate_function_body_sizes): Update.
31631 (estimate_edge_size_and_time): Update.
31632 (estimate_calls_size_and_time): Update.
31633 (estimate_node_size_and_time): Update.
31634 (inline_merge_summary): Update.
31635 (inline_update_overall_summary): Update.
31636 (estimate_time_after_inlining): Update.
31637 (inline_read_section): Update.
31638 (inline_write_summary): Update.
31639 * ipa-inline.c (compute_uninlined_call_time): Update.
31640 (compute_inlined_call_time): Update.
31641 (recursive_inlining): Update.
31642 (inline_small_functions): Update.
31643 (dump_overall_stats): Update.
31644 * ipa-inline.h: Include sreal.h.
31645 (size_time_entry): Turn time to sreal.
31646 (inline_summary): Turn self_time nad time to sreal.
31648 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31650 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31652 (sreal::stream_out, sreal::stream_in): New.
31653 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31655 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31657 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31660 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31663 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31665 (dimode_scalar_chain::compute_convert_gain): Ditto.
31666 (dimode_scalar_chain::make_vector_copies): Ditto.
31667 (dimode_scalar_chain::convert_reg): Ditto.
31668 (dimode_scalar_chain::convert_insn): Ditto.
31669 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31670 (VI248_AVX512BW_1): New mode iterator.
31671 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31672 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31675 2017-04-25 Martin Sebor <msebor@redhat.com>
31677 PR tree-optimization/80497
31678 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31679 constants are representable in HOST_WIDE_INT.
31680 (parse_directive): Ditto.
31682 2017-04-25 Martin Sebor <msebor@redhat.com>
31685 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31686 (new_zero_array): Adjust signature.
31687 (dom_info::dom_init): Used unsigned rather that size_t.
31688 (dom_info::dom_info): Same.
31690 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31691 Jakub Jelinek <jakub@redhat.com>
31694 * config/arm/arm.c: Include gimple.h.
31695 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31696 returns negative, increment ncrn only if it returned positive.
31697 (arm_needs_doubleword_align): Return int instead of bool,
31698 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31699 members, but if there is any such non-FIELD_DECL
31700 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31701 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31702 returns negative, increment nregs only if it returned positive.
31703 (arm_setup_incoming_varargs): Likewise.
31704 (arm_function_arg_boundary): Emit -Wpsabi note if
31705 arm_needs_doubleword_align returns negative, return
31706 DOUBLEWORD_ALIGNMENT only if it returned positive.
31708 2017-04-25 Marek Polacek <polacek@redhat.com>
31711 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31712 first argument to type.
31714 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
31717 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
31718 type checks to test for compatibility instead of equality.
31720 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31721 Jakub Jelinek <jakub@redhat.com>
31724 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
31726 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
31727 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
31728 the alignment computation, but return their maximum in warn_alignment.
31729 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31730 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
31732 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
31733 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31736 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31738 * config/arc/simdext.md (dmpyh): Fix typo.
31740 2017-04-25 Richard Biener <rguenther@suse.de>
31742 PR tree-optimization/80492
31743 * alias.c (compare_base_decls): Handle registers with asm
31744 specification conservatively.
31745 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31746 compare_base_decls returning dont-know properly.
31748 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31750 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31751 (legitimate_offset_address_p): New function.
31752 (arc_legitimate_address_p): Use above function.
31754 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31756 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31758 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31760 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31761 ACCH registers whenever they are available.
31763 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31765 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31766 double regs fix when not used.
31768 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31770 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31772 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31773 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31775 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31777 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31778 when emitting short ADD instructions.
31780 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31782 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31784 (cmpsi_cc_c_insn): Likewise.
31785 (cbranchsi4_scratch): Compute proper instruction length using
31786 compact_hreg_operand.
31787 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31789 2017-04-25 Richard Biener <rguenther@suse.de>
31791 PR middle-end/80509
31792 * passes.c (pass_manager::pass_manager): Initialize
31793 m_name_to_pass_map.
31795 2017-04-25 Richard Biener <rguenther@suse.de>
31797 PR tree-optimization/79201
31798 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31800 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31803 * config/s390/vector.md: Split MEM->GPR vector moves for
31804 non-s_operand addresses.
31806 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31809 * config/s390/predicates.md (reload_const_wide_int_operand): New
31811 * config/s390/s390.md ("movti"): Remove d/P alternative.
31812 ("movti_bigconst"): New pattern definition.
31814 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31817 * s390-protos.h (s390_expand_cs_hqi): Removed.
31818 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31819 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31820 modes as well as CCZ1mode and CCZmode.
31821 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31822 signature of s390_emit_compare_and_swap.
31823 (s390_expand_cs_hqi): Likewise, make static.
31824 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31825 compare-and-swap, in some cases.
31826 (s390_expand_cs): Wrapper function.
31827 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31829 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31830 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31831 patterns for small and large integers. Forbid symref memory operands.
31832 Move expander to s390.c. Require cc register.
31833 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31834 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31835 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31836 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31837 symref memory operands. Remove CC mode and call s390_match_ccmode
31839 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31841 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31843 * config/s390/s390.md (define_peephole2): New peephole to help
31844 combining the load-and-test pattern with volatile memory.
31846 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31848 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31849 with CCZmode for TARGET_Z196.
31851 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31853 PR rtl-optimization/80501
31854 * combine.c (make_compound_operation_int): Set subreg_code to SET
31855 even for AND with mask of the sign bit of mode.
31857 PR rtl-optimization/80500
31858 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31859 sum's initial value.
31861 2017-04-25 Julian Brown <julian@codesourcery.com>
31862 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31864 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31866 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31868 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31870 2017-04-25 Julian Brown <julian@codesourcery.com>
31871 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31873 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31874 (thunderx2t99_sha): New Reservation.
31876 2017-04-25 Julian Brown <julian@codesourcery.com>
31877 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31879 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31880 type for 1-element load.
31882 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31884 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31886 2017-04-24 Martin Jambor <mjambor@suse.cz>
31888 PR tree-optimization/80293
31889 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31890 char arrays not totally scalarizable if it is false.
31891 (analyze_all_variable_accesses): Pass correct value in the new
31892 parameter. Add a statistics counter.
31894 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31896 PR middle-end/79931
31897 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31899 2017-04-24 Richard Biener <rguenther@suse.de>
31901 PR tree-optimization/80494
31902 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31903 out for complex types.
31905 2017-04-24 Richard Biener <rguenther@suse.de>
31907 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31908 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31909 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31910 (DFS): Adjust and never fail.
31911 (sccvn_dom_walker::fail): Remove.
31912 (sccvn_dom_walker::before_dom_children): Adjust.
31913 (run_scc_vn): Likewise and never fail.
31914 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31915 (pass_fre::execute): Likewise.
31917 2017-04-24 Richard Biener <rguenther@suse.de>
31919 PR tree-optimization/79725
31920 * tree-ssa-sink.c (statement_sink_location): Return whether
31921 failure reason was zero uses. Move that check later.
31922 (sink_code_in_bb): Deal with zero uses by removing the stmt
31925 2017-04-24 Richard Biener <rguenther@suse.de>
31928 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31929 pointer-based references.
31931 2017-04-24 Richard Biener <rguenther@suse.de>
31934 * pass_manager.h (pass_manager::operator new): Remove.
31935 (pass_manager::operator delete): Likewise.
31936 * passes.c (pass_manager::operator new): Remove.
31937 (pass_manager::operator delete): Likewise.
31938 (pass_manager::pass_manager): Zero individual pass members.
31940 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31943 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31944 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31945 Check "XEXP (src, 1)" operand here.
31946 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31947 Check "XEXP (src, 1)" operand here.
31948 (dimode_scalar_chain::make_vector_copies): Detect count register
31949 of a shift instruction. Zero extend count register from QImode
31950 to DImode to satisfy vector shift pattern count operand predicate.
31951 Substitute vector shift count operand with a DImode copy.
31952 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31955 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31957 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31958 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31959 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31960 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31961 (UNSPEC_NOREX_MEM): Remove definition.
31963 2017-04-21 Richard Biener <rguenther@suse.de>
31965 PR tree-optimization/79547
31966 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31967 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31968 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31969 without any constraints.
31971 2017-04-21 Richard Biener <rguenther@suse.de>
31973 PR tree-optimization/78847
31974 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31976 2017-04-21 Richard Biener <rguenther@suse.de>
31978 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31979 (build_distinct_type_copy): Likewise.
31980 (build_variant_type_copy): Likewise.
31981 * tree.c (build_qualified_type): Pass down mem-stat info.
31982 (build_distinct_type_copy): Likewise.
31983 (build_variant_type_copy): Likewise.
31985 2017-04-21 Richard Biener <rguenther@suse.de>
31987 PR tree-optimization/80237
31988 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31990 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31991 for a simplified result.
31993 2016-04-21 Richard Biener <rguenther@suse.de>
31995 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31996 sth as strict as a simple_iv but a chrec without symbols and an
31997 operand defined in the loop we are peeling (and not some subloop).
31998 (propagate_constants_for_unrolling): Propagate all constants.
32000 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32003 * config/i386/i386.c (print_reg): Remove assert for disalowed
32004 regno values, call output_operand_lossage instead.
32006 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32009 * config/i386/constraints.md (Yc): New register constraint.
32010 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32011 Use Yc constraint for alternative 2 of operand 0. Remove
32012 preferred_for_speed attribute.
32014 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32016 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32017 lastprivate clauses in SIMT case.
32019 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32021 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32023 2017-04-20 Richard Biener <rguenther@suse.de>
32025 PR tree-optimization/57796
32026 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32027 as N scalar stores.
32028 (vect_model_load_cost): Cost gathers as N scalar loads.
32030 2017-04-20 Richard Biener <rguenther@suse.de>
32032 * ggc-page.c (ggc_allocated_p): Rename to ...
32033 (safe_lookup_page_table_entry): ... this and return the lookup
32035 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32037 2017-04-20 Richard Biener <rguenther@suse.de>
32039 PR tree-optimization/80453
32040 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32041 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32042 from the conditions.
32043 (vn_phi_eq): Pass them down.
32044 (vn_phi_lookup): Record them.
32045 (vn_phi_insert): Likewise.
32047 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32049 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32050 uninitialized variable warning to avoid buffer overrun.
32052 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32055 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32056 is suppressed for '{ 0 }' in C.
32058 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32060 * BASE-VER: Set to 8.0.0.
32062 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32064 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32065 priority .init_array and .fini_array section with SECTION_NOTYPE
32068 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32070 PR middle-end/80423
32071 * tree.h (build_array_type): Add typeless_storage default argument.
32072 * tree.c (type_cache_hasher::equal): Also compare
32073 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32074 (build_array_type): Add typeless_storage argument, set
32075 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32077 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32078 (build_array_type): Likewise. Add typeless_storage argument.
32080 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32081 Jakub Jelinek <jakub@redhat.com>
32083 PR tree-optimization/80426
32084 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32085 operation on symbolic operands, also compute the overflow for the
32086 invariant part when the operation degenerates into a negation.
32088 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32091 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32092 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32095 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32097 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32100 * config/avr/avr.c (tree.h): Include it.
32101 (cgraph.h): Include it.
32102 (avr_encode_section_info): Don't warn for uninitialized progmem
32103 variable if it's just an alias.
32105 2017-04-19 Richard Biener <rguenther@suse.de>
32108 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32109 when needed by AutoPGO.
32111 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32114 * doc/lto.texi: Remove an extra 'that'.
32116 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32118 PR rtl-optimization/80429
32119 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32120 are only used in debug insns.
32122 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32123 Vladimir Makarov <vmakarov@redhat.com>
32125 * config/sparc/predicates.md (input_operand): Add comment. Return
32126 true for any memory operand when LRA is in progress.
32127 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32129 2017-04-18 Jeff Law <law@redhat.com>
32132 * mips.md ({return,simple_return}_internal): Do not overwrite
32135 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32137 PR tree-optimization/80443
32138 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32139 instead of adding 1, subtract -1 and similarly instead of subtracting
32142 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32144 PR rtl-optimization/80357
32145 * haifa-sched.c (tmp_bitmap): New variable.
32146 (model_recompute): Handle duplicate use records.
32147 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32148 (free_global_sched_pressure_data): Free it.
32150 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32153 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32154 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32155 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32156 instead of SYSTEM_HEADER_DIR.
32158 2017-04-18 Jeff Law <law@redhat.com>
32160 PR middle-end/80422
32161 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32162 predecessors after walking up the insn chain.
32164 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32167 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32168 sizetype type into debug info.
32170 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32173 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32174 unneeded test for TARGET_UPPER_REGS_SF.
32175 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32177 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32180 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32181 instead of gsi_after_labels.
32183 2017-04-18 Jeff Law <law@redhat.com>
32185 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32189 2017-04-13 Jeff Law <law@redhat.com>
32190 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32191 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32193 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32196 * config/avr/avr.c (intl.h): Include it.
32197 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32199 2017-04-18 Martin Liska <mliska@suse.cz>
32201 PR gcov-profile/78783
32202 * gcov-tool.c (gcov_output_files): Validate that destination
32203 file is either removed by the tool or by a user.
32205 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32206 Guy Benyei <guybe@mellanox.com>
32208 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32209 block, and do not negate it, the stored id is already negative.
32211 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32213 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32215 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32218 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32219 masks of options that should be turned off if the VSX vector
32220 options are turned off.
32221 (OTHER_P8_VECTOR_MASKS): Likewise.
32222 (OTHER_VSX_VECTOR_MASKS): Likewise.
32223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32224 rs6000_disable_incompatible_switches to validate no type switches
32226 (rs6000_incompatible_switch): New function to disallow turning on
32227 other vector options if -mno-vsx, -mno-power8-vector, or
32228 -mno-power9-vector are specified.
32230 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32232 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32234 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32236 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32237 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32238 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32239 (ARG_POINTER_CFA_OFFSET): Likewise.
32241 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32243 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32244 conditions to take advantage of various optimizations.
32246 2017-04-13 Jeff Law <law@redhat.com>
32248 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32249 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32250 (zero_extendsidi2_dext): Likewise.
32252 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32255 * fold-const.c (fold_ternary_loc): Revert
32256 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32259 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32261 PR rtl-optimization/80343
32262 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32263 new scratch pseudo.
32265 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32268 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32269 to ubsan_encode_value.
32271 2017-04-13 Jeff Law <law@redhat.com>
32273 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32274 appearing in DEBUG_INSNs.
32276 2017-04-13 Martin Liska <mliska@suse.cz>
32278 PR gcov-profile/80413
32279 * gcov-io.c (gcov_write_string): Copy to buffer just when
32280 allocated size is greater than zero.
32282 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32285 * dwarf2out.c (decls_for_scope): Ignore declarations of
32286 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32288 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32291 * ipa-visibility.c (non_local_p): Fix typos.
32292 (localize_node): When localizing symbol in same comdat group,
32293 dissolve the group only when we know external symbols are going
32295 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32297 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32299 PR tree-optimization/79390
32300 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32301 order does not result in usable sequence, retry with reversed operand
32307 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32308 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32309 op0 instead of fold_convert_loc (loc, type, arg0).
32311 2017-04-12 Jeff Law <law@redhat.com>
32313 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32314 has a delay slot in the generated code.
32316 * config/cris/cris.md (cris_preferred_reload_class): Return
32317 GENNONACR_REGS rather than GENERAL_REGS.
32319 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32322 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32323 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32324 signedness of the result type.
32326 2017-04-12 Richard Biener <rguenther@suse.de>
32327 Jeff Law <law@redhat.com>
32329 PR tree-optimization/80359
32330 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32331 trim stores to TARGET_MEM_REFs.
32333 2017-04-12 Richard Biener <rguenther@suse.de>
32335 PR tree-optimization/79390
32336 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32337 threading case even more.
32339 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32342 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32343 for quad_address_p for TImode, instead of just not indexed_address.
32345 2017-04-12 Richard Biener <rguenther@suse.de>
32346 Bernd Edlinger <bernd.edlinger@hotmail.de>
32348 PR middle-end/79671
32349 * alias.c (component_uses_parent_alias_set_from): Handle
32350 TYPE_TYPELESS_STORAGE.
32351 (get_alias_set): Likewise.
32352 * tree-core.h (tree_type_common): Add typeless_storage flag.
32353 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32354 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32355 for types containing members with TYPE_TYPELESS_STORAGE.
32356 (place_field): Likewise.
32357 (layout_type): Likewise for ARRAY_TYPE.
32358 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32359 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32360 TYPE_TYPELESS_STORAGE.
32361 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32363 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32366 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32367 first argument to type.
32369 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32373 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32374 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32375 (rs6000_expand_binop_builtin): Likewise.
32376 (rs6000_expand_ternop_builtin): Likewise; also add missing
32377 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32378 vshasigma built-ins.
32379 * doc/extend.texi: Document that vec_xxpermdi's third argument
32380 must be a constant.
32382 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32384 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32385 Use shift_const cost parameter when calculating gain of STV shifts.
32387 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32389 PR rtl-optimization/70478
32390 * lra-constraints.c (process_alt_operands): Check memory for
32391 disfavoring memory insn operand.
32393 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32395 PR middle-end/80100
32396 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32397 left shift in unsigned HOST_WIDE_INT type.
32399 PR rtl-optimization/80385
32400 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32401 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32404 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32405 if they have any depend clauses.
32407 2017-04-11 Martin Liska <mliska@suse.cz>
32410 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32411 * ipa-split.c (split_function): Create a local comdat symbol
32412 if caller is in a comdat group.
32414 2017-04-11 Martin Liska <mliska@suse.cz>
32417 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32420 2017-04-11 Martin Sebor <msebor@redhat.com>
32422 PR middle-end/80364
32423 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32424 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32426 (directive::set_width, directive::set_precision, format_character):
32428 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32431 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32434 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32435 conflict, set target->arch_name instead of target->cpu_name.
32437 2017-04-11 Richard Biener <rguenther@suse.de>
32439 PR tree-optimization/80374
32440 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32441 build_zero_cst, remove fold_convertible_p check again.
32443 2017-04-11 Martin Liska <mliska@suse.cz>
32446 * ubsan.c (instrument_object_size): Do not instrument register
32449 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32452 * config/i386/i386-builtin-types.def
32453 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32454 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32455 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32456 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32457 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32458 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32459 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32460 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32461 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32462 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32463 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32464 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32465 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32466 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32467 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32468 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32469 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32470 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32471 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32472 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32473 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32474 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32475 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32476 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32477 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32478 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32479 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32480 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32481 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32482 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32483 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32484 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32485 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32486 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32487 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32488 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32489 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32490 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32491 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32492 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32493 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32494 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32495 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32496 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32497 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32498 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32500 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32501 flag to second_arg_count, handle 4 argument function type _COUNT
32502 aliases, handle second_arg_count on second argument rather than last.
32504 2017-04-10 Jeff Law <law@redhat.com>
32506 PR tree-optimization/80374
32507 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32508 record anything if we can not convert integer_zero_node to the
32511 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32514 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32515 Enhance special handling given to the TARGET_P9_MINMAX option in
32516 relation to certain other options.
32518 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32520 PR tree-optimization/80153
32521 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32522 remove POINTER_PLUS_EXPR's base part directly, rather than through
32525 2017-04-10 Richard Biener <rguenther@suse.de>
32526 Bin Cheng <bin.cheng@arm.com>
32528 PR tree-optimization/80153
32529 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32530 the first element of pointer type aff_tree. Build result expr in
32532 (add_elt_to_tree): Convert to type unconditionally. Remove other
32533 fold_convert calls.
32534 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32535 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32537 2017-04-10 Richard Biener <rguenther@suse.de>
32539 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32542 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32544 PR rtl-optimization/70478
32545 * lra-constraints.c (curr_small_class_check): New.
32546 (update_and_check_small_class_inputs): New.
32547 (process_alt_operands): Update curr_small_class_check. Disfavor
32548 alternative insn memory operands. Check available regs for small
32551 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32554 * config/mips/mips.opt (-mvirt): Update description.
32555 * doc/invoke.texi (-mvirt): Likewise.
32557 2017-04-10 Richard Biener <rguenther@suse.de>
32559 PR middle-end/80362
32560 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32561 looking for NEGATE_EXPR in -A / -B to A / B folding.
32563 2017-04-10 Martin Liska <mliska@suse.cz>
32565 PR gcov-profile/80224
32566 * gcov.c (print_usage): Fix usage string.
32567 (get_gcov_intermediate_filename): Remove.
32568 (output_gcov_file): Use both for normal and intermediate format.
32569 (generate_results): Do not initialize special file for
32570 intermediate format.
32572 2017-04-10 Richard Biener <rguenther@suse.de>
32574 PR tree-optimization/80304
32575 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32578 2017-04-10 Nathan Sidwell <nathan@acm.org>
32581 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32582 (rs6000_init_builtins): Use it.
32584 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32586 * config/arm/arm.md (<mrc>): Add mode to SET source.
32587 (<mrrc>): Likewise.
32589 2017-04-10 Richard Biener <rguenther@suse.de>
32591 PR middle-end/80344
32592 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32594 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32597 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32598 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32599 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32600 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32601 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32602 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32603 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32604 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32605 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32606 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32607 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32608 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32609 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32610 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32611 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32612 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32613 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32614 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32615 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32616 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32617 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32618 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32619 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32621 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32623 PR rtl-optimization/70478
32624 * lra-constraints.c: Reverse the last patch.
32626 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32628 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32629 Add comment for WCHAR_T.
32631 2017-04-08 Martin Liska <mliska@suse.cz>
32634 2017-04-07 Martin Liska <mliska@suse.cz>
32637 * ipa-split.c (split_function): Add function part to a same comdat
32640 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32643 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32645 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32647 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32648 (vec_pairing): ...to this.
32649 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32650 (rs6000_sched_init): Adjust for name change.
32651 (struct rs6000_sched_context): Likewise.
32652 (rs6000_init_sched_context): Likewise.
32653 (rs6000_set_sched_context): Likewise.
32655 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32661 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32663 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32664 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32665 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32667 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32669 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32671 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32673 PR rtl-optimization/70703
32674 * ira-color.c (update_conflict_hard_regno_costs): Use
32675 int64_t instead of HOST_WIDE_INT.
32677 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32679 PR rtl-optimization/70478
32680 * lra-constraints.c (process_alt_operands): Disfavor alternative
32681 insn memory operands.
32683 2017-04-07 Jeff Law <law@redhat.com>
32685 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32686 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32688 2017-04-07 Martin Liska <mliska@suse.cz>
32691 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32692 Show error message instead of an ICE.
32694 2017-04-07 Martin Liska <mliska@suse.cz>
32697 * ipa-split.c (split_function): Add function part to a same comdat
32700 2017-04-07 Richard Biener <rguenther@suse.de>
32702 PR middle-end/80341
32703 * tree.c (get_unwidened): Also handle ! for_type case for
32705 * convert.c (do_narrow): Split out from ...
32706 (convert_to_integer_1): ... here. Do not pass final truncation
32707 type to get_unwidened for TRUNC_DIV_EXPR.
32709 2017-04-07 Richard Biener <rguenther@suse.de>
32711 * tree-affine.c (wide_int_ext_for_comb): Take type rather
32713 (aff_combination_const): Adjust.
32714 (aff_combination_scale): Likewise.
32715 (aff_combination_add_elt): Likewise.
32716 (aff_combination_add_cst): Likewise.
32717 (aff_combination_convert): Likewise.
32718 (add_elt_to_tree): Likewise. Remove unused argument.
32719 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
32721 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
32723 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
32725 * config/arm/arm.c (arm_default_short_enums): Use
32726 ARM_DEFAULT_SHORT_ENUMS.
32727 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
32729 2017-04-06 Jakub Jelinek <jakub@redhat.com>
32732 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
32733 members with redundant out-of-class redeclaration.
32735 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32738 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32739 * config/i386/i386.md (*zero_extendsidi2):
32740 Add (?*x,*x) and (?*v,*v) alternatives.
32742 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32745 * config/i386/i386.c (ix86_expand_builtin)
32746 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32747 mode from insn data. Convert operands to insn operand mode.
32748 Copy operands that don't satisfy insn predicate to a register.
32750 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32752 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32755 2017-04-06 Richard Biener <rguenther@suse.de>
32757 PR tree-optimization/80334
32758 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32759 preserve alignment of accesses.
32761 2017-04-06 Richard Biener <rguenther@suse.de>
32763 PR tree-optimization/80262
32764 * tree-sra.c (build_ref_for_offset): Preserve address-space
32766 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32767 Drop useless address-space information on MEM_REF offsets.
32769 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32771 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32773 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32775 PR rtl-optimization/70703
32776 * ira-color.c (update_conflict_hard_regno_costs): Use
32777 HOST_WIDE_INT instead of long.
32779 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32782 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32783 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32785 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32786 for x86_64 target. Handle -m3dnowa option.
32788 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32790 PR rtl-optimization/70703
32791 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32792 (update_conflict_hard_regno_costs): Use long instead of unsigned
32793 arithmetic for cost calculation.
32795 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32796 Bernd Edlinger <bernd.edlinger@hotmail.de>
32799 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32802 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32805 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32806 ptr_mode with Pmode throughout.
32807 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32808 into probe_stack_range and use DImode.
32810 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32813 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32814 call_eh_return is true.
32816 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32818 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32819 Initialize last_match_fntype_index.
32821 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32824 * tree-nvr.c: Include internal-fn.h.
32825 (pass_return_slot::execute): Ignore internal calls without
32828 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32829 Richard Biener <rguenther@suse.de>
32832 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32833 captures used multiple times, except for the last use.
32834 * generic-match-head.c: Include gimplify.h.
32836 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32838 PR tree-optimization/79390
32839 * target.h (struct noce_if_info): Declare.
32840 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32841 * target.def (noce_conversion_profitable_p): New target hook.
32842 * ifcvt.h (struct noce_if_info): New type, moved from ...
32843 * ifcvt.c (struct noce_if_info): ... here.
32844 (noce_conversion_profitable_p): Renamed to ...
32845 (default_noce_conversion_profitable_p): ... this. No longer
32847 (noce_try_store_flag_constants, noce_try_addcc,
32848 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32849 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32850 instead of noce_conversion_profitable_p.
32851 * config/i386/i386.c: Include ifcvt.h.
32852 (ix86_option_override_internal): Don't override
32853 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32854 (ix86_noce_conversion_profitable_p): New function.
32855 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32856 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32857 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32858 * doc/tm.texi: Regenerated.
32860 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32862 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32865 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32868 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32869 instructions for small multiply cores.
32871 2017-04-04 Jeff Law <law@redhat.com>
32873 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32875 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32876 that are not set by the loop over the elements.
32878 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32881 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32882 int mode, convert_modes it to mode as unsigned, otherwise use
32883 lowpart_subreg to mode rather than SImode.
32884 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32885 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32886 Use DImode instead of SImode for the shift count operand.
32887 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32890 2017-04-04 Richard Biener <rguenther@suse.de>
32892 PR middle-end/80281
32893 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32894 arithmetic done for the negate or the plus. Simplify.
32895 (A - (-B) -> A + B): Likewise.
32896 * fold-const.c (split_tree): Make sure to not negate pointers.
32898 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32900 PR rtl-optimization/60818
32901 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32902 a compare of comparisons with the thing compared if this results
32903 in a different machine mode.
32905 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32907 * alias.c (base_alias_check): Fix typo in comment.
32908 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32909 * cgraphunit.c (symbol_table::compile): Likewise.
32910 * collect2.c (maybe_run_lto_and_relink): Likewise.
32911 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32912 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32913 * config/avr/avr.c (avr_map_op_t): Likewise.
32914 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32915 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32916 * config/epiphany/epiphany.md (movcc): Likewise.
32917 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32918 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32920 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32921 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32922 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32923 * config/sh/sh.c (sh_rtx_costs): Likewise.
32924 * fold-const.c (fold_truth_andor): Likewise.
32925 * genautomata.c (collapse_flag): Likewise.
32926 * gengtype.h (struct type::u::s): Likewise.
32927 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32928 * input.c (FORMAT_AMOUNT): Likewise.
32929 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32930 (known_aggs_to_agg_replacement_list): Likewise.
32931 * ipa-inline-analysis.c: Likewise.
32932 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32933 * ipa-polymorphic-call.c
32934 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32935 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32936 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32938 * modulo-sched.c (apply_reg_moves): Likewise.
32939 * omp-expand.c (build_omp_regions_1): Likewise.
32940 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32941 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32942 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32943 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32944 * value-prof.c: Likewise.
32945 * var-tracking.c (val_reset): Likewise.
32947 2017-04-03 Richard Biener <rguenther@suse.de>
32949 PR tree-optimization/80275
32950 * fold-const.c (split_address_to_core_and_offset): Handle
32953 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32955 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32956 descriptors is at least equal to that of functions.
32958 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32960 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32962 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32965 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32966 (mov<IMOD4:mode>): New expander.
32967 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32969 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32971 PR rtl-optimization/79405
32972 * fwprop.c (propagations_left): New variable.
32973 (forward_propagate_into): Decrement it.
32974 (fwprop_init): Initialize it.
32975 (fw_prop): If the variable has reached zero, stop propagating.
32976 (fwprop_addr): Ditto.
32978 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32981 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32982 a FUNCTION_DECL, pass it as decl instead of origin to
32985 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32987 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32990 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32993 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32994 TARGET_VSX_SMALL_INTEGER.
32996 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32998 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32999 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33001 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33003 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33004 extraction from odd-numbered MSA register.
33006 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33008 PR middle-end/80173
33009 * expmed.c (store_bit_field_1): Don't attempt to create
33010 a word subreg out of hard registers wider than word if they
33011 have HARD_REGNO_NREGS of 1 for their mode.
33013 PR middle-end/80163
33014 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33015 conversions to integer types wider than word and pointer.
33018 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33019 (rtx_equal_for_cselib_p): Pass 0 to it.
33020 * cselib.c (cselib_hasher::equal): Likewise.
33021 (rtx_equal_for_cselib_1): Add depth argument. If depth
33022 is 128, don't look up VALUE locs and punt. Increment
33023 depth in recursive calls when walking VALUE locs.
33025 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33027 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33028 (make_gcov_file_name): Use the canonical path name for generating
33030 (read_line): Fix handling of files with ascii null bytes.
33032 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33034 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33035 to initialise a vector register instead
33036 of using a const_int.
33038 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33040 PR translation/80189
33041 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33042 diagnostic messages.
33044 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33047 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33048 (dfp_diex_<mode>): Update mode of operand 1.
33049 * doc/extend.texi (dxex, dxexq): Document change to return type.
33050 (diex, diexq): Document change to argument type.
33052 2017-03-30 Martin Jambor <mjambor@suse.cz>
33055 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33056 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33057 it reflects the signature changes performed at the callee side.
33058 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33059 to cgraph_build_function_type_skip_args.
33060 (build_function_decl_skip_args): Adjust call to the above function.
33062 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33065 * config/i386/sse.md
33066 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33067 register as dest whenever it is a MEM not rtx_equal_p to the
33068 corresponding dup operand, and when forcing into reg move the
33069 reg into the memory afterwards.
33070 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33071 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33072 for the force_reg mode.
33073 (avx512vl_vextractf128<mode>): Use register as dest either
33074 always when a MEM, or when it is a MEM not rtx_equal_p to the
33075 corresponding dup operand, or even not when it is a CONST_VECTOR
33076 depending on the mode and lo vs. hi.
33077 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33079 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33080 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33081 Likewise. Require that operands[2] is even.
33082 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33083 Remove extraneous parens. Require that operands[2] is a multiple
33085 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33086 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33087 disallow memory then.
33089 2017-03-30 Richard Biener <rguenther@suse.de>
33091 PR tree-optimization/77498
33092 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33093 to non-constants over backedges.
33095 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33097 PR rtl-optimization/80233
33098 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33099 as last_combined_insn. Do not test for BARRIER_P separately.
33101 2017-03-29 Andreas Schwab <schwab@suse.de>
33104 * calls.c (prepare_call_address): Convert funexp to Pmode before
33105 copying to temp reg.
33107 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33109 PR tree-optimization/80158
33110 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33111 Handle possible future case of more than one alternate
33113 (replace_rhs_if_not_dup): Likewise.
33114 (replace_one_candidate): Likewise.
33116 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33118 PR rtl-optimization/80193
33119 * ira.c (ira): Do not check allocation for LRA.
33121 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33123 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33124 (nvptx_output_simt_exit): Declare.
33125 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33126 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33127 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33128 (nvptx_declare_function_name): ...here. Emit declaration of local
33129 memory space buffer for omp_simt_enter insn.
33130 (nvptx_output_unisimt_switch): New.
33131 (nvptx_output_softstack_switch): New.
33132 (nvptx_output_simt_enter): New.
33133 (nvptx_output_simt_exit): New.
33134 * config/nvptx/nvptx.h (struct machine_function): New fields
33135 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33136 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33137 (UNSPECV_SIMT_EXIT): Ditto.
33138 (omp_simt_enter_insn): New insn.
33139 (omp_simt_enter): New expansion.
33140 (omp_simt_exit): New insn.
33141 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33143 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33144 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33145 (expand_GOMP_SIMT_EXIT): New.
33146 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33147 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33148 (GOMP_SIMT_EXIT): Ditto.
33149 * target-insns.def (omp_simt_enter): New insn.
33150 (omp_simt_exit): Ditto.
33151 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33153 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33154 (lower_rec_input_clauses): Likewise.
33155 (lower_lastprivate_clauses): Handle SIMT privatization.
33157 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33158 (ompdevlow_adjust_simt_enter): New.
33159 (find_simtpriv_var_op): New.
33160 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33161 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33163 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33164 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33165 (copy_decl_for_dup_finish): Ditto.
33167 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33169 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33172 * config/i386/i386.c (ix86_option_override_internal): Always
33173 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33175 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33177 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33179 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33181 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33182 mark new edge's irreducible flag accordign to it.
33183 (vect_do_peeling): Check loop preheader edge's irreducible flag
33184 and pass it to function slpeel_add_loop_guard.
33186 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33188 PR tree-optimization/80218
33189 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33190 Update block frequencies and counts.
33192 2017-03-28 Richard Biener <rguenther@suse.de>
33194 PR tree-optimization/78644
33195 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33196 of a simplification result we may not use it at all.
33198 2017-03-28 Richard Biener <rguenther@suse.de>
33201 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33202 without arguments, generate default definition of a SSA name.
33204 2017-03-28 Richard Biener <rguenther@suse.de>
33206 PR middle-end/80222
33207 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33208 TYPE_REF_CAN_ALIAS_ALL references.
33209 * fold-const.c (fold_indirect_ref_1): Likewise.
33211 2017-03-28 Martin Liska <mliska@suse.cz>
33214 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33215 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33217 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33218 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33220 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33221 (EXTRA_SPECS): Define.
33222 (SUBTARGET_EXTRA_SPECS): Likewise.
33223 (SUBTARGET_CPP_SPEC): Likewise.
33224 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33225 SUBTARGET_EXTRA_SPECS.
33226 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33228 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33230 * config/arc/simdext.md (vst64_insn): Update pattern.
33231 (vld32wh_insn): Likewise.
33232 (vld32wl_insn): Likewise.
33233 (vld64_insn): Likewise.
33234 (vld32_insn): Likewise.
33236 2017-03-28 Marek Polacek <polacek@redhat.com>
33239 * fold-const.c (fold_comparison): Use protected_set_expr_location
33240 instead of SET_EXPR_LOCATION.
33242 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33244 * tree.c (add_expr): Avoid name lookup warning.
33246 2017-03-27 Jeff Law <law@redhat.com>
33248 PR tree-optimization/80216
33249 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33250 function name. Limit recursion depth.
33251 (record_temporary_equivalences): Corresponding changes.
33253 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33255 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33258 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33261 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33263 * cfgcleanup.c (reg_note_cfa_p): New array.
33264 (insns_have_identical_cfa_notes): New function.
33265 (old_insns_match_p): Don't cross-jump in between /f
33266 and non-/f instructions. If both i1 and i2 are frame related,
33267 verify all CFA notes, their order and content.
33269 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33272 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33273 HImode and SImode with zero extend to DImode to one insn.
33274 (bswap<mode>2_extenddi): Likewise.
33275 (bswapsi2_extenddi): Likewise.
33276 (bswaphi2_extendsi): Likewise.
33277 (bswaphi2): Combine bswap HImode and SImode into one insn.
33278 Separate memory insns from swapping register.
33279 (bswapsi2): Likewise.
33280 (bswap<mode>2): Likewise.
33281 (bswaphi2_internal): Delete, no longer used.
33282 (bswapsi2_internal): Likewise.
33283 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33284 store, and gpr<-gpr swap insns.
33285 (bswap<mode>2_store): Likewise.
33286 (bswaphi2_reg): Register only splitter, combine with the splitter.
33287 (bswaphi2 splitter): Likewise.
33288 (bswapsi2_reg): Likewise.
33289 (bswapsi2 splitter): Likewise.
33290 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33291 the insns into load, store, and register/register insns.
33292 (bswapdi2_ldbrx): Likewise.
33293 (bswapdi2_load): Likewise.
33294 (bswapdi2_store): Likewise.
33295 (bswapdi2_reg): Likewise.
33297 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33299 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33300 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33302 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33305 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33307 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33308 special handling for target option conflicts between dform
33309 options (-mpower9-dform, -mpower9-dform-vector,
33310 -mpower9-dform-scalar) and -mno-direct-move.
33312 2017-03-27 Richard Biener <rguenther@suse.de>
33314 PR tree-optimization/80181
33315 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33317 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33319 * config/arc/predicates.md (move_double_src_operand): Replace the
33320 call to move_double_src_operand with a call to address_operand.
33322 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33324 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33325 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33326 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33328 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33330 * config/arc/predicates.md (long_immediate_loadstore_operand):
33331 Consider scaled addresses cases.
33333 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33335 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33336 restored when in interrupt.
33337 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33338 doesn't have delay slot.
33340 2017-03-27 Richard Biener <rguenther@suse.de>
33343 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33344 inlined thunk clones.
33346 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33349 * asan.c (instrument_derefs): Copy over last operand from
33350 original COMPONENT_REF to the new COMPONENT_REF with
33351 DECL_BIT_FIELD_REPRESENTATIVE.
33352 * ubsan.c (instrument_object_size): Likewise.
33354 2017-03-27 Richard Biener <rguenther@suse.de>
33356 PR tree-optimization/80170
33357 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33358 sure DR/SCEV didnt fold in constants we do not see when looking
33359 at the reference base alignment.
33361 2017-03-27 Richard Biener <rguenther@suse.de>
33363 PR middle-end/80171
33364 * gimple-fold.c (fold_ctor_reference): Properly guard against
33365 NULL return value from canonicalize_constructor_val.
33367 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33370 * config/i386/i386.c (ix86_expand_builtin)
33371 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33372 flags reg setting and flags reg using instructions.
33373 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33374 clobbering instructions to zero extend op2.
33376 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33378 * doc/install.texi (Configuration) <--with-aix-soname>:
33379 Update link to AIX ld.
33381 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33383 PR rtl-optimization/80160
33384 PR rtl-optimization/80159
33385 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33386 reg_alternate_class into account.
33388 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33391 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33392 to consider in curr_insn_transform.
33394 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33396 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33397 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33398 and emit_mode_inner.
33400 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33402 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33403 argument to the overloaded builtin variants. Use the new flag to
33404 deprecate certain builtin variants.
33405 * config/s390/s390-builtin-types.def: Add new builtin types.
33406 * config/s390/s390-builtins.h: Support new flags field for
33407 overloaded builtins.
33408 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33409 (s390_macro_to_expand): Enable vector float data type.
33410 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33411 builtins by incrementing the __VEC__ version number.
33412 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33414 (s390_resolve_overloaded_builtin): Emit error messages depending
33415 on the builtin flags.
33416 * config/s390/s390.c (s390_expand_builtin): Support additional
33417 flags argument. Change error message to match the messages
33418 emitted in s390-c.c.
33419 * config/s390/s390.md: New UNSPEC_* constants.
33420 (op_type): Add new instruction types.
33421 * config/s390/vecintrin.h: Add new builtins and test data class
33423 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33424 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33425 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33426 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33427 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33428 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33430 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33431 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33432 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33433 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33435 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33436 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33437 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33438 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33440 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33441 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33442 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33443 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33444 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33445 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33446 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33448 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33449 ("vec_scatter_element<V_HW_4:mode>_DI")
33450 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33451 ("vec_fpint<mode>", "vflls")
33452 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33453 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33454 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33455 ("*vec_cmphe<mode>_cc"): ... these.
33457 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33458 mode constant instead of magic value.
33460 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33462 * config/s390/s390.c (s390_expand_vec_compare): Support other
33463 vector floating point modes than just V2DF.
33464 (s390_expand_vcond): Likewise.
33465 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33466 (s390_cannot_change_mode_class): Prevent mode changes between TF
33467 and V1TF in vector registers.
33468 * config/s390/s390.md (DF, SF): New mode attributes.
33469 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33470 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33471 SFmode support for VRs.
33472 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33474 (VFT, VF_HW): New mode iterators.
33475 (vw, sdx): New mode attributes.
33476 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33477 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33478 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33479 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33480 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33481 also the new vector floating point modes. Renaming to ...
33483 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33484 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33485 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33486 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33487 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33488 ("vec_unordered<mode>"): ... these.
33490 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33491 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33492 ("*vec_extendv2df"): New insn definitions.
33494 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33496 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33497 ("mulditi3_2", "*muldi3_sign"): New patterns.
33498 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33499 rename the pattern definition.
33501 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33503 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33505 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33509 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33510 instruction if possible.
33511 * config/s390/vector.md (vec_halfnumelts): New mode
33513 ("*vec_vllezlf<mode>"): New pattern.
33515 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33517 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33518 ("popcountv4si2", "popcountv2di2"): Rename to ...
33519 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33520 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33522 ("popcount<mode>2_vxe"): New pattern.
33524 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33526 * common/config/s390/s390-common.c (processor_flags_table): Add
33528 * config.gcc: Add arch12.
33529 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33530 Default to arch12 for unknown CPU model numbers.
33531 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33532 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33533 PROCESSOR_max sanity check.
33534 * config/s390/s390-opts.h (enum processor_type): Add
33536 * config/s390/s390.c (processor_table): Add arch12.
33537 (s390_expand_builtin): Add check for B_VXE flag.
33538 (s390_issue_rate): Add PROCESSOR_ARCH12.
33539 (s390_get_sched_attrmask): Likewise.
33540 (s390_get_unit_mask): Likewise.
33541 (s390_sched_score): Enable z13 scheduling for arch12.
33542 (s390_sched_reorder): Likewise.
33543 (s390_sched_variable_issue): Likewise.
33544 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33546 (s390_tune_attr): Use z13 scheduling also for arch12.
33547 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33548 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33549 (TARGET_VXE_P): New macros.
33550 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33551 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33552 * config/s390/s390.opt: Add arch12 as processor_type.
33554 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33556 * config/s390/s390.md
33557 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33558 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33559 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33561 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33562 Rename expanders to ...
33564 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33565 ("fixuns_truncdddi2_emu"): ... these.
33567 ("fixuns_trunc<mode>si2_emu"): New expander.
33569 ("*fixuns_truncdfdi2_z13"): Rename to ...
33570 ("*fixuns_truncdfdi2_vx"): ... this.
33572 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33574 * config/s390/2964.md: Remove the single element vector compare
33575 instructions which are no longer used.
33576 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33578 (s390_canonicalize_comparison): Remove handling of DFmode
33580 (s390_expand_vec_compare_scalar): Remove function.
33581 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33582 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33584 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33586 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33588 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33590 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33591 will anyway by matched by mov<mode>_64dfp.
33593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33595 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33596 vlef/vstef. Add missing operand to vleif.
33598 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33600 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33601 pair for all vector types with 64 bit elements.
33602 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33603 * config/s390/vector.md (V_HW_64): ... here.
33604 (V_128_NOSINGLE): New mode iterator.
33605 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33606 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33607 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33608 ("*vec_load_pairv2di"): Change to ...
33609 ("*vec_load_pair<mode>"): ... this one.
33611 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33613 * config/s390/constraints.md: Add comments.
33614 (jKK): Reject element sizes > 8 bytes.
33615 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33617 * config/s390/s390.md: Add the s_operand checks formerly in
33618 s390_split_ok_p to various splitters where they are still
33620 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33621 for 128 bit vectors. Plus two splitters.
33623 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33625 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33631 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33632 error if the boundary argument is not constant.
33634 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33636 PR rtl-optimization/80112
33637 * loop-doloop.c (doloop_condition_get): Don't check condition
33638 if cmp isn't SET with IF_THEN_ELSE src.
33640 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33642 PR tree-optimization/80158
33643 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33644 replacing a candidate statement, also replace it for the
33645 candidate's alternate interpretation.
33646 (replace_rhs_if_not_dup): Likewise.
33647 (replace_one_candidate): Likewise.
33649 2017-03-24 Richard Biener <rguenther@suse.de>
33651 PR tree-optimization/80167
33652 * graphite-isl-ast-to-gimple.c
33653 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33655 (translate_isl_ast_to_gimple::get_rename): Likewise.
33657 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33659 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33660 handling of certain combinations of target options, including the
33661 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33662 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33664 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33667 * config/arm/arm.md (*load_multiple): Add reload_completed to
33668 matching condition.
33670 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33671 Richard Biener <rguenth@suse.de>
33673 PR tree-optimization/79908
33674 PR tree-optimization/80136
33675 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33676 been cast away, gimplify_and_add suffices.
33678 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33680 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33682 2017-03-23 Richard Biener <rguenther@suse.de>
33684 PR tree-optimization/80032
33685 * gimplify.c (gimple_push_cleanup): Forced unconditional
33686 cleanups still have to go to the conditional_cleanups
33689 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33691 PR tree-optimization/80072
33692 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33694 (next_operand_entry_id): Change type to unsigned int.
33695 (sort_by_operand_rank): Make sure to return the right return value
33696 even if unsigned fields are bigger than INT_MAX.
33697 (struct oecount): Change cnt and id type to unsigned int.
33698 (oecount_hasher::equal): Formatting fix.
33699 (oecount_cmp): Make sure to return the right return value
33700 even if unsigned fields are bigger than INT_MAX.
33701 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33704 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33705 TREE_READONLY on result if writing it more than once.
33708 * doc/invoke.texi (-fsanitize=thread): Document that with
33709 -fnon-call-exceptions atomics are not able to throw
33713 * tsan.c: Include tree-eh.h.
33714 (instrument_builtin_call): Call maybe_clean_eh_stmt or
33715 maybe_clean_or_replace_eh_stmt where needed.
33716 (instrument_memory_accesses): Add cfg_changed argument.
33717 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
33718 if it returned true.
33719 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
33721 PR rtl-optimization/63191
33722 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
33723 wrapper function, moved the whole old content into ...
33724 (ix86_delegitimize_address_1): ... this. New inline function.
33725 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
33726 true as last argument instead of ix86_delegitimize_address.
33728 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33730 * config/aarch64/aarch64.c (generic_branch_cost): Copy
33731 cortexa57_branch_cost.
33733 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33735 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33737 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33740 * doc/md.texi (Constraints): Document wA constraint.
33741 * config/rs6000/constraints.md (wA): New.
33742 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33743 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33744 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33745 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33747 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33750 * gimplify.c (is_oacc_declared): New function.
33751 (oacc_default_clause): Use it to set default flags for acc declared
33752 variables inside parallel regions.
33753 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33754 declared variables.
33755 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33756 declare attribute to any decl as necessary.
33758 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33761 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33762 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33763 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33764 (arm_arch_lpae): This.
33765 * config/arm/arm.c (arm_arch7ve): Rename into ...
33766 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33767 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33770 2017-03-22 Martin Liska <mliska@suse.cz>
33773 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33774 error message instead of an ICE.
33776 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33778 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33780 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33782 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33784 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33787 2017-03-21 Martin Sebor <msebor@redhat.com>
33789 * doc/extend.texi: Use "cannot" instead of "can't."
33790 * doc/hostconfig.texi: Same.
33791 * doc/install.texi: Same.
33792 * doc/invoke.texi: Same.
33793 * doc/loop.texi: Same.
33794 * doc/md.texi: Same.
33795 * doc/objc.texi: Same.
33796 * doc/rtl.texi: Same.
33797 * doc/tm.texi: Same.
33798 * doc/tm.texi.in: Same.
33799 * doc/trouble.texi: Same.
33801 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33804 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33805 (collect_checksum_attributes): Set it.
33806 (die_checksum_ordered): Use it.
33808 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33810 PR tree-optimization/79908
33811 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33812 change: For a VA_ARG whose LHS has been cast away, use
33813 force_gimple_operand to construct the side effects.
33815 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33817 PR translation/80001
33818 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33819 more amenable to translation.
33820 (oacc_loop_auto_partitions): Likewise.
33822 2017-03-21 Marek Polacek <polacek@redhat.com>
33823 Martin Sebor <msebor@redhat.com>
33825 PR tree-optimization/80109
33826 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33827 on INTEGRAL_TYPE_P.
33829 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33830 Segher Boessenkool <segher@kernel.crashing.org>
33833 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33834 check reg_used_between_p between insn and one of succ or succ2
33835 depending on if succ is artificial insn not inserted into insn
33838 2017-03-21 Martin Liska <mliska@suse.cz>
33840 PR gcov-profile/80081
33841 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33842 * doc/gcc.texi: Include gcov-dump stuff.
33843 * doc/gcov-dump.texi: New file.
33845 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33847 PR rtl-optimization/79150
33848 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33849 conditional jump, if the jump is the last insn of the loop.
33851 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33852 Richard Biener <rguenth@suse.de>
33854 PR tree-optimization/79908
33855 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33856 been cast away, use force_gimple_operand to construct the side
33859 2017-03-21 Martin Liska <mliska@suse.cz>
33861 PR libfortran/79956
33862 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33865 2017-03-21 Brad Spengler <spender@grsecurity.net>
33868 * plugin.c (htab_hash_plugin): New function.
33869 (add_new_plugin): Use it and adjust.
33870 (parse_plugin_arg_opt): Adjust.
33871 (init_one_plugin): Likewise.
33873 2017-03-21 Richard Biener <rguenther@suse.de>
33875 PR tree-optimization/80032
33876 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33877 if set force the cleanup to happen unconditionally.
33878 (gimplify_target_expr): Push inserted clobbers with force_uncond
33879 to avoid them being removed by control-dependent DCE.
33881 2017-03-21 Richard Biener <rguenther@suse.de>
33883 PR tree-optimization/80122
33884 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33885 va_arg_pack_len when the inlined call stmt requires pack
33887 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33889 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33892 * tsan.c (instrument_builtin_call): If the memory model argument
33893 is not a constant, assume it is valid.
33896 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33899 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33901 PR rtl-optimization/79910
33902 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33903 if its dest is used by an insn before I2 (other than the combined
33904 insns themselves, which are properly handled already).
33906 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33909 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33911 * combine.c (record_used_regs): New static function.
33912 (try_combine): Handle situations where there is an additional
33913 instruction between I2 and I3 which needs to have a LOG_LINK
33917 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33919 * combine.c (try_combine): Delete redundant i1 test. Call
33920 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33922 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33925 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33926 alternatives 13/14.
33928 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33930 PR tree-optimization/80054
33931 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33932 the optimization if a PHI or any of its arguments is not dominated
33933 by the candidate's basis. Use gphi* rather than gimple* as
33935 (replace_profitable_candidates): Clean up a gimple* variable that
33936 should be a gphi* variable.
33938 2017-03-20 Martin Sebor <msebor@redhat.com>
33941 * doc/extend.texi (attribute constructor): Document present limitation.
33943 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33946 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33947 __POWER9_VECTOR__ #ifdef control, change template definition to
33948 use Power9-specific built-in function.
33949 (vec_any_eq): Likewise.
33950 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33951 to control outcomes from this test.
33952 (vector_ae_<mode>p): For VEC_F modes, likewise.
33954 2017-03-20 Ian Lance Taylor <iant@google.com>
33956 * config/i386/i386.c (ix86_function_regparm): Save an extra
33957 register for -fsplit-stack with DECL_STATIC_CHAIN.
33959 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33962 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33963 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33965 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33967 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33969 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33972 2017-03-20 Marek Polacek <polacek@redhat.com>
33975 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33977 2017-03-20 Richard Biener <rguenther@suse.de>
33979 PR tree-optimization/80113
33980 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33981 allocate extra SSA name for PHI def.
33982 (add_close_phis_to_outer_loops): Likewise.
33983 (add_close_phis_to_merge_points): Likewise.
33984 (copy_loop_close_phi_args): Likewise.
33985 (copy_cond_phi_nodes): Likewise.
33987 2017-03-20 Martin Liska <mliska@suse.cz>
33989 PR middle-end/79753
33990 * tree-chkp.c (chkp_build_returned_bound): Do not build
33991 returned bounds for a LHS that's not a BOUNDED_P type.
33993 2017-03-20 Martin Liska <mliska@suse.cz>
33997 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33998 COMPLEX_CST and VECTOR_CST.
34000 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34003 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34004 target operand. A new splitter adds the clobber statement in case
34005 the target operand is dead anyway.
34007 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34009 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34010 to age-old versions of binutils and glibc.
34012 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34014 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34016 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34018 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34020 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34022 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34023 requirement for binutils 2.13.
34025 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34027 * combine.c (try_combine): Delete redundant i1 test. Call
34028 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34030 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34032 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34033 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34035 <riscv64-*-elf>: Re-arrange section
34036 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34037 <riscv32-*-linux>: Likewise.
34038 <riscv64-*-elf>: Likewise
34039 <riscv64-*-linux>: Likewise.
34041 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34044 * aarch64.opt(verbose-cost-dump): Fix typo.
34046 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34049 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34050 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34052 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34054 * reload.c (find_reloads): When reloading a nonoffsettable address,
34055 use RELOAD_OTHER for it and its address reloads.
34057 PR rtl-optimization/79910
34058 * combine.c (record_used_regs): New static function.
34059 (try_combine): Handle situations where there is an additional
34060 instruction between I2 and I3 which needs to have a LOG_LINK
34063 2017-03-17 Jeff Law <law@redhat.com>
34065 PR tree-optimization/71437
34066 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34067 conditional in the hash table first.
34068 (vrp_dom_walker::before_dom_children): Extract condition from
34069 ASSERT_EXPR. Record condition, its inverion and any implied
34070 conditions as well.
34072 2017-03-17 Marek Polacek <polacek@redhat.com>
34073 Markus Trippelsdorf <markus@trippelsdorf.de>
34075 PR tree-optimization/80079
34076 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34079 2017-03-17 Richard Biener <rguenther@suse.de>
34081 PR middle-end/80075
34082 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34083 Properly verify the LHS before the RHS possibly claims to be
34085 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34088 2017-03-17 Martin Jambor <mjambor@suse.cz>
34090 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34091 (List of -O2 options): Likewise.
34092 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34095 2017-03-17 Tom de Vries <tom@codesourcery.com>
34097 * gcov-dump.c (print_usage): Print bug_report_url.
34099 2017-03-17 Richard Biener <rguenther@suse.de>
34101 PR middle-end/80050
34102 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34103 (parser::peek): Likewise.
34105 2017-03-17 Richard Biener <rguenther@suse.de>
34107 PR tree-optimization/80048
34108 * sese.c (free_sese_info): Properly release rename_map and
34109 copied_bb_map elements.
34111 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34113 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34114 Add linked-list forward and backlinks. Insert on
34115 construction, remove on destruction.
34116 (class pass_store_merging): Add m_stores_head field.
34117 (pass_store_merging::terminate_and_process_all_chains):
34118 Iterate over m_stores_head list.
34119 (pass_store_merging::terminate_all_aliasing_chains):
34121 (pass_store_merging::execute): Check for debug stmts first.
34122 Push new chains onto the m_stores_head stack.
34124 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34127 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34128 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34129 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34131 2017-03-16 Jeff Law <law@redhat.com>
34133 PR tree-optimization/71437
34134 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34135 member function. Implementation moved into after_dom_children
34136 member function and into the threader's thread_outgoing_edges
34138 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34139 some code into new thread_outgoing_edges.
34140 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34141 definition. Simplify marker handling (do it here). Assume we always
34142 have the available expression and the const/copies tables.
34143 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34145 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34146 * tree-vrp.c (equiv_stack): No longer file scoped.
34147 (vrp_dom_walker): New class.
34148 (vrp_dom_walker::before_dom_children): New member function.
34149 (vrp_dom_walker::after_dom_children): Likewise.
34150 (identify_jump_threads): Setup domwalker. Use it rather than
34151 walking edges in a random order by hand. Simplify setup/finalization.
34152 (finalize_jump_threads): Remove.
34153 (vrp_finalize): Do not call identify_jump_threads here.
34154 (execute_vrp): Do it here instead and call thread_through_all_blocks
34157 PR tree-optimization/71437
34158 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34160 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34162 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34163 (dom_opt_dom_walker::thread_across_edge): Remove
34164 handle_dominating_asserts argument. All callers changed.
34165 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34166 changes. Remove calls to lhs_of_dominating_assert. Other
34167 uses of handle_dominating_asserts turn into unconditional code
34168 (simplify_control_stmt_condition_1): Likewise.
34169 (simplify_control_stmt_condition): Likewise.
34170 (thread_through_normal_block, thread_across_edge): Likewise.
34171 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34172 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34173 object if it is not an SSA_NAME.
34174 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34175 before calling into the VRP specific simplifiers.
34176 (identify_jump_threads): Remove handle_dominating_asserts
34179 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34182 * tree-diagnostic.c (default_tree_printer): No longer static.
34183 * tree-diagnostic.h (default_tree_printer): New prototype.
34185 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34187 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34188 Change ins into fmov.
34190 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34192 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34193 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34194 Use h_con constraint for operand 1.
34195 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34196 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34198 2017-03-15 Jeff Law <law@redhat.com>
34200 PR tree-optimization/71437
34201 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34202 (record_temporary_equivalences): Use it.
34204 PR tree-optimization/71437
34205 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34206 tree-ssa-scopedtables.
34207 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34208 (record_conditions, record_cond, vuse_eq): Likewise.
34209 (record_edge_info): Adjust to API tweak of record_conditions.
34210 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34211 (record_temporary_equivalences, optimize_stmt): Likewise.
34212 (eliminate_redundant_computations): Likewise.
34213 (record_equivalences_from_stmt): Likewise.
34214 * tree-ssa-scopedtables.c: Include options.h and params.h.
34215 (vuse_eq): New function, moved from tree-ssa-dom.c
34216 (build_and_record_new_cond): Likewise.
34217 (record_conditions): Likewise. Accept vector of conditions rather
34218 than edge_equivalence structure for first argument.
34219 for the first argument.
34220 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34221 from tree-ssa-dom.c.
34222 (avail_exprs_stack::record_cond): Likewise.
34223 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34224 from tree-ssa-dom.c.
34225 (avail_exprs_stack): Add new member functions lookup_avail_expr
34227 (record_conditions): Declare.
34229 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34232 * lra-constraints.c (process_alt_operands): Increase reject for
34233 reloading an input/output operand.
34235 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34238 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34239 insns to convert from signed/unsigned char/short to IEEE 128-bit
34241 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34243 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34246 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34247 subreg of inner mode for values already in registers.
34249 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34251 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34252 iteration reg is used after the loop.
34254 2017-03-14 Martin Sebor <msebor@redhat.com>
34256 PR tree-optimization/79800
34257 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34258 precision in negative-positive range.
34259 (format_floating): Call non-const overload with adjusted precision.
34261 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34264 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34267 2017-03-14 Martin Sebor <msebor@redhat.com>
34269 PR middle-end/80020
34270 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34271 * builtins.def (aligned_alloc): Use it.
34274 * Makefile.in (GTFILES): Add calls.c.
34275 * calls.c: Include "gt-calls.h".
34277 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34279 PR rtl-optimization/79728
34280 * regs.h (struct target_regs): New field
34281 x_contains_allocatable_regs_of_mode.
34282 (contains_allocatable_regs_of_mode): New macro.
34283 * reginfo.c (init_reg_sets_1): Initialize it, and change
34284 contains_reg_of_mode so it includes global regs as well.
34285 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34286 rather than contains_regs_of_mode.
34288 2017-03-14 Martin Liska <mliska@suse.cz>
34290 * doc/invoke.texi: Document options that can't be combined with
34291 -fcheck-pointer-bounds.
34293 2017-03-14 Martin Liska <mliska@suse.cz>
34295 PR middle-end/79831
34296 * doc/invoke.texi (-Wchkp): Document the option.
34298 2017-03-14 Martin Liska <mliska@suse.cz>
34300 * Makefile.in: Install gcov-dump.
34302 2017-03-14 Martin Liska <mliska@suse.cz>
34304 * multiple_target.c (expand_target_clones): Bail out for
34305 an invalid attribute.
34307 2017-03-14 Richard Biener <rguenther@suse.de>
34309 * alias.c (struct alias_set_entry): Pack properly.
34310 * cfgloop.h (struct loop): Likewise.
34311 * cse.c (struct set): Likewise.
34312 * ipa-utils.c (struct searchc_env): Likewise.
34313 * loop-invariant.c (struct invariant): Likewise.
34314 * lra-remat.c (struct cand): Likewise.
34315 * recog.c (struct change_t): Likewise.
34316 * rtl.h (struct address_info): Likewise.
34317 * symbol-summary.h (function_summary): Likewise.
34318 * tree-loop-distribution.c (struct partition): Likewise.
34319 * tree-object-size.c (struct object_size_info): Likewise.
34320 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34321 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34322 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34323 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34324 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34325 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34326 (struct _stmt_vec_info): Likewise.
34328 2017-03-14 Martin Liska <mliska@suse.cz>
34331 * multiple_target.c (create_dispatcher_calls): Check that
34332 a target can create a function dispatcher.
34334 2017-03-14 Martin Liska <mliska@suse.cz>
34337 * multiple_target.c (expand_target_clones): Drop local.local
34338 flag for default implementation.
34340 2017-03-14 Richard Biener <rguenther@suse.de>
34342 PR tree-optimization/80030
34343 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34345 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34347 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34348 gcc_fallthrough() instead of __attribute__((fallthrough));
34350 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34352 * doc/gcc.texi: Remove "up" link to (DIR).
34353 * doc/gccint.texi: Ditto.
34355 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34357 * doc/install.texi (Specific) <avr>: Remove reference to
34360 2017-03-13 Jeff Law <law@redhat.com>
34362 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34363 attribute rather than comments.
34365 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34366 match_scratch operand is highest.
34368 2017-03-13 Martin Liska <mliska@suse.cz>
34370 PR middle-end/78339
34371 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34372 is a CHKP clone, use original declaration.
34374 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34376 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34377 (arc_conditional_register_usage): Use a different allocation order
34378 when optimizing for size.
34379 * common/config/arc/arc-common.c (arc_option_optimization_table):
34380 Section anchors default on when optimizing for size.
34382 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34384 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34386 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34388 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34389 * config/arc/arc.md (cpu_facility): Add cd variant.
34390 (*movqi_insn): Add code density variant.
34391 (*movhi_insn): Likewise.
34392 (*movqi_insn): Likewise.
34393 (*addsi3_mixed): Likewise.
34394 (subsi3_insn): Likewise.
34396 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34398 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34400 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34402 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34403 expressions with MINUS and UNARY ops.
34405 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34408 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34410 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34411 between vec_select and vector argument.
34412 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34413 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34414 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34415 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34416 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34417 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34419 2017-03-13 Richard Biener <rguenther@suse.de>
34422 * params.def (vect-max-peeling-for-alignment): Fix typo.
34424 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34426 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34427 issue that only occurred with binutils below 2.18.
34429 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34431 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34432 refer to binutils 2.11/2.12 minimum.
34434 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34436 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34437 ftp.kernel.org and simplify binutils requirement.
34439 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34441 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34443 (Optimize Options): Ditto. Also remove redundancy.
34445 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34447 PR translation/79848
34448 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34450 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34451 to G_ to avoid double translation.
34453 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34455 PR translation/79923
34456 * auto-profile.c (get_combined_location): Convert leading
34457 character of diagnostics to lower case and remove trailing period.
34458 (read_profile): Likewise for various diagnostics.
34459 * config/arm/arm.c (arm_option_override): Remove trailing period
34460 from various diagnostics.
34461 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34462 (msp430_expand_delay_cycles): Likewise.
34464 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34467 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34468 full command-line argument, rather than just "str".
34469 (aarch64_validate_march): Likewise.
34470 (aarch64_validate_mtune): Likewise.
34472 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34474 PR rtl-optimization/78911
34475 * lra-assigns.c (must_not_spill_p): New function.
34476 (spill_for): Use it.
34478 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34480 PR tree-optimization/79981
34481 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34482 ATOMIC_COMPARE_EXCHANGE ifn result.
34483 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34484 IFN_ATOMIC_COMPARE_EXCHANGE.
34486 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34489 * opts.c (parse_sanitizer_options): Add missing question mark to
34490 "did you mean" message.
34492 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34494 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34496 (VMULEUH_UNS): Likewise.
34497 (VMULOUB_UNS): Likewise.
34498 (VMULOUH_UNS): Likewise.
34499 * config/rs6000/rs6000.c (builtin_function_type): Remove
34500 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34502 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34505 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34506 x with result of extra_parsing_for_operand_code_0.
34507 (function_reader::extra_parsing_for_operand_code_0): Convert
34508 return type from void to rtx, returning x. When reading
34509 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34510 larger size containing struct block_symbol.
34512 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34514 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34515 -mfloat128-hardware without -m64.
34517 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34520 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34521 entries to the case statement that marks unsigned arguments to
34522 overloaded functions.
34524 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34526 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34527 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34529 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34532 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34533 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34535 2017-03-10 Martin Liska <mliska@suse.cz>
34539 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34540 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34543 2017-03-10 Olivier Hainque <hainque@adacore.com>
34545 * tree-switch-conversion (array_value_type): Start by resetting
34546 candidate type to it's main variant.
34548 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34550 PR rtl-optimization/79909
34551 * combine.c (try_combine): Use simplify_replace_rtx on individual
34552 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34553 of the whole CALL_INSN_FUNCTION_USAGE.
34555 PR tree-optimization/79972
34556 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34557 get_range_info on SSA_NAMEs. Formatting fixes.
34559 2017-03-10 Richard Biener <rguenther@suse.de>
34560 Jakub Jelinek <jakub@redhat.com>
34562 PR tree-optimization/77975
34563 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34564 edge to be constant.
34565 (get_val_for): For constant x return it. Formatting fix.
34566 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34567 would use the same bases as the current one.
34569 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34571 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34572 instead of vec_select for V1TImode.
34573 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34575 (VSX_LE_128): Add V1TI to this mode iterator.
34576 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34577 (*vsx_le_perm_store_<mode>): Likewise.
34578 (pre-reload splitter for VSX stores): Likewise.
34579 (post-reload splitter for VSX stores): Likewise.
34580 (*vsx_xxpermdi2_le_<mode>): Likewise.
34581 (*vsx_lxvd2x2_le_<mode>): Likewise.
34582 (*vsx_stxvd2x2_le_<mode>): Likewise.
34584 2017-03-09 Michael Eager <eager@eagercon.com>
34586 Correct failures with --enable-checking=yes,rtl.
34588 * config/microblaze/microblaze.c (microblaze_expand_shift):
34589 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34590 test for const0_rtx.
34591 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34592 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34594 2017-03-09 Richard Biener <rguenther@suse.de>
34596 PR tree-optimization/79977
34597 * graphite-scop-detection.c (scop_detection::merge_sese):
34598 Handle the case of extra exits to blocks dominating the entry.
34600 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34602 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34605 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34607 PR rtl-optimization/79949
34608 * lra-constraints.c (process_alt_operands): Check memory when
34609 trying to predict a cycle. Print about the overall increase.
34611 2017-03-09 Richard Biener <rguenther@suse.de>
34613 PR middle-end/79971
34614 * gimple-expr.c (useless_type_conversion_p): Preserve
34615 TYPE_SATURATING for fixed-point types.
34617 2017-03-09 Richard Biener <rguenther@suse.de>
34620 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34621 alignment of BLKmode params.
34623 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34626 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34627 (VALL_NO_V2Q): Likewise.
34629 * config/aarch64/aarch64-simd.md
34630 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34632 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34633 VALL_NO_V2Q mode iterator.
34634 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34636 2017-03-09 Martin Liska <mliska@suse.cz>
34638 PR tree-optimization/79631
34639 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34640 tree_int_cst_sign_bit just for INTEGER constants.
34642 2017-03-09 Martin Liska <mliska@suse.cz>
34646 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34649 2017-03-09 Marek Polacek <polacek@redhat.com>
34652 * tree.c (inchash::add_expr): Handle TREE_VEC.
34654 2017-03-09 Martin Liska <mliska@suse.cz>
34657 (chkp_narrow_size_and_offset): New function.
34658 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34659 (void chkp_parse_bit_field_ref): New function.
34660 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34661 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34663 2017-03-09 Martin Liska <mliska@suse.cz>
34666 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34667 (chkp_find_bounds_1): Remove gcc_unreachable.
34669 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34672 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34673 BUILT_IN_SYNC*, determine the access type from the size suffix and
34674 always build a MEM_REF with that type. Handle forgotten
34675 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34678 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34679 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34680 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34681 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34682 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34683 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34684 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34685 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34686 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34687 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34688 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34689 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34690 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34691 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34692 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34693 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34694 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34695 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34696 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34697 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34698 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34699 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34700 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34701 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34702 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34703 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34704 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34705 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34706 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34707 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34708 definitions outside of __OPTIMIZE__ guarded section.
34711 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34712 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
34713 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
34714 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
34717 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34719 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
34720 ("vfenez<mode>"): Add missing constraints.
34722 2017-03-08 Martin Sebor <msebor@redhat.com>
34725 * config/nds32/nds32.c (nds32_option_override):
34726 Fix misspelled diagnostic.
34728 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34731 * gimplify.c (gimplify_omp_for): Replace index var in outer
34732 taskloop statement with an artificial variable and add
34733 OMP_CLAUSE_PRIVATE clause for it.
34735 2017-03-08 Richard Biener <rguenther@suse.de>
34737 PR tree-optimization/79955
34738 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34739 for accesses that are completely outside of the variable.
34741 2017-03-08 Andrew Haley <aph@redhat.com>
34743 PR tree-optimization/79943
34744 * tree-ssa-loop-split.c (compute_new_first_bound): When
34745 calculating the new upper bound, (END-BEG) should be added, not
34748 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34750 * config/avr/avr.md (setmemhi): Make sure match_dup
34751 operand number comes before match_scratch.
34753 2017-03-08 Richard Biener <rguenther@suse.de>
34755 PR tree-optimization/79920
34756 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34757 with ncopies == 1 to ...
34758 (vect_transform_slp_perm_load): ... here. Properly compute
34759 all element loads by iterating VF times over the group. Do
34760 not handle ncopies (computed in a broken way) in
34761 vect_create_mask_and_perm.
34763 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34766 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34767 is a uniform vector, use uniform_vector_p return value instead of
34768 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34770 2017-03-07 Marek Polacek <polacek@redhat.com>
34772 PR middle-end/79809
34773 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34774 (alloca_call_type): Likewise.
34776 2017-03-07 Martin Liska <mliska@suse.cz>
34778 * gcov.c (process_args): Put comment to correct location.
34780 2017-03-07 Martin Liska <mliska@suse.cz>
34782 PR middle-end/68270
34783 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34784 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34785 (chkp_narrow_bounds_for_field): Likewise.
34786 (chkp_parse_array_and_component_ref): Pass one more argument to
34789 2017-03-07 Richard Biener <rguenther@suse.de>
34791 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34794 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34796 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34797 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34799 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34802 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34803 to end of description.
34804 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34806 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34808 PR rtl-optimization/79901
34809 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34811 (*avx512f_<code><mode>3<mask_name>): ... this.
34812 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34813 iterator instead of VI8_AVX2_AVX512BW.
34815 PR rtl-optimization/79901
34816 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34817 min/max expander, expand it using expand_vec_cond_expr.
34820 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34823 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34826 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34827 to void * for PCH reasons.
34828 * dwarf2out.c (output_loc_operands, output_die): Cast
34829 v.val_vec.array to unsigned char *.
34831 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34834 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34837 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34839 PR rtl-optimization/79571
34840 * lra-constraints.c (process_alt_operands): Calculate static
34841 reject and subtract it from overall when only addresses will be
34844 2017-03-06 Julia Koval <julia.koval@intel.com>
34847 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34848 incoming stack boundary to 128 for 64-bit targets.
34850 2017-03-06 Richard Biener <rguenther@suse.de>
34852 PR tree-optimization/79894
34853 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34854 to NULL after folding it.
34856 2017-03-06 Richard Biener <rguenther@suse.de>
34858 PR tree-optimization/79824
34859 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34860 check disabling peeling for gaps.
34862 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34864 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34865 attributes): Document gettimeofday.
34867 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34869 * config/s390/s390.c (s390_option_override_internal): Set
34870 PARAM_MIN_VECT_LOOP_BOUND
34872 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34874 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34875 * config/s390/s390.md: Likewise.
34877 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34880 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34881 (<avx2_avx512>_perm<mode>): Rename to ...
34882 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34884 (<avx512>_perm<mode>_mask): Rename to ...
34885 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34887 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34888 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34889 instead of VI8F_256_512.
34890 (avx512f_perm<mode>): New define_expand.
34891 (avx512f_perm<mode>_mask): Likewise.
34892 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34893 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34895 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34897 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34898 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34900 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34902 2017-03-06 Martin Liska <mliska@suse.cz>
34905 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34906 when having a SSA NAME w/o VAR_DECL assigned to it.
34908 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34910 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34911 msa_dpsub_<su>_d): Fix MODE for vec_select.
34913 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34915 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34917 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34919 2017-03-06 Richard Biener <rguenther@suse.de>
34921 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34922 * plugin.c (register_plugin_info): Likewise.
34923 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34925 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34927 * config/i386/sse.md (sse_storehps, sse_storelps,
34928 avx_<castmode><avxsizesuffix>_<castmode>,
34929 avx512f_<castmode><avxsizesuffix>_<castmode>,
34930 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34931 in condition that at least one operand is not a MEM.
34933 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34935 PR middle-end/79805
34936 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34937 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34939 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34940 gimple_call_nothrow_p flag based on whether original builtin can throw.
34941 If it can, emit following stmts on the fallthrough edge.
34942 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34943 don't create new bb if inserting just debug stmts on the edge, try to
34944 insert them on the fallthru bb or just reset debug stmts.
34946 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34949 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34950 restore recog_data (including the operand rtxes inside it) around
34951 the call to get_insn_template.
34953 2017-03-03 Martin Sebor <msebor@redhat.com>
34955 PR tree-optimization/79699
34956 * context.c (context::~context): Free MPFR caches to avoid
34957 a memory leak on program exit.
34959 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34961 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34962 Use wide_int::ulow () instead of .elt (0).
34964 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34966 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34967 (*pushxf): Limit oF constraint to 32bit targets and add oC
34968 constraint for 64bit targets.
34969 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34970 (*pushdf): Change rmF constraint to rmC.
34972 2017-03-03 Martin Liska <mliska@suse.cz>
34974 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34975 Remove unused variable.
34977 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34980 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34981 is a memory operand, increase num_memory.
34982 (ix86_expand_args_builtin): Likewise.
34984 2017-03-03 Jan Hubicka <jh@suse.cz>
34987 * ipa-devirt.c (maybe_record_node): Properly handle
34988 __cxa_pure_virtual visibility.
34990 2017-03-03 Martin Liska <mliska@suse.cz>
34992 PR tree-optimization/79803
34993 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34995 (pass_loop_prefetch::execute): Disabled optimization if an
34996 assumption about L1 cache size is not met.
34998 2017-03-03 Martin Liska <mliska@suse.cz>
35000 PR rtl-optimization/79574
35001 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35002 (hash_scan_set): Likewise.
35003 (dump_hash_table): Likewise.
35004 (hoist_code): Likewise.
35006 2017-03-03 Richard Biener <rguenther@suse.de>
35008 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35009 in place of uhigh ().
35010 (fixed_convert_from_real): Likewise.
35012 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35015 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35017 2017-03-03 Richard Biener <rguenther@suse.de>
35019 PR middle-end/79818
35020 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35021 TYPE_OVERFLOW_UNDEFINED check.
35023 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35025 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35027 (vector_ae_<mode>_p): Likewise.
35028 (vector_nez_<mode>_p): Likewise.
35029 (vector_ne_v2di_p): Likewise.
35030 (vector_ae_v2di_p): Likewise.
35031 (vector_ne_<mode>_p): Likewise.
35032 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35034 (vsx_tsqrt<mode>2_fe): Likewise.
35036 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35039 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35041 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35043 PR rtl-optimization/79780
35044 * cprop.c (one_cprop_pass): When second and further conditional trap
35045 in a single basic block is turned into an unconditional trap, turn it
35046 into a deleted note to avoid RTL verification failures.
35048 2017-03-02 Richard Biener <rguenther@suse.de>
35050 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35052 2017-03-02 Richard Biener <rguenther@suse.de>
35054 PR tree-optimization/79345
35056 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35057 param and abort the walk, returning -1 if it is hit.
35058 (walk_aliased_vdefs): Take a limit param and pass it on.
35059 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35060 defaulting to 0 and return a signed int.
35061 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35062 (check_defs): New helper.
35063 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35064 about uninitialized memory.
35065 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35066 bogus uninitialized warning.
35067 (fixed_convert_from_real): Likewise.
35069 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35071 PR tree-optimization/66768
35072 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35073 iv_use if base object can't be determined.
35075 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35077 PR tree-optimization/79345
35078 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35079 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35080 (get_pattern_stats): Initialize it.
35081 * genemit.c (gen_expand): Verify match_scratch numbers come after
35082 match_operand/match_dup numbers.
35083 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35084 match_scratch numbers.
35085 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35087 * config/s390/s390.md (trunctdsd2): Likewise.
35089 2017-03-02 Richard Biener <rguenther@suse.de>
35091 * wide-int.h (wide_int_storage::operator=): Implement in terms
35094 2017-03-02 Richard Biener <rguenther@suse.de>
35096 PR tree-optimization/79777
35097 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35098 the to insert expression to sth existing.
35100 2017-03-01 Martin Sebor <msebor@redhat.com>
35102 PR middle-end/79692
35103 * gimple-ssa-sprintf.c
35104 (directive::known_width_and_precision): New function.
35105 (format_integer): Use it.
35106 (get_mpfr_format_length): Consider the full range of precision
35107 when computing %g output with the # flag. Set the likely byte
35108 count to 3 rather than 1 when precision is indeterminate.
35109 (format_floating): Correct the lower bound of precision.
35111 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35113 * doc/invoke.texi: Document default code model for 64-bit Linux.
35115 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35118 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35119 udiv rather than div since input pattern is unsigned.
35121 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35123 * config/i386/i386.c (print_reg): Warn for values of
35124 unsupported size in integer register.
35126 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35129 * config/rs6000/predicates.md (current_file_function_operand): Do
35130 not allow self calls to be local if the function is replaceable.
35132 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35135 * config/rs6000/altivec.h (vec_ctz and others): Change the
35136 preprocessor macro that controls conditional compilation from
35137 _ARCH_PWR9 to __POWER9_VECTOR__.
35138 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35139 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35140 control (instead of _ARCH_PWR9 control) so that template
35141 definition uses power9-specific function.
35142 (vec_any_eq): Likewise.
35143 (vec_all_ne): Change macro definition to use a power9-specific
35144 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35145 _ARCH_PWR9 control).
35146 (vec_any_eq) Likewise.
35147 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35148 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35149 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35150 support for xvcmpnedp instruction.
35151 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35152 macro expansion so that Power9 implementation of vec_all_ne does
35153 not use the AltiVec predicate framework.
35154 (VCMPNEH_P): Likewise.
35155 (VCMPNEW_P): Likewise.
35156 (VCMPNED_P): Likewise.
35157 (VCMPNEFP_P): Likewise.
35158 (VCMPNEDP_P): Likewise.
35159 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35160 implementation of vec_any_eq to not use AltiVec predicate
35162 (VCMPAEH_P): Likewise.
35163 (VCMPAEW_P): Likewise.
35164 (VCMPAED_P): Likewise.
35165 (VCMPAEFP_P): Likewise.
35166 (VCMPAEDP_P): Likewise.
35167 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35168 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35169 not use the AltiVec predicate framework.
35170 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35171 of vec_any_eq to not use AltiVec predicate framework.
35172 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35173 support for predefined __POWER9_VECTOR__ macro to indicate that
35174 Power9 instruction selection is enabled.
35175 (altivec_overloaded_builtins): Remove extraneous
35176 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35177 function argument types RS6000_BTI_bool_V16QI and
35178 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35179 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35180 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35181 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35182 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35183 Power9 for implementations of vec_cmpne. Change the signature for
35184 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35185 (representing vec_all_ne) to remove the previously described first
35186 argument of type RS6000_BTI_INTSI, as this was an artifact of
35187 reliance on the AltiVec predicate framework, which is no longer
35188 used in the implementation of these functions. Add
35189 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35190 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35191 since, unlike the AltiVec predicate framework implementation, we
35192 do not share function descriptors between vec_alle and vec_anyeq.
35193 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35194 set of modes that receive special treatment even when
35195 TARGET_P9_VECTOR is true. The special treatment emits code that
35196 does not depend on Power9 instructions.
35197 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35198 define_expand to not rely on AltiVec predicate framework.
35199 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35201 (vector_ne_v2di_p): Change this define_expand to not rely on
35202 AltiVec predicate framework.
35203 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35205 (vector_ne_<mode>_p): Change this define_expand to not rely on
35206 AltiVec predicate framework.
35207 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35209 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35210 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35211 define_insn pattern.
35212 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35213 define_insn pattern because the xvcmpne<VSs>. instruction is not
35215 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35216 instruction is not supported.
35218 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35220 * config/nvptx/nvptx.c: Include intl.h.
35222 2017-03-01 Martin Jambor <mjambor@suse.cz>
35225 * ipa-prop.h (ipa_bits): Removed field known.
35226 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35227 to pointers. Adjusted their comments to warn about their sharing.
35228 (ipcp_transformation_summary): Change bits to a vector of pointers.
35229 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35230 (ipa_get_ipa_bits_for_value): Declare.
35231 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35232 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35233 (ipa_bits_hash_table): Likewise.
35234 (ipa_vr_ggc_hash_traits): Likewise.
35235 (ipa_vr_hash_table): Likewise.
35236 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35237 being pointers and vr_known being removed.
35238 (ipa_set_jf_unknown): Likewise.
35239 (ipa_get_ipa_bits_for_value): New function.
35240 (ipa_set_jfunc_bits): Likewise.
35241 (ipa_get_value_range): New overloaded functions.
35242 (ipa_set_jfunc_vr): Likewise.
35243 (ipa_compute_jump_functions_for_edge): Use the above functions to
35244 construct bits and vr parts of jump functions.
35245 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35246 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35248 (ipcp_grow_transformations_if_necessary): Also allocate
35249 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35251 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35252 them. Fix too long lines.
35253 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35254 vr_known being removed.
35255 (ipa_read_jump_function): Use new setter functions to construct bits
35256 and vr parts of jump functions or set them to NULL.
35257 (write_ipcp_transformation_info): Adjust for bits being pointers.
35258 (read_ipcp_transformation_info): Likewise.
35259 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35261 Include gt-ipa-prop.h.
35262 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35264 (ipcp_store_bits_results): Likewise.
35265 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35266 Do not write to existing jump functions but use a temporary instead.
35268 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35271 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35272 attempt to use its first operand as BIT_FIELD_REF base.
35274 2017-03-01 Richard Biener <rguenther@suse.de>
35276 PR middle-end/79721
35277 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35278 interpolating formula in wrapping arithmetic.
35279 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35281 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35283 PR tree-optimization/79734
35284 * tree-vect-generic.c (expand_vector_condition): Optimize
35285 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35286 Handle VEC_COND_EXPR where comparison has different inner width from
35287 type's inner width.
35289 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35291 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35292 markup, and similar issues. Remove @opindex entries for things
35293 that aren't options. Add missing -mmpy-option entries.
35295 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35297 PR tree-optimization/79737
35298 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35299 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35300 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35301 instead of byte_size. Formatting fix.
35302 (shift_bytes_in_array_right): Formatting fix.
35304 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35307 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35308 condition on optimize for the leaf function test.
35310 2017-02-28 Martin Liska <mliska@suse.cz>
35313 * read-rtl-function.c (function_reader::handle_unknown_directive):
35314 Bail out when one uses -flto.
35316 2017-02-28 Martin Liska <mliska@suse.cz>
35318 * common.opt: Replace space with tabular for options of <number>
35320 * config/i386/i386.opt: Show <number> value for
35321 -mlarge-data-threshold.
35322 * opts.c (print_filtered_help): Do not display number in hexadecimal
35325 2017-02-28 Martin Liska <mliska@suse.cz>
35327 * common.opt: Fix --help=option -Q for options which are of
35330 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35332 * config/i386/i386.c (print_reg): Error out for values
35333 of 8-bit size in invalid integer register.
35335 2017-02-28 Martin Sebor <msebor@redhat.com>
35337 PR tree-optimization/79691
35338 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35340 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35343 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35344 gcc_unreachable with output_operand_lossage.
35346 2017-02-28 Richard Biener <rguenther@suse.de>
35348 PR tree-optimization/79740
35349 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35351 (visit_nary_op): Insert the nary into the hashtable if we
35352 pattern-matched sth.
35353 * tree-ssa-pre.c (eliminate_insert): Robustify.
35355 2017-02-28 Richard Biener <rguenther@suse.de>
35357 PR middle-end/79731
35358 * fold-const.c (decode_field_reference): Reject out-of-bound
35361 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35363 * config/i386/i386.c: Include intl.h.
35364 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35365 instead of just cond ? "..." : "...".
35366 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35367 * coverage.c (read_counts_file): Likewise.
35368 * omp-offload.c: Include intl.h.
35369 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35370 of just cond ? "..." : "...".
35371 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35372 of just cond ? "..." : "...".
35374 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35377 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35379 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35380 'tune for' CPU name.
35381 * config/arm/arm-cpu-data.h: Regenerated.
35383 2017-02-28 Richard Biener <rguenther@suse.de>
35385 PR tree-optimization/79732
35386 * tree-inline.c (expand_call_inline): Do not shadow var.
35388 2017-02-28 Richard Biener <rguenther@suse.de>
35390 PR tree-optimization/79723
35391 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35392 address-space properly.
35394 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35396 * doc/optinfo.texi (Optimization groups): Fix option used for
35398 * doc/invoke.texi (-fopt-info): Document "omp".
35399 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35400 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35401 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35402 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35403 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35405 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35407 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35408 OPTGROUP_OMP with "omp".
35410 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35413 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35414 for arithmetic shift of unsigned V2DI.
35416 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35418 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35419 arc/linux.h headers.
35420 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35421 (LINK_SPEC): Likewise.
35422 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35423 (EXTRA_SPECS): Likewise.
35424 (STARTFILE_SPEC): Likewise.
35425 (ENDFILE_SPEC): Likewise.
35426 (LIB_SPEC): Likewise.
35427 (TARGET_SDATA_DEFAULT): Likewise.
35428 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35429 (MULTILIB_DEFAULTS): Likewise.
35430 (DWARF2_UNWIND_INFO): Likewise.
35431 * config/arc/big.h: New file.
35432 * config/arc/elf.h: Likewise.
35433 * config/arc/linux.h: Likewise.
35434 * config/arc/t-uClibc: Remove.
35436 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35438 PR tree-optimization/77536
35439 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35440 (tree_transform_and_unroll_loop): Use above function to compute the
35441 estimated niter of unrolled loop and use it when scaling profile.
35442 Also use count info rather than frequency if it's non-zero.
35443 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35444 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35445 (vect_transform_loop): Call above function.
35447 2017-02-27 Richard Biener <rguenther@suse.de>
35449 PR tree-optimization/45397
35450 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35451 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35452 (visit_nary_op): Add pattern matching for CSEing sign-changed
35453 or truncated operations with wider ones.
35455 2017-02-27 Richard Biener <rguenther@suse.de>
35457 PR tree-optimization/79690
35458 * tree-vect-stmts.c (vectorizable_store): Use vector type
35459 built from the DR with address-space.
35461 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35463 * doc/invoke.texi (Optimize Options): Refine the description
35464 of asan-use-after-return.
35466 2017-02-25 Alan Modra <amodra@gmail.com>
35468 PR rtl-optimization/79584
35469 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35470 base, not ad->base_term, the reg within base. Remove assertion
35471 that ad->base == ad->base_term. Replace gen_int_mode using
35472 bogus mode with const0_rtx.
35474 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35476 PR middle-end/79396
35477 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35478 FMA_EXPR like tcc_binary or tcc_unary.
35480 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35483 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35485 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35486 (output_loc_operands): Handle DW_OP_call_ref and
35487 DW_OP_GNU_variable_value.
35488 (struct variable_value_struct): New type.
35489 (struct variable_value_hasher): Likewise.
35490 (variable_value_hash): New variable.
35491 (string_types): Remove.
35492 (copy_loc_descr): New function.
35493 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35494 (prepend_loc_descr_to_each): New function.
35495 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35496 instead of add_loc_descr_to_each if the first argument is single
35497 location list and the second has multiple.
35498 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35499 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35500 when looking for variable value which doesn't have other location info.
35501 (loc_list_from_tree): Formatting fix.
35502 (gen_array_type_die): Simplify DW_AT_string_length handling.
35503 (adjust_string_types): Remove.
35504 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35505 string_types. Call resolve_variable_values.
35506 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35507 (resolve_addr_in_expr): Likewise. Add A argument.
35508 (copy_deref_exprloc): Remove deref argument. Adjust for the
35509 original expression being DW_OP_GNU_variable_value with optionally
35510 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35511 optionally after it.
35512 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35513 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35514 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35515 (variable_value_hasher::hash, variable_value_hasher::equal): New
35517 (resolve_variable_value_in_expr, resolve_variable_value,
35518 resolve_variable_values, note_variable_value_in_expr,
35519 note_variable_value): New functions.
35520 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35523 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35526 * opts.h (handle_generated_option): Add GENERATED_P argument.
35527 * opts-common.c (handle_option): Adjust function comment.
35528 (handle_generated_option): Add GENERATED_P argument, pass it to
35530 (control_warning_option): Pass false to handle_generated_option
35532 * opts.c (maybe_default_option): Pass true to handle_generated_option
35534 * optc-gen.awk: Likewise.
35536 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35538 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35539 a REG, look at the REG it is a SUBREG of.
35540 (splitter for cmpeqsi_t): Ditto.
35542 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35544 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35545 the special USEs with the pattern of the insn, not the insn itself.
35547 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35550 * doc/invoke.texi: Document -mload-store-pairs.
35552 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35553 Sandra Loosemore <sandra@codesourcery.com>
35555 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35556 argument isn't a CONST_INT.
35557 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35558 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35559 (nios2_validate_compare): Bypass alternate compare logic if *op2
35560 is not a CONST_INT.
35561 (ldstwm_operation_p): Return false if first_base is not a REG or
35562 if first_offset is not a CONST_INT.
35564 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35566 * config/cris/cris.md: Use correct operand in a define_peephole2.
35568 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35570 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35572 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35574 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35575 this_insn if it is an INSN or JUMP_INSN.
35576 (force_offsettable): Look at base, not at addr.
35577 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35578 on things that aren't necessarily CONST_INTs.
35580 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35582 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35583 -mfpmath=sse is the default also for x86-32 targets with SSE2
35584 instruction set when @option{-ffast-math} is enabled
35586 2017-02-24 Jeff Law <law@redhat.com>
35588 PR rtl-optimizatoin/79286
35589 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35592 2017-02-24 Richard Biener <rguenther@suse.de>
35594 PR tree-optimization/79389
35595 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35598 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35600 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35601 function comment to reflect reality.
35602 (loop_exits_before_overflow): Fix typo in function description.
35604 2017-02-24 Richard Biener <rguenther@suse.de>
35606 PR tree-optimization/79389
35607 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35608 properly that a threading opportunity exists. Detect conditional
35609 copy/constant propagation opportunities.
35611 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35613 * config/visium/visium.md (type): Add trap.
35614 (b): New mode attribute.
35615 (*btst): Rename into...
35616 (*btst<mode>): ...this and adjust.
35617 (*cbranchsi4_btst_insn): Rename into...
35618 (*cbranch<mode>4_btst_insn): ...this and adjust.
35619 (trap): New define_insn.
35621 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35623 PR tree-optimization/79389
35624 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35625 (noce_reversed_cond_code): New function.
35626 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35627 reversed_comparison_code. Formatting fix.
35628 (noce_try_store_flag): Test rev_cond != NULL in addition to
35629 reversed_comparison_code.
35630 (noce_try_store_flag_constants): Likewise.
35631 (noce_try_store_flag_mask): Likewise.
35632 (noce_try_addcc): Use rev_cond if non-NULL instead of
35633 reversed_comparison_code.
35634 (noce_try_cmove_arith): Likewise. Formatting fixes.
35635 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35636 (noce_find_if_block): Initialize rev_cond.
35637 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35638 instead of false as last argument never attempt to reverse it
35641 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35643 PR tree-optimization/79663
35644 * tree-predcom.c (combine_chains): Process refs in reverse order
35645 only for ZERO length chains, and add explaining comment.
35647 2017-02-23 Jeff Law <law@redhat.com>
35649 PR tree-optimization/79578
35650 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35651 in call to operand_equal_p.
35653 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35656 * config/i386/cpuid.h: Fix another undefined behavior.
35658 2017-02-23 Richard Biener <rguenther@suse.de>
35660 PR tree-optimization/79683
35661 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35662 vector types for data-refs.
35664 2017-02-23 Martin Liska <mliska@suse.cz>
35666 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35668 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35670 PR middle-end/79665
35671 * internal-fn.c (get_range_pos_neg): Moved to ...
35672 * tree.c (get_range_pos_neg): ... here. No longer static.
35673 * tree.h (get_range_pos_neg): New prototype.
35674 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35675 are known to be in between 0 and signed maximum inclusive, try to
35676 expand both unsigned and signed divmod and use the cheaper one from
35679 2017-02-22 Jeff Law <law@redhat.com>
35681 PR tree-optimization/79578
35682 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35683 to compare base operands.
35685 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35688 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35689 gpc_reg_operand instead of fpr_reg_operand.
35691 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35693 * config/mips/mips.c (mips_return_in_memory): Force FP
35694 vector types to be returned in memory for o32 ABI.
35696 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35698 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35699 instead of DW_TAG_member for static data member declarations and don't
35700 set no_linkage_name for static inline data members.
35701 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35704 2017-02-22 Martin Liska <mliska@suse.cz>
35706 * doc/invoke.texi: Replace inequality signs with square brackets
35709 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35712 * lra-constraints.c (simplify_operand_subreg): Handle
35713 WORD_REGISTER_OPERATIONS targets.
35715 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35718 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
35719 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
35720 elimination by swapping fld*.
35722 2017-02-22 Richard Biener <rguenther@suse.de>
35724 PR tree-optimization/79673
35725 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
35726 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
35727 irrelevant address-space qualifiers and avoiding a
35728 ADDR_SPACE_CONVERT_EXPR from fold_convert.
35730 2017-02-22 Richard Biener <rguenther@suse.de>
35732 PR tree-optimization/79666
35733 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
35734 to not symbolically negate if that may introduce undefined
35737 2017-02-22 Martin Liska <mliska@suse.cz>
35740 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35741 * data-streamer-out.c (streamer_write_gcov_count_stream):
35743 * value-prof.c (stream_out_histogram_value): Make assert more
35744 precise based on type of counter.
35746 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35749 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35750 Use nonimmediate_operand instead of memory_operand for operand 1.
35751 (float-extend standard_x87sse_constant_load splitter): Ditto.
35753 2017-02-21 Jeff Law <law@redhat.com>
35755 PR tree-optimization/79621
35756 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35757 blocks with edges to themselves.
35759 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35762 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35763 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35764 Use gimple_call_builtin_p.
35767 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35768 on temporarily removed DEBUG_INSNs.
35770 PR tree-optimization/79649
35771 * tree-loop-distribution.c (classify_partition): Give up on
35772 non-generic address space loads/stores.
35774 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35776 * doc/loop.texi (Loop manipulation): Remove nonexistent
35777 tree_ssa_loop_version from the documentation.
35778 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35780 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35783 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35784 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35785 * config/rs6000/rs6000.c: Include except.h.
35786 (rs6000_expand_split_stack_prologue): Call
35787 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35789 2017-02-21 Martin Jambor <mjambor@suse.cz>
35792 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35793 have been analyzed.
35795 2017-02-21 Martin Jambor <mjambor@suse.cz>
35797 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35798 for backward compatibility only.
35799 * doc/invoke.texi (Option Summary): Remove all references to
35800 -fipa-cp-alignment.
35802 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35806 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35808 * lra-constraints.c (curr_insn_transform): Handle
35809 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35811 2017-02-21 Martin Liska <mliska@suse.cz>
35813 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35815 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35817 PR translation/79638
35818 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35820 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35823 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35824 (arm_function_ok_for_sibcall): Return false for an indirect call by
35825 descriptor if all the argument registers are used.
35826 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35827 alignment of the function.
35829 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35831 PR tree-optimization/61441
35832 * simplify-rtx.c (simplify_const_unary_operation): For
35833 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35834 the sNaN unmodified.
35836 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35838 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35839 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35840 instead of SYSTEM_HEADER_DIR.
35842 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35843 Martin Liška <mliska@suse.cz>
35845 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35846 Fix typos and grammar, use active voice, and clarify.
35848 2017-02-20 Marek Polacek <polacek@redhat.com>
35850 PR middle-end/79537
35851 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35854 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35856 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35859 * config/i386/i386.c (ix86_expand_builtin): Handle
35860 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35861 ix86_builtins_isa[fcode].isa as a requirement of those
35862 flags and any other flag in the bitmask.
35863 (ix86_init_mmx_sse_builtins): Use 0 instead of
35864 ~OPTION_MASK_ISA_64BIT as mask.
35865 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35866 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35867 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35868 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35870 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35873 * lra-constraints.c (split_reg): Check requested split mode
35874 is supported by the register.
35876 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35878 * lra-constraints.c (simplify_operand_subreg): Remove early
35881 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35884 * lra-constraints.c (curr_insn_transform): Tighten condition
35885 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35887 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35890 * lra-constraints.c (curr_insn_transform): Handle
35891 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35893 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35896 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35898 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35900 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35903 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35906 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35908 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35911 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35913 * final.c (last_columnnum, override_columnnum): New variables.
35914 (final_start_function): Set last_columnnum, pass it to begin_prologue
35915 hook and pass 0 to dwarf2out_begin_prologue.
35916 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35917 to source_line debug hook.
35918 (notice_source_line): Compute last_columnnum and for debug_column_info
35919 return true on column changes.
35920 * debug.h (struct gcc_debug_hooks): Add column argument to
35921 source_line and begin_prologue hooks.
35922 (debug_nothing_int_charstar_int_bool): Remove prototype.
35923 (debug_nothing_int_int_charstar,
35924 debug_nothing_int_int_charstar_int_bool): New prototypes.
35925 (dwarf2out_begin_prologue): Add column argument.
35926 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35927 begin_prologue hooks.
35928 (debug_nothing_int_charstar_int_bool): Remove.
35929 (debug_nothing_int_int_charstar,
35930 debug_nothing_int_int_charstar_int_bool): New functions.
35931 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35932 through to dwarf2out_source_line.
35933 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35934 (dwarf2out_source_line): Add column argument, emit it if requested.
35935 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35937 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35938 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35939 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35940 through to dwarf2out_begin_prologue.
35941 (vmsdbgout_source_line): Add column argument, pass it through to
35942 dwarf2out_source_line.
35943 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35944 dbxout_source_line caller.
35945 (dbxout_source_line): Add column argument.
35947 * common.opt (gno-column-info, gcolumn-info): New options.
35948 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35949 (check_die): Also test for multiple DW_AT_decl_column attributes.
35950 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35951 DW_AT_decl_column if requested.
35952 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35954 (gen_variable_die): Likewise.
35955 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35956 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35959 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35960 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35961 (ix86_handle_option): Handle OPT_m3dnowa.
35962 * doc/invoke.texi (-m3dnowa): Document.
35963 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35964 -m3dnowa instead of -m3dnow -march=athlon.
35967 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35968 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35970 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35973 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35974 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35975 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35976 generator for vsx_xxpermdi_<mode>_be.
35977 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35978 force big-endian semantics.
35979 (vsx_xxpermdi_<mode>_be): New define_expand with same
35980 implementation as previous version of vsx_xxpermdi_<mode>.
35982 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35984 PR tree-optimization/79327
35985 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35986 variable, its initialization and use.
35988 2017-02-17 Julia Koval <julia.koval@intel.com>
35990 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35991 (OPTION_MASK_ISA_PKU_UNSET): New.
35992 (ix86_handle_option): Handle -mrdpid.
35993 * config/i386/cpuid.h (bit_RDPID): New.
35994 * config/i386/driver-i386.c (host_detect_local_cpu):
35995 Detect RDPID feature.
35996 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35997 * config/i386/i386-c.c (ix86_target_macros_internal):
35999 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36000 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36001 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36002 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36003 * config/i386/i386.md (define_insn "rdpid"): New.
36004 * config/i386/i386.opt Add -mrdpid.
36005 * config/i386/immintrin.h (_rdpid_u32): New.
36007 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36009 PR rtl-optimization/79541
36010 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36011 instead of transforming it into USE.
36013 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36015 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36016 If HONOR_SNANS (SFmode) force the input to a register.
36017 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36018 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36019 an frsp or similar insn.
36021 2017-02-17 Martin Liska <mliska@suse.cz>
36023 PR rtl-optimization/79577
36024 * params.def (selsched-max-sched-times): Increase minimum to 1.
36026 2017-02-17 Martin Liska <mliska@suse.cz>
36028 PR rtl-optimization/79574
36029 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36031 2017-02-17 Martin Liska <mliska@suse.cz>
36033 PR tree-optimization/79529
36034 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36035 ssa_defined_default_def_p to handle cases which are implicitly
36037 * tree-ssa.c (ssa_defined_default_def_p): New function.
36038 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36039 which are implicitly defined.
36040 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36042 2017-02-17 Richard Biener <rguenther@suse.de>
36044 PR middle-end/79576
36045 * params.def (max-ssa-name-query-depth): Limit to 10.
36047 2017-02-17 Richard Biener <rguenther@suse.de>
36049 PR tree-optimization/79552
36050 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36053 2017-02-17 Richard Biener <rguenther@suse.de>
36056 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36058 2017-02-17 Marek Polacek <polacek@redhat.com>
36060 PR middle-end/79536
36061 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36062 (fold_negate_expr): New wrapper.
36064 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36066 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36067 Correct terminology and de-emphasize pre-standard behavior.
36069 2017-02-16 Alan Modra <amodra@gmail.com>
36071 PR rtl-optimization/79286
36072 * ira.c (def_dominates_uses): New function.
36073 (update_equiv_regs): Don't create an equivalence for insns that
36074 may trap where the register def does not dominate the use.
36076 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36078 PR rtl-optimization/78127
36079 * lra.c (lra): Call lra_eliminate before finish the loop after
36082 2017-02-16 Richard Biener <rguenther@suse.de>
36084 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36086 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36087 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36088 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36089 (isl_val_int_from_wi): New function.
36090 (extract_affine_gmp): Rename to ...
36091 (extract_affine_wi): ... this, take a widest_int.
36092 (extract_affine_int): Just wrap extract_affine_wi.
36093 (add_param_constraints): Use isl_val_int_from_wi.
36094 (add_loop_constraints): Likewise, and extract_affine_wi.
36096 2017-02-15 Jeff Law <law@redhat.com>
36098 PR middle-end/79521
36099 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36100 ira_init_register_move_cost_if_necessary.
36102 2017-02-15 Martin Sebor <msebor@redhat.com>
36104 PR middle-end/32003
36105 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36106 removed in a prior commit.
36108 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36110 PR tree-optimization/79347
36111 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36112 counters during peeling.
36114 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36116 * Makefile.in (site.exp): Remove "set ISLVER".
36118 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36121 * real.c (real_from_integer): Call real_convert even for decimal.
36123 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36126 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36128 2017-02-14 Andrew Pinski <apinski@cavium.com>
36130 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36131 cores and change the partno/implementer to be correct.
36132 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36133 the 'B" as the implementer.
36134 * config/aarch64/aarch64-tune.md: Regenerate.
36136 2017-02-14 Carl Love <cel@us.ibm.com>
36138 * config/rs6000/rs6000.c: Add case statement entry to make the
36139 xvcvuxdsp built-in argument unsigned.
36140 * config/rs6000/vsx.md: Fix the source and return operand types so they
36141 match the instruction definitions from the ISA document. Fix typo
36142 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36145 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36148 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36149 member early_clobber_alts.
36150 * lra-lives.c (reg_early_clobber_p): New.
36151 (process_bb_lives): Use it.
36152 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36153 (debug_operand_data): Initialize early_clobber_alts.
36154 (setup_operand_alternative): Set up early_clobber_alts.
36155 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36156 alternatives to new_insn_reg.
36157 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36159 (lra_update_insn_regno_info): Pass the new arg.
36161 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36163 PR middle-end/79505
36164 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36165 (new_oacc_loop_raw): Don't clear already cleared fields.
36168 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36169 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36170 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36172 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36175 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36177 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36180 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36181 the extra instruction to the right place to store 128-bit constant
36184 2017-02-14 Martin Sebor <msebor@redhat.com>
36186 PR middle-end/79448
36187 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36188 warning for strings of unknown length.
36190 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36192 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36194 2017-02-14 Jeff Law <law@redhat.com>
36197 * ira-costs.c (scan_one_insn): Initialize register move costs
36198 for pseudos seen in USE/CLOBBER insns.
36200 PR tree-optimization/79095
36201 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36202 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36203 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36204 if the operands are known to be not equal, then the resulting range
36206 (intersect_ranges): If the new range is ~[0,0] and the old range is
36207 wide, then prefer ~[0,0].
36208 * tree-vrp.c (overflow_comparison_p_1): New function.
36209 (overflow_comparison_p): New function.
36210 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36211 if NAME is used in an overflow test.
36212 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36213 overflow check that can be expressed as an equality test, then adjust
36214 ops to be that equality test.
36216 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36218 * config/s390/s390-builtin-types.def: Remove flags argument.
36219 * config/s390/s390.c (s390_init_builtins): Likewise.
36221 2017-02-14 Martin Liska <mliska@suse.cz>
36223 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36224 vector. Fix trailing white spaces.
36226 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36228 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36231 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36233 PR rtl-optimization/68664
36234 * config/arm/arm.c (arm_sched_can_speculate_insn):
36235 New function. Declare prototype.
36236 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36238 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36240 PR rtl-optimization/68664
36241 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36243 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36245 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36247 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36248 max skip bytes for function, loop and jump.
36250 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36252 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36253 ABS_EXPR for gimple dump.
36255 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36258 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36260 PR tree-optimization/79408
36261 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36262 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36263 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36264 also if rhs1 is INTEGER_CST.
36266 2017-02-14 Richard Biener <rguenther@suse.de>
36268 PR middle-end/79432
36269 * tree-into-ssa.c (insert_phi_nodes): When the function can
36270 have abnormal edges rewrite SSA names with broken use-def
36271 dominance out of SSA and register them for PHI insertion.
36273 2017-02-13 Martin Sebor <msebor@redhat.com>
36275 PR middle-end/79496
36276 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36277 clearing info.nowrite flag when snprintf size argument is a range.
36279 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36281 * cprop.c (cprop_jump): Add missing space in string literal.
36282 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36283 (get_constraint_for_component_ref): Likewise.
36284 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36285 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36286 * lra-constraints.c (process_alt_operands): Likewise.
36287 * ipa-inline.c (inline_small_functions): Likewise.
36288 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36289 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36290 * trans-mem.c (diagnose_tm_1_op): Likewise.
36291 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36292 (grid_parallel_clauses_gridifiable): Likewise.
36294 * config/nvptx/mkoffload.c (process): Add space in between
36297 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36298 "MOD4_SSE_REGS" and "ALL_REGS".
36300 * spellcheck.c (test_data): Add , in between "foo" and "food".
36302 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36305 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36306 boundary crossing check and subsequent code generation agree.
36308 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36310 * config/aarch64/aarch64.c (has_memory_op): Delete.
36311 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36314 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36316 PR rtl-optimization/79388
36317 PR rtl-optimization/79450
36318 * combine.c (distribute_notes): When removing TEM_INSN for which
36319 corresponding dest has last value recorded, invalidate that last
36322 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36324 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36325 of explicit '@'. Add missing assembly comment marker on branch costs
36328 2017-02-13 Nathan Sidwell <nathan@acm.org>
36330 * gengtype-lex.l (<in_struct>): Add '/'.
36332 2017-02-13 Martin Liska <mliska@suse.cz>
36335 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36337 2017-02-13 Richard Biener <rguenther@suse.de>
36339 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36341 * configure: Re-generate.
36342 * config.in: Likewise.
36343 * graphite-dependences.c: Simplify as if
36344 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36345 * graphite-isl-ast-to-gimple.c: Likewise.
36346 * graphite-optimize-isl.c: Likewise.
36347 * graphite-poly.c: Likewise.
36348 * graphite-sese-to-poly.c: Likewise.
36349 * graphite.h: Likewise.
36350 * toplev.c: Include isl/version.h and use isl_version () for
36351 printing the ISL version.
36352 * doc/install.texi: Update ISL requirement.
36354 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36356 * doc/standards.texi (Standards): Update reference to
36359 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36361 * doc/extend.texi (Named Address Spaces): sourceware.org now
36363 * doc/install.texi (Binaries): Ditto.
36366 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36368 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36369 standard terminology "stringize"/"stringizing" throughout.
36370 * doc/cppinternals.texi: Likewise.
36372 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36374 * doc/extend.texi: Fix some spelling mistakes and typos.
36375 * doc/invoke.texi: Likewise.
36377 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36380 * params.def (inline-min-speedup) Change from 10 to 8.
36382 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36384 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36387 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36390 * ipa-inline-analysis.c (get_minimal_bb): New function.
36391 (record_modified): Use it.
36392 (remap_edge_change_prob): Handle also ancestor functions.
36394 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36396 * doc/contrib.texi (Contributors): Remove broken link into
36397 the Mauve CVS repository.
36399 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36401 PR middle-end/79454
36402 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36403 result computation whenever lhs doesn't have vector mode, not
36404 just when it has BLKmode.
36406 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36408 * doc/makefile.texi (profiledbootstrap): Refer to the
36409 installation instructions only in textual form.
36411 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36414 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36416 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36418 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36419 (Specific): Update mingw-w64 reference.
36421 (Specific): Remove broken link to Renesas RX processor.
36423 2017-02-10 Richard Biener <rguenther@suse.de>
36425 * toplev.c (process_options): Do not mention obsolete graphite
36426 options when printing sorry message about missing graphite support.
36427 Mention -floop-nest-optimize.
36429 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36431 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36432 (vtst_p16): Likewise.
36433 (vtstq_p8): Likewise.
36434 (vtstq_p16): Likewise.
36436 (vtstq_p64): Likewise.
36437 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36438 (vset_lane_p64): New.
36439 (vsetq_lane_p64): New.
36441 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36443 PR tree-optimization/79411
36444 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36445 stmt operands are SSA_NAMEs used in abnormal phis.
36446 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36449 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36452 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36455 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36457 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36459 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36461 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36465 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36466 not arbitrary TREE_CONSTANT.
36469 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36470 "omp declare target link" attribute unless is_global_var.
36471 * omp-offload.c (find_link_var_op): Likewise.
36473 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36474 Chung-Lin Tang <cltang@codesourcery.com>
36476 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36478 (gimplify_adjust_omp_clauses): Don't delete TILE.
36479 (gimplify_omp_for): Deal with TILE.
36480 * internal-fn.c (expand_GOACC_TILE): New function.
36481 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36483 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36484 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36486 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36487 avoid DIV for outermost collapse var.
36488 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36489 Remove out of date comments, fix whitespace.
36490 * omp-general.c (omp_extract_for_data): Deal with tiling.
36491 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36492 adjust OLF_DIM_BASE value.
36493 (struct omp_for_data): Add tiling field.
36494 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36495 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36496 for auto loops. Remove default auto determining, moved to
36497 oacc_loop_fixed_partitions.
36498 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36499 stmts, add e_mask field.
36500 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36501 (oacc_thread_numbers): Use oacc_dim_call.
36502 (oacc_xform_tile): New.
36503 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36504 (finish_oacc_loop): Adjust for ifns vector.
36505 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36506 add case for GOACC_TILE fns.
36507 (oacc_loop_xform_loop): Delete.
36508 (oacc_loop_process): Iterate over call list directly, and add
36509 handling for GOACC_TILE fns.
36510 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36512 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36513 vector partitioning to outer loops. Assign 2 partitions to loops
36514 when available. Add TILE handling.
36515 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36516 (execite_oacc_device_lower): Process GOACC_TILE fns,
36517 ignore unknown specs.
36518 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36519 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36520 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36522 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36524 * configure.ac (ACX_BUGURL): Update.
36525 * configure: Regenerate.
36527 2017-02-09 Richard Biener <rguenther@suse.de>
36529 PR tree-optimization/69823
36530 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36531 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36533 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36535 * config/arc/arc-c.def: Add __NPS400__ definition.
36536 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36537 (TARGET_NPS400): Define.
36539 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36541 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36543 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36544 pointer, arch_info.
36545 (arc_cpu_types): Fill the arch_info field with a pointer into the
36546 arc_arch_types table.
36547 (arc_selected_cpu): Declare.
36548 * config/arc/arc.c (arc_selected_cpu): Make global.
36549 (arc_selected_arch): Delete.
36550 (arc_base_cpu): Delete.
36551 (arc_override_options): Remove references to deleted variables,
36552 update access to arch information.
36553 (ARC_OPT): Update access to arch information.
36554 (ARC_OPTX): Likewise.
36555 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36556 (TARGET_ARC600): Update access to arch information.
36557 (TARGET_ARC601): Likewise.
36558 (TARGET_ARC700): Likewise.
36559 (TARGET_EM): Likewise.
36560 (TARGET_HS): Likewise.
36561 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36564 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36567 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36568 condition/operands for integer GE/LE/GEU/LEU operations.
36570 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36572 PR translation/79397
36573 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36576 2017-02-08 Martin Jambor <mjambor@suse.cz>
36579 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36580 whether allocation happened.
36581 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36582 nothing was allocated.
36584 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36586 PR tree-optimization/79408
36587 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36588 constant, but SSA_NAME with a known integer range, use the minimum
36589 of that range instead of op1 to determine if modulo can be replaced
36590 with its first operand.
36592 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36594 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36596 2017-02-08 Richard Biener <rguenther@suse.de>
36598 PR tree-optimization/71824
36599 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36600 Check all loops contained in the merged region.
36602 2017-02-07 Andrew Pinski <apinski@cavium.com>
36604 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36606 2017-02-07 Andrew Pinski <apinski@cavium.com>
36608 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36609 (thunderxt88): Likewise.
36610 (thunderxt81): Disable LSE and change v8.1 to v8.
36611 (thunderxt83): Likewise.
36613 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36614 Richard Biener <rguenther@suse.de>
36616 PR middle-end/79399
36617 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36618 type from int to size_t.
36619 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36621 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36623 PR rtl-optimization/79386
36624 * cprop.c (bypass_conditional_jumps): Initialize
36625 bypass_last_basic_block already before splitting bbs after
36626 unconditional traps...
36627 (bypass_conditional_jumps): ... rather than here.
36630 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36631 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36632 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36633 fix -masm=intel patterns.
36635 2017-02-07 Richard Biener <rguenther@suse.de>
36637 PR tree-optimization/79256
36638 PR middle-end/79278
36639 * builtins.c (get_object_alignment_2): Use min_align_of_type
36640 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36641 and ADJUST_FIELD_ALIGN.
36643 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36645 * doc/tm.texi: Regenerate.
36646 * stor-layout.c (layout_decl): Adjust.
36647 (update_alignment_for_field): Likewise.
36648 (place_field): Likewise.
36649 (min_align_of_type): Likewise.
36650 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36651 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36652 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36653 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36654 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36655 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36656 * config/i386/i386.c (x86_field_alignment): Likewise.
36657 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36658 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36659 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36660 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36661 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36662 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36666 2017-01-30 Richard Biener <rguenther@suse.de>
36668 PR tree-optimization/79256
36669 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36670 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36673 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36675 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36676 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36677 builtins to SImode and emit a zero-extend, if necessary.
36679 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36681 * docs/invoke.texi (RISC-V Options): Alphabetize.
36683 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36685 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36688 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36690 * config/riscv/riscv.c: New file.
36691 * common/config/riscv/riscv-common.c: Likewise.
36692 * config.gcc: Likewise.
36693 * config/riscv/constraints.md: Likewise.
36694 * config/riscv/elf.h: Likewise.
36695 * config/riscv/generic.md: Likewise.
36696 * config/riscv/linux.h: Likewise.
36697 * config/riscv/multilib-generator: Likewise.
36698 * config/riscv/peephole.md: Likewise.
36699 * config/riscv/pic.md: Likewise.
36700 * config/riscv/predicates.md: Likewise.
36701 * config/riscv/riscv-builtins.c: Likewise.
36702 * config/riscv/riscv-c.c: Likewise.
36703 * config/riscv/riscv-ftypes.def: Likewise.
36704 * config/riscv/riscv-modes.def: Likewise.
36705 * config/riscv/riscv-opts.h: Likewise.
36706 * config/riscv/riscv-protos.h: Likewise.
36707 * config/riscv/riscv.h: Likewise.
36708 * config/riscv/riscv.md: Likewise.
36709 * config/riscv/riscv.opt: Likewise.
36710 * config/riscv/sync.md: Likewise.
36711 * config/riscv/t-elf-multilib: Likewise.
36712 * config/riscv/t-linux: Likewise.
36713 * config/riscv/t-linux-multilib: Likewise.
36714 * config/riscv/t-riscv: Likewise.
36715 * configure.ac: Likewise.
36716 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
36717 Waterman as RISC-V maintainers.
36718 * doc/install.texi: Add RISC-V entries.
36719 * doc/invoke.texi: Add RISC-V options section.
36720 * doc/md.texi: Add RISC-V constraints section.
36721 * configure: Regenerated.
36723 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
36726 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
36727 false values to be constant vectors with all 0 or all 1 bits set.
36728 (vcondu<mode><mode>): Likewise.
36729 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
36731 (fpmask_comparison_operator): Update comment.
36732 (vecint_comparison_operator): New predicate.
36733 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
36734 vector conditionals when the true and false values are constant
36735 vectors with all 0 bits or all 1 bits set.
36737 2017-02-06 Martin Sebor <msebor@redhat.com>
36739 PR tree-optimization/79376
36740 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36742 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36744 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36745 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36746 to simplify split condition.
36748 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36750 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36751 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36754 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36756 PR rtl-optimization/68664
36757 * target.def (can_speculate_insn): New hook.
36758 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36759 * doc/tm.texi: Regenerate.
36760 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36761 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36762 (rs6000_sched_can_speculate_insn): New function.
36764 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36766 PR tree-optimization/79284
36767 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36768 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36769 vectorizable_mask_load_store, vectorizable_operation,
36770 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36771 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36772 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36773 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36774 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36775 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36776 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36777 is_gimple_assign (stmt). Replace another such test with
36778 is_gimple_assign (stmt).
36780 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36783 * config/avr/avr.c (rtl-iter.h): Include it.
36784 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36785 (avr_legitimate_combined_insn): ...and implementation.
36787 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36789 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36790 * config/s390/s390.c (s390_const_operand_ok)
36791 (s390_canonicalize_comparison, s390_extract_part)
36792 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36793 (s390_contiguous_bitmask_p, s390_rtx_costs)
36794 (legitimize_pic_address): Likewise.
36795 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36796 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36797 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36798 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36799 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36801 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36803 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36804 REGNO($0) == REGNO($1).
36806 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36808 * config/s390/linux.h(SIZE_TYPE): Add comment.
36810 2017-02-06 Julian Brown <julian@codesourcery.com>
36811 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36812 Virendra Pathak <virendra.pathak@broadcom.com>
36814 * config/aarch64/aarch64-cores.def: Change the scheduler
36816 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36817 * config/aarch64/thunderx2t99.md: New file.
36819 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36821 * doc/standards.texi (Go Language): Update link to language
36824 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36826 * tree-eh.c (lower_resx): Sanitize profile.
36827 (cleanup_empty_eh_move_lp): Likewise.
36829 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36832 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36834 * cfgloopmanip.h (loop_version): Update prototype.
36835 * modulo-sched.c (sms_schedule): Update call of loop_version.
36836 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36837 * tree-parloops.c (gen_parallel_loop): Likewise.
36838 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36839 * tree-ssa-loop-split.c (split_loop): Likewise.
36840 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36841 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36843 2017-02-05 Martin Liska <mliska@suse.cz>
36846 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36848 (print_operand_address): Initialize a struct to zero.
36850 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36852 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36853 garbage collector only in textual form.
36855 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36857 * doc/extend.texi (x86 specific memory model extensions for
36858 transactional memory): Simplify a phrase.
36860 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36863 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36864 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36865 (atomic_storedi_1): Likewise.
36867 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36869 PR tree-optimization/79338
36870 * tree-parloops.c (gather_scalar_reductions): Don't call
36871 vect_analyze_loop_form for loop->inner before destroying loop's
36874 2017-02-03 Martin Sebor <msebor@redhat.com>
36876 PR tree-optimization/79327
36877 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36878 when precision has resulted in leading zeros.
36879 (format_integer): Adjust the likely counter to assume an unknown
36880 argument that may be zero is non-zero.
36882 2017-02-03 Jason Merrill <jason@redhat.com>
36885 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36886 avoid copying non-taken branch.
36888 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36890 PR tree-optimization/79340
36891 * tree-vect-loop.c (vectorizable_reduction): Release
36892 vec_defs elements after safe_splicing them into other vectors.
36895 PR tree-optimization/79327
36896 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36897 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36899 (format_integer): Use wide_int_to_tree instead of build_int_cst
36900 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36901 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36902 of shortest and longest sequence.
36904 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36906 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36907 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36909 2017-02-03 Walter Lee <walt@tilera.com>
36912 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36913 after initial stackframe link reg save.
36914 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36916 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36919 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36920 wu for stxssp alternative.
36922 2017-02-03 Martin Sebor <msebor@redhat.com>
36924 PR tree-optimization/79352
36925 * gimple-fold.c (get_range_strlen): Add argument.
36926 (get_range_strlen): Change return type to bool.
36927 (get_maxval_strlen): Pass in a dummy argument.
36928 * gimple-fold.h (get_range_strlen): Change return type to bool.
36929 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36930 * tree.h (array_at_struct_end_p): Add argument.
36931 * tree.c (array_at_struct_end_p): Handle it.
36933 2017-02-03 Martin Liska <mliska@suse.cz>
36936 * multiple_target.c (create_dispatcher_calls): Redirect edge
36937 from a caller of a dispatcher.
36938 (expand_target_clones): Make the clones local.
36939 (ipa_target_clone): Do both target clones and resolvers.
36940 (ipa_dispatcher_calls): Remove the pass.
36941 (pass_dispatcher_calls::gate): Likewise.
36942 (make_pass_dispatcher_calls): Likewise.
36943 * passes.def (pass_target_clone): Put as very first IPA early
36946 2017-02-03 Martin Liska <mliska@suse.cz>
36948 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36949 in case of a function with ifunc attribute.
36951 2017-02-03 Martin Liska <mliska@suse.cz>
36953 * cgraph.c (cgraph_node::dump): Dump function version info.
36954 * symtab.c (symtab_node::dump_base): Add missing new line.
36956 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36958 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36959 (ifcombine_ifandif): Use it.
36961 2017-02-03 Martin Liska <mliska@suse.cz>
36963 * doc/invoke.texi: Document default value for
36964 use-after-scope-direct-emission-threshold.
36966 2017-02-03 Martin Liska <mliska@suse.cz>
36968 PR tree-optimization/79339
36969 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36970 (format_floating): Likewise.
36972 2017-02-03 Martin Liska <mliska@suse.cz>
36975 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36977 (ipa_node_params_t::remove): Likewise.
36978 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36979 initialization from removed ipa_node_params_t::insert.
36980 (ipa_node_params::~ipa_node_params): Move from removed
36981 ipa_node_params_t::release.
36982 * symbol-summary.h (symbol_summary::m_released): New member.
36983 Do not release a summary twice. Do not allow to call finalizer
36984 for types of a summary that live in GGC memory.
36986 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36988 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36991 2017-02-02 Martin Sebor <msebor@redhat.com>
36993 PR middle-end/79275
36994 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36995 (format_string): Tighten up the range of output for non-constant
36996 strings and correct the expected range for wide non-constant strings.
36998 2017-02-02 Martin Sebor <msebor@redhat.com>
37000 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37002 PR middle-end/32003
37003 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37005 (-fdump-tree-@var): Add to index and document how to come up
37006 with pass-specific option and dump file names.
37007 (-fdump-passes): Clarify where to look for output.
37009 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37011 PR middle-end/77445
37012 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37013 statistics of the analyzed path; allow threading for speed when
37014 any of BBs along the path are optimized for speed.
37016 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37018 PR middle-end/78468
37019 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37020 settings of the virtual registers.
37023 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37025 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37026 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37029 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37031 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37032 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37034 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37036 * config/s390/s390.md: Add missing comments with the expanded
37038 * config/s390/vector.md: Likewise.
37039 * config/s390/vx-builtins.md: Likewise.
37041 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37044 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37045 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37046 conditions on a single line.
37048 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37050 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37051 __S390_VX__ to __VX__.
37053 2017-02-01 Andrew Pinski <apinski@cavium.com>
37055 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37056 stmt_info to record_stmt_cost.
37057 (vect_get_known_peeling_cost): Pass stmt_info if known to
37059 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37060 cpu_vector_cost field into
37061 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37062 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37063 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37064 splitting of scalar_stmt_cost and vec_stmt_cost.
37065 (thunderx_vector_cost): Likewise.
37066 (cortexa57_vector_cost): LIkewise.
37067 (exynosm1_vector_cost): Likewise.
37068 (xgene1_vector_cost): Likewise.
37069 (thunderx2t99_vector_cost): Improve after the splitting of the two
37071 (aarch64_builtin_vectorization_cost): Update for the splitting of
37072 scalar_stmt_cost and vec_stmt_cost.
37074 2017-02-01 Torvald Riegel <triegel@redhat.com>
37075 Richard Henderson <rth@redhat.com>
37077 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37078 conditional on existance of a fast atomic load.
37079 * optabs-query.c (can_atomic_load_p): New function.
37080 * optabs-query.h (can_atomic_load_p): Declare it.
37081 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37082 no fast atomic load is available for the particular size of access.
37083 (expand_atomic_compare_and_swap): Likewise.
37084 (expand_atomic_load): Likewise.
37085 (expand_atomic_store): Likewise.
37086 (expand_atomic_fetch_op): Likewise.
37087 * testsuite/lib/target-supports.exp
37088 (check_effective_target_sync_int_128): Remove x86 because it provides
37089 no fast atomic load.
37090 (check_effective_target_sync_int_128_runtime): Likewise.
37092 2017-02-01 Richard Biener <rguenther@suse.de>
37094 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37095 (graphite_transform_loops): Provide opt-info for optimized nests.
37096 * tree-parloop.c (parallelize_loops): Provide opt-info for
37097 parallelized loops.
37099 2017-02-01 Richard Biener <rguenther@suse.de>
37101 PR middle-end/79315
37102 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37103 was not set before.
37105 2017-02-01 Richard Biener <rguenther@suse.de>
37107 PR tree-optimization/71824
37108 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37109 Verify the loops are valid in the merged SESE region.
37110 (scop_detection::can_represent_loop_1): Check analyzing the
37111 evolution of the number of iterations in the region succeeds.
37113 2017-01-31 Ian Lance Taylor <iant@golang.org>
37115 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37116 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37118 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37120 PR preprocessor/79210
37121 * input.c (get_substring_ranges_for_loc): Replace line_width
37122 assertion with error-handling.
37124 2017-01-31 Richard Biener <rguenther@suse.de>
37126 PR tree-optimization/77318
37127 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37128 (create_pw_aff_from_tree): Take loop parameter.
37129 (add_condition_to_pbb): Pass loop of the condition to
37130 create_pw_aff_from_tree.
37132 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37134 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37135 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37137 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37141 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37143 (convert_int_to_float128): Likewise.
37144 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37145 (convert_int_to_float128): Likewise.
37146 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37147 (UNSPEC_IEEE128_CONVERT): Likewise.
37148 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37149 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37150 Use local variables for IBM extended format.
37151 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37152 (fix_trunc<mode>si2_fprs): Likewise.
37153 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37154 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37155 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37156 to know that we can now have integers of all sizes in vector
37158 (fix<uns>_<mode>di2_hw): Likewise.
37159 (float<uns>_<mode>si2_hw): Likewise.
37160 (fix_<mode>si2_hw): Likewise.
37161 (fixuns_<mode>si2_hw): Likewise.
37162 (float<uns>_<mode>di2_hw): Likewise.
37163 (float_<mode>di2_hw): Likewise.
37164 (float_<mode>si2_hw): Likewise.
37165 (floatuns_<mode>di2_hw): Likewise.
37166 (floatuns_<mode>si2_hw): Likewise.
37167 (xscvqp<su>wz_<mode>): Delete, no longer used.
37168 (xscvqp<su>dz_<mode>): Likewise.
37169 (xscv<su>dqp_<mode>): Likewise.
37170 (ieee128_mfvsrd_64bit): Likewise.
37171 (ieee128_mfvsrd_32bit): Likewise.
37172 (ieee128_mfvsrwz): Likewise.
37173 (ieee128_mtvsrw): Likewise.
37174 (ieee128_mtvsrd_64bit): Likewise.
37175 (ieee128_mtvsrd_32bit): Likewise.
37177 2017-01-31 Martin Liska <mliska@suse.cz>
37180 * ipa-prop.c (ipa_free_all_node_params): Call release method
37181 instead of ~sumbol_summary to not to trigger double times
37184 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37186 PR tree-optimization/71691
37187 * bitmap.h (class auto_bitmap): New.
37188 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37189 is_maybe_undefined instead of ssa_undefined_value_p.
37191 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37193 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37194 __S390_ARCH_LEVEL__ to __ARCH__.
37196 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37198 PR tree-optimization/79267
37199 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37200 if should_remove_lhs_p is true.
37202 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37205 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37206 (add_alignment_attribute): New.
37207 (base_type_die): Add alignment attribute.
37208 (subrange_type_die): Likewise.
37209 (modified_type_die): Likewise.
37210 (gen_array_type_die): Likewise.
37211 (gen_descr_array_type_die: Likewise.
37212 (gen_enumeration_type_die): Likewise.
37213 (gen_subprogram_die): Likewise.
37214 (gen_variable_die): Likewise.
37215 (gen_field_die): Likewise.
37216 (gen_ptr_to_mbr_type_die): Likewise.
37217 (gen_struct_or_union_type_die): Likewise.
37218 (gen_subroutine_type_die): Likewise.
37219 (gen_typedef_die): Likewise.
37220 (base_type_cmp): Compare alignment attribute.
37222 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37225 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37226 (setb_unsigned) New pattern for setb with CCUNS.
37227 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37228 subfc./subfe sequence to avoid overflow problems. Generate a
37229 shorter sequence with cmpld/setb for power9.
37230 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37231 for generating subfc. instruction.
37232 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37233 now uses this instruction.
37235 2017-01-30 Ian Lance Taylor <iant@google.com>
37238 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37239 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37241 2017-01-30 Martin Sebor <msebor@redhat.com>
37243 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37244 Move constant to the right of a relational operator.
37245 (get_mpfr_format_length, format_character, format_string): Ditto.
37246 (should_warn_p, maybe_warn): Same.
37248 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37250 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37253 * asan.c (get_translation_unit_decl): Remove function.
37254 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37256 2017-01-30 Martin Liska <mliska@suse.cz>
37258 PR gcov-profile/79259
37259 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37260 -fprofile-generate.
37262 2017-01-30 Martin Liska <mliska@suse.cz>
37265 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37266 Initialize variables with NULL value.
37268 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37271 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37273 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37275 2017-01-30 Richard Biener <rguenther@suse.de>
37277 PR tree-optimization/79276
37278 * tree-vrp.c (process_assert_insertions): Properly adjust common
37279 when removing a duplicate.
37281 2017-01-30 Richard Biener <rguenther@suse.de>
37283 PR tree-optimization/79256
37284 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37285 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37287 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37289 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37292 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37293 ("*r<noxa>sbg_<mode>_sll_bitmask")
37294 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37295 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37296 Use contiguous_bitmask_nowrap_operand.
37298 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37301 * config/rs6000/altivec.h (vec_xl): Revise #define.
37302 (vec_xst): Likewise.
37304 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37306 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37308 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37310 PR rtl-optimization/79194
37311 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37312 traps before call to bypass_conditional_jumps.
37314 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37316 PR tree-optimization/71374
37317 * lra-constraints.c (check_conflict_input_operands): New.
37318 (match_reload): Use it.
37320 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37323 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37324 account to calculate conflict_set.
37326 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37328 PR rtl-optimization/78559
37329 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37330 other_insn in combine.
37332 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37334 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37335 uint16_type_node for BT_UINT16.
37337 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37339 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37340 "RTL Tests" to menu.
37341 (GIMPLE Tests): New node.
37342 (RTL Tests): New node.
37344 2017-01-27 Richard Biener <rguenther@suse.de>
37346 PR tree-optimization/79245
37347 * tree-loop-distribution.c (distribute_loop): Apply cost
37348 modeling also to detected patterns.
37350 2017-01-27 Richard Biener <rguenther@suse.de>
37352 PR tree-optimization/71433
37353 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37354 (compare_assert_loc): New function.
37355 (process_assert_insertions): Sort and optimize assert locations
37356 to remove duplicates and push down identical assertions on
37357 edges to their destination block.
37359 2017-01-27 Richard Biener <rguenther@suse.de>
37361 PR tree-optimization/79244
37362 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37363 out SSA names even if abnormal.
37365 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37367 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37368 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37369 instead of MPFR_RNDN.
37371 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37374 * arm.c (arm_option_override): Don't call build_target_option_node
37375 until after doing all option overrides.
37376 (arm_valid_target_attribute_tree): Likewise.
37378 2017-01-27 Martin Liska <mliska@suse.cz>
37380 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37381 for {cd}tors and C++ {cd}tors.
37383 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37385 * config/s390/s390.md ("*setmem_long_and")
37386 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37388 2017-01-26 Martin Sebor <msebor@redhat.com>
37390 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37393 2017-01-26 Martin Sebor <msebor@redhat.com>
37395 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37396 HAVE_DFmode before using XFmode or DFmode.
37397 (parse_directive): Avoid using the z length modifier to avoid
37398 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37400 PR middle-end/78703
37401 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37402 to accept adjustment as an array.
37403 (get_int_range): New function.
37404 (struct directive): Make width and prec arrays.
37405 (directive::set_width, directive::set_precision): Call get_int_range.
37406 (format_integer, format_floating): Handle width and precision ranges.
37407 (format_string, parse_directive): Same.
37409 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37412 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37413 comdat_type_p set, just clone them, but keep the children in the
37417 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37418 which have direct callers with -fvar-tracking-assignments enabled
37420 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37421 inside of type units.
37423 2017-01-26 Martin Sebor <msebor@redhat.com>
37425 PR middle-end/78703
37426 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37428 (struct format_result): Replace number_chars, number_chars_min,
37429 and number_chars_max with a single member of struct result_range.
37431 (format_result::operator+=): Adjust.
37432 (struct fmtresult): Remove bounded. Handle likely and unlikely
37434 (fmtresult::adjust_for_width_or_precision): New function.
37435 (fmtresult:type_max_digits): New function.
37436 (bytes_remaining): Handle likely and unlikely counters.
37437 (min_bytes_remaining): Remove.
37438 (format_percent): Simplify.
37439 (format_integer, format_floating): Set likely and unlikely counters.
37440 (get_string_length, format_character, format_string): Same.
37441 (format_plain, should_warn_p): New function.
37442 (maybe_warn): Call should_warn_p. Update diagnostic messages
37443 and handle those for all directives, including plain strings.
37444 (format_directive): Handle likely and unlikely counters.
37445 Remove unnecessary quoting from diagnostics. Add an informational
37447 (add_bytes): Remove.
37448 (pass_sprintf_length::compute_format_length): Simplify.
37449 (try_substitute_return_value): Handle likely and unlikely counters.
37451 2017-01-26 Carl Love <cel@us.ibm.com>
37453 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37454 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37456 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37459 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37460 endianess for subregs into account.
37461 * lra-constraints.c (lra_constraints): Do risky transformations
37462 always on the first iteration.
37463 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37464 last_call_used_reg_set.
37465 (process_bb_lives): Define and use last_call_used_reg_set.
37466 * lra.c (lra): Always continue after lra_constraints on the first
37469 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37471 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37473 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37475 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37477 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37478 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37479 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37480 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37481 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37482 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37483 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37484 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37485 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37487 2017-01-26 Marek Polacek <polacek@redhat.com>
37490 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37491 for the third operand.
37493 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37495 PR middle-end/79236
37496 * omp-low.c (struct omp_context): Add simt_stmt field.
37497 (scan_omp_for): Return omp_context *.
37498 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37499 context to the _simt_ SIMD stmt.
37500 (lower_omp_for): For combined SIMD with sibling _simt_
37501 SIMD, make sure to use the same decls in _looptemp_
37502 clauses as in the sibling.
37504 2017-01-26 David Sherwood <david.sherwood@arm.com>
37506 PR middle-end/79212
37507 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37510 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37513 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37514 emit fld b; fld a; if possible.
37516 * brig-builtins.def: Update copyright years.
37517 * config/arm/arm_acle_builtins.def: Update copyright years.
37519 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37522 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37523 constraint instead of o for the stxsd instruction.
37525 2017-01-25 Carl Love <cel@us.ibm.com>
37527 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37528 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37530 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37532 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37534 2017-01-25 Richard Biener <rguenther@suse.de>
37536 PR tree-optimization/69264
37537 * target.def (vector_alignment_reachable): Improve documentation.
37538 * doc/tm.texi: Regenerate.
37539 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37541 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37542 earlier changes with respect to TYPE_USER_ALIGN.
37543 (vector_alignment_reachable_p): Likewise. Improve dumping.
37545 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37548 * config/arm/arm.md (xordi3): Force constant operand into a register
37551 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37553 * doc/invoke.texi (-fstore-merging): Correct default optimization
37554 levels at which it is enabled.
37555 (-O): Move -fstore-merging from list to...
37558 2017-01-25 Richard Biener <rguenther@suse.de>
37561 * omp-expand.c: Include debug.h.
37562 (expand_omp_taskreg): Make sure to generate early debug before
37563 outlining anything from a function.
37564 (expand_omp_target): Likewise.
37565 (grid_expand_target_grid_body): Likewise.
37567 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37570 * asan.c (get_translation_unit_decl): New function.
37571 (asan_add_global): Extract modules file name from globals
37572 TRANSLATION_UNIT_DECL name.
37574 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37577 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37578 for long calls with APCS frame and VFP.
37580 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37582 * cfg.c (original_copy_tables_initialized_p): New function.
37583 * cfg.h (original_copy_tables_initialized_p): New decl.
37584 * cfgrtl.c (relink_block_chain): Guard the call to
37585 free_original_copy_tables with a call to
37586 original_copy_tables_initialized_p.
37587 * cgraph.h (symtab_node::native_rtl_p): New decl.
37588 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37589 (symtab_node::needed_p): Don't assert for early assembly output
37590 for __RTL functions.
37591 (cgraph_node::finalize_function): Set "force_output" for __RTL
37593 (cgraph_node::analyze): Bail out early for __RTL functions.
37594 (analyze_functions): Update assertion to support __RTL functions.
37595 (cgraph_node::expand): Bail out early for __RTL functions.
37596 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37598 * function.h (struct function): Update comment for field
37600 * gimple-expr.c: Include "tree-pass.h".
37601 (gimple_has_body_p): Return false for __RTL functions.
37602 * Makefile.in (OBJS): Add run-rtl-passes.o.
37603 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37605 (gcc::pass_manager::get_clean_slate): New accessor.
37606 * passes.c: Include "insn-addr.h".
37607 (should_skip_pass_p): Add logging. Update logic for running
37608 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37609 property-provider override so it is only done for gimple passes.
37611 (skip_pass): New function.
37612 (execute_one_pass): Call skip_pass when skipping passes.
37613 * read-md.c (md_reader::read_char): Support filtering
37614 the input to a subset of line numbers.
37615 (md_reader::md_reader): Initialize fields
37616 m_first_line and m_last_line.
37617 (md_reader::read_file_fragment): New function.
37618 * read-md.h (md_reader::read_file_fragment): New decl.
37619 (md_reader::m_first_line): New field.
37620 (md_reader::m_last_line): New field.
37621 * read-rtl-function.c (function_reader::create_function): Only
37622 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37623 curr_properties. Set DECL_INITIAL to a dummy block.
37624 (read_rtl_function_body_from_file_range): New function.
37625 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37627 * run-rtl-passes.c: New file.
37628 * run-rtl-passes.h: New file.
37630 2017-01-24 Jeff Law <law@redhat.com>
37632 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37635 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37637 PR tree-optimization/79159
37638 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37639 (record_nonwrapping_iv): Improve boundary using above function if no
37640 value range information.
37642 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37643 Martin Jambor <mjambor@suse.cz>
37645 * brig-builtins.def: New file.
37646 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37647 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37648 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37649 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37650 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37651 * builtin-types.def (BT_INT8): New.
37652 (BT_INT16): Likewise.
37653 (BT_UINT8): Likewise.
37654 (BT_UINT16): Likewise.
37655 (BT_FN_ULONG): Likewise.
37656 (BT_FN_UINT_INT): Likewise.
37657 (BT_FN_UINT_ULONG): Likewise.
37658 (BT_FN_UINT_LONG): Likewise.
37659 (BT_FN_UINT_PTR): Likewise.
37660 (BT_FN_ULONG_PTR): Likewise.
37661 (BT_FN_INT8_FLOAT): Likewise.
37662 (BT_FN_INT16_FLOAT): Likewise.
37663 (BT_FN_UINT32_FLOAT): Likewise.
37664 (BT_FN_UINT16_FLOAT): Likewise.
37665 (BT_FN_UINT8_FLOAT): Likewise.
37666 (BT_FN_UINT64_FLOAT): Likewise.
37667 (BT_FN_UINT16_UINT32): Likewise.
37668 (BT_FN_UINT32_UINT16): Likewise.
37669 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37670 (BT_FN_INT_PTR_INT): Likewise.
37671 (BT_FN_UINT_PTR_UINT): Likewise.
37672 (BT_FN_LONG_PTR_LONG): Likewise.
37673 (BT_FN_ULONG_PTR_ULONG): Likewise.
37674 (BT_FN_VOID_UINT64_UINT64): Likewise.
37675 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37676 (BT_FN_INT8_INT8_INT8): Likewise.
37677 (BT_FN_INT16_INT16_INT16): Likewise.
37678 (BT_FN_INT_INT_INT): Likewise.
37679 (BT_FN_UINT_FLOAT_UINT): Likewise.
37680 (BT_FN_FLOAT_UINT_UINT): Likewise.
37681 (BT_FN_ULONG_UINT_UINT): Likewise.
37682 (BT_FN_ULONG_UINT_PTR): Likewise.
37683 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37684 (BT_FN_UINT_UINT_UINT): Likewise.
37685 (BT_FN_VOID_UINT_PTR): Likewise.
37686 (BT_FN_UINT_UINT_PTR: Likewise.
37687 (BT_FN_UINT32_UINT64_PTR): Likewise.
37688 (BT_FN_INT_INT_UINT_UINT): Likewise.
37689 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37690 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37691 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37692 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37693 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37694 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37695 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37696 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37697 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37698 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37699 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37700 * doc/frontends.texi: List BRIG FE.
37701 * doc/install.texi (Testing): Add BRIG tesring requirements.
37702 * doc/invoke.texi (Overall Options): Mention BRIG.
37703 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37705 2017-01-24 Richard Biener <rguenther@suse.de>
37707 PR translation/79208
37708 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37710 2017-01-24 Martin Jambor <mjambor@suse.cz>
37713 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
37714 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
37715 and known_contexts.
37717 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
37719 PR middle-end/79123
37720 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
37721 casts from signed to unsigned really don't have a range.
37723 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
37725 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
37726 GMP_RNDx for compatiblity.
37728 2017-01-24 Martin Liska <mliska@suse.cz>
37731 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
37732 that would prevent us to call alloca with -1 as argument.
37734 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37736 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37737 Avoid -Wformat-security warning.
37739 2017-01-23 Andrew Pinski <apinski@cavium.com>
37741 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37744 2017-01-23 Martin Sebor <msebor@redhat.com>
37746 PR middle-end/78703
37747 * gimple-ssa-sprintf.c (warn_level): New global.
37748 (format_integer): Use it here and throughout the rest of the file.
37749 Use the same switch to compute sign as base.
37750 (maybe_warn): New function.
37751 (format_directive): Factor out warnings into maybe_warn.
37752 Add debugging output. Use warn_level.
37753 (add_bytes): Use warn_level.
37754 (pass_sprintf_length::compute_format_length): Add debugging output.
37755 (try_substitute_return_value): Same.
37756 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37758 PR middle-end/78703
37759 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37760 (struct fmtresult, format_integer, format_floating): Adjust.
37761 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37762 (get_string_length, format_string,format_directive): Same.
37763 (pass_sprintf_length::compute_format_length): Same.
37764 (try_substitute_return_value): Simplify slightly.
37766 PR middle-end/78703
37767 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37768 (fmtresult::operator+=): Outlined.
37769 (struct fmtresult): Add ctors.
37770 (struct conversion_spec): Rename...
37771 (struct directive): ...to this. Add and remove data members.
37772 (directive::set_width, directive::set_precision): New functions.
37773 (format_percent): Use fmtresult ctor.
37774 (get_width_and_precision): Remove.
37775 (format_integer): Make naming changes. Avoid computing width and
37777 (format_floating): Same. Adjust indentation.
37778 (format_character, format_none): New functions.
37779 (format_string): Moved character handling to format_character.
37780 (format_directive): Remove arguments, change return type.
37781 (parse_directive): New function.
37782 (pass_sprintf_length::compute_format_length): Move directive
37783 parsing to parse_directive.
37785 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37787 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37788 (assign_assembler_name_if_needed): ... this.
37789 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37790 (assign_assembler_name_if_needed): ... this.
37791 (free_lang_data_in_cgraph): Adjust callers.
37792 * cgraphunit.c (cgraph_node::analyze): Likewise.
37793 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37796 2017-01-23 Richard Biener <rguenther@suse.de>
37798 PR tree-optimization/79088
37799 PR tree-optimization/79188
37800 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37801 resetting loop bounds after last path deletion. Reset loop
37802 bounds of the target loop, make code match the comments.
37803 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37804 Make sure loops need no fixups.
37806 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37808 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37809 exponent support with double type for first argument.
37810 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37811 type returned by __builtin_vec_extract_sig,
37812 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37813 functions from "vector int" to "vector unsigned int" or from
37814 "vector long long int" to "vector unsigned long long int".
37815 Changed type returned by __builtin_vec_extract_exp,
37816 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37817 functions from "vector int" to "vector unsigned int" or from
37818 "vector long long int" to "vector unsigned long long int".
37819 Changed return type of __builtin_vec_test_data_class,
37820 __builtin_vec_test_data_class_sp, and
37821 __builtin_vec_test_data_class_dp from "vector int" to
37822 "vector bool int" or from "vector long long int" to "vector bool
37823 long long int" and changed second argument type from "unsigned
37824 int" to "int". Added new overloaded function forms "vector float
37825 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37826 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37827 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37828 double, vector unsigned long long int)" and "vector double
37829 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37830 long int)". Changed return type of
37831 __builtin_scalar_test_data_class and
37832 __builtin_scalar_test_data_class_sp and
37833 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37834 int" and changed second argument from "unsigned int" to "int".
37835 Changed type returned by __builtin_scalar_test_neg,
37836 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37837 from "int" to "bool int". Added new overloaded function form
37838 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37839 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37840 exponent double-precision with floating point first argument.
37841 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37842 documentation of scalar_test_data_class, scalar_test_neg,
37843 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37844 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37845 vec_test_data_class built-in functions to reflect refinements in
37846 their type signatures.
37848 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37850 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37852 (aarch64_elf_asm_destructor): Likewise.
37854 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37856 PR rtl-optimization/78634
37857 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37858 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37859 * ifcvt.c (noce_try_cmove): Add missing cost check.
37861 PR rtl-optimization/71724
37862 * combine.c (if_then_else_cond): Look for situations where it is
37863 beneficial to undo the work of one of the recursive calls.
37865 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37867 PR tree-optimization/70754
37868 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37869 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37870 combined stmt before it if not NULL.
37871 (combine_chains): Process refs reversely and compute dominance point
37874 2017-01-23 Martin Liska <mliska@suse.cz>
37876 PR tree-optimization/79196
37877 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37878 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37880 (strlen_optimize_stmt): Call the renamed function.
37882 2017-01-23 Michael Matz <matz@suse.de>
37884 PR tree-optimization/78384
37885 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37887 2017-01-23 Richard Biener <rguenther@suse.de>
37889 PR tree-optimization/79186
37890 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37891 both incoming edges before moving an assert.
37893 2017-01-23 Martin Jambor <mjambor@suse.cz>
37895 * ipa-prop.c (load_from_param_1): Removed.
37896 (load_from_unmodified_param): Bits from load_from_param_1 put back
37898 (load_from_param): Removed.
37899 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37900 with stmt. Reverted back to use of load_from_unmodified_param.
37902 2017-01-23 Martin Jambor <mjambor@suse.cz>
37905 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37906 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37907 field a pointer to garbage collected vector, mark lattices and
37908 ipcp_orig_node with GTY((skip)).
37909 (ipa_get_param_count): Adjust to descriptors being a pointer.
37910 (ipa_get_param): Likewise.
37911 (ipa_get_type): Likewise.
37912 (ipa_get_param_move_cost): Likewise.
37913 (ipa_set_param_used): Likewise.
37914 (ipa_get_controlled_uses): Likewise.
37915 (ipa_set_controlled_uses): Likewise.
37916 (ipa_is_param_used): Likewise.
37917 (ipa_node_params_t): Move into garbage collector. New methods insert
37919 (ipa_node_params_sum): Annotate wth GTY(()).
37920 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37922 (ipa_load_from_parm_agg): Adjust declaration.
37923 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37924 * ipa-profile.c (ipa_profile): Likewise.
37925 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37926 (ipa_populate_param_decls): Make descriptors parameter garbage
37928 (ipa_dump_param): Adjust to descriptors being a pointer.
37929 (ipa_alloc_node_params): Likewise.
37930 (ipa_initialize_node_params): Likewise.
37931 (load_from_param_1): Make descriptors parameter garbage collected.
37932 (load_from_unmodified_param): Likewise.
37933 (load_from_param): Likewise.
37934 (ipa_load_from_parm_agg): Likewise.
37935 (ipa_node_params::~ipa_node_params): Removed.
37936 (ipa_free_all_node_params): Remove call to delete operator.
37937 (ipa_node_params_t::insert): New.
37938 (ipa_node_params_t::remove): Likewise.
37939 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37940 copy known_csts and known_contexts vectors.
37941 (ipa_read_node_info): Adjust to descriptors being a pointer.
37942 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37944 (ipcp_transform_function): Make descriptors variable garbage
37947 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37949 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37950 * config/i386/avx512dqintrin.h: Ditto.
37951 * config/i386/avx512fintrin.h: Ditto.
37952 * config/i386/i386.c: Handle new builtins.
37953 * config/i386/i386-builtin.def: Add new builtins.
37954 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37955 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37957 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37958 Martin Liska <mliska@suse.cz>
37960 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37961 * asan.c (asan_expand_poison_ifn): Support stores and use
37962 appropriate ASAN report function.
37963 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37964 * internal-fn.def (ASAN_POISON_USE): Declare.
37965 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37966 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37967 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37968 ASAN_POISON calls w/o LHS.
37969 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37970 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37971 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37972 * gimplify.c (asan_poison_variables): Add attribute
37973 use_after_scope_memory to variables that really needs to live
37975 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37976 having the attribute.
37978 2017-01-23 Martin Liska <mliska@suse.cz>
37980 * asan.c (create_asan_shadow_var): New function.
37981 (asan_expand_poison_ifn): Likewise.
37982 * asan.h (asan_expand_poison_ifn): New declaration.
37983 * internal-fn.c (expand_ASAN_POISON): Likewise.
37984 * internal-fn.def (ASAN_POISON): New builtin.
37985 * sanopt.c (pass_sanopt::execute): Expand
37986 asan_expand_poison_ifn.
37987 * tree-inline.c (copy_decl_for_dup_finish): Make function
37989 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37990 * tree-ssa.c (is_asan_mark_p): New function.
37991 (execute_update_addresses_taken): Rewrite local variables
37992 (identified just by use-after-scope as addressable) into SSA.
37994 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37996 * doc/install.texi (Specific): opensource.apple.com uses https
37997 now. Remove trailing slash.
37999 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38001 * README.Portability: Remove note on an Irix compatibility issue.
38003 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38005 * gcov.c (INCLUDE_ALGORITHM): Define.
38006 (INCLUDE_VECTOR): Define.
38007 No longer include <vector> and <algorithm> directly.
38009 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38011 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38013 * doc/invoke.texi (Code Gen Options): Ditto.
38015 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38018 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38020 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38022 rtl-optimization/79125
38023 * cprop.c (local_cprop_pass): Handle cases where we make an
38024 unconditional trap.
38026 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38030 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38031 read from, for big endian.
38033 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38035 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38036 register pauth builtins for LP64 only.
38038 2017-01-20 Marek Polacek <polacek@redhat.com>
38041 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38044 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38046 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38048 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38049 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38050 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38052 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38055 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38056 in big-endian mode when they are not a single duplicated value.
38058 2017-01-20 Richard Biener <rguenther@suse.de>
38060 * BASE-VER: Bump to 7.0.1.
38062 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38064 * omp-low.c (omplow_simd_context): New struct. Use it...
38065 (lower_rec_simd_input_clauses): ...here and...
38066 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38067 references to idx, lane, max_vf, is_simt.
38069 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38071 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38074 2017-01-20 Martin Jambor <mjambor@suse.cz>
38076 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38077 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38079 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38080 (GTFILES): Rename hsa.c to hsa-common.c.
38081 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38082 * hsa-dump.c: Likewise.
38083 * hsa-gen.c: Likewise.
38084 * hsa-regalloc.c: Likewise.
38085 * ipa-hsa.c: Likewise.
38086 * omp-expand.c: Likewise.
38087 * omp-low.c: Likewise.
38088 * toplev.c: Likewise.
38090 2017-01-20 Marek Polacek <polacek@redhat.com>
38093 * doc/invoke.texi: Document -Wduplicated-branches.
38094 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38095 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38096 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38097 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38098 return 0 only when not OEP_LEXICOGRAPHIC.
38099 (fold_build_cleanup_point_expr): Use the expression
38100 location when building CLEANUP_POINT_EXPR.
38101 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38102 * tree.c (add_expr): Handle error_mark_node.
38104 2017-01-20 Martin Liska <mliska@suse.cz>
38107 * tree-profile.c (init_ic_make_global_vars): Do not call
38109 (gimple_init_gcov_profiler): Likewise.
38111 2017-01-20 Martin Liska <mliska@suse.cz>
38114 * cgraph.h (maybe_create_reference): Remove argument and
38116 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38118 * ipa-cp.c (create_specialized_node): Likewise.
38119 * symtab.c (symtab_node::maybe_create_reference): Handle
38120 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38122 2017-01-20 Martin Liska <mliska@suse.cz>
38124 * read-rtl-function.c (function_reader::create_function): Use
38125 build_decl instread of build_decl_stat.
38127 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38129 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38130 * config/i386/avx512dqintrin.h: Ditto.
38131 * config/i386/avx512fintrin.h: Ditto.
38132 * config/i386/i386-builtin-types.def: Add new types.
38133 * config/i386/i386.c: Handle new types.
38134 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38135 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38136 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38137 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38138 (__builtin_ia32_kshiftridi): New.
38139 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38141 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38145 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38146 define to rs6000_init_stack_protect_guard.
38147 (rs6000_init_stack_protect_guard): New function.
38149 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38150 Yunqiang Su <yunqiang.su@imgtec.com>
38152 * config.gcc (supported_defaults): Add madd4.
38153 (with_madd4): Add validation.
38154 (all_defaults): Add madd4.
38155 * config/mips/mips.opt (mmadd4): New option.
38156 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38158 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38160 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38161 (ISA_HAS_FUSED_MADD4): Likewise.
38162 * doc/invoke.texi (-mmadd4): Document the new option.
38163 * doc/install.texi (--with-madd4): Document the new option.
38165 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38167 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38168 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38169 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38170 (aarch64_init_pauth_hint_builtins): New.
38171 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38172 (aarch64_expand_builtin): Expand new builtins.
38174 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38176 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38177 * combine-stack-adj.c (no_unhandled_cfa): Handle
38178 REG_CFA_TOGGLE_RA_MANGLE.
38179 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38180 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38181 info for return address signing.
38182 (aarch64_expand_epilogue): Likewise.
38184 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38186 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38187 * config/aarch64/aarch64-protos.h
38188 (aarch64_return_address_signing_enabled): New declaration.
38189 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38191 (aarch64_expand_prologue): Sign return address before it's pushed onto
38193 (aarch64_expand_epilogue): Authenticate return address fetched from
38195 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38196 (aarch64_attributes): New function attributes for "sign-return-address".
38197 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38198 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38199 ("*do_return"): Generate combined instructions according to key index.
38200 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38201 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38203 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38204 * config/aarch64/aarch64.opt (msign-return-address=): New.
38205 * doc/extend.texi (AArch64 Function Attributes): Documents
38206 "sign-return-address=".
38207 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38209 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38211 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38212 overall option summary.
38214 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38216 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38217 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38218 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38219 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38221 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38223 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38224 -mpower9-minmax by default for -mcpu=power9.
38225 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38226 128-bit floating point.
38228 2017-01-20 Alan Modra <amodra@gmail.com>
38230 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38231 optimizing for size.
38233 2017-01-20 Alan Modra <amodra@gmail.com>
38236 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38237 for strcmp and strncmp from corresponding builtin decl.
38239 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38241 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38242 instead of i386/rtems-64.h.
38243 * config/i386/rtems-64.h: Remove.
38245 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38249 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38251 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38253 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38255 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38256 Change int to HOST_WIDE_INT.
38257 * config/aarch64/aarch64-protos.h
38258 (aarch64_simd_gen_const_vector_dup): Likewise.
38259 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38261 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38263 * langhooks-def.h (lhd_type_for_size): New decl.
38264 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38265 * langhooks.c (lhd_type_for_size): New function, taken from
38268 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38270 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38271 define_bypass for CR latency.
38272 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38273 (power9-alu2): Add define_bypass for CR latency.
38275 (power9-mul): Update insn latency.
38276 (power9-mul-compare): Update insn latency, bypass latency and remove
38279 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38281 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38283 * config/aarch64/aarch64.md
38284 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38285 aarch64_nopcrelative_literal_loads.
38286 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38288 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38290 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38291 TARGET_LOONGSON_3A.
38292 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38294 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38297 * config.gcc (supported_defaults): Add lxc1-sxc1.
38298 (with_lxc1_sxc1): Add validation.
38299 (all_defaults): Add lxc1-sxc1.
38300 * config/mips/mips.opt (mlxc1-sxc1): New option.
38301 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38303 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38304 __mips_no_lxc1_sxc1.
38305 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38306 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38307 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38309 2017-01-19 Richard Biener <rguenther@suse.de>
38311 PR tree-optimization/72488
38312 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38313 sure to restore SSA info.
38314 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38316 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38318 PR rtl-optimization/79121
38319 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38320 of the inner type when shifting an extended value.
38322 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38325 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38326 interposable aliases.
38328 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38331 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38332 Use the evmergelohi instruction.
38333 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38334 (mov_sitf_e500_subreg8_2_be): Likewise.
38335 (mov_sitf_e500_subreg12_2_le): Likewise.
38336 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38337 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38338 (mov_sitf_e500_subreg8_2_le): Likewise.
38339 (mov_sitf_e500_subreg12_2_be): Likewise.
38341 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38343 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38344 attribute from vecsimple to vecperm.
38345 (altivec_vbpermq2): Likewise.
38347 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38350 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38352 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38353 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38354 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38355 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38356 case where N arg is SIZE_MAX.
38357 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38358 (cmpstrsi): Add pattern.
38360 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38362 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38363 __builtin_vec_revb builtins.
38364 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38365 built-in functions to support generation of the ISA 3.0 XXBR<x>
38366 vector byte reverse instructions.
38367 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38368 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38369 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38370 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38371 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38372 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38373 (P9V_BUILTIN_VEC_REVB): Likewise.
38374 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38375 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38376 (p9_xxbrq_v16qi): Likewise.
38377 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38378 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38379 (p9_xxbrh_v8hi): Likewise.
38380 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38381 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38382 vec_revb built-in functions.
38384 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38386 PR rtl-optimization/78952
38387 * config/i386/i386.md (any_extract): New code iterator.
38388 (*insvqi_2): Use any_extract for source operand.
38389 (*insvqi_3): Use any_shiftrt for source operand.
38391 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38393 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38395 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38397 2017-01-18 Matthias Klose <doko@ubuntu.com>
38399 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38401 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38403 * config/rs6000/altivec.h (vec_bperm): Change #define.
38404 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38405 (altivec_vbpermq2): New define_insn.
38406 (altivec_vbpermd): Likewise.
38407 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38408 function interface.
38409 (VBPERMD): Likewise.
38410 (VBPERM): New polymorphic function interface.
38411 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38412 Add entries for P9V_BUILTIN_VEC_VBPERM.
38413 * doc/extend.texi: Add interfaces for vec_bperm.
38415 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38417 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38418 first letter of error messages.
38419 (s390_resolve_overloaded_builtin): Likewise.
38420 * config/s390/s390.c (s390_expand_builtin): Likewise.
38421 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38422 (s390_valid_target_attribute_inner_p): Likewise.
38423 * config/s390/s390.md ("tabort"): Likewise.
38425 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38427 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38428 (ISA_AVOID_DIV_HILO): New macro.
38429 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38430 (ISA_HAS_DDIV): Likewise.
38432 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38434 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38436 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38438 * doc/invoke.texi (fabi-version): Spelling fix.
38440 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38443 * doc/invoke.texi (fabi-version): Mention mangling fix for
38446 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38449 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38451 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38454 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38455 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38457 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38459 (stack_protect_test): Ditto.
38460 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38461 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38463 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38464 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38465 -mstack-protector-guard-offset=.
38466 (RS/6000 and PowerPC Options): Ditto.
38468 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38470 * config/i386/i386.h (MASK_CLASS_P): New define.
38471 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38472 there are no registers from different register sets also when
38473 mask registers are used. Update function comment.
38474 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38475 to (*k/*r) and (*k/*km) alternatives.
38477 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38479 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38480 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38481 (EH_RETURN_HANDLER_RTX): New define.
38482 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38483 Force frame pointer in EH return functions.
38484 (aarch64_expand_epilogue): Add barrier for eh_return.
38485 (aarch64_final_eh_return_addr): Remove.
38486 (aarch64_eh_return_handler_rtx): New function.
38487 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38489 (aarch64_eh_return_handler_rtx): New prototype.
38491 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38493 * config/rs6000/altivec.h (vec_rlmi): New #define.
38494 (vec_vrlnm): Likewise.
38495 (vec_rlnm): Likewise.
38496 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38497 (UNSPEC_VRLNM): Likewise.
38498 (VIlong): New mode iterator.
38499 (altivec_vrl<VI_char>mi): New define_insn.
38500 (altivec_vrl<VI_char>nm): Likewise.
38501 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38503 (VRLDNM): Likewise.
38504 (RLNM): New polymorphic function entry.
38505 (VRLWMI): New monomorphic function entry.
38506 (VRLDMI): Likewise.
38507 (RLMI): New polymorphic function entry.
38508 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38509 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38510 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38513 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38516 * dwarf2out.c (field_byte_offset): Restore the
38517 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38518 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38519 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38522 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38525 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38527 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38530 * dwarf2out.c (add_data_member_location_attribute): For constant
38531 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38532 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38533 DW_AT_byte_size attributes.
38535 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38537 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38538 after forcing to constant memory when the code model is medium.
38540 2017-01-17 Julia Koval <julia.koval@intel.com>
38543 * config/i386/avx512fintrin.h
38544 (_mm512_i32gather_ps): Change __addr type to void const*.
38545 (_mm512_mask_i32gather_ps): Ditto.
38546 (_mm512_i32gather_pd): Ditto.
38547 (_mm512_mask_i32gather_pd): Ditto.
38548 (_mm512_i64gather_ps): Ditto.
38549 (_mm512_mask_i64gather_ps): Ditto.
38550 (_mm512_i64gather_pd): Ditto.
38551 (_mm512_mask_i64gather_pd): Ditto.
38552 (_mm512_i32gather_epi32): Ditto.
38553 (_mm512_mask_i32gather_epi32): Ditto.
38554 (_mm512_i32gather_epi64): Ditto.
38555 (_mm512_mask_i32gather_epi64): Ditto.
38556 (_mm512_i64gather_epi32): Ditto.
38557 (_mm512_mask_i64gather_epi32): Ditto.
38558 (_mm512_i64gather_epi64): Ditto.
38559 (_mm512_mask_i64gather_epi64): Ditto.
38560 (_mm512_i32scatter_ps): Change __addr type to void*.
38561 (_mm512_mask_i32scatter_ps): Ditto.
38562 (_mm512_i32scatter_pd): Ditto.
38563 (_mm512_mask_i32scatter_pd): Ditto.
38564 (_mm512_i64scatter_ps): Ditto.
38565 (_mm512_mask_i64scatter_ps): Ditto.
38566 (_mm512_i64scatter_pd): Ditto.
38567 (_mm512_mask_i64scatter_pd): Ditto.
38568 (_mm512_i32scatter_epi32): Ditto.
38569 (_mm512_mask_i32scatter_epi32): Ditto.
38570 (_mm512_i32scatter_epi64): Ditto.
38571 (_mm512_mask_i32scatter_epi64): Ditto.
38572 (_mm512_i64scatter_epi32): Ditto.
38573 (_mm512_mask_i64scatter_epi32): Ditto.
38574 (_mm512_i64scatter_epi64): Ditto.
38575 (_mm512_mask_i64scatter_epi64): Ditto.
38576 * config/i386/avx512pfintrin.h
38577 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38578 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38579 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38580 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38581 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38582 (_mm512_prefetch_i32scatter_ps): Ditto.
38583 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38584 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38585 (_mm512_prefetch_i64scatter_pd): Ditto.
38586 (_mm512_prefetch_i64scatter_ps): Ditto.
38587 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38588 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38589 * config/i386/avx512vlintrin.h
38590 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38591 (_mm_mmask_i32gather_ps): Ditto.
38592 (_mm256_mmask_i32gather_pd): Ditto.
38593 (_mm_mmask_i32gather_pd): Ditto.
38594 (_mm256_mmask_i64gather_ps): Ditto.
38595 (_mm_mmask_i64gather_ps): Ditto.
38596 (_mm256_mmask_i64gather_pd): Ditto.
38597 (_mm_mmask_i64gather_pd): Ditto.
38598 (_mm256_mmask_i32gather_epi32): Ditto.
38599 (_mm_mmask_i32gather_epi32): Ditto.
38600 (_mm256_mmask_i32gather_epi64): Ditto.
38601 (_mm_mmask_i32gather_epi64): Ditto.
38602 (_mm256_mmask_i64gather_epi32): Ditto.
38603 (_mm_mmask_i64gather_epi32): Ditto.
38604 (_mm256_mmask_i64gather_epi64): Ditto.
38605 (_mm_mmask_i64gather_epi64): Ditto.
38606 (_mm256_i32scatter_ps): Change __addr type to void*.
38607 (_mm256_mask_i32scatter_ps): Ditto.
38608 (_mm_i32scatter_ps): Ditto.
38609 (_mm_mask_i32scatter_ps): Ditto.
38610 (_mm256_i32scatter_pd): Ditto.
38611 (_mm256_mask_i32scatter_pd): Ditto.
38612 (_mm_i32scatter_pd): Ditto.
38613 (_mm_mask_i32scatter_pd): Ditto.
38614 (_mm256_i64scatter_ps): Ditto.
38615 (_mm256_mask_i64scatter_ps): Ditto.
38616 (_mm_i64scatter_ps): Ditto.
38617 (_mm_mask_i64scatter_ps): Ditto.
38618 (_mm256_i64scatter_pd): Ditto.
38619 (_mm256_mask_i64scatter_pd): Ditto.
38620 (_mm_i64scatter_pd): Ditto.
38621 (_mm_mask_i64scatter_pd): Ditto.
38622 (_mm256_i32scatter_epi32): Ditto.
38623 (_mm256_mask_i32scatter_epi32): Ditto.
38624 (_mm_i32scatter_epi32): Ditto.
38625 (_mm_mask_i32scatter_epi32): Ditto.
38626 (_mm256_i32scatter_epi64): Ditto.
38627 (_mm256_mask_i32scatter_epi64): Ditto.
38628 (_mm_i32scatter_epi64): Ditto.
38629 (_mm_mask_i32scatter_epi64): Ditto.
38630 (_mm256_i64scatter_epi32): Ditto.
38631 (_mm256_mask_i64scatter_epi32): Ditto.
38632 (_mm_i64scatter_epi32): Ditto.
38633 (_mm_mask_i64scatter_epi32): Ditto.
38634 (_mm256_i64scatter_epi64): Ditto.
38635 (_mm256_mask_i64scatter_epi64): Ditto.
38636 (_mm_i64scatter_epi64): Ditto.
38637 (_mm_mask_i64scatter_epi64): Ditto.
38638 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38639 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38640 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38641 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38642 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38643 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38644 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38645 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38646 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38647 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38648 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38649 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38650 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38651 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38652 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38653 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38654 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38655 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38656 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38657 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38658 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38659 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38660 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38661 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38662 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38663 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38664 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38665 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38666 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38667 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38668 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38669 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38670 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38671 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38672 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38673 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38674 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38675 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38676 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38677 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38678 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38679 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38680 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38681 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38682 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38683 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38684 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38685 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38686 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38687 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38688 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38689 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38690 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38691 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38692 definitions accordingly.
38694 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38695 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38698 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38701 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38704 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38705 hard regs for inner regno.
38707 2017-01-17 Martin Liska <mliska@suse.cz>
38710 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38711 assumption and add comment.
38713 2017-01-17 Nathan Sidwell <nathan@acm.org>
38715 * ipa-visibility.c (localize_node): New function, broken out of ...
38716 (function_and_variable_visibility): ... here. Call it.
38718 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38720 PR middle-end/77445
38721 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
38722 correctly set frequency of oudgoing edge.
38723 (duplicate_thread_path): Fix profile updating.
38725 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38728 * configure.ac: Add GCC_BASE_VER.
38729 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
38730 version from BASE-VER file.
38731 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
38732 (gcc.o): Depend on $(BASEVER).
38733 * common.opt (dumpfullversion): New option.
38734 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38735 * doc/invoke.texi: Document -dumpfullversion.
38736 * doc/install.texi: Document --with-gcc-major-version-only.
38737 * configure: Regenerated.
38739 2017-01-17 Richard Biener <rguenther@suse.de>
38741 PR tree-optimization/71433
38742 * tree-vrp.c (register_new_assert_for): Merge same asserts
38743 on all incoming edges.
38744 (process_assert_insertions_for): Handle insertions at the
38747 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38749 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38750 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38752 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38755 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38758 2017-01-17 Alan Modra <amodra@gmail.com>
38761 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38762 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38763 symbolic stack limit when pic.
38765 2017-01-16 Martin Sebor <msebor@redhat.com>
38767 PR tree-optimization/78608
38768 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38770 2017-01-16 Jeff Law <law@redhat.com>
38773 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38774 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38775 for several include directories that may be relative to sysroot.
38776 * config/i386/x-mingw32 (gplus_includedir): Define.
38777 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38778 (native_system_includedir): Likewise.
38779 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38780 override if TARGET_SYSTEM_ROOT is defined.
38781 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38783 PR tree-optimization/79090
38784 PR tree-optimization/33562
38785 PR tree-optimization/61912
38786 PR tree-optimization/77485
38787 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38788 and computed trims into the dump file.
38790 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38792 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38794 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38797 * gimplify.c (gimplify_init_constructor): If want_value and
38798 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38802 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38803 sequence. Formatting fixes.
38804 (doloop_optimize): Formatting fixes.
38807 * gcc.c (debug_level_greater_than_spec_func): New function.
38808 (static_spec_functions): Add debug-level-gt spec function.
38809 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38811 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38812 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38813 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38814 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38815 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38816 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38818 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38820 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38821 QImode fixups to general and mask registers only.
38823 2017-01-16 Carl Love <cel@us.ibm.com>
38825 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38826 for built-in functions
38827 vector signed char vec_nabs (vector signed char)
38828 vector signed short vec_nabs (vector signed short)
38829 vector signed int vec_nabs (vector signed int)
38830 vector signed long long vec_nabs (vector signed long long)
38831 vector float vec_nabs (vector float)
38832 vector double vec_nabs (vector double)
38833 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38835 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38836 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38837 * doc/extend.texi: Update the documentation file for the new built-in
38840 2017-01-16 Martin Sebor <msebor@redhat.com>
38842 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38845 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38847 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38848 UNSPEC_VSX__XXSPLTD to require special splat handling.
38850 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38853 * system.h: Poison strndup.
38855 2017-01-16 Alan Modra <amodra@gmail.com>
38858 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38861 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38863 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38865 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38867 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38868 call recog here. Assert that INSN_CODE (insn) is non-negative.
38870 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38873 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38875 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38876 in the currently scheduled RTL fragment.
38878 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38880 PR rtl-optimization/78751
38881 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38884 2017-01-14 Jeff Law <law@redhat.com>
38886 PR tree-optimization/79090
38887 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38888 variable length stores.
38889 (compute_trims): Delete dead assignment to *trim_tail.
38890 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38893 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38895 PR rtl-optimization/78626
38896 PR rtl-optimization/78727
38897 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38898 of a block, and split such blocks after everything else is finished.
38900 2017-01-14 Alan Modra <amodra@gmail.com>
38903 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38904 target legitimate_combined_insn.
38905 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38906 (rs6000_legitimate_combined_insn): New function.
38907 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38909 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38910 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38911 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38913 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38915 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38917 2017-01-13 Jeff Law <law@redhat.com>
38919 PR tree-optimization/33562
38920 PR tree-optimization/61912
38921 PR tree-optimization/77485
38922 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38924 (delete_dead_assignment): Likewise.
38925 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38926 statement to delete_dead_call and delete_dead_assignment.
38928 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38931 * substring-locations.c (format_warning_va): Strengthen case 1 so
38932 that both endpoints of the substring must be within the format
38933 range for just the substring to be printed.
38935 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38937 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38938 * config/i386/i386.c (ix86_target_string): Add missing options
38939 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38940 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38941 flags_other and ix86_target_other to flags2_other. Display unknown
38943 (ix86_valid_target_attribute_inner_p): Add missing options and
38944 reorder options by implied ISAs, as in ix86_target_string.
38946 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38948 * hash-table.h (hash_table::too_empty_p): New function.
38949 (hash_table::expand): Use it.
38950 (hash_table::traverse): Likewise.
38951 (hash_table::empty_slot): Use sizeof (value_type) instead of
38952 sizeof (PTR) to convert bytes to elements. Shrink the table
38953 if the current size is excessive for the current number of
38956 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38958 * ira-costs.c (record_reg_classes): Break from the inner loop
38959 early once alt_fail is known to be true. Update outer loop
38960 handling accordingly.
38962 2017-01-13 Jeff Law <law@redhat.com>
38964 * tree-ssa-dse.c (decrement_count): New function.
38965 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38966 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38967 when we know the partially dead statement is a mem* function.
38969 PR tree-optimization/61912
38970 PR tree-optimization/77485
38971 * tree-ssa-dse.c: Include expr.h.
38972 (maybe_trim_constructor_store): New function.
38973 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38975 PR tree-optimization/33562
38976 PR tree-optimization/61912
38977 PR tree-optimization/77485
38978 * doc/invoke.texi: Document new dse-max-object-size param.
38979 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38980 * tree-ssa-dse.c: Include params.h.
38981 (dse_store_status): New enum.
38982 (initialize_ao_ref_for_dse): New, partially extracted from
38984 (valid_ao_ref_for_dse, normalize_ref): New.
38985 (setup_live_bytes_from_ref, compute_trims): Likewise.
38986 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38987 (maybe_trim_partially_dead_store): Likewise.
38988 (maybe_trim_complex_store): Likewise.
38989 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38990 Track what bytes live from the original store. Return tri-state
38991 for dead, partially dead or live.
38992 (dse_dom_walker): Add constructor, destructor and new private members.
38993 (delete_dead_call, delete_dead_assignment): New extracted from
38995 (dse_optimize_stmt): Make a member of dse_dom_walker.
38996 Use initialize_ao_ref_for_dse.
38998 PR tree-optimization/33562
38999 PR tree-optimization/61912
39000 PR tree-optimization/77485
39001 * sbitmap.h (bitmap_count_bits): Prototype.
39002 (bitmap_clear_range, bitmap_set_range): Likewise.
39003 * sbitmap.c (bitmap_clear_range): New function.
39004 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39006 2017-01-13 Martin Liska <mliska@suse.cz>
39009 * function.c (set_cfun): Add new argument force.
39010 * function.h (set_cfun): Likewise.
39011 * ipa-inline-transform.c (inline_call): Use the function when
39012 strict alising from is dropped for function we inline to.
39014 2017-01-13 Richard Biener <rguenther@suse.de>
39016 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39017 for dumping GIMPLE INTEGER_CSTs.
39019 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39021 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39022 to 201112L since C++17.
39024 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39027 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39028 if -fsanitize=kernel-address is present.
39030 2017-01-13 Richard Biener <rguenther@suse.de>
39032 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39033 as _Literal ( type ) number in case usual suffixes do not
39034 preserve all information.
39036 2017-01-13 Richard Biener <rguenther@suse.de>
39038 PR tree-optimization/77283
39039 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39040 and ssa-iterators.h.
39041 (is_feasible_trace): Implement a cost model based on joiner
39044 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39047 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39048 char or short to __float128/_Float128 directly.
39050 2017-01-12 Martin Sebor <msebor@redhat.com>
39052 to -Wformat-overflow.
39053 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39054 (min_bytes_remaining): Same.
39055 (get_string_length): Same.
39056 (format_string): Same.
39057 (format_directive): Same.
39059 (pass_sprintf_length::handle_gimple_call): Same.
39061 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39063 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39064 info.nowrite calls with no lhs that can't throw. Return bool
39065 whether gsi_remove has been called or not.
39066 (pass_sprintf_length::handle_gimple_call): Return bool whether
39067 try_substitute_return_value called gsi_remove. Formatting fix.
39068 (pass_sprintf_length::execute): Don't use gsi_remove if
39069 handle_gimple_call returned true.
39072 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39073 be removed due to side-effects, don't remove following barrier nor
39074 turn the successor edge into fallthru edge.
39076 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39079 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39080 element-reversing loads and stores as not swappable.
39082 2017-01-12 Nathan Sidwell <nathan@acm.org>
39083 Nicolai Stange <nicstange@gmail.com>
39085 * combine.c (try_combine): Don't ignore result of overlap checking
39086 loop. Combine overlap & asm check into single loop.
39088 2017-01-12 Richard Biener <rguenther@suse.de>
39090 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39091 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39093 2017-01-12 Richard Biener <rguenther@suse.de>
39095 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39096 and TS_TARGET_OPTION directly derive from TS_BASE.
39097 * tree-core.h (tree_optimization_option): Derive from tree_base.
39098 (tree_target_option): Likewise.
39100 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39102 * config/i386/i386.c (memory_address_length): Increase len
39103 only when rip_relative_addr_p returns false.
39105 2017-01-11 Julia Koval <julia.koval@intel.com>
39107 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39108 (OPTION_MASK_ISA_SGX_SET): New.
39109 (ix86_handle_option): Handle OPT_msgx.
39110 * config.gcc: Added sgxintrin.h.
39111 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39112 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39113 * config/i386/i386.c (ix86_target_string): Add -msgx.
39115 (ix86_option_override_internal): Handle new options.
39116 (ix86_valid_target_attribute_inner_p): Add sgx.
39117 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39118 * config/i386/i386.opt: Add msgx.
39119 * config/i386/sgxintrin.h: New file.
39120 * config/i386/x86intrin.h: Add sgxintrin.h.
39122 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39125 * fold-const.c (maybe_nonzero_address): Return 1 for function
39127 (tree_single_nonzero_warnv_p): Don't handle function local objects
39131 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39134 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39137 * opts.c: Include "spellcheck.h"
39138 (struct string_fragment): New struct.
39139 (struct edit_distance_traits<const string_fragment &>): New
39141 (get_closest_sanitizer_option): New function.
39142 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39144 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39146 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39148 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39149 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39150 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39151 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39152 for initial die_offset if dwarf_split_debug_info.
39153 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39154 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39156 (output_skeleton_debug_sections): Formatting fix. Use
39157 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39158 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39160 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39162 * config/arm/cortex-a53.md: Add bypasses for
39163 cortex_a53_r2f_cvt.
39164 (cortex_a53_r2f): Only use for transfers.
39165 (cortex_a53_f2r): Likewise.
39166 (cortex_a53_r2f_cvt): Add reservation for conversions.
39167 (cortex_a53_f2r_cvt): Likewise.
39169 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39171 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39172 to all inlined functions, change static to extern.
39174 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39177 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39179 (arm_assemble_integer): Likewise.
39181 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39183 * config.gcc: Use new awk script to check CPU, FPU and architecture
39184 parameters for --with-... options.
39185 * config/arm/parsecpu.awk: New file
39186 * config/arm/arm-cpus.in: New file.
39187 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39189 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39191 * config/arm/t-arm: Update dependency rules.
39192 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39193 of processing .def files.
39194 * config/arm/genopt.sh: Deleted.
39195 * config/arm/gentune.sh: Deleted.
39196 * config/arm/arm-cores.def: Deleted.
39197 * config/arm/arm-arches.def: Deleted.
39198 * config/arm/arm-fpus.def: Deleted.
39199 * config/arm/arm-tune.md: Regenerated.
39200 * config/arm/arm-tables.opt: Regenerated.
39201 * config/arm/arm-cpu.h: New generated file.
39202 * config/arm/arm-cpu-data.h: New generated file.
39203 * config/arm/arm-cpu-cdata.h: New generated file.
39205 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39208 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39210 (input_varpool_node): Unpack dynamically_initialized bit.
39212 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39214 PR rtl-optimization/79032
39215 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39216 the alignment of the adjusted memory reference against that of MODE,
39217 instead of the alignment of the original memory reference.
39219 2017-01-11 Martin Jambor <mjambor@suse.cz>
39221 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39223 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39224 decorated functions.
39226 2017-01-11 Richard Biener <rguenther@suse.de>
39228 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39229 set range/nonnull info for PHI results. Do not set it on
39230 stmts marked for removal.
39232 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39234 * expr.c (store_field): In the bitfield case, fetch the return value
39235 from the registers before applying a single big-endian adjustment.
39236 Always do a final load for a BLKmode value not larger than a word.
39238 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39241 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39242 that we correctly handle column numbers greater than
39243 LINE_MAP_MAX_COLUMN_NUMBER.
39245 2017-01-10 Martin Sebor <msebor@redhat.com>
39247 PR middle-end/78245
39248 * gimple-ssa-sprintf.c (get_destination_size): Call
39249 {init,fini}object_sizes.
39250 * tree-object-size.c (addr_object_size): Adjust.
39251 (pass_through_call): Adjust.
39252 (pass_object_sizes::execute): Adjust.
39253 * tree-object-size.h (fini_object_sizes): Declare.
39255 2017-01-10 Martin Sebor <msebor@redhat.com>
39257 PR tree-optimization/78775
39258 * builtins.c (get_size_range): Move...
39259 * calls.c: ...to here.
39260 (alloc_max_size): Accept zero argument.
39261 (operand_signed_p): Remove.
39262 (maybe_warn_alloc_args_overflow): Call get_size_range.
39263 * calls.h (get_size_range): Declare.
39265 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39267 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39268 from TI's devices.csv file as of September 2016.
39269 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39271 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39273 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39274 * doc/invoke.texi: Likewise.
39275 * doc/md.texi: Likewise.
39276 * doc/objc.texi: Likewise.
39278 2017-01-10 Joshua Conner <joshconner@google.com>
39280 * config/arm/fuchsia-elf.h: New file.
39281 * config/fuchsia.h: New file.
39282 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39283 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39285 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39287 2016-01-10 Richard Biener <rguenther@suse.de>
39289 PR tree-optimization/79034
39290 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39291 Propagate out degenerate PHIs in the joiner.
39293 2017-01-10 Martin Liska <mliska@suse.cz>
39295 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39296 (sort_congruence_classes_by_decl_uid): Likewise.
39297 (sort_congruence_class_groups_by_decl_uid): Likewise.
39298 (sem_item_optimizer::merge_classes): Sort class, groups in these
39299 classes and members in the groups by DECL_UID of declarations.
39300 This would make merge operations stable.
39302 2017-01-10 Martin Liska <mliska@suse.cz>
39304 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39305 usage of m_classes_vec.
39306 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39307 (sem_item_optimizer::get_group_by_hash): Likewise.
39308 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39309 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39310 (sem_item_optimizer::verify_classes): Likewise.
39311 (sem_item_optimizer::process_cong_reduction): Likewise.
39312 (sem_item_optimizer::dump_cong_classes): Likewise.
39313 (sem_item_optimizer::merge_classes): Likewise.
39314 * ipa-icf.h (congruence_class_hash): Rename from
39315 congruence_class_group_hash. Remove declaration of m_classes_vec.
39317 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39319 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39320 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39321 * config.gcc: Add avx512vpopcntdqintrin.h.
39322 * config/i386/avx512vpopcntdqintrin.h: New.
39323 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39324 * config/i386/i386-builtin-types.def: Add new types.
39325 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39326 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39327 __builtin_ia32_vpopcountq_v8di_mask): New.
39328 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39329 __AVX512VPOPCNTDQ__.
39330 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39331 (PTA_AVX512VPOPCNTDQ): Define.
39332 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39333 TARGET_AVX512VPOPCNTDQ_P): Define.
39334 * config/i386/i386.opt: Add mavx512vpopcntdq.
39335 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39336 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39338 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39340 PR middle-end/77484
39341 * predict.def (PRED_CALL): Set to 67.
39343 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39345 * expr.c (store_field): In the bitfield case, if the value comes from
39346 a function call and is of an aggregate type returned in registers, do
39347 not modify the field mode; extract the value in all cases if the mode
39348 is BLKmode and the size is not larger than a word.
39350 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39353 * config/i386/cpuid.h: Fix undefined behavior.
39355 2017-01-04 Jeff Law <law@redhat.com>
39357 PR tree-optimization/79007
39358 PR tree-optimization/67955
39359 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39360 conservative for pt.null when flag_non_call_exceptions is on.
39362 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39364 PR translation/79019
39365 PR translation/79020
39366 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39367 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39368 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39370 * config/avr/avr.opt (maccumulate-args): Likewise.
39371 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39372 * common.opt (freport-bug): Likewise.
39373 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39374 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39375 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39376 translatable string.
39377 * config/i386/i386.c (function_value_32): Likewise.
39378 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39379 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39381 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39382 * common/config/msp430/msp430-common.c (msp430_handle_option):
39384 * symtab.c (symtab_node::verify_base): Likewise.
39385 * opts.c (set_debug_level): Likewise.
39386 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39387 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39388 missing whitespace to translatable strings.
39389 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39390 * config/sh/superh.h: Likewise.
39391 * config/i386/xopintrin.h: Likewise.
39392 * config/i386/znver1.md: Likewise.
39393 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39394 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39395 * double-int.h (struct double_int): Likewise.
39396 * double-int.c (div_and_round_double): Likewise.
39397 * wide-int.cc: Likewise.
39398 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39399 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39400 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39401 (crossjumps_occurred): ... this.
39402 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39405 PR tree-optimization/78899
39406 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39407 returning bool return struct loop *, NULL for failure and the new
39409 (versionable_outer_loop_p): Don't version outer loop if it has
39410 dont_vectorized bit set.
39411 (tree_if_conversion): When versioning outer loop, ensure
39412 tree_if_conversion is performed also on the inner loop of the
39413 non-vectorizable outer loop copy.
39414 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39415 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39416 prevent vectorization of it.
39417 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39418 the outer loop vectorization of the non-scalar version is attempted
39419 before vectorization of the inner loop in scalar version. If
39420 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39421 vectorization of its inner loop.
39422 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39423 has 2 inner loops, rename also on edges from bb whose single pred
39424 is outer_loop->header. Fix typo in function comment.
39426 2017-01-09 Martin Sebor <msebor@redhat.com>
39429 * asan.c (asan_emit_stack_protection): Increase local buffer size
39430 to avoid snprintf truncation warning.
39432 2017-01-09 Andrew Pinski <apinski@cavium.com>
39434 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39435 to reference thunderx2t99 for the tuning structure
39436 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39438 (thunderx2t99_extra_costs): This.
39439 * config/aarch64/aarch64-tune.md: Regenerate.
39440 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39441 (vulcan_addrcost_table): This.
39442 (vulcan_regmove_cost): Rename to ...
39443 (thunderx2t99_regmove_cost): This.
39444 (vulcan_vector_cost): Rename to ...
39445 (thunderx2t99_vector_cost): this.
39446 (vulcan_branch_cost): Rename to ...
39447 (thunderx2t99_branch_cost): This.
39448 (vulcan_tunings): Rename to ...
39449 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39450 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39452 2017-01-09 Martin Jambor <mjambor@suse.cz>
39456 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39457 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39458 (propagate_vr_accross_jump_function): Use the above function for all
39459 value range computations for pass-through jump functions and type
39460 converasion from explicit value range values.
39461 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39462 parameters from TYPE_ARG_TYPES.
39463 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39464 (ipa_write_node_info): Stream type of the actual argument.
39465 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39467 2017-01-09 Martin Liska <mliska@suse.cz>
39470 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39471 (lookup_compiler): Do not show error message with have_E.
39473 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39475 PR tree-optimization/78938
39476 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39477 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39478 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39479 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39482 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39484 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39487 2017-01-09 Richard Biener <rguenther@suse.de>
39489 PR tree-optimization/78997
39490 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39491 name condition properly.
39493 2017-01-09 Richard Biener <rguenther@suse.de>
39496 * dwarf2out.c (is_cxx): New overload with context.
39497 (is_naming_typedef_decl): Use it.
39499 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39501 * invoke.texi (Option Summary): Correct spacing in option lists
39502 and add line breaks to fix over-long lines.
39504 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39506 PR middle-end/17660
39508 * extend.texi (Common Variable Attributes): Add xref to GCC
39509 Internals manual to explain mode attribute keywords.
39511 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39514 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39515 and Preprocessor Options.
39516 (Options for Linking): Document -pthread here....
39517 (RS/6000 and PowerPC Options): ...not here.
39518 (Solaris 2 Options): ...or here.
39519 * doc/cppopts.texi: Document -pthread.
39521 2017-01-08 Martin Sebor <msebor@redhat.com>
39523 PR middle-end/77708
39524 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39525 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39526 New member functions.
39527 (format_directive): Used them.
39529 (pass_sprintf_length::handle_gimple_call): Same.
39530 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39531 to avoid truncation for any argument.
39532 (extract_affine_mul): Same.
39533 * tree.c (get_file_function_name): Same.
39535 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39537 PR middle-end/77484
39538 * predict.def (PRED_INDIR_CALL): Set to 86.
39540 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39542 PR preprocessor/54124
39543 * doc/cppopts.texi: Reformat -d subtable to list the full name
39544 of the options. Add cross-reference to the docs for the general
39545 compiler -d options.
39546 * doc/invoke.texi (Developer Options): Add cross-reference to the
39547 preprocessor-specific -d option documentation.
39549 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39551 PR preprocessor/13498
39552 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39553 redudant material, and reflect new command-line options.
39554 (System Headers): Likewise.
39556 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39558 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39559 -isystem, and -idirafter. Copy-edit.
39560 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39561 default for -ftrack-macro-expansion. Delete obsolete and
39562 badly-formatted implementation details about -fdebug-cpp output.
39563 * doc/cppwarnopts.texi: Copy-edit.
39565 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39568 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39569 that the transition from a max line width >= 1<<10 to narrower
39570 lines works correctly.
39572 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39574 * doc/options.texi (PerFunction): New.
39575 * opt-functions.awk (switch_flags): Map both Optimization and
39576 PerFunction to CL_OPTIMIZATION.
39577 * opth-gen.awk: Test for PerFunction flag along with
39579 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39580 it only when the latter is present. Skip those that don't in
39581 the hash function generator.
39582 * common.opt (fvar-tracking): Mark as PerFunction instead of
39584 (fvar-tracking-assignments): Likewise.
39585 (fvar-tracking-assignments-toggle): Likewise.
39586 (fvar-tracking-uninit): Likewise.
39588 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39590 PR translation/79018
39591 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39594 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39597 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39598 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39599 TARGET_LONG_JUMP_TABLE_OFFSETS.
39600 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39601 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39602 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39603 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39604 * config/m68k/m68k.md (tablejump expander): Likewise.
39605 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39606 TARGET_LONG_JUMP_TABLE_OFFSETS.
39607 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39608 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39610 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39611 David Holsgrove <david.holsgrove@xilinx.com>
39613 * common/config/microblaze/microblaze-common.c
39614 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39615 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39617 * config/microblaze/microblaze.c (microblaze_must_save_register)
39618 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39620 (microblaze_eh_return): New function.
39621 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39622 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39623 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39624 * config/microblaze/microblaze.md (eh_return): New pattern.
39626 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39628 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39629 GCC_DIAGNOSTIC_STRINGIFY): Define.
39631 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39633 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39635 * config/arm/arm.md (<mcrr>): New.
39637 * config/arm/arm.c (arm_arch5te): New.
39638 (arm_option_override): Set arm_arch5te.
39639 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39641 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39642 (arm_mcrr_qualifiers): ... this. New.
39643 (MRRC_QUALIFIERS): Define to...
39644 (arm_mrrc_qualifiers): ... this. New.
39645 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39647 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39648 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39649 (MRRCI, mrrc, MRRC): New.
39650 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39651 VUNSPEC_MRRC2): New.
39653 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39655 * config/arm/arm.md (<mcr>): New.
39657 * config/arm/arm.c (arm_coproc_builtin_available): Add
39658 support for mcr, mrc, mcr2 and mrc2.
39659 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39660 (arm_mcr_qualifiers): ... this. New.
39661 (MRC_QUALIFIERS): Define to ...
39662 (arm_mrc_qualifiers): ... this. New.
39663 (MCR_QUALIFIERS): Define to ...
39664 (arm_mcr_qualifiers): ... this. New.
39665 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39667 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39668 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39669 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39670 VUNSPEC_MRC2): New.
39672 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39674 * config/arm/arm.md (*ldc): New.
39678 * config/arm/arm.c (arm_coproc_builtin_available): Add
39679 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39680 (arm_coproc_ldc_stc_legitimate_address): New.
39681 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39682 'qualifier_const_pointer'.
39683 (LDC_QUALIFIERS): Define to...
39684 (arm_ldc_qualifiers): ... this. New.
39685 (STC_QUALIFIERS): Define to...
39686 (arm_stc_qualifiers): ... this. New.
39687 * config/arm/arm-protos.h
39688 (arm_coproc_ldc_stc_legitimate_address): New.
39689 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39690 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39691 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39692 stc2, stcl, stc2l): New.
39693 * config/arm/constraints.md (Uz): New.
39694 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39695 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39696 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39697 VUNSPEC_STC2L): New.
39699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39701 * config/arm/arm.md (<cdp>): New.
39702 * config/arm/arm.c (neon_const_bounds): Rename this ...
39703 (arm_const_bounds): ... this.
39704 (arm_coproc_builtin_available): New.
39705 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39706 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39707 (CDP_QUALIFIERS): Define to...
39708 (arm_cdp_qualifiers): ... this. New.
39710 (arm_expand_builtin_args): Add case for 6 arguments.
39711 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39712 (arm_const_bounds): ... this.
39713 (arm_coproc_builtin_available): New.
39714 * config/arm/arm_acle.h (__arm_cdp): New.
39716 * config/arm/arm_acle_builtins.def (cdp): New.
39718 * config/arm/iterators.md (CDPI,CDP,cdp): New.
39719 * config/arm/neon.md: Rename all 'neon_const_bounds' to
39720 'arm_const_bounds'.
39721 * config/arm/types.md (coproc): New.
39722 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
39723 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
39724 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
39725 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
39727 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39729 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
39730 (UBINOP_QUALIFIERS): New.
39732 (acle_builtin_data): New. Change comment.
39733 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
39734 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39735 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39736 arm_acle_builtins.def.
39737 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39738 (arm_init_acle_builtins): New.
39739 (CRC32_BUILTIN): Remove.
39740 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39741 crc32cb, crc32ch and crc32cw.
39742 (arm_init_crc32_builtins): Remove.
39743 (arm_init_builtins): Use arm_init_acle_builtins rather
39744 than arm_init_crc32_builtins.
39745 (arm_expand_acle_builtin): New.
39746 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39747 * config/arm/arm_acle_builtins.def: New.
39749 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39751 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39752 (arm_builtin_datum): ... this.
39753 (arm_init_neon_builtin): Rename to ...
39754 (arm_init_builtin): ... this. Add a new parameters PREFIX
39755 and USE_SIG_IN_NAME.
39756 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39757 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39758 'arm_builtin_datum'.
39759 (arm_init_vfp_builtins): Likewise.
39760 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39761 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39762 (arm_expand_neon_args): Rename to ...
39763 (arm_expand_builtin_args): ... this. Rename builtin_arg
39764 enum values and differentiate between ARG_BUILTIN_MEMORY
39765 and ARG_BUILTIN_NEON_MEMORY.
39766 (arm_expand_neon_builtin_1): Rename to ...
39767 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39768 values, arm_expand_builtin_args and add bool parameter NEON.
39769 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39770 (arm_expand_vfp_builtin): Likewise.
39771 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39773 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39775 PR middle-end/77484
39776 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39777 * predict.c (tree_estimate_probability_bb): Reverse direction of
39778 polymorphic call predictor.
39780 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39782 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39783 (determine_pass_name_match): ...this new function and...
39784 (should_skip_pass_p): ...this new function.
39786 2017-01-06 Nathan Sidwell <nathan@acm.org>
39788 * ipa-visibility.c (function_and_variable_visibility): Reformat
39789 comments and long lines. Remove extrneous if.
39790 * symtab.c (symtab_node::make_decl_local): Fix code format.
39791 (symtab_node::set_section_for_node): Fix comment typo.
39793 2017-01-06 Martin Liska <mliska@suse.cz>
39796 * lra-constraints.c: Rename invariant to lra_invariant.
39797 * predict.c (set_even_probabilities): Initialize e to NULL.
39799 2017-01-05 Martin Sebor <msebor@redhat.com>
39801 PR tree-optimization/78910
39802 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39803 (format_integer): Correct off-by-one error in the handling
39804 of precision with negative numbers in signed conversions..
39806 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39808 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39810 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39812 PR tree-optimization/71016
39813 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39814 factor_out_conditional_conversion. Formatting fix.
39815 (factor_out_conditional_conversion): Add cond_stmt argument.
39816 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39817 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39820 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39822 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39823 read-rtl-function.o, and selftest-rtl.o.
39824 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39825 (selftest::aarch64_test_loading_full_dump): New function.
39826 (selftest::aarch64_run_selftests): New function.
39827 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39828 selftest::aarch64_run_selftests.
39829 * config/i386/i386.c
39830 (selftest::ix86_test_loading_dump_fragment_1): New function.
39831 (selftest::ix86_test_loading_call_insn): New function.
39832 (selftest::ix86_test_loading_full_dump): New function.
39833 (selftest::ix86_test_loading_unspec): New function.
39834 (selftest::ix86_run_selftests): Call the new functions.
39835 * emit-rtl.c (maybe_set_max_label_num): New function.
39836 * emit-rtl.h (maybe_set_max_label_num): New decl.
39837 * function.c (instantiate_decls): Guard call to
39838 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39839 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39841 * gensupport.c (gen_reader::gen_reader): Pass "false"
39842 for new "compact" param of rtx_reader.
39843 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39844 rather than an empty string for NULL strings.
39845 * read-md.c: Potentially include config.h rather than bconfig.h.
39846 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39847 (have_error): New global, copied from errors.c.
39848 (md_reader::read_name): Rename to...
39849 (md_reader::read_name_1): ...this, adding "out_loc" param,
39850 and converting "missing name or number" to returning false, rather
39852 (md_reader::read_name): Reimplement in terms of read_name_1.
39853 (md_reader::read_name_or_nil): New function.
39854 (md_reader::read_string): Handle "(nil)" by returning NULL.
39855 (md_reader::md_reader): Add new param "compact".
39856 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39857 (md_reader::read_file): New method.
39858 * read-md.h (md_reader::md_reader): Add new param "compact".
39859 (md_reader::read_file): New method.
39860 (md_reader::is_compact): New accessor.
39861 (md_reader::read_name): Convert return type from void to file_location.
39862 (md_reader::read_name_or_nil): New decl.
39863 (md_reader::read_name_1): New decl.
39864 (md_reader::m_compact): New field.
39865 (noop_reader::noop_reader): Pass "false" for new "compact" param
39867 (rtx_reader::rtx_reader): Add new "compact" param.
39868 (rtx_reader::read_rtx_operand): Make virtual and convert return
39869 type from void to rtx.
39870 (rtx_reader::read_until): New decl.
39871 (rtx_reader::handle_any_trailing_information): New virtual function.
39872 (rtx_reader::postprocess): New virtual function.
39873 (rtx_reader::finalize_string): New virtual function.
39874 (rtx_reader::m_in_call_function_usage): New field.
39875 (rtx_reader::m_reuse_rtx_by_id): New field.
39876 * read-rtl-function.c: New file.
39877 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39878 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39879 (selftest::verify_three_block_rtl_cfg): New decl.
39880 * read-rtl-function.h: New file.
39881 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39882 For host, include function.h, memmodel.h, and emit-rtl.h.
39883 (one_time_initialization): New function.
39884 (struct compact_insn_name): New struct.
39885 (compact_insn_names): New array.
39886 (find_code): Handle insn codes in compact dumps.
39887 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39888 (bind_subst_iter_and_attr): Likewise.
39889 (add_condition_to_string): Likewise.
39890 (add_condition_to_rtx): Likewise.
39891 (apply_attribute_uses): Likewise.
39892 (add_current_iterators): Likewise.
39893 (apply_iterators): Likewise.
39894 (initialize_iterators): Guard usage of apply_subst_iterator with
39895 #ifdef GENERATOR_FILE.
39896 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39897 (md_reader::read_mapping): Likewise.
39898 (add_define_attr_for_define_subst): Likewise.
39899 (add_define_subst_attr): Likewise.
39900 (read_subst_mapping): Likewise.
39901 (check_code_iterator): Likewise.
39902 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39904 (one_time_initialization): New function.
39905 (rtx_reader::read_until): New method.
39906 (read_flags): New function.
39907 (parse_reg_note_name): New function.
39908 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39909 Handle reuse_rtx ids.
39910 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39911 Add parsing support for RTL dumps, mirroring the special-cases in
39912 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39913 values, and calling handle_any_trailing_information.
39914 (rtx_reader::read_rtx_operand): Convert return type from void
39915 to rtx, returning return_rtx. Handle case 'e'. Call
39916 finalize_string on XSTR and XTMPL fields.
39917 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39918 "(nil)" values were omitted. Call the postprocess vfunc on the
39920 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39921 class ctor. Initialize m_in_call_function_usage. Call
39922 one_time_initialization.
39923 * rtl-tests.c (selftest::test_uncond_jump): Call
39924 set_new_first_and_last_insn.
39925 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39926 * selftest-rtl.c: New file.
39927 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39928 (selftest::get_insn_by_uid): New decl.
39929 * selftest-run-tests.c (selftest::run_tests): Call
39930 read_rtl_function_c_tests.
39931 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39932 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39935 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39937 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39938 operands in a special way. Assert that pos+len <= mode precision.
39940 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39942 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39943 3 argument Alias with unlimited for the negative form.
39944 (fno-vect-cost-model): Removed.
39946 2017-01-05 Martin Liska <mliska@suse.cz>
39948 * hsa-gen.c (gen_hsa_divmod): New function.
39949 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39951 2017-01-05 Martin Liska <mliska@suse.cz>
39954 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39957 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39959 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39960 small constant length operands.
39962 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39964 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39965 between loop iterations.
39967 2017-01-05 Martin Liska <mliska@suse.cz>
39970 * gimplify.c (gimplify_decl_expr): Compare to
39971 asan_poisoned_variables instread of checking flags.
39972 (gimplify_target_expr): Likewise.
39973 (gimplify_expr): Likewise.
39974 (gimplify_function_tree): Conditionally initialize
39975 asan_poisoned_variables.
39977 2017-01-04 Jeff Law <law@redhat.com>
39979 PR tree-optimizatin/78812
39980 * rtl.h (contains_mem_rtx_p): Prototype.
39981 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39982 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39983 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39984 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39985 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39987 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39989 * input.c (assert_char_at_range): Default-initialize actual_range.
39991 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39993 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39994 to match the caller.
39996 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39998 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39999 insns after final jump in test to emit dummy move.
40001 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40003 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40004 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40006 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40008 * multiple_target.c (create_dispatcher_calls): Init e_next.
40009 * tree-ssa-loop-split.c (split_loop): Init border.
40010 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40013 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40018 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40019 (altivec_register_operand): Do not return true if the operand
40020 contains a SUBREG mixing SImode and SFmode.
40021 (vsx_register_operand): Likewise.
40022 (vsx_reg_sfsubreg_ok): New predicate.
40023 (vfloat_operand): Do not return true if the operand contains a
40024 SUBREG mixing SImode and SFmode.
40025 (vint_operand): Likewise.
40026 (vlogical_operand): Likewise.
40027 (gpc_reg_operand): Likewise.
40028 (int_reg_operand): Likewise.
40029 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40030 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40031 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40033 (rs6000_emit_move_si_sf_subreg): New helper function.
40034 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40035 fixup SUBREGs involving SImode and SFmode.
40036 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40037 numbers for the new peephole2 optimization.
40038 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40039 the GLIBC math library that do AND/IOR/XOR operations on single
40040 precision floating point.
40041 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40042 target macros to say whether we need to avoid SUBREGs mixing
40044 (TARGET_ALLOW_SF_SUBREG): Likewise.
40045 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40046 (UNSPEC_SI_FROM_SF): Likewise.
40047 (iorxor): Change spacing.
40048 (and_ior_xor): New iterator for AND, IOR, and XOR.
40049 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40050 (movdi_from_sf_zero_ext): Likewise.
40051 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40052 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40053 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40054 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40055 (fms<mode>4): Likewise.
40056 (fnma<mode>4): Likewise.
40057 (fnms<mode>4): Likewise.
40058 (nfma<mode>4): Likewise.
40059 (nfms<mode>4): Likewise.
40061 2017-01-04 Marek Polacek <polacek@redhat.com>
40064 * doc/invoke.texi: Document -Wpointer-compare.
40066 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40068 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40071 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40072 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40075 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40078 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40079 documentation of the powerpc_popcntb_ok attribute.
40080 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40081 code to issue warning messages if a requested CPU configuration is
40082 not supported by the binary (assembler and loader) toolchain.
40083 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40084 made to define a built-in function that has been disabled.
40085 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40086 made to define a built-in function that has been disabled.
40087 (altivec_init_builtins): Add comment explaining why definition
40088 of the DST built-in functions is not preceded by an assertion
40089 check. Add assertions to prevent ICE if attempts are made to
40090 define an altivec predicate or an abs* built-in function that has
40092 (htm_init_builtins): Add comment explaining why definition of the
40093 htm built-in functions is not preceded by an assertion check.
40095 2017-01-04 Jeff Law <law@redhat.com>
40097 PR tree-optimizatin/67955
40098 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40099 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40100 the points-to solution does not include pt_null. Use DECL_PT_UID
40103 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40105 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40106 Use gen_int_mode instead of gen_lopwart for const_int operands.
40108 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40110 PR tree-optimization/71563
40111 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40112 out of range value - has low bits known to be zero.
40114 2017-01-04 Alan Modra <amodra@gmail.com>
40116 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40117 * configure: Regenerate.
40118 * config.in: Regenerate.
40120 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40123 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40124 a substring of the message, but strcmp with the whole message. Ifdef
40125 ENABLE_NLS, translate the message first using dgettext.
40127 2017-01-03 Jeff Law <law@redhat.com>
40129 PR tree-optimizatin/78856
40130 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40131 (mark_threaded_blocks): Remove code to truncate thread paths that
40132 cross multiple loop headers. Instead invalidate the cached loop
40133 iteration information and handle case of a thread path walking
40134 into an irreducible region.
40136 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40139 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40140 assertions. Add support for doing the signbit if the IEEE 128-bit
40141 floating point value is in a GPR.
40142 * config/rs6000/rs6000.md (Fsignbit): Delete.
40143 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40144 Update the length attribute if the value is in a GPR.
40145 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40146 the sign or zero extension instruction, since the value is always 0/1.
40147 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40150 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40151 extracting SImode to a GPR register so that we can generate a
40152 store, limit the vector to be in a traditional Altivec register
40153 for the vextuwrx instruction.
40155 2017-01-03 Ian Lance Taylor <iant@google.com>
40157 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40159 2017-01-03 Martin Sebor <msebor@redhat.com>
40161 PR tree-optimization/78696
40162 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40163 precision. Use MPFR for %f for greater fidelity. Correct handling
40165 (pass_sprintf_length::compute_format_length): Set width and precision
40166 specified by asrerisk to void_node for vararg functions.
40167 (try_substitute_return_value): Adjust dump output.
40169 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40171 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40173 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40175 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40176 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40177 -mlra/-mno-lra was passed to the compiler.
40179 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40181 PR rtl-optimization/65618
40182 * emit-rtl.c (try_split): Move initialization of "before" and
40183 "after" to just before the call to emit_insn_after_setloc.
40185 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40187 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40189 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40191 * dwarf2out.c (gen_enumeration_type_die): When
40192 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40194 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40196 PR tree-optimization/78965
40197 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40198 Change first argument from const call_info & to call_info &. For %n
40199 set info.nowrite to false.
40201 PR middle-end/78901
40202 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40203 possibly throwing calls.
40205 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40206 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40207 and fns handling, rather than in a separate case SSA_NAME.
40209 2017-01-02 Jeff Law <law@redhat.com>
40211 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40212 fixes for first_period and second_period variables.
40214 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40217 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40218 (*insvqi_1): New insn pattern.
40219 (*insvqi_1_mem_rex64): Ditto.
40220 (*insvqi_2): Ditto.
40221 (*insvqi_3): Rename from *insvqi.
40223 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40225 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40227 * doc/cfg.texi (Edges): Remove reference to Java.
40228 (Maintaining the CFG): Ditto.
40230 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40232 PR middle-end/77674
40233 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40234 transparent aliases.
40236 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40238 PR middle-end/77484
40239 * predict.def (PRED_CALL): Update hitrate.
40240 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40241 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40242 into direct/indirect/polymorphic variants.
40244 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40246 Update copyright years.
40248 * gcc.c (process_command): Update copyright notice dates.
40249 * gcov-dump.c (print_version): Ditto.
40250 * gcov.c (print_version): Ditto.
40251 * gcov-tool.c (print_version): Ditto.
40252 * gengtype.c (create_file): Ditto.
40253 * doc/cpp.texi: Bump @copying's copyright year.
40254 * doc/cppinternals.texi: Ditto.
40255 * doc/gcc.texi: Ditto.
40256 * doc/gccint.texi: Ditto.
40257 * doc/gcov.texi: Ditto.
40258 * doc/install.texi: Ditto.
40259 * doc/invoke.texi: Ditto.
40261 Copyright (C) 2017 Free Software Foundation, Inc.
40263 Copying and distribution of this file, with or without modification,
40264 are permitted in any medium without royalty provided the copyright
40265 notice and this notice are preserved.